Phosphatidylinositol 3-Phosphate-dependent and -independent Functions of p40phox in Activation of the Neutrophil NADPH Oxidase*

In response to bacterial infection, the neutrophil NADPH oxidase assembles on phagolysosomes to catalyze the transfer of electrons from NADPH to oxygen, forming superoxide and downstream reactive oxygen species (ROS). The active oxidase is composed of a membrane-bound cytochrome together with three cytosolic phox proteins, p40phox, p47phox, and p67phox, and the small GTPase Rac2, and is regulated through a process involving protein kinase C, MAPK, and phosphatidylinositol 3-kinase. The role of p40phox remains less well defined than those of p47phox and p67phox. We investigated the biological role of p40phox in differentiated PLB-985 neutrophils, and we show that depletion of endogenous p40phox using lentiviral short hairpin RNA reduces ROS production and impairs bacterial killing under conditions where p67phox levels remain constant. Biochemical studies using a cytosol-reconstituted permeabilized human neutrophil cores system that recapitulates intracellular oxidase activation revealed that depletion of p40phox reduces both the maximal rate and total amount of ROS produced without altering the KM value of the oxidase for NADPH. Using a series of mutants, p47PX-p40phox chimeras, and deletion constructs, we found that the p40phox PX domain has phosphatidylinositol 3-phosphate (PtdIns(3)P)-dependent and -independent functions. Translocation of p67phox requires the PX domain but not 3-phosphoinositide binding. Activation of the oxidase by p40phox, however, requires both PtdIns(3)P binding and an Src homology 3 (SH3) domain competent to bind to poly-Pro ligands. Mutations that disrupt the closed auto-inhibited form of full-length p40phox can increase oxidase activity ∼2.5-fold above that of wild-type p40phox but maintain the requirement for PX and SH3 domain function. We present a model where p40phox translocates p67phox to the region of the cytochrome and subsequently switches the oxidase to an activated state dependent upon PtdIns(3)P and SH3 domain engagement.

Neutrophils are phagocytic polymorphonuclear white blood cells (PMNs) 3 of the innate immune system and represent one of the first lines of defense against invading microorganisms (1,2). The neutrophil NADPH oxidase enzyme catalyzes the transfer of electrons from NADPH to oxygen to form superoxide within pathogen-containing phagosomes and at the plasma membrane. The crucial role of NADPH oxidase activity in host defense is evidenced by patients with chronic granulomatous disease (CGD) whose neutrophils lack a functional NADPH oxidase resulting in frequent and persistent infections because of an inability to kill microbes efficiently (3)(4)(5).
The NADPH oxidase is a multisubunit enzyme made up of a membrane-spanning heterodimer of gp91 phox and p22 phox , which forms the catalytic core (cytochrome b 558 ), as well as four cytosolic components, p47 phox , p67 phox , p40 phox , and the small G-protein Rac2 (2,6). Upon stimulation, the cytosolic components translocate to and activate the cytochrome (7). p47 phox and p67 phox have relatively well established roles in activation as follows: p47 phox is essential for localization of the cytosolic phox components to the cytochrome (8); p67 phox has been implicated as the cytosolic NADPH-binding protein (9,10) and facilitates electron transport by the cytochrome (11). Mutations in either of these proteins results in CGD. In contrast, a definitive role for endogenous p40 phox in NADPH oxidase activation and bacterial killing within neutrophils has been difficult to establish, and no CGD mutations within p40 phox have yet been reported. Hawkins and co-workers (12) recently generated p40 phoxϪ/Ϫ mice and reported that murine bone marrowderived neutrophils from these animals were substantially defective in NADPH oxidase activity in response to some, but not all, stimuli examined, as well as in their ability to kill serumopsonized Staphylococcus aureus. However, the genetic knockout of p40 phox resulted in the concomitant ϳ60% decrease in the levels of p67 phox , complicating interpretation of their data. Additional insights into the function of p40 phox from in cellulo studies has primarily involved non-neutrophil cell types. For example, Suh et al. (13) examined p40 phox in a monkey kidney COS7 cell system stably transfected with cytochrome b 558 , p47 phox , and p67 phox transgenes along with the Fc␥R receptor (COS phox -Fc␥R cells), and they found that p40 phox was necessary for activation of the NADPH oxidase on Fc␥IIA receptorinduced phagosomes, suggesting that similar functions might occur in neutrophils or neutrophil-like cells. Similarly, Kuribayashi et al. (14) transfected p40 phox into K562 erythroleukemia cells that stably express cytochrome b 558 , p47 phox , and p67 phox and observed increased superoxide production in the p40 phoxtransfected cells compared with vector-transfected controls upon stimulation with PMA or a muscarinic receptor agonist peptide, whereas Sathyamoorthy et al. (15) reported