Cytosolic Phospholipase A2α Is Targeted to the p47phox-PX Domain of the Assembled NADPH Oxidase via a Novel Binding Site in Its C2 Domain*

We have previously demonstrated a physical interaction between cytosolic phospholipase A2α (cPLA2) and the assembled NADPH oxidase on plasma membranes following neutrophil stimulation. The aim of the present study was to define the exact binding sites between these two enzymes. Here we show, based on blot overlay experiments, Förster resonance energy transfer analysis and studies in neutrophils from patients with chronic granulomatous disease deficient in p67phox or p47phox, that cPLA2 specifically binds to p47phox and that p47phox is sufficient to anchor cPLA2 to the assembled oxidase on the plasma membranes upon stimulation. Blot overlay and affinity binding experiments using subfragments of cPLA2 and p47phox demonstrated that the cPLA2-C2 domain and the p47phox-PX domain interact to form a complex that is resistant to high salt. Computational docking was used to identify hydrophobic peptides within these two domains that inhibited the association between the two enzymes and NADPH oxidase activity in electro-permeabilized neutrophils. These results were used in new docking computations that produced an interaction model. Based on this model, cPLA2-C2 domain mutations were designed to explore its interaction p47phox in neutrophil lysates. The triple mutant F35A/M38A/L39A of the cPLA2-C2 domain caused a slight inhibition of the affinity binding to p47phox, whereas the single mutant I67A was highly effective. The double mutant M59A/H115A of the p47phox-PX domain caused a significant inhibition of the affinity binding to cPLA2. Thus, Ile67 of the cPLA2-C2 domain is identified as a critical, centrally positioned residue in a hydrophobic interaction in the p47phox-PX domain.

We have previously demonstrated a physical interaction between cytosolic phospholipase A 2 ␣ (cPLA 2 ) and the assembled NADPH oxidase on plasma membranes following neutrophil stimulation. The aim of the present study was to define the exact binding sites between these two enzymes. Here we show, based on blot overlay experiments, Förster resonance energy transfer analysis and studies in neutrophils from patients with chronic granulomatous disease deficient in p67 phox or p47 phox , that cPLA 2 specifically binds to p47 phox and that p47 phox is sufficient to anchor cPLA 2 to the assembled oxidase on the plasma membranes upon stimulation. Blot overlay and affinity binding experiments using subfragments of cPLA 2 and p47 phox demonstrated that the cPLA 2 -C2 domain and the p47 phox -PX domain interact to form a complex that is resistant to high salt. Computational docking was used to identify hydrophobic peptides within these two domains that inhibited the association between the two enzymes and NADPH oxidase activity in electro-permeabilized neutrophils. These results were used in new docking computations that produced an interaction model. Based on this model, cPLA 2 -C2 domain mutations were designed to explore its interaction p47 phox in neutrophil lysates. The triple mutant F35A/M38A/L39A of the cPLA 2 -C2 domain caused a slight inhibition of the affinity binding to p47 phox , whereas the single mutant I67A was highly effective. The double mutant M59A/H115A of the p47 phox -PX domain caused a significant inhibition of the affinity binding to cPLA 2 . Thus, Ile 67 of the cPLA 2 -C2 domain is identified as a critical, centrally positioned residue in a hydrophobic interaction in the p47 phox -PX domain.
The NADPH oxidase is a multicomponent electron carrier that transfers electrons from NADPH to molecular oxygen to form superoxide, a precursor of microbicidal oxidants. NADPH oxidase subunits include four cytoplasmic components, p47 phox , p67 phox , p40 phox , and Rac2, and a hetero-dimeric transmembrane glycoprotein flavocytochrome b 558 composed of gp91 phox and p22 phox (for reviews see Refs. 1, 2). Essential roles for at least four of the phox components is evident from studies on chronic granulomatous disease (CGD) 2 patients, who suffer from inherited defects in NADPH oxidase-dependent microbial killing due to defect in the genes encoding these oxidase proteins (3). In resting cells, p47 phox , p67 phox , and p40 phox exist as a tight cytosolic complex dissociated from the membranebound flavocytochrome. Upon stimulation, the cytosolic components translocate to the plasma membrane and associate with the flavocytochrome b 558 to form the assembled active oxidase. Based on studies of p47 phox -deficient cells from chronic CGD patients (4), p47 phox appears to have a key role in translocation of the cytosolic subunits. p47 phox possesses a phox homology (PX) domain, tandem SH3 domains, a series of basic residues, and phosphorylation targets that is also referred to as an autoinhibitory region, and a proline-rich region in the C terminus. In resting cells, p47 phox is found in an autoinhibited state due to intramolecular interactions among the PX domain, the tandem Src homology 3 (SH3) domains, and polybasic sequences of p47 phox , thereby preventing its binding to membranes (5). In stimulating cells, the restrictive conformation of the autoinhibitory region of p47 phox is released through phosphorylation of several critical serine residues within its polybasic region (6). Conformational changes in p47 phox following phosphorylation result in unfolding and exposing the now interactive SH3 domains that direct its translocation to the membranes by binding to specific targets in p22 phox and its PX domain that specifically binds phosphatidylinositol 3,4bisphosphate and phosphatidic acid on the plasma membrane (7)(8)(9). Neither p40 phox nor p67 phox is able to translocate to * This work was supported by the Israel Sciences Foundation founded by the Israel Academy of Sciences and Humanities (Grant 438/03). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. □ S The on-line version of this article (available at http://www.jbc.org) contains supplemental text. 1 To whom correspondence should be addressed. membranes in the absence of p47 phox , as evidenced by the cytoplasmic location of p40 phox and p67 phox in stimulated cells from CGD patients who lack a functional p47 phox (10). However, it appears that p40 phox functions in retention of p47 phox and p67 phox on phagosome membranes through interactions of its PX domain with phosphatidylinositol 3-phosphate, as well as interactions of its PB1 and SH3 domains with p47 phox and p67 phox , respectively (11). Cytosolic phospholipase A 2 ␣ (cPLA 2 ), which hydrolyzes phospholipids containing arachidonate at the sn-2 position (12), has been implicated as the major enzyme in the formation of eicosanoids. cPLA 2 has two functionally distinct domains: an N-terminal C2 domain necessary for Ca 2ϩ -dependent phospholipid binding, and a C-terminal Ca 2ϩ -independent catalytic region (13). It was shown that cPLA 2 translocates from the cytosol to the nuclear membrane and to the endoplasmic reticulum by an increase in cytoplasmic [Ca 2ϩ ] in a variety of cells (14). The C2 domain of cPLA 2 has been cocrystallized with calcium (15), and the structure revealed that it binds two calcium ions at one end of the domain between three loops, called calcium binding regions (CBRs): CBR1, CBR2, and CBR3 (16). The calcium ions neutralize the negative electrostatic potential surrounding the hydrophobic residues located at the tips of CBRs 1 and 3, thereby facilitating their interactions with the hydrophobic portions of the lipid head groups of PC-enriched membranes (16). Phosphorylation of cPLA 2 is important for regulating release of arachidonic acid in cells, but this process is not clearly understood. The catalytic domain of cPLA 2 contains several functionally important phosphorylation sites (for review see Ref. 17), Ser 505 , Ser 727 , and Ser 515 , which are phosphorylated by mitogen-activated protein kinases (MAPKs), mitogen-activated protein kinase interacting kinase (MNK1), or a related MAPK-activated protein kinase, and calmodulin kinase II, respectively. It is suggest that, depending on the cell type and agonist used for activation, phosphorylation may function to regulate cPLA 2 catalytic activity and membrane binding.
