p47phox Phox Homology Domain Regulates Plasma Membrane but Not Phagosome Neutrophil NADPH Oxidase Activation*

The assembly of cytosolic subunits p47phox, p67phox, and p40phox with flavocytochrome b558 at the membrane is required for activating the neutrophil NADPH oxidase that generates superoxide for microbial killing. The p47phox subunit plays a critical role in oxidase assembly. Recent studies showed that the p47phox Phox homology (PX) domain mediates phosphoinositide binding in vitro and regulates phorbol ester-induced NADPH oxidase activity in a K562 myeloid cell model. Because the importance of the p47phox PX domain in neutrophils is unclear, we investigated its role using p47phox knock-out (KO) mouse neutrophils to express human p47phox and derivatives harboring R90A mutations in the PX domain that result in loss of phosphoinositide binding. Human p47phox proteins were expressed at levels similar to endogenous murine p47phox, with the exception of a chronic granulomatous disease-associated R42Q mutant that was poorly expressed, and wild type human p47phox rescued p47phox KO mouse neutrophil NADPH oxidase activity. Plasma membrane NAPDH oxidase activity was reduced in neutrophils expressing p47phox with Arg90 substitutions, with substantial effects on responses to either phorbol ester or formyl-Met-Leu-Phe and more modest effects to particulate stimuli. In contrast, p47phox Arg90 mutants supported normal levels of intracellular NADPH oxidase activity during phagocytosis of a variety of particles and were recruited to phagosome membranes. This study defines a differential and agonist-dependent role of the p47phox PX domain for neutrophil NADPH oxidase activation.

The assembly of cytosolic subunits p47 phox , p67 phox , and p40 phox with flavocytochrome b 558 at the membrane is required for activating the neutrophil NADPH oxidase that generates superoxide for microbial killing. The p47 phox subunit plays a critical role in oxidase assembly. Recent studies showed that the p47 phox Phox homology (PX) domain mediates phosphoinositide binding in vitro and regulates phorbol ester-induced NADPH oxidase activity in a K562 myeloid cell model. Because the importance of the p47 phox PX domain in neutrophils is unclear, we investigated its role using p47 phox knock-out (KO) mouse neutrophils to express human p47 phox and derivatives harboring R90A mutations in the PX domain that result in loss of phosphoinositide binding. Human p47 phox proteins were expressed at levels similar to endogenous murine p47 phox , with the exception of a chronic granulomatous disease-associated R42Q mutant that was poorly expressed, and wild type human p47 phox rescued p47 phox KO mouse neutrophil NADPH oxidase activity. Plasma membrane NAPDH oxidase activity was reduced in neutrophils expressing p47 phox with Arg 90 substitutions, with substantial effects on responses to either phorbol ester or formyl-Met-Leu-Phe and more modest effects to particulate stimuli. In contrast, p47 phox Arg 90 mutants supported normal levels of intracellular NADPH oxidase activity during phagocytosis of a variety of particles and were recruited to phagosome membranes. This study defines a differential and agonist-dependent role of the p47 phox PX domain for neutrophil NADPH oxidase activation.
The reduced nicotinamide adenine dinucleotide phosphate (NADPH) oxidase of phagocytic leukocytes plays a key role in innate host defense against bacterial and fungal infections (1)(2)(3). The phagocyte NADPH oxidase is composed of mem-brane-integrated flavocytochrome b 558 (a heterodimer composed of gp91 phox (NOX2) and p22 phox ) and four cytosolic components: p47 phox , p67 phox , p40 phox , and Rac2 (2)(3)(4). Upon activation by either soluble or particulate stimuli, the cytosolic subunits translocate to flavocytochrome b 558 to form the activated NADPH oxidase complex, resulting in electron transfer from cytosolic NADPH through FAD and heme groups to extracellular or phagosome-located oxygen, from which superoxide is generated (2)(3)(4). Genetic defects in any of the five phox subunits of the NADPH oxidase complex result in chronic granulomatous disease (CGD), 2 which is characterized by absent or deficient NADPH oxidase activity, recurrent pyogenic infections, and granulomatous inflammation (1,5,6).
The assembly of the NADPH oxidase complex is essential for activation of superoxide production, and p47 phox plays a central role in this assembly (2)(3)(4)(7)(8)(9)(10)(11)(12). From the N terminus to the C terminus, p47 phox contains a Phox homology (PX) domain, two tandemly arranged Src homology 3 (SH3) domains, an autoinhibitory region (AIR), and a proline-rich region (PRR; Fig. 1A). In the resting state, p47 phox is autoinhibited via intramolecular interactions of the PX and two SH3 domains with the AIR and adjacent region (4,10,(13)(14)(15). p47 phox forms a heterotrimeric complex with p67 phox and p40 phox via a "tail-to-tail" interaction between the C-terminal SH3 domain of p67 phox and prolinerich region of p47 phox and a PB1-PB1 association between p67 phox and p40 phox (4,16). Upon cell stimulation, p47 phox is phosphorylated on multiple serine residues in the AIR, which acts as a molecular switch to liberate its autoinhibited structure and release the PX and tandem SH3 domains, with the latter binding to the proline-rich region of membrane-bound p22 phox (8,14,17,18). The p47 phox -p22 phox interaction mediates the recruitment of the heterotrimeric phox complex, and neither p67 phox nor p40 phox undergoes membrane translocation in the absence of p47 phox (7,19).