that similar transfection studies of p40 phox into K562 cells resulted in an ϳ40% decrease in NADPH oxidase activity following PMA stimulation. Biochemical analysis of p40 phox function has been equally difficult because classical cell-free systems for analyzing the NADPH oxidase either do not require p40 phox for activity (16,17) or show only minor effects of p40 phox under typical assay conditions (18). Thus, many details of how p40 phox functions are still unclear. p40 phox contains an N-terminal PtdIns(3)P-binding PX domain (19,20), a central SH3 domain capable of interacting with a proline-rich region in p47 phox in vitro (21,22), and a C-terminal PB1 domain that is required for constitutive interaction with p67 phox (23). Recently, the structure of full-length p40 phox was solved (24). The structure, together with a companion cell biology-based study, showed that p40 phox exists in a "closed" state where the PX and PB1 domains interact with each other, preventing the PX domain from interacting with PtdIns(3)P-containing membranes (25). Mutating residues in the interface between the PX and PB1 domains resulted in an "open" conformation that allowed the protein to bind to PtdIns(3)P. However, the consequences of this conformational change in the activation of the oxidase and the generation of superoxide remain unknown.
In this study we have investigated the role of p40 phox in the activation of the NADPH oxidase utilizing two separate systems. First we use RNA interference in human PLB-985-derived neutrophils to show that p40 phox is important for maximum superoxide production and that p40 phox has a physiologically important role in the killing of pathogenic bacteria under conditions where the levels of p67 phox are unchanged. We then used a permeabilized neutrophil system developed previously in our laboratory to biochemically investigate the mechanism by which p40 phox acts on the NADPH oxidase (26). This system depletes purified human neutrophils of cytosol by permeabilization with the bacterial toxin streptolysin-O, generating cytosol-free neutrophil "cores" that can then be repleted with previously manipulated neutrophil cytosol. This system offers several advantages for studying the NADPH oxidase compared with traditional recombinant protein-based in vitro assays and with transfected non-neutrophil cell line-based systems, because the assays are performed using primary human neutrophils. Importantly, the permeabilized neutrophils maintain many of the intracellular structures on which the NADPH oxidase is thought to assemble and allows for the detection of intracellularly produced superoxide (26).
Additionally, because the system requires repletion with prepared cytosol, traditional biochemical techniques can be applied to perform structure-function studies in the setting of the primary neutrophil.
Using this reconstituted cores assay, we report that p40 phox positively regulates the NADPH oxidase through all three of its modular signaling domains in a manner that extends beyond its role in facilitating p67 phox localization. Finally, we show that the open form of p40 phox causes the oxidase to generate significantly more superoxide than the closed form, and we propose a multistep mechanism of action for how p40 phox activates the oxidase in both a PtdIns(3)P-independent and PtdIns(3)P-dependent manner.

EXPERIMENTAL PROCEDURES
Materials-Streptolysin-O, ATP, luminol, diisopropyl fluorophosphate, creatine kinase, creatine phosphate, and dimethyl formamide (DMF) were purchased from Sigma. All other protease inhibitors were purchased from American Bioanalytical. Endotoxin-free Dulbecco's modified PBS without calcium and magnesium (DMPBS) as well as Dulbecco's modified PBS containing calcium and magnesium (DMPBSϩ) were purchased from Invitrogen; protein A-Sepharose beads and Ficoll-Paque were from GE Healthcare; GTP, GTP␥S, NADPH, and Nutridoma-SP were from Roche Applied Science, and wortmannin was from EMD Biosciences. Polyclonal antibodies against p40 phox and p67 phox were generated by peptide immunization of rabbits and have been described previously (27). A polyclonal antibody against p22 phox was a gift from Dr. Katrin Rittinger. A mouse monoclonal anti-penta-His antibody was purchased from Qiagen.
Bacterial Killing Assays-Bacterial killing assays were performed following the procedure of Silverstein and co-workers (29). In brief, S. aureus Rosenbach (ATCC 25923) was cultured in tryptic soy broth to early log phase, collected by centrifugation, and resuspended at 2 ϫ 10 8 colony-forming units/ml in DMPBSϩ containing 1 mg/ml bovine serum albumin. Bacteria were incubated at a ratio of 5:1 with 1 ϫ 10 6 differentiated PLB-985 cells in 75% human serum in a total volume of 1 ml for 90 min at 37°C with intermittent mixing. Following incubation, duplicate aliquots (350 l) were lysed by mixing with 650 l of sterile water, pH 11, followed by incubation at 37°C with shaking for 5 min. Lysates were serially diluted, plated on tryptic soy agar plates, grown overnight at 37°C, and the number of surviving bacterial colonies counted.