We have previously demonstrated an essential requirement for cPLA 2 in activation of the assembled phagocyte NADPH oxidase (18), the oxidase-associated H ϩ channel (19), and oxidase-associated diaphorase activity (20). The absolute requirement of cPLA 2 for oxidase activation is in line with other studies (21-23) utilizing inhibitors and antisense molecules. In contrast, phagocytes from cPLA 2 -deficient mice showed a normal stimulated superoxide release (24) that could be attributed to the effect of other compensating isoenzymes, a response frequently observed in knockout animal models. Our most recent study (25) demonstrated that in peripheral blood neutrophils and granulocyte-like PLB-985 cells, cPLA 2 translocates to the plasma membrane by interacting with the assembled oxidase complex in addition to its translocation to nuclear membranes. Thus, the ability of cPLA 2 to colocalize in two different compartments in the same cells enables it to participate in both eicosanoid production and to regulate NADPH oxidase activation. The activation and translocation of cPLA 2 by PMA (25,26), which does not induce an increase in cytoplasmic [Ca 2ϩ ], together with its translocation to the plasma membrane suggest the existence of alternative pathways for inducing translocation of cPLA 2 that are distinct from the C2 domain phospholipidbinding mechanism. In agreement with our results, it was recently reported (27) that during phagocytosis of zymosan, cPLA 2 translocates in a Ca 2ϩ -independent manner to the forming phagosomes in kinetics similar to acquisition of the plasma membrane and prior to phagolysosome fusion. The aim of the present study was to explore the nature of the interaction between cPLA 2 and the assembled NADPH oxidase, which anchors cPLA 2 to the plasma membrane upon stimulation of neutrophils or granulocyte-like PLB cells.

EXPERIMENTAL PROCEDURES
Neutrophil Purification-Neutrophils from healthy volunteers or from CGD patients were separated by Ficoll/Hypaque centrifugation, dextran sedimentation, and hypotonic lysis of erythrocytes (18). p47 phox -deficient CGD patients and p67 phoxdeficient CGD patients lacking expression of these cytosolic oxidase proteins due to mutations in the genes for neutrophil cytosolic factor 1 on chromosome 7q11.23 and in the gene for neutrophil cytosolic factor 2, located on 1q25, respectively, were enrolled in the study. The study was approved by the institutional Human Research Committee of the Soroka University Medical Center.
Cell Culture and Differentiation-PLB-985 leukemic cell lines and gp91 phox -deficient PLB-985 cells lacking normal expression of normal gp91 phox (X-CGD), provided by M. C. Dinauer (James Whitcomb Riley Hospital for Children, Indianapolis, IN), were grown in stationary suspension culture in RPMI 1640 and differentiated toward the granulocyte phenotype with 10 Ϫ6 M retinoic acid as described earlier (18) .
Superoxide Anion Measurements-The production of superoxide anion (O 2 . ) by intact cells was measured as the superoxide dismutase inhibitable reduction of ferricytochrome c (18).

Isolation of Membrane and Cytosol
Fractions-Isolation of membrane and cytosol fractions was performed exactly as described earlier (29).
Coimmunoprecipitation-Immunoprecipitation was performed as described earlier (25). The detection of cPLA 2 or the NADPH oxidase components after SDS-PAGE electrophoresis was analyzed as described previously (26).
Overlay Assay-Recombinant p47 phox and p67 phox were a kind gift from Prof. Edgar Pick (Tel-Aviv University, Israel). The different cPLA 2 fusion proteins or cell lysates of 5 ϫ 10 7 (25) were separated on SDS-PAGE gel without ␤-mercaptoethanol and boiling. Protein renaturation was performed by incubation in 25% isopropanol solution for 30 min before immunoblotting (20).
Affinity Binding Assay-GST or GST fusion proteins were added to lysates of resting or stimulated neutrophils or to recombinant p47 phox N-terminal in phosphate-buffered saline, and were tumbled end-over-end for 1 h at room temperature. The samples were washed six times with phosphate-buffered saline, boiled in SDS sample buffer, and separated SDS-PAGE before immunoblotting.