The PX domain is a phosphoinositide binding module that was first described as a motif in the p47 phox and p40 phox subunits of the NADPH oxidase complex (20 -24). Binding of the p40 phox PX domain to its target, PI3P, plays a critical role in NADPH oxidase activity in neutrophil phagosomes (6,(25)(26)(27)(28). Unlike the p40 phox PX domain, which has a single binding pocket with high affinity for PI3P, the PX domain of p47 phox has two distinct lipid binding pockets. The main pocket prefers PI(3,4)P 2 but also weakly binds other phosphoinositides (23, 29 -31). The p47 phox PX domain has a shallow second pocket with affinity for phosphatidic acid or phosphatidylserine, and both pockets can simultaneously and synergistically bind to their lipid ligands (30,32,33). Arg 43 and Arg 90 in the p47 phox PX domain mediate binding to P(3,4)P 2 via interaction with the 3and 4-phosphates, respectively, based on crystallography (30) and mutagenesis studies (23,32,34). The PX domain of fulllength p47 phox is masked in unstimulated cells but exposed upon activation-induced p47 phox phosphorylation of the AIR (34). In a whole-cell model using K562 cells, an R90K mutation in p47 phox markedly reduced phorbol ester-induced recruitment of p47 phox to membranes and NADPH oxidase activity (34). The NOXO1 (Nox-organizing protein 1) homolog of p47 phox also has a PX domain, which binds to PI(3,5)P 2 , PI5P, and PI4P (35) . The PX domain in NOXO1, which lacks an AIR, does not appear to be masked and mediates the constitutive localization of NOXO1 to the plasma membrane and its activation of the NOX1 homolog of gp91 phox in an HEK293 cell model (35,36).
The physiological function of the p47 phox PX domain in phagocytic leukocytes remains unknown. In this study, we introduced PX domain mutations that impair phosphoinositide binding into full-length p47 phox and examined the impact on NADPH oxidase activity elicited by soluble and particulate stimuli. We took advantage of the p47 phox knock-out (KO) mouse (37) to express wild type human p47 phox and derivatives. Human p47 phox , which is 82% identical to murine p47 phox , can rescue phorbol ester-elicited NADPH oxidase activity in p47 phox KO mouse neutrophils (38). Here, we showed that human p47 phox or a derivative tagged at its C terminus with YFP could rescue NADPH oxidase activity in response to PMA, fMLF, and particulate stimuli, including IgG-opsonized latex beads, serum-opsonized zymosan (SOZ), serum-opsonized Staphylococcus aureus, and sterilized Aspergillus fumigatus hyphae. Mutations in the PX domain of p47 phox that impair phosphoinositide binding led to impaired neutrophil NADPH oxidase activation on the plasma membrane but had little effect on intracellular reactive oxygen species (ROS) production during phagocytosis, thus defining a differential role for the p47 phox PX domain.

EXPERIMENTAL PROCEDURES
Reagents and Antibodies-Chemicals were purchased from Sigma-Aldrich unless otherwise stated. Phosphate-buffered sodium (PBS), pH 7.2, penicillin/streptomycin, neomycin, and RPMI 1640 were from Invitrogen; fetal calf serum (FCS) was from HyClone Laboratory (Logan, UT). G418 was purchased from Calbiochem. The ECL detection kit came from Pierce. Polyclonal antibody against DsRed (red fluorescent protein) was obtained from Clontech (catalog no. 632496). Latex beads (3.3 m) were from Bangs laboratory Inc. (Fisher, IN). Polyclonal antibody against green fluorescent protein (GFP) was obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Rabbit polyclonal antibodies against p40 phox and p47 phox were from Upstate Biotechnology, Inc. (Lake Placid, NY), and monoclonal antibody against p67 phox was from BD Biosciences. Monoclonal antibodies 54.1 and NS2 against gp91 phox and p22 phox , respectively, were gifts from Dr. Jesaitis (Montana State University, Bozeman, MT). Rabbit polyclonal antibody against p47 phox was a gift from Dr. Lambeth and Dr. Uhlinger (Emory University, Atlanta, GA). 7D5 mAb (anti-gp91 phox ) was collected from hybridoma cells kindly provided by M. Nakamura (Nagasaki University, Japan). The Amaxa kit V was from Amaxa Biosystems (Cologne, Germany). S. aureus Wood 46 was purchased from ATCC (Manassas, VA).