PLB-985 Chemiluminescence Assays-For chemiluminescence assays following S. aureus infection, 2 ϫ 10 5 p40KD or LucKD differentiated PLB-985 cells were resuspended in DMPBS containing 100 g/ml bovine serum albumin and 0.15 mM luminol (PBS/B/L) and incubated in 75% human serum. S. aureus was added to the PLB-985 suspension at a ratio of 5:1, and ROS production was monitored as luminol-dependent chemiluminescence (CL), in duplicate, using an AutoLumat LB953 luminometer (Berthold Technologies, Oak Ridge, TN) at 37°C. Unless otherwise indicated, CL is reported as total events recorded during the indicated time after subtracting the machine background, which typically averaged less than 1-5% of the total signal. When PMA was used as the agonist, experiments were performed by resuspending 1.5 ϫ 10 5 differentiated PLB-985 cells in PBS/B/L, and the reaction was initiated by the addition of PMA at the indicated final concentrations.
Preparation of Permeabilized Neutrophils and PMN Cytosol-Anticoagulated whole blood was obtained from healthy human volunteers following an Institutional Review Board-approved protocol. PMNs were isolated by centrifugation through Ficoll-Paque followed by dextran sedimentation and water lysis of residual red blood cells as described earlier (26). PMNs were resuspended to a final concentration of 1 ϫ 10 7 /ml in DMPBS containing 4 g/ml 4-(2-aminoethyl)benzenesulfonyl fluoride, incubated on ice for 10 min, pelleted, and then resuspended in RB-EBL buffer (RB buffer (10 mM PIPES, pH 7.3, with KOH, 100 mM K ϩ , 3 mM Na ϩ , 3.5 mM Mg 2ϩ ) containing 0.3 mM EGTA, 100 g/ml bovine serum albumin, and 0.15 mM luminol).
Permeabilized PMN cores were prepared by incubating 1 ϫ 10 6 freshly isolated cells in 300 l of RB-EBL buffer containing 1000 units of reduced streptolysin-O for 5 min at 0°C. Cores were recovered by centrifugation at 280 ϫ g for 10 min, resuspended in RB-EBL, and used within 60 min. PMN cytosol was prepared by nitrogen cavitation of 2 ϫ 10 8 cells/ml in RB buffer containing 0.3 mM EGTA and 5.6 mM diisopropyl fluorophosphate by pressurization to 400 p.s.i. for 20 min at 0°C prior to release. The cavitate was centrifuged at 2500 ϫ g for 10 min at 4°C followed by centrifugation of the resulting supernatant at 200,000 ϫ g for 1 h at 4°C. The high speed supernatant was flash-frozen as aliquots in liquid N 2 and stored at Ϫ80°C. 2 ϫ 10 8 PMNs typically yielded ϳ3 mg of protein; cytosol was thawed on ice immediately prior to use. To immunodeplete cytosol of p40 phox , anti-p40 phox antibody beads were prepared by incubating 50 l of protein A-Sepharose beads with 100 g (50 l) of p40 phox antiserum overnight at 4°C. Beads were washed twice with RB buffer, four times with RB containing 1 M NaCl, twice more with RB buffer and added to 60 l of freshly thawed cytosol. Following a 2.5-h incubation at 4°C with gentle rocking, the p40 phox -depleted cytosol was recovered by pelleting the beads and recovering the supernatant. Mock-depleted cytosols were prepared in an identical manner using protein A-Sepharose beads coated with nonspecific rabbit IgG.