FRET Analysis-Glass-adherent neutrophils or granulocytelike PLB cells were fixed with 3.7% formaldehyde in phosphatebuffered saline, either resting or after stimulation for 3 min with 50 ng/ml PMA. The cells were co-immunostained with anti-cPLA 2 mouse monoclonal antibodies (Santa Cruz Biotechnology, Santa Cruz, CA) and either p67 phox or p47 phox polyclonal rabbit antibodies (31) and with Cy3-conjugated (anti-rabbit IgG) and Cy5-conjugated (anti-goat IgG) secondary antibodies, respectively (Jackson ImmunoResearch Laboratories). Primary and secondary antibodies were diluted 1:500/1:4000 for neutrophils and 1:400/1:500 for granulocyte-like PLB cells, respectively. Cells were imaged with a Zeiss LSM 510 laser scanning confocal microscope using the 543 nm Green HeNe laser line and a 560 -615 nm band-pass filter for Cy3 and the 633 nm Red HeNe laser line and a 650 long pass filter for Cy5. Under these conditions, bleed-through between the Cy3 and Cy5 channels was insignificant. Cy3-sensitized emission of Cy5 was measured directly in the Cy5 channel while exciting Cy3 with the 543 nm line. A relative FRET index was calculated by normalizing the intensity of Cy3-sensitized Cy5 emission by the direct Cy3 emission intensity. FRET was further validated by donor recovery after acceptor photobleaching (32), by comparing Cy3 emission intensity before and after selective photobleaching of Cy5 with the 633 nm laser line. Mean fluorescence intensities were measured by calculating the average gray level of a circular region of interest encompassing each analyzed cell, correcting each channel for the background (offset), the average autofluorescence (determined in unstained samples), and the low bleed through between channels (calculated using singly stained samples).
Neutrophil Permeabilization and Superoxide Production-Neutrophils were electroporated exactly as described earlier (20) based on a previous study (34). The effect of the synthetic peptides on binding and NADPH oxidase activity was studied as described (35). The cell suspension was supplemented with 800 M synthetic peptides (Sigma) incubated on ice for 30 min, centrifuged, and resuspended in the same supplemented buffer containing 200 M peptide and 150 mM cytochrome c, stimulated for immediate measurement of superoxide production at 22°C.
Molecular Modeling-p47 phox -PX with bound phosphatidylinositol 3-phosphate was docked to the C2 domain of cPLA 2 (36) using the protein-protein docking program MolFit (37)(38)(39)(40). The initial step consisted of MolFit docking of the p47 phox -PX in complex with phosphatidylinositol 3-phosphate and cPLA 2 -C2 domain (36) (PDB code 1rlw), employing translational interval of 1.05 Å and rotational interval of 12°. The docking models were filtered, selecting models in which the active site in the catalytic domain of intact cPLA 2 (15) pointed toward the membrane, as deduced from the position of phosphatidylinositol 3-phosphate. In the second docking stage, the interactions of inhibitory peptides identified experimentally in this study were up-weighted. A similar filtering procedure for the position of the active site of cPLA 2 was employed as in the first docking stage, and the accepted models were clustered. The final models were energy-minimized. Details of the procedure are given as supplemental text.
Mutagenesis of Expression Vectors-pGEX-4T-2 expression vector encoding the cDNA of cPLA 2 -C2 domain or pGEX-3X expression vector encoding the cDNA of p47 phox -PX domain were used as a template, to generate the desired mutations introduced according to the molecular modeling, and generated by the overlap extension PCR (41). The PCR reactions, using appropriate complementary synthetic oligonucleotides FIGURE 1. cPLA 2 does not bind to flavocytochrome b 558 . A, cPLA 2 did not translocate to membranes in undifferentiated X-CGD-gp91 phox PLB cells. Undifferentiated or differentiated X-CGD-gp91 phox cells and differentiated X-CGD cells were stimulated by 50 ng/ml PMA for 3 min, before plasma membranes separation. The expression of gp91 phox in the plasma membrane and the detection of cPLA 2 and p47 phox translocated to the plasma membrane were evaluated by immunoblot analysis. B, the presence of a protein kinase C inhibitor attenuated the assembly of the oxidase and the translocation of cPLA 2 to the membranes of activated neutrophils. Neutrophils were preincubated for 5 min at 37°C without or with different concentrations of GF-109203X (GFX), and stimulated with 50 ng/ml PMA for 3 min before plasma membrane separation. The levels of cPLA 2 and p47 phox translocated to the plasma membrane were evaluated by Western blot analysis. For A and B, 2 ϫ 10 7 cell membrane equivalent were applied per lane, and the results shown are from one representative experiment out of three that gave identical results.
introducing the desired mutation and two additional primers at the ends of the cPLA 2 -C2 fragment and of p47 phox -PX, were performed with Red Load Taq Master/high yield using Thermostable DNA polymerase (Larova, Germany) with appropri-ate complementary synthetic oligonucleotides that introduced the desired mutation and two additional primers at the ends of the cPLA 2 -C2 fragment. The mutated products were digested with EcoRI and XhoI or with BamH1 and EcoRI and cloned into pGEX-4T-2 or pGEX-3X expression vectors, respectively, digested with the same enzymes. The vectors were then transformed into Escherichia coli DH-101. The mutated fragments were sequenced using the ABI3100 Genetic Analyzer. The first three mutations in cPLA 2 -C2 (F35A plus M38A plus L39A) were mutated in combination to generate triple mutation. The I67A mutation was mutated alone or in addition to the three mutations shown above. M59A and H115A in the p47 phox -PX were mutated alone or together.
Statistical Analysis-The mean differences were analyzed by Student's t test.