Plasmid Construction-The human p47 phox cDNA was subcloned into a pEYFP-N1 vector to generate a fluorescently tagged p47 phox probe. The cDNA encoding wild type (WT) p47 phox was amplified from pRK5-p47 phox (gift from Dr. Lambeth, Emory University Medical School, Atlanta, GA) by polymerase chain reaction (PCR) and cloned into EcoRI and KpnI sites of pEYFP-N1 (BD Biosciences Clontech) to generate p47YFP. Site-directed mutagenesis was performed in p47YFP using the QuikChange site-directed mutagenesis kit (Stratagene; La Jolla, CA). The constructs were confirmed by sequencing. The cDNA fragment of p47YFP or mutants was then excised by digesting with NotI, blunting, and then digesting with EcoRI; pMSCV (Clontech) was digested with ClaI, followed with blunting, and then digested with EcoRI. The cDNA for p47YFP or mutants was ligated to pMSCV. A cDNA for p47 phox in pRK5 vector was subcloned into pMSCVpac using EcoRI to generate pMSCV-pac-p47 phox . The cDNA fragment of p67 phox from p67YFP was subcloned into mCherry-N vector (gift from J. Swanson, University of Michigan) at the XhoI and HindIII sites to generate p67Cherry (39,40). Retroviral vectors were packaged as described previously (28). Other plasmids for expression of phox subunits have been described previously (28,41).
Retroviral Transduction of p47 phox KO Mouse Bone Marrow (BM) and Neutrophil Differentiation-Retroviral transduction of p47 phox KO mouse BM cells with MSCV-p47YFP or mutants or with MSCV-pac-p47 phox or mutants was performed as described (41,42). Transduced BM cells were differentiated p47 phox PX Domain in NADPH Oxidase Activation into neutrophils in ␣-minimum essential medium with 20% heat-inactivated FCS, 1% penicillin/streptomycin, 50 ng/ml human G-CSF and 50 units/ml mIL-3. The first day of starting differentiation was considered as day 0. Cells were counted and replated to a concentration of 0.5 ϫ 10 6 /ml in fresh differentiation medium every 2 days; activity and live images were analyzed on days 6 and 7. Puromycin (1 g/ml) was used to select the MSCV-pac-p47 phox , R90K, or R42Q mutant transduced cells. After differentiation, transduction efficiency was determined by flow cytometry (FACSCalibur, BD Biosciences). Except for p47-YFP-R42Q, the YFP-positive cells were ϳ30 -50% of the total, and mean fluorescent intensity was 150 -400 relative units.
Analysis of phox Subunit Expression-Cell lysates were prepared from K562 cells or mouse neutrophils using 1% Triton X-100, and 15 or 30 g were subjected to SDS-PAGE and immunoblotting using ECL detection (40). ImageJ (available from the National Institutes of Health Web site) was used for densitometry analysis of YFP-tagged or non-YFP-tagged p47 phox or mutants. In some experiments, the Triton X-100insoluble pellet was also analyzed by SDS-PAGE and immunoblotting as described (28,44). YFP expression was also analyzed by flow cytometry (FACSCalibur, BD Biosciences) (40).
NADPH Oxidase Activation in Intact Cells-NADPH oxidase activity was assayed using chemiluminescence enhanced by luminol or isoluminol, which is membrane-impermeable; both compounds detect ROS in a peroxidase-dependent reaction (40,45,46). PMA (300 ng/ml) or hIgG-latex beads was used to activate 2 ϫ 10 5 K562-gp91/p67Cherry cells, co-transfected with p40 phox and p47YFP or mutants in the presence of 20 M isoluminol and 20 units/ml horseradish peroxidase (HRP). An Lmax microplate luminometer (Molecular Devices, Sunnyvale, CA) was used to record luminescence every 1-1.5 min at 37°C for a total of 46 readings. A similar protocol was used to measure extracellular ROS release in 2.5 ϫ 10 5 BM differentiated neutrophils after PMA, fMLF, or particulate stimulation. Intracellular ROS production during synchronized phagocytosis of hIgG-latex or SOZ was measured in the presence of luminol and superoxide dismutase (28), and intracellular ROS generation during synchronized phagocytosis of serum-opsonized or heat-inactivated S. aureus or hyphae was measured in the presence of luminol as described by Hawkins' group (25). Activity was normalized for p47 phox expression determined by densitometry.
Live Images by Confocal Video Microscopy-SOZ-induced phagocytosis in p47 phox KO neutrophils expressing p47YFP or mutants was filmed using a spinning disk (CSU10) confocal system mounted on a Nikon TE-2000U inverted microscope with an Ixon air-cooled EMCCD camera (Andor Technology, South Windsor, CT) and a Nikon Plan Apo ϫ100 1.4 numerical aperture objective as described previously (28,40). All images were analyzed with Metamorph software (Universal Imaging; Downington, PA). Each type of experiment was performed on at least three independent occasions.
ImageJ (National Institutes of Health) was used to analyze the accumulation of fluorescent probes during phagocytosis relative to their cytoplasmic distribution as described (28,40). Phagosomes were monitored and analyzed at each stage (cup, closure (time of sealing), and postinternalization (200 -300 s)) with this method, including determination of the mean Ϯ S.E.