Expression and Purification of Recombinant p40 phox -For bacterial expression of human p40 phox , the full-length cDNA was cloned into the EcoRI and XhoI sites of pET28a to provide an N-terminal His 6 tag. Constructs expressing R58Q, W207R, D293K, and E259A as well as R58Q/W207R, E259A/R58Q, E259A/W207R, and E259A/D293K p40 phox mutants were constructed using the QuikChange site-directed mutagenesis kit (Stratagene). The ⌬PX domain construct lacking amino acids 1-135 and the p47PXp40 phox fusion containing amino acids 1-121 of p47 phox fused to amino acids 136 -339 of p40 phox were both constructed using PCR. The p47PX-R43A site mutant was constructed using the QuikChange site-directed mutagenesis kit using the p47PXp40 phox construct as a template. All constructs were verified by dideoxy sequencing and transformed into BL21(DE3) Escherichia coli cells. Recombinant proteins were produced by inducing 1 liter of late log phase cultures (A 1.0 -1.2) with 1 mM isopropyl 1-thio-␤-D-galactopyranoside overnight at room temperature. Bacteria were harvested by centrifugation, resuspended in 25 ml of lysis buffer (50 mM HEPES, pH 8.0, containing 1 mM MgCl 2 , 300 mM NaCl, 7 mM ␤-mercaptoethanol, and 4 g/ml each of leupeptin, pepstatin, 4-(2-aminoethyl)benzenesulfonyl fluoride, and aprotinin), and disrupted by sonication. Lysates were clarified at 150,000 ϫ g for 30 -60 min at 4°C, incubated in batch with 1 ml of Ni-NTA beads, and mixed end-over-end for 1 h at 4°C. The beads were then loaded into an fast protein liquid chromatography column, washed with 100 ml of lysis buffer containing 10 mM imidazole at a flow rate of 1 ml/min, and eluted with 50 ml of lysis buffer containing 300 mM imidazole. Fractions (8 ml) were analyzed for p40 phox content by SDS-PAGE, pooled, and snap-frozen at Ϫ80°C. Prior to use in core reconstitution reactions, the proteins were thawed and dialyzed against RB buffer.
Reconstitution Reactions and Chemiluminescence Assay-Reconstitution reactions were performed in 100 l of RB-EBL buffer containing 8 ϫ 10 4 freshly prepared neutrophil cores, 30 g of cytosol, 2 mM ATP, 200 M GTP, 200 M GTP␥S, 10 mM creatine phosphate, and 25 g/ml creatine kinase. Unless otherwise indicated, reactions were preincubated at 37°C for 10 min to induce permeabilization, followed by addition of 400 M NADPH and 100 ng/ml PMA, and assayed for ROS production by measuring luminol-dependent CL in duplicate using an AutoLumat LB953 luminometer at 37°C. Where appropriate, 400 ng of recombinant p40 phox protein was added prior to the preincubation step (unless otherwise noted); 400 ng gave the maximal recovery of ROS production in this assay with p40 phox -depleted cytosol (data not shown).
Cores Association Assay-To examine movement of p67 phox to the core membranes, 8 ϫ 10 5 cores in 1 ml of RB buffer were incubated with 200 g of p40 phox immunodepleted cytosol, in the presence or absence of 3 g of purified wild-type or mutant p40 phox , 60 ng/ml PMA, and 120 M NADPH for 10 min at 37°C. Cores were recovered by centrifugation (2,400 ϫ g for 2 min) and rinsed once with RB buffer. To minimize nonspecific protein adherence, cores were then incubated in RB buffer containing 5 g/ml of cytochalasin B at 37°C for 10 min and then rinsed twice with RB containing 0.1% Brij-35. Proteins were resolved by SDS-PAGE and immunoblotted for p67 phox .
Structural Analysis-Structures of the p40 phox and p47 phox PX domains and full-length p40 phox protein were visualized using SwissPDB viewer (30) and GRASP (31). Figures were constructed using PyMol.

p40 phox Is Required for Optimal Killing of S. aureus and ROS
Production in Differentiated PLB-985 Cells-To investigate the role of p40 phox in the activation of the NADPH oxidase, we used RNA interference to knock down p40 phox in PLB-985 cells, a human myeloid cell line that can undergo differentiation into neutrophils (32), and examined their ability to generate ROS. Undifferentiated PLB-985 cells were infected with lentiviruses expressing either an shRNA against the 3Ј-untranslated region of p40 phox (p40KD) or a control shRNA against luciferase (lucKD), together with a puromycin resistance gene. Following selection in puromycin-containing media, the cells were induced to undergo granulocyte differentiation. When immunoblotted 8 days following the induction of differentiation, the lucKD control cells showed robust expression of p40 phox (Fig.  1A), whereas in the p40KD cells, the levels of p40 phox were reduced by 63 Ϯ 2.4%. Importantly, the levels of p67 phox in both the differentiated lucKD cells and the p40KD cells were identical (97.8 Ϯ 4.0% in p40KD compared with control).
Differentiated lucKD cells and p40KD cells were activated using PMA, and ROS production was measured using luminoldependent chemiluminescence. As shown in Fig. 1B, the p40KD cells showed a statistically significant defect in PMAdependent ROS generation