RESULTS AND DISCUSSION
Flavocytochrome b 558 Alone Is Not Sufficient to Anchor cPLA 2 to Plasma Membranes after Cell Stimulation-To define the exact binding domains between cPLA 2 and the assembled NADPH oxidase we first explored the oxidase subunits responsible for recruitment of cPLA 2 to the plasma membrane. To study whether cPLA 2 binds the cytochrome b 558 , cPLA 2 translocation to the plasma membranes was examined in undifferentiated X-CGD PLB cells transduced with full-length gp91 phox (XCGD-gp91) as reported in our previous studies (20,25). As shown in Fig. 1A, membranes were prepared from XCGD-gp91 PLB cells expressing gp91 phox protein, and its level did not change after differentiation. Stimulation with PMA did not cause any translocation of cPLA 2 to the plasma membrane of undifferentiated XCGD-gp91 phox PLB cells that do not express any other cytosolic oxidase subunits, indicating that cPLA 2 does not bind the gp91 phox itself. During differentiation the cells acquire the NADPH oxidase subunits as reported previously (18). In the presence of the cytosolic oxidase sub- A, translocation p47 phox and cPLA 2 to plasma membrane in stimulated p67 phox -deficient neutrophils expressing cPLA 2 and p47 phox but not p67 phox . Stimulation was induced by fMLP (5 ϫ 10 Ϫ7 M) for 3 min at 37°C. Similar results were obtained when cells were stimulated by PMA (not shown). B, no translocation of p47 phox or cPLA 2 to plasma membrane in stimulated p47 phoxdeficient neutrophils expressing cPLA 2 and p67 phox but not p47 phox . Cell stimulation as in A. C, translocation p47 phox , p67 phox , and cPLA 2 to plasma membrane in stimulated neutrophils expressing cPLA 2 and both p47 phox and p67 phox . Cell stimulation as in A. D, coimmunoprecipitation of p47 phox with cPLA 2 in p67 phox -deficient neutrophil membranes and in healthy controls membranes. The cell-solubilized membranes were subjected to immunoprecipitation with anti-cPLA 2 antibodies, separated by SDS-PAGE, and immunoblotted with anti-cPLA 2 , anti-p67 phox , and anti-p47 phox antibodies. E, p47 phox and p67 phox protein expression in cytosols from PLB-985 cells differentiated for 2 and 6 days with 10 Ϫ6 M retinoic acid. F, immunoblot analysis of cytosolic oxidase components and of cPLA 2 was performed to detect their translocation to the membranes of differentiated PLB cells after stimulation by 50 ng/ml PMA or 5 ϫ 10 Ϫ7 M fMLP. G, coimmunoprecipitation of p47 phox with cPLA 2 in membranes of differentiated PLB cells for 2 or 6 days. Solubilized membranes of unstimulated and stimulated differentiated PLB-985 cells (as in F) were subjected to immunoprecipitation with anti-cPLA 2 antibodies, applied to SDS-PAGE, and followed by immunoblotting with anti-p67 phox , anti-p47 phox , or anti-cPLA 2 antibodies. For both cell types the results are from a representative experiment out of three. 2 ϫ 10 6 cell cytosol equivalents or 2 ϫ 10 7 cell membrane equivalents were applied per lane. When cPLA 2 immunoprecipitation was performed with preimmune serum, cPLA 2 and oxidase components were not detected (not shown).  Fig. 2) were separated on 7% SDS gel, renatured, transblotted to nitrocellulose, and incubated with 5 g/ml recombinant p47 phox (A) or 5 g/ml recombinant p67 phox (B) followed by detection with antibodies against the recombinant protein (overlay). Western blot analysis (W.B.) with anti-cPLA 2 (right gels), anti-p47 phox , or anti-p67 phox (middle gels) antibodies was performed to indicate the location of cPLA 2 , p47 phox , and p67 phox and to demonstrate the specificity of the antibodies excluding nonspecific detection of other proteins. The results are from a representative experiment out of three with identical results. Similar results were obtained when cells were stimulated by PMA (not shown).   (Cy5, red). FRET was determined by measuring Cy5 sensitized emission induced by exciting Cy3. Sensitized emission could be observed only in the stimulated cells, and only when using the cPLA 2 and p47 phox antibodies. B and F, FRET index (mean Ϯ S.E.) measured in cells such as those shown in A and E, respectively (n Ͼ 20, three independent experiments). C and G, representative images of recovery of Cy3 fluorescence after Cy5 photobleaching using cPLA 2 (Cy3) and p47 phox (Cy5) labeling in neutrophils and granulocyte-like PLB-985 cells, respectively. Images show Cy3 fluorescence before (left) and after (right) photobleaching of Cy5. An increase (recovery) in Cy3 fluorescence, indicative of FRET, was observed only in stimulated cells. For clarity, Cy3 fluorescence intensity is shown in pseudocolor, using the scale supplied to the right of panel C. D and H, mean percentage of increase (recovery) of Cy3 fluorescence after Cy5 photobleaching, as in C and G, respectively (n Ͼ 20, three independent experiments). In all panels, ** indicates statistical significance at the 0.01 confidence level, as determined by the Student's t test. manner (Fig. 1B), consistent with our previous study reporting that GF-109203X causes the same dose response inhibition of p47 phox phosphorylation, oxidase assembly, and activity (26). Taken together, these results suggest that cPLA 2 does not bind directly to flavocytochrome b 558 but, rather, binds to the assembled oxidase complex. cPLA 2 Is Bound to p47 phox in the Assembled Oxidase-To determine which cytosolic component in the assembled oxidase is bound to cPLA 2 after stimulation, neutrophils from p67 phox -deficient CGD patients were studied, as shown by immunoblot of their cytosol ( Fig. 2A). Activation of these cells resulted in translocation of p47 phox and cPLA 2 to the membrane fractions ( Fig. 2A) similar to the translocation of these proteins in neutrophils from healthy donors, who express all oxidase components (Fig. 2C). In contrast, when neutrophils from p47 phox -deficient CGD patients were stimulated, there was no translocation of either p67 phox or cPLA 2 (Fig. 2B). The lack of p67 phox translocation is expected, because it is dependent on the translocation of p47 phox , which is missing in these cells. Addition of antibodies against cPLA 2 to the membrane fractions of activated neutrophils from p67 phox -deficient CGD patients caused a significant immunoprecipitation of cPLA 2 and coimmunoprecipitation of p47 phox similar to that observed in stimulated neutrophils from healthy controls (Fig. 2D). Similar results were obtained in PLB-985 cells differentiated with retinoic acid, which, similar to HL-60 cells (29), serve as a model for p67 phox -deficient cells, because the induction of p47 phox precedes that of p67 phox . As shown in Fig. 2C, PLB-985 cells differentiated for 2 days with retinoic acid express only p47 phox and not p67 phox and were not capable of producing superoxide following stimulation by PMA (0.2 Ϯ 0.07 nmol/10 6 /min), while after 6 days of differentiation, with the appearance of p67 phox protein (Fig. 2E), they produced significant stimulated superoxide (11.5 Ϯ 1.7 nmol/10 6 /min). The translocation of cPLA 2 to the plasma membranes was studied in differentiated PLB-985 cells for 2 and 6 days, which express similar levels of p47 phox but differ in their expression of p67 phox . As shown in Fig. 2F, cPLA 2 translocated to the plasma membranes after stimulation with fMLP or PMA, which correlated with p47 phox translocation, regardless of the presence of p67 phox . Immunoprecipitation of cPLA 2 from membrane fractions of stimulated cells resulted in coimmunoprecipitation of p47 phox in cells lacking or expressing p67 phox protein (Fig. 2G). These results show a direct binding between p47 phox and cPLA 2 in stimulated neutrophils and granulocyte-like PLB cells and that p47 phox is sufficient for the translocation of cPLA 2 to the plasma membranes, whereas p67 phox is not involved in this process. The absence of cPLA 2 translocation in the absence of the assembly of the oxidase suggests that cPLA 2 does not bind the oxidase membrane components, consistent with the results in Fig. 1B.