p47 phox PX Domain Is Important for Supporting Plasma
Membrane NADPH Oxidase Activity in K562 Cells in Response to PMA and hIgG-Latex Beads-K562-gp91 phox cells, a primitive myeloid cell line that expresses endogenous p22 phox and is engineered to express flavocytochrome b by insertion of a stable transgene for gp91 phox , were used to validate the function of fluorescence-tagged cytosolic phox subunit probes. K562-gp91 phox cells that were transiently co-transfected for expression of p47 phox tagged at its C terminus with YFP (p47YFP) (Fig.  1A) and p67 phox produced similar amounts of ROS induced by PMA, compared with cells transfected with untagged p47 phox and p67 phox (supplemental Fig. 1, A-C), showing that p47YFP is fully functional. In contrast, p47 phox tagged at its N terminus (YFP-p47 phox ) supported only 50% activity compared with p47 phox in response to PMA (data not shown). No ROS production was observed in K562-gp91 phox expressing YFP alone in the presence of p67 phox (data not shown). For subsequent studies, we generated K562 cells stably expressing both gp91 phox and p67Cherry, and a population expressing both transgenes was isolated by FACS. The function of p67Cherry was validated by showing that K562-gp91 phox cells co-transfected for expression of p67 phox or p67Cherry with p47YFP showed similar ROS release capacities upon PMA stimulation (data not shown).
To examine the role of phosphoinositide binding to the p47 phox PX domain, we mutated two amino acids (Arg 43 and Arg 90 ) that are critical for interactions with the 3-and 4-phosphates, respectively, of PI(3,4)P 2 (Fig. 1A) (30,32,47). We tested these mutants in K562-gp91/p67Cherry cells stimulated with either PMA or IgG-opsonized particles following co-transfection of plasmids for expression of p47YFP WT or derivatives, along with p40 phox ; co-expression of p40 phox enhances oxidase activity in response to PMA (48), and we found that p40 phox was required for response to hIgG-latex beads (data not shown). For comparison with p47 phox PX domain mutants, we assayed in parallel a p47 phox derivative with a W193R mutation in the N-terminal SH3 domain, which is unable to translocate to the membrane because it cannot bind to the flavocytochrome b via p22 phox (8,18,49) (Fig. 1A). YFPtagged p47 phox and mutants were similarly expressed in K562-gp91/p67Cherry cells (Fig. 1B), and protein levels of p40 phox , p67Cherry and gp91 phox , and endogenous p22 phox were also similar. p47 phox PX Domain in NADPH Oxidase Activation NOVEMBER 5, 2010 • VOLUME 285 • NUMBER 45 Fig. 1 are representative assays and a summary of three independent experiments in which the activity of YFPtagged p47 phox mutants relative to p47YFP was normalized to protein expression level. As expected, a W193R mutation in the SH3b domain of p47 phox abolished NADPH oxidase activity in response to PMA and hIgG-latex ( Fig. 1, C-E). The p47 phox R43A and p47 phox R90A mutations resulted in a 50 -70% reduction, respectively, in ROS production in response to PMA (Fig.  1, C and E), confirming a previous report on the importance of the p47 phox PX domain for PMA-induced oxidase activity in the K562 model (34). We further showed that these mutations also significantly reduced ROS production in response to hIgG-latex beads compared with p47 phox WT (Fig. 1, D and E). The double mutant p47 phox R43A/R90A reduced ROS generation by 80 -85% (Fig. 1, C-E), a modest decrease over the R90A mutation alone. The significant difference in ROS generation was also observed in R43A versus R90A and R90A versus R43A/ R90A in response to both PMA and hIgG-latex beads. Both PMA-and hIgG-latex bead-induced ROS production were fully superoxide dismutase-sensitive (not shown), indicating that the detected ROS was released at the plasma membrane, as expected, because this K562 cell line does not ingest IgG particles. 3 Non-YFP-tagged p47 phox -R90K was also tested in K562-gp91/p67Cherry cells, with similar results (data not shown).

Shown in
Thus, our results suggest that Arg 43 and Arg 90 are each important for supporting NADPH oxidase activity on the plasma membrane in response to either PMA or hIgG-latex beads in K562 cells. Although Arg 90 is conserved in human and mouse p47 phox , there is a lysine rather than an arginine at position 43 in mouse p47 phox (50). Because p47 phox R90A is both more conserved and showed a greater reduction in ROS production than R43A in the K562 cell model (Fig. 1), we therefore focused on this mutant in subsequent studies.
Expression and Function of Human p47 phox in Murine Neutrophils-Human and murine p47 phox are 82% identical, and human p47 phox partially rescued NADPH oxidase activity in PMA-simulated p47 phox KO mouse neutrophils (38). To further evaluate the function of human p47 phox in p47 phox KO mouse neutrophils, we used an MSCV-p47 phox vector containing a puromycin-resistant cassette and selected for transduced cells undergoing neutrophil differentiation from myeloid progenitors. Human p47 phox was expressed in puromycin-selected mouse p47 phox KO mouse neutrophils at levels somewhat higher than endogenous p47 phox in WT mouse neutrophils (supplemental Fig. 2A). Endogenous p47 phox was absent in p47 phox KO neutrophils, as expected (supplemental Fig. 2A). Other phox proteins were expressed at similar levels (supplemental Fig. 2A). Human p47 phox partially rescued ROS generation (41 Ϯ 11% of WT mouse neutrophils) in response to PMA (supplemental Fig. 2, B and J) and fully rescued NADPH oxidase activity (98 Ϯ 30%) in response to fMLF (supplemental Fig. 2, C and J). We next characterized ROS production in response to model particles and to microbial stimuli. Human p47 phox expression completely rescued NADPH oxidase activity in p47 phox -deficient mouse neutrophils in response to hIgG-latex beads, SOZ, serum-opsonized S. aureus, heat-inactivated S. aureus, and sterilized A. fumigatus hyphae (supplemental Fig. 2, D-K), further extending previous findings (38) that human p47 phox can replace murine p47 phox in supporting NADPH oxidase activity and that the p47 phox KO neutrophils are a good model for studying the function of human p47 phox .