Although p67 phox is not required to anchor cPLA 2 to the assembled oxidase, there is a possibility of interaction between these two proteins within the assembled oxidase after the translocation of cPLA 2 . Overlay experiments were performed to determine whether cPLA 2 binds only p47 phox or also binds p67 phox , because in the absence of p47 phox subunit, in neutrophils of p47 phox -deficient CGD patients, there is no translocation of p67 phox to the membranes. As shown by the overlay experiments (detailed in the legends) presented in Fig. 3A, recombinant p47 phox was bound to p67 phox in lysates of both resting and stimulated cells (as expected) and was bound to cPLA 2 only in the lysates of activated cells. In contrast, recombinant p67 phox did not bind cPLA 2 in resting or stimulated cells, although it could bind p47 phox (Fig. 3B), indicating that there is no direct binding between cPLA 2 and p67 phox . The binding of recombinant-autoinhibited p47 phox to cPLA 2 in lysates of stimulated cells suggests that p47 phox does not need to be phosphorylated to bind cPLA 2 , whereas cPLA 2 has to be phosphorylated for binding p47 phox , which is in accordance with our previous study (25).
The binding of cPLA 2 to p47 phox but not to p67 phox was further supported by FRET measurements. Because efficient FRET occurs only at a distance of a few nanometers, its detection indicates a molecular range interaction between the labeled proteins (32,42). FRET between Cy3-labeled cPLA 2 and Cy5labeled p47 phox or p67 phox was performed, and FRET indices were consistent with binding between cPLA 2 and p47 phox , but not p67 phox , both in neutrophils and in granulocyte-like cells after stimulation (Fig. 4, A, B, E, and F), in correspondence with FIGURE 5. Binding between cPLA 2 and the p47 phox N-terminal. A, schematic presentation of the p47 phox domains as GST fusion proteins. The ball symbolizes GST. Amino acid residue numbers are shown in parentheses. B, affinity binding assay using GST-p47 phox C-terminal (47CT), GST-p47 phox N-terminal (47NT), or GST-p47 phox center region (47SH3). The GST fusion proteins were added to lysates of unstimulated (Ϫ) or stimulated neutrophils (with 50 ng/ml PMA or 1 mg/ml opsonized zymosan (OZ)), and subjected to Western blot analysis for detection of cPLA 2 . Detection of the different GST fusion domains of p47 phox was performed using anti-GST antibodies, because anti-p47 phox antibodies recognize only the N-terminal region. The results are from a representative experiment out of three. C, affinity binding assay in high salt concentrations using GST-p47NT and lysates of unstimulated or PMA-stimulated neutrophils. GST-47NT pulled down cPLA 2 similarly in the absence or presence of 0.5 M or 1 M NaCl. Shown are results of one representative experiment out of three. cPLA 2 -C2 and p47 phox -PX Interaction NOVEMBER 14, 2008 • VOLUME 283 • NUMBER 46 JOURNAL OF BIOLOGICAL CHEMISTRY 31903 the binding between cPLA 2 and p47 phox , but not p67 phox (Fig.  3). FRET was not detectable in unstimulated cells further supporting our previous finding (25) that though these proteins are both located in the cytosol, they do not interact prior to stimulation. Moreover, there is only partial translocation of p47 phox (ϳ10%) (43) and of cPLA 2 (25) to the plasma membrane. FRET is detected only in the cell periphery and not in the cytosol, consistent with our suggestion (25) that the binding of cPLA 2 occurs after the assembly of the oxidase. To further validate our FRET measurements, we tested for donor recovery after acceptor photobleaching (32). Indeed, photobleaching of Cy5-labeled p47 phox resulted in a 28% increase in Cy3-labeled cPLA 2 fluorescence in both types of stimulated cells, but not in unstimulated cells (Fig. 4, C, D, G, and H). Thus we show here by two different methods (Figs. 3 and 4) that cPLA 2 is anchored to the assembled oxidase upon stimulation through its interaction with p47 phox but not with p67 phox .
In Vitro Interaction between the Cytosolic p47 phox Component N-terminal Domain and the cPLA 2 -C2 Domain-To determine the p47 phox binding domains for cPLA 2 , three GST fusion constructs of p47 phox were expressed (Fig. 5A): GST-p47 phox N-terminal (1-150 aa) (GST-p47NT); GST-p47 phox SH3 domains (151-284 aa) (GST-p47SH3); and GST-p47 phox C-terminal (280 -390 aa) (GST-p47CT) (30). These GST fusion proteins were used to pull down cPLA 2 from cell lysates of unstimulated neutrophils or neutrophils stimulated by PMA or opsonized zymosan. As shown in Fig. 5B, cPLA 2 was pulled down from lysates of stimulated neutrophils only with GST-p47NT, but not with the other two constructs, suggesting that the binding between cPLA 2 and p47 phox occurs via the N-terminal domain of p47 phox . A significant binding to cPLA 2 was detected in lysates of stimulated neutrophils, indicating that the binding occurs only when cPLA 2 is in its phosphorylated form, consistent with the results in Fig. 3A. The low and not significant binding in lysates of unstimulated neutrophils probably results from some degree of neutrophil activation during preparation. The binding between cPLA 2 and GST-p47NT was not disrupted by the addition of high concentrations of NaCl (Fig. 5C), indicating that the binding between the two proteins is not mediated by electrostatic interactions.
To determine which domain in cPLA 2 binds the p47 phox N-terminal domain, four different GST fusion constructs of cPLA 2 were engineered (Fig. 6A) (Fig. 6B, left  blot). The blotted fusion proteins were overlaid with recombinant p47 phox N-terminal followed by probing the blot with goat anti-p47 phox (Fig. 6B, right blot), which recognizes the p47 phox N-terminal (30). The p47 phox N-terminal was bound exclusively to the C2 region of cPLA 2 , suggesting that these two domains are responsible for the interaction between cPLA 2 and the assembled NADPH oxidase after cell stimulation. Likewise, addition of GST-cPLA 2 -C2 to the p47 phox N-terminal recombinant protein (GST-C2ϩNT) pulled down the p47 phox N-terminal domain (Fig. 6C, right blot), whereas addition of GST to the p47 phox N-terminal (GSTϩNT) did not, indicating the specificity of the binding between cPLA 2 -C2 and the p47 phox N-terminal domains. The presence and location of the GST constructs were verified by immunoblotting with anti-GST antibodies (Fig. 6C, right blot). The binding between cPLA 2 -C2 and the whole p47 phox protein was also demonstrated in Fig. 6D. were separated on 10% SDS gels. The right blot was incubated with 5 g/ml recombinant p47 phox N-terminal followed by immunoblotting with anti-p47 phox antibodies. The location of the different GST fusion domains of cPLA 2 was determined using anti-GST antibodies (left gel). p47 phox N-terminal was bound only to the cPLA 2 -C2 domain. C, affinity-binding GST-cPLA 2 -C2 and p47 phox N-terminal domains GST-cPLA 2 -C2 or GST recombinant proteins attached to glutathione-Sepharose beads were added to recombinant p47 phox N-terminal (GST-C2ϩNT or GSTϩNT, respectively). The samples were separated on 10% SDS gel and subjected to Western blot analysis with anti-p47 phox (right gel). In addition, in each gel GST-cPLA 2 -C2 and p47 phox N-terminal proteins were loaded to verify their location (GST-C2 and NT, respectively). The left gel was immunoblotted with anti-GST antibodies to indicate the location and amount of GST-cPLA 2 -C2 and GST used. The right gel was immunoblotted with anti-p47 phox antibodies, and its location was verified by comparison with the migration of the p47 phox N-terminal (NT). In the right lane, in which only GST-C2 was loaded, p47 phox antibodies did not react, excluding nonspecific identification of other proteins. The results are from a representative experiment out of three. D, affinity binding assay between GST-cPLA 2 -C2 and whole p47 phox protein from lysates of neutrophils. The GST-cPLA 2 -C2 domain (GST-C2) or GST alone attached to glutathione-Sepharose beads were added to lysates of unstimulated neutrophils (Ϫ) or neutrophils stimulated by fMLP (ϩ). The samples were separated on 10% SDS gel and subjected to Western blot analysis with anti-p47 phox (upper gel) or anti-GST antibodies (lower gel). The results are from a representative experiment out of three. Similar results were obtained when cells were stimulated by PMA (not shown).
Addition of GST-cPLA 2 -C2 to lysates of unstimulated or stimulated neutrophils ( (Fig. 6D, lower gel) pulled down the cytosolic oxidase component p47 phox (Fig. 6D, upper gel), indicating that cPLA 2 -C2, similar to the whole cPLA 2 , can bind both forms of p47 phox , the autoinhibited unphosphorylated form, as shown for the recombinant p47 phox (Fig. 3A), and the phosphorylated-open form in stimulated cells. The affinity binding between recombinant p47 phox and cPLA 2 was detected only in lysates of stimulated neutrophils (Fig. 3), and FRET between p47 phox and cPLA 2 occurred only after cell stimulation (Fig. 4), indicating that cPLA 2 has to be in its phosphorylated form to bind p47 phox . In contrast, binding of GST-cPLA 2 -C2 to p47 phox N-terminal (Fig. 6) appears to be facilitated by the absence of the rest of cPLA 2 , bypassing the need for phosphorylation-dependent conformational changes required for binding of the complete cPLA 2 .