The PX Domain of p47 phox Is Important for Supporting Plasma Membrane NADPH Oxidase Activity in Neutrophils in Response to PMA and fMLF-We took advantage of the p47 phox KO mouse to evaluate the function of human p47 phox derivatives with mutations in the PX domain. We first evaluated NADPH oxidase activity in p47 phox KO neutrophils expressing either YFP-tagged p47 phox or mutants. Both immunoblotting ( Fig. 2A) and flow cytometry (not shown) showed that the expression of p47YFP-R90A was slightly lower than that of p47YFP; therefore, NADPH oxidase activity was normalized to YFP-tagged p47 phox protein expression. All other NADPH oxi- p47 phox PX Domain in NADPH Oxidase Activation dase components were expressed at similar levels in p47 phox KO non-transduced or transduced cells ( Fig. 2A). The p47YFP-R90A mutation resulted in a severe reduction in PMA-stimulated plasma membrane ROS production by 84 Ϯ 6% compared with p47YFP (Fig. 2B). This is consistent with results in the K562 cell model (Fig. 1, C and E) (34). The p47 phox R90A mutation also produced a 72 Ϯ 13% reduction in ROS generation upon fMLF stimulation (Fig. 2C). Similar results were observed in puromycin-selected neutrophils expressing the p47 phox -R90K mutant (without YFP tag) as compared with WT p47 phox (Fig. 2, D-F). These results indicate that Arg 90 in p47 phox , a residue that plays a critical role in phosphoinositide binding to its PX domain, is a positive regulator of both phorbol ester-and chemoattractant-induced NADPH oxidase activity on neutrophil plasma membranes.
Effect of PI3K Inhibition by Wortmannin-To evaluate whether the effect of the PX domain mutation on reduced p47 phox function in PMA-stimulated neutrophils is mediated by PI3K-derived phosphoinositides, we examined the effect of wortmannin. As reported for K562 cells (where wortmannin also did not affect translocation) (34), PMA-induced activity in mouse neutrophils was not inhibited by wortmannin (supplemental Fig. 3), suggesting that phosphoinositides with a 3Ј-phosphate are not required for the p47 phox PX domain to regulate PMA-induced ROS production. Although PI3K inhib-itors abolish neutrophil NADPH oxidase activity in response to fMLF (51), PI3K plays multiple roles downstream of this agonist, including Rac activation, and it is not possible to tie its inhibition of ROS production to effects on binding targets of the p47 phox PX domain.
Effect of p47 phox Arg 90 Mutations on Neutrophil Plasma Membrane and Intracellular NADPH Oxidase Activity Induced by Model Particles and Microbial Stimuli-We next examined the impact of mutations in the p47 phox PX domain on NADPH oxidase activation by particulate stimuli. The kinetics of particle-induced extracellular ROS production in mouse p47 phox KO neutrophils expressing p47YFP-R90A was similar to the kinetics of neutrophils expressing p47YFP. However, total integrated extracellular ROS generation was reduced by 20 -40% in response to hIgG-latex beads (Fig. 3, A and B), SOZ (Fig. 3, A  and B), serum-opsonized S. aureus (Fig. 3, A and B), heat-inactivated S. aureus (data not shown) (Fig. 3B), and hyphae (Fig. 3,  A and B), respectively. In contrast, the time course and amount of intracellular ROS elicited during phagocytosis were comparable in neutrophils expressing either p47YFP or p47YFP-R90A (Fig. 3, A and B). Similar results were observed using the p47 phox -R90K mutant (without YFP tag) compared with p47 phox WT (Fig. 3C), although there was small but statistically significant reduction in serum-opsonized S. aureus-induced intracellular ROS generation in neutrophils expressing p47 phox -R90K (Fig. 3C). Taken together, our data suggest that although phosphoinositide binding to the p47 phox PX domain plays an important role in regulating extracellular ROS responses, particularly for chemoattractants and phorbol ester, it is not involved in regulating intracellular NADPH oxidase activity during phagocytosis.