Determining the Binding Sites between cPLA 2 -C2 and the p47 phox N-terminal-Findings based on the reported structure of p47 phox in the assembled oxidase (2) and our results demonstrating that cPLA 2 translocates and binds to the NADPH oxidase only after its assembly (Figs. 1B and 4 and Ref. 25) and that the binding is located in the p47 phox -PX domain (Fig. 5) suggest that the binding site is located on an exposed surface of the p47 phox -PX domain. After stimulation, the structure of p47 phox is opened due to its phosphorylation, which allows binding to the C-terminal domains of gp91 phox (44) and p22 phox through its SH3 domains (6,45) and to phosphatidylinositol 3,4bisphosphate and phosphatidic acid on the plasma membrane through two basic pockets within its PX domain (6,9). Thus, other exposed PX domain surfaces appear to function as a docking site for cPLA 2 on the assembled NADPH oxidase, without interfering with the interaction of p47 phox with the plasma membrane, with the flavocytochrome b 558 , or with p67 phox . This notion is supported by the observed binding between the recombinant p47 phox protein and cPLA 2 (Fig. 3) and by affinity binding between p47 phox in lysates of unstimulated neutrophils and the GST-cPLA 2 -C2 construct (Fig. 6D). In both cases, p47 phox is found in its autoinhibited, folded conformation (1,6,30,46,47) in which the sole region of the p47 phox N-terminal domain available for binding cPLA 2 is its outer surface. This is because the inner surface that binds the membranes via its phosphatidylinositol 3,4-bisphosphate lipid binding pockets is normally masked by intramolecular interactions with the C-terminal domains in the autoinhibited form of p47 phox (2,9).
A computer-based docking search to predict the binding interface between p47 phox and cPLA 2 was conducted as described under "Experimental Procedures." In the first docking stage, interactions involving p47 phox -PX residues implicated in membrane binding were disregarded, because cPLA 2 binds p47 phox after oxidase assembly. In addition, models that did not position the active site of the catalytic domain of cPLA 2 near the cellular membranes (48) were discarded. The search produced several clusters of models with similar geometric electrostatic hydrophobic complementarity scores. A few linear peptides that were buried at the p47 phox -PX and cPLA 2 -C2 interfaces in these models were selected to determine the binding sites by competitive binding experiments. Electroporated neutrophils were shown to be permeable to small peptides used to inhibit NADPH oxidase assembly and activation (35). We used this methodology to determine the effects of the different peptides on binding of cPLA 2 to the assembled NADPH oxidase and on the production of superoxide by the oxidase after cell stimulation. Of all the peptides analyzed (Table 1), only two were found to be inhibitory: peptide 33 GAFGDML 39 , which corresponds to CBR1 of the cPLA 2 -C2 domain, and peptide 58 EMFPIEAGA 66 , derived from the p47 phox -PX domain. These two peptides significantly inhibited cPLA 2 translocation to the plasma membranes to same extent (62 Ϯ 10% inhibition compared with untreated cells) without affecting the assembly of the NADPH oxidase, as demonstrated by the detection of p47 phox translocation (Fig. 7A). Likewise, as shown in Fig. 7B, they caused a marked attenuation of affinity binding between the GST-p47 phox N-terminal and cPLA 2 in lysates of stimulated cells (82 Ϯ 11% inhibition, compared with untreated cells). The effects of the various peptides were analyzed on superoxide production in electro-permeabilized neutrophils. The rate of superoxide production in control electro-permeabilized neutrophils stimulated with 50 ng/ml PMA or 5 ϫ 10 Ϫ7 M fMLP was: 9.2 Ϯ 1.8 or 8.1 Ϯ 1.1 nmol/10 min/10 6 cells, respectively.

TABLE 1
The effects of peptides derived from cPLA 2 -C2 and p47 phox -PX domains on binding between cPLA 2 and NADPH oxidase and on superoxide production Shown are the sequences and topographical locations of the PX and C2 domain peptides used in the assay. Neutrophils were electropermeabilized in the presence of 0.8 mM concentrations of each peptide (as described under "Experimental Procedures") before stimulating with PMA (50 ng/ml). Superoxide production is expressed as % of control Ϯ S.E. of seven different experiments performed in duplicates. Similar results were obtained when the cells were stimulated with the inhibition of binding between cPLA 2 and NADPH oxidase shown in an example in Fig. 7  (A and B) is summarized in the right column (Binding Inhibition). The two peptides ( 33 GAFGDML 39 and 58 EMFPIEAGA 66 ) were also the only ones that inhibited stimulated superoxide production in electroporated neutrophils (Table 1), emphasizing the requirement for cPLA 2 binding for NADPH oxidase activation. Extension of the inhibitory peptides 33-39 to 32-40 aa resulting with addition of one charged amino acid on each side (Lys and Asp) caused a loss of the inhibitory effect, supporting the results (Fig. 5C) demonstrating that the interaction is primarily of a hydrophobic nature. All the other p47 phox peptides examined had no effect (Table 1), suggesting their corresponding protein domains do not play a significant role in the protein interaction. The PX domain of p47 phox has a flat, compact shape that consists of three ␤ strands, four ␣ helixes, and an exposed proline-rich segment held by the two helixes ␣1 and ␣2 that has been shown to bind SH3 domains in other proteins (33,49). However, the peptide competition experiments suggest that the exposed proline-rich region of the p47 phox PX domain, 71 NRIIPHLPAPK 81 aa, is not involved in binding to cPLA 2 . Fur-  Table 1. Similar inhibition of translocation and of affinity binding was obtained only in the presence of 58 EMFPIEAGA 66 (not shown). C, affinity binding assay between mutated GST-cPLA 2 -C2 and whole p47 phox protein from lysates of neutrophils. The different GST-cPLA 2 -C2 domains: wild type (WT), with substitution of Phe 35 , Met 38 , and Leu 39 to Ala (FML/A), with substitution of ILe 67 to Ala (I/A) and with substitution of the four amino acids to Ala (FMLI/A) or GST alone attached to glutathione-Sepharose beads were incubated with lysates of neutrophils. The samples were then separated on 10% SDS gel and subjected to Western blot analysis first with anti-p47 phox and then with anti-GST antibodies. The results are from a representative experiment out of seven independent experiments. The intensity of each band of p47 phox protein was divided by the intensity of each band of GST protein and expressed as percentage of WT. The mean Ϯ S.E. of the analysis of the seven experiments is presented. D, affinity binding assay between mutated GST-p47 phox -PX and cPLA 2 protein from lysates of stimulated neutrophils (with 50 ng/ml PMA). The different GST-p47 phox -PX domains: wild type (WT), with substitution of Met 59 to Ala (M/A), with substitution of His 115 to Ala (H/A) and with substitution of the two amino acids to Ala (MH/A) or GST alone attached to glutathione-Sepharose beads were incubated with lysates of stimulated neutrophils. The samples were then separated on 10% SDS gel and subjected to Western blot analysis with anti-cPLA 2 and with anti-GST antibodies. The results are from a representative experiment out of three independent experiments. The intensity of each band of cPLA 2 protein was divided by the intensity of each band of GST protein and expressed as percentage of WT. The mean Ϯ S.E. of the analysis of the three experiments is presented. E, model structure depicting the interaction between p47 phox -PX and cPLA 2 C2 domains. The N-terminal domain of PX is shown as a ribbon diagram in gray except for peptide 58 EMFPIEAGA 66 (colored green). Shown are the side chains of residues that bind phosphatidylinositol 3-phosphate or are suggested to interact with the membrane (blue) and residues that interact with cPLA 2 according to our docking model (green and gray). The phosphatidylinositol 3-phosphate molecule bound to PX is shown as a ball and stick model in cyan. cPLA 2 is shown as a ribbon diagram in dark blue except for peptide 33 GAFG-DML 39 , which is colored yellow, and the lid peptide, which is colored cyan. Shown are side chains of residues that interact with PX according to our docking model (yellow and dark blue), side chains of residues implicated in membrane binding (ball and stick model, cyan) and active site residues (ball and stick, red). Note that the catalytic site faces the membrane; the proposed rotation of the catalytic domain of cPLA 2 with respect to the C2 domain (15) will place the catalytic site closer to the membrane. The enlarged inset focuses on the predicted PX-cPLA 2 interaction contact site. It is rotated by ϳ90°compared with the whole view and shows the surface of PX in gray except for residues Pro 114 , His 115 , and Met 59 that bind to residue Ile 67 of cPLA 2 , which are emphasized in light brown, and residues 58 -66 (the inhibitory peptide), which are colored in green. cPLA 2 is shown as a dark blue and yellow ribbon (yellow for the inhibitory peptide) with Ile 67 emphasized (ball and stick model).
thermore, the region containing 3 MDGTFIR 9 aa and 10 HIALLGF 16 aa, covering most of the ␤ strand (Ile 6 -Lys 16 ), does not mediate binding to cPLA 2 . Neither do most of the ␣4 helix nor the C terminus of the PX domain, because peptides 106 LPTKISRC 113 and 132 QTKKPET 138 aa had no effect on binding or superoxide production.
The experimental results presented in Fig. 7 (A and B) and Table 1 were then used in the second docking stage in which interactions involving either peptide 58 EMFPIEAGA 66 of PX or peptide 33 GAFGDML 39 of C2 were favored (see "Experimental Procedures"). This led to the formation of a single cluster of six models in which these two peptides were at the interface, whereas the other tested peptides were not. Notably, the biased docking procedure produced more models in which these two peptides contributed to the interface, but it did not require that they interact with each other. Hence, the direct interaction between these peptides (Fig. 7E) is an independent result, not enforced by the weighting scheme employed in the computations. The six models in the final cluster, although similar, form two subgroups that differ in the details of the interaction in that in one subgroup residues Phe 35 , Met 38 , and Leu 39 of cPLA 2 are centrally positioned in the interface and in the other subgroup the centrally positioned residue is Ile 67 . The interface in both subgroups is mostly hydrophobic, in correspondence with the experimental results (Fig. 5C). To examine these models and to define the relative role of the cPLA 2 -C2 amino acid residues participating in the binding to p47 phox -PX, mutated GST constructs were engineered, and their efficiency for pulling down p47 phox from neutrophil lysates was analyzed. As shown in Fig.  7C, substitution of Phe 35 , Met 38 , and Leu 39 to Ala resulted in a slight decrease in binding affinity to p47 phox (of 30 Ϯ 7%) compared with the wild type, whereas substitution of Ile 67 to Ala was very efficient in inhibiting binding to p47 phox , reaching ϳ93 Ϯ 3% inhibition. Substitution of all four amino acids to Ala caused a 96 Ϯ 2% inhibition of binding to p47 phox , which was not significantly different from the effect of Ile 67 substitution alone. The mutagenesis results clearly favor one of the two docking models, in which Ile 67 is centrally positioned in the interface. Thus, Ile 67 , the first amino acid of the ␤4 strand, resides in a hydrophobic pocket on the surface of the PX domain and interacts with its residues Pro 114 and His 115 in the ␣4 helix, and Met 59 in the end of the ␣1 helix (the brown patch in Fig. 7E). Substitution of His 115 to Ala in the GST construct of p47 phox -PX domain did not affect the binding affinity to cPLA 2 in stimulated neutrophil lysate compared with the wild type, whereas substitution of Met 59 to Ala cause significant inhibition of the binding to cPLA 2 (32 Ϯ 8%). Substitution of both His 115 and Met 59 to Ala was very efficient in inhibiting the binding to cPLA 2 in stimulated neutrophil lysate reaching about 60% inhibition. Pro 114 was not mutated as it would have changed the PX domain conformation. The two inhibitory peptides (Fig. 7, A and B, and Table 1), highlighted in green and yellow in Fig. 7E, are at the edge of the interface. Thus, it is likely that they interfere with the hydrophobic interaction between cPLA 2 -C2 and p47 phox -PX, preventing the formation of the complex. In addition, the PX domain peptide 58 EMFPIEAGA 66 probably directly inhibits the binding, because it contains Met 59 , which participates in the binding with cPLA 2 -C2 Ile 67 (Fig. 7D).
In conclusion, we have clearly demonstrated that the binding between cPLA 2 to the assembled oxidase upon activation of neutrophils or granulocyte-like PLB-985 cells is mediated by the p47 phox cytosolic subunit. Our study shows that the cPLA 2 -C2 domain, which is responsible for anchoring cPLA 2 to the membrane, also contains other binding sites that engage cPLA 2 with the assembled oxidase and promote its activity.