The p47 phox R90A Mutation Does Not Impair the Recruitment of p47 phox during SOZ Phagocytosis-To investigate the localization of YFP-tagged p47 phox and YFP-tagged p47 phox R90A during serum-opsonized zymosan phagocytosis, we took advantage of time lapse confocal video microscopy. p47YFP, which was cytosolic in unstimulated cells, accumulated on the cup prior to phagosome sealing and persisted after sealing for at least 200 s during SOZ phagocytosis (Fig. 4A and supplemental Movie 1). Like p47YFP, p47YFP-90A was cytosolic in resting cells, and during SOZ phagocytosis it was recruited to the phago-some cup and visible on the phagosome for at least 200 s ( Fig.  4A and supplemental Movie 2). As previously observed for accumulation of p67YFP in PLB-985 granulocytes during hIgGzymosan phagocytosis (40), only approximately one-half of the phagocytic cups and internalized phagosomes accumulated p47YFP (Fig. 4B). Although cells expressing p47 phox R90A demonstrated an ϳ20% reduction in extracellular ROS production (Fig. 3, A and B), there was no effect on the frequency of p47FYP-R90A-positive phagocytic cups or phagosomes compared with p47YFP in comparable studies (Fig. 4B). The kinetics and relative amount of p47YFP and p47YFP-R90A accumulating on individual phagosomes were also similar (Fig. 4C). The p47YFP-W193R mutant (Fig. 1A), unable to bind to flavocytochrome b, was also examined as a negative control. p47YFP-W193R was distributed in the nucleus and cytosol in the resting neutrophils (data not shown), consistent with obser-  (52), and did not accumulate on phagosome membranes during SOZ phagocytosis (Fig. 4C).
Analysis of an R42Q p47 phox Mutant Identified in p47 phoxdeficient CGD-Noack et al. (53) studied two p47 phox CGD patients with absent O 2 . production who had point mutations in NCF1 (neutrophil cytosolic factor-1), the gene encoding p47 phox , predicting an R42Q substitution in the phosphoinositide binding pocket. Both patients were compound heterozygotes, with the second allele in each patient harboring small deletions that result in frameshift and premature stop codons. The introduction of the R42Q mutation into the isolated p47 phox PX domain abolished its phosphoinositide binding in vitro (23,29). However, the above mentioned patients had no p47 phox expression by immunoblot, suggesting that the R42Q renders the full-length protein unstable (54).
To explore the behavior of this CGD-related p47 phox mutation, we first used K562-gp91/p67Cherry cells. As detected by flow cytometry following transient transfection with plasmid expression vectors, p47YFP-R42Q was expressed at only 7% of wild type p47YFP levels (Fig. 5A). Increasing the amount of the pEYFP-N1-p47 phox -R42Q plasmid did not significantly increase the expression of the mutant protein (data not shown). Immunoblots using an anti-GFP polyclonal antibody showed that the p47YFP-R42Q was trapped in the Triton X-100-insoluble fraction (Fig. 5B). Interestingly, co-expressed p40 phox and p67Cherry were also partly retained in the insoluble fraction (Fig. 5B), suggesting that these three cytosolic phox proteins can form a trimeric complex in K562 cells, as in neutrophils (4,16). Under confocal microscopy, we observed a punctate distribution of p47YFP-R42Q (Fig. 5C) in many cells, whereas p47YFP was distributed uniformly in the cytosol (Fig. 5C). A punctate distribution of p67Cherry was also observed in cells with this distribution of p47YFP-R42Q (Fig. 5C). PMA-stimulated cells expressing p47YFP-R42Q supported only ϳ1% of the ROS release seen in cells expressing p47YFP (Fig. 5, D and F). No ROS release was detected in p47YFP-R42Q in K562 cells stimulated with hIgG-latex beads (Fig. 5, E and F).
We next expressed p47YFP-R42Q in mouse p47 phox KO neutrophils using retroviral transduction of myeloid progenitors, as above. Flow cytometry showed a 10-fold reduction in the level of expression of p47YFP-R42Q compared with p47YFP (Fig. 6A). We also saw markedly reduced expression of p47YFP-R42Q by Western blot of Triton-soluble neutrophil extracts (Fig. 6B). Other phox proteins in neutrophils expressing p47YFP or p47YFP-R42Q were present at comparable levels (Fig. 6B), suggesting that bone marrow progenitors differentiated similarly into neutrophils. To try to enhance expression of p47 phox -R42Q, we used a retroviral vector with a marker for puromycin selection, but protein expression of p47 phox -R42Q was still much lower than that of p47 phox WT (Fig. 6C). Similar to K562 cells (Fig. 5C), neutrophil p47YFP-R42Q showed a punctate distribution (Fig. 6D). For reasons that are uncertain, we did not detect p47YFP-R42Q (supplemental Fig. 4) or p47 phox -R42Q (not shown) in the Triton X-100-insoluble fraction from neutrophil extracts, in contrast to K562 cells expressing p47YFP-R42Q.
We used confocal video microscopy to examine whether p47YFP-R42Q could accumulate on SOZ phagosomes. We did not detect any membrane translocation ( Fig. 6E and supplemental Movie 3), although the interpretation of this observation is difficult given the low level of protein expression and abnormal distribution (Fig. 6, A-D). We also examined NADPH oxidase activity in neutrophils expressing p47 phox -R42Q. Extracellular ROS in response to PMA, fMLF, or different particles was not detectable (data not shown). However, low but detectable intracellular ROS were observed in response to hIgG-latex beads, serum-opsonized S. aureus, and sterile hyphae (Fig. 6, F-H).

DISCUSSION
PX domains are important motifs for binding membrane phosphoinositides that regulate the localization and activity of proteins and are contained in two regulatory subunits of the NADPH oxidase. Unlike the p40 phox PX domain, which is specific for PI3P (23,33), the p47 phox PX domain binds preferentially to PI(3,4)P 2 and can also bind with lower affinity to other phosphoinositides (23,29,30,32,47). The current study identifies a differential and agonist-dependent role of the p47 phox PX domain during neutrophil NADPH oxidase activation in response to soluble and particulate stimuli, showing that the p47 phox PX domain regulates ROS production on the plasma membrane, particularly in response to fMLF and phorbol ester, but not intracellular ROS production during phagocytosis.
An arginine residue at amino acid 90 in the p47 phox PX domain plays a critical role in ligating the 4-position phosphate in target phosphoinositides (30). In initial studies, we confirmed the importance of p47 phox Arg 90 for PMA-activated p47 phox PX Domain in NADPH Oxidase Activation NOVEMBER 5, 2010 • VOLUME 285 • NUMBER 45 ROS release by K562 cells on the plasma membrane (34) and extended findings in this model to show that p47 phox Arg 90 regulates IgG particle-induced NADPH oxidase activity on the plasma membrane. We also studied the effect of mutating Arg 43 , which normally interacts with the 3-phosphate of PI(3,4)P 2 in the phosphoinositide-binding pocket of human p47 phox . A R43Q mutation in the isolated p47 phox PX domain resulted in a more severe impairment of phosphoinositide binding in vitro compared with R90A (30). However, in the K562 model, the p47 phox R43A mutant had less effect on PMAinduced ROS generation than the p47 phox R90A mutant. This disparity could result from the complexity of the in vivo microenvironment compared with the simplicity of in vitro lipid binding studies.
In mouse p47 phox KO neutrophils, p47 phox R90A or R90K mutants showed substantial defects in both PMA-and fMLFinduced release of ROS at the plasma membrane. Upon fMLF stimulation, PI(3,4)P 2 and PI(3,4,5)P 3 increase and PI(4,5)P 2 and PI3P decrease in human PMNs (55), as a result of activation of Class I PI3Ks (51,56). Thus, fMLF-induced increases in plasma membrane PI(3,4)P 2 and PI(3,4,5)P 3 could stimulate NADPH oxidase activity via the p47 phox PX domain. However, PMA-induced plasma membrane NADPH oxidase activity is insensitive to wortmannin while still dependent on p47 phox Arg 90 , which ligates the 4-position phosphate of target phosphoinositides. This suggests that the p47 phox PX domain can regulate ROS production via phosphoinositides lacking a 3Ј-phosphate, such as PI(4,5)P 2 , which is ϳ25-fold more abundant in the plasma membrane compared with PI(3,4)P 2 (57), or other targets. Interaction with PI(4,5)P 2 may also account for the observation that a fluorescently tagged probe derived from the p47 phox PX domain was reported to accumulate on the plasma membrane of resting neutrophils (58), which only have low levels of PI(3,4)P 2 and PI(3,4,5)P 3 .
To examine the impact of p47 phox Arg 90 mutants on neutrophil NADPH oxidase activity during phagocytosis, we used IgG-latex beads, serum-opsonized zymosan, and sterilized Aspergillus hyphae as model particles as well as complementopsonized S. aureus and heat-killed S. aureus. These particulate stimuli bind to receptors that activate phagocytosis and ROS production through overlapping pathways that include activation of Class I and III PI3Ks (51). Results from all partic-

p47 phox PX Domain in NADPH Oxidase Activation
ulate stimuli tested were similar. There was a modest but statistically significant decrease in the release of oxidants at the plasma membrane when p47 phox Arg 90 was replaced with either alanine or lysine, with the greatest effect seen with stimulation of IgG-opsonized particles. However, p47 phox Arg 90 substitutions had little effect on intracellular ROS production, indicating that the p47 phox PX domain regulates oxidase activity only before phagosome sealing. We were also unable to detect reduced translocation of p47 phox -R90A during ingestion of serumopsonized particles, including to the phagocytic cup formed prior to internalization.
The differential influence of p47 phox R90A on extracellular and intracellular NADPH oxidase activity and the more marked sensitivity of the former to soluble ligands is likely to reflect differences in the composition of the plasma and phagosome membranes. In activated neutrophils, the spatiotemporal dynamics of membrane phosphoinositides during phagocytosis reflect the sequential activity of Class I and III PI3Ks in concert with lipid phosphatases, such as SHIP-1 (Src homology 2-containing inositol phosphatase) and PTEN (phosphatase and tensin homolog) (59,60). Although the p47 phox PX domain has a variety of targets, the lipid with the highest affinity is PI(3,4)P 2 , largely derived from Class I PI3Kgenerated PI(3,4,5)P 3 . The absence of a fluorescently tagged p47 phox PX domain probe from phagosomes (58) and the lack of dependence of phagosome NADPH oxidase activity on p47 phox Arg 90 are consistent with prior studies showing that although PI(3,4,5)P 3 or PI(3,4)P 2 accumulates in the phagocytic cup during ingestion of SOZ-or IgG-opsonized particles by macrophage or in neutrophilic HL60 cells, these disappear after phagosome sealing (59,61).
The secondary binding pocket for phosphatidic acid or phosphatidylserine in the p47 phox PX domain synergizes with ligand binding to the primary pocket to increase overall affinity and penetration of the isolated p47 phox PX domain into the membrane (30,32). As a result, the otherwise poor binding affinity of p47 phox PX R90A to PI(3,4)P 2 is enhanced by the presence of phosphoserine or phosphatidic acid (30,32). Phosphatidylserine is present both on plasma membranes and phagosomes (62). In the current study, we did not specifically study mutations in this secondary pocket (amino acids 55 and 70) as an independent factor; however, the p47 phox PX R90A mutation alone was sufficient to profoundly impair neutrophil plasma membrane NADPH oxidase activity in response to PMA or fMLF. . CGD-associated p47 phox R42Q mutant expressed in p47 phox KO mouse PMNs. A, YFP fusion protein expression in p47 phox KO mouse PMNs was measured by flow cytometry. B, phox protein expression in p47 phox KO mouse PMN transduced with p47YFP WT or R42Q mutant. 15 g of protein was loaded in each lane, except 30 g of protein from R42Q mutant-transduced PMNs was loaded for probing p47 phox . C, phox protein expression in p47 phox KO mouse PMN transduced with p47 phox WT or R42Q mutant under puromycin selection. A vertical line was added when the sample was loaded in a non-continuous lane. D, distribution of p47YFP and p47YFP-R42Q in resting p47 phox KO mouse PMNs. N, nucleus. E, live image of p47 phox KO mouse PMNs expressing p47YFP-R42Q during SOZ phagocytosis. The arrow indicates the cup of phagosomes, and asterisks indicate the internalized phagosomes. Bar, 5 m. Intracellular ROS production induced by hIgG-latex beads (F), serumopsonized S. aureus (G), or hyphae (H), was measured in p47 phox KO mouse PMNs transduced with p47YFP or p47YFP-R42Q. The inset shows the intracellular phagosome ROS production measured in p47YFP-R42Q mutant induced by serum-opsonized S. aureus. ࡗ, p47YFP-transduced p47 phox KO neutrophils; छ, p47-YFP R42Q-transduced p47 phox KO neutrophils; *, p47 phox KO nutrophils (n ϭ 3). p47 phox PX Domain in NADPH Oxidase Activation NOVEMBER 5, 2010 • VOLUME 285 • NUMBER 45 p47 phox is also reported to bind to the cytoskeletal protein moesin via its PX domain as assayed in vitro (63), although the molecular details of this interaction are not further characterized. Because of insensitivity to p47 phox PX domain mutations affecting phosphoinositide binding, interaction with moesin was proposed to mediate p47 phox plasma membrane translocation after insulin-like growth factor-1 stimulation in COS cells (29), although there was no direct evidence for this conclusion. The PX domain of p47 phox also binds to cytosolic phospholipase A 2 ␣, which is required for maximal response to fMLF or PMA stimulation in neutrophils or PLB-985 granulocytes (64). However, the interaction of p47 phox and cytosolic phospholipase A 2 ␣ involves the face of the p47 phox PX domain opposite the phosphoinositide binding pocket (64).
A G125A mutation in NCF1, predicting a point substitution R42Q in p47 phox , was identified in two p47 phox CGD patients who lacked p47 phox protein and neutrophil NADPH oxidase activity (53,54). Phosphoinositides do not bind to the isolated p47 phox PX domain harboring a R42Q mutation, which has been taken as evidence for phosphoinositide-mediated regulation of NADPH oxidase activity by the p47 phox PX domain (23,29). However, we observed very low levels of full-length p47 phox R42Q and an abnormal subcellular distribution when expressed in either K562 cells or neutrophils. Despite this abnormal expression, small amounts of intracellular but not extracellular ROS release during phagocytosis were detected in murine neutrophils expressing p47YFP-R42Q, consistent with the concept that phosphoinositide binding to the p47 phox PX domain does not regulate intracellular NADPH oxidase activity. Taken together with prior studies of CGD patients with p47 phox -R42Q mutations (53,54), these results indicate that this NCF1 mutant behaves as a null allele due to deleterious effects on neutrophil expression of p47 phox rather than deficient oxidant production from loss of p47 phox function.
The differential effect of the p47 phox PX domain on NADPH oxidase activity on plasma membrane versus phagosomes is consistent with prior studies showing that the environment and regulation of the NADPH oxidase differ in these two compartments. For example, the tail-to-tail interaction of p47 phox / p67 phox becomes dissociated after phagosome sealing during ingestion of IgG-opsonized zymosan (40). In addition, whereas the binding of PI3P to the p40 phox PX domain is not required for NADPH oxidase activity on the plasma membrane, PI3P is a strong positive regulator of NADPH oxidase activity after phagosome closure and is also required to retain p40 phox on the phagosome (6,28).
In conclusion, we establish for the first time that the p47 phox PX domain plays a role in positively regulating neutrophil NADPH oxidase activity but that this role is differentially dependent on the membrane compartment and the agonist. Substitutions at residues critical for the ability of p47 phox to bind phosphoinositides substantially decrease plasma membrane enzyme activity in response to fMLF and PMA and, to a lesser extent, particulate stimuli. However, intracellular ROS production during phagocytosis was not affected by p47 phox PX mutations, in contrast to the important role for PI3P, binding to the p40 phox PX domain for oxidant production in phagosomes (6,25,26,28,51,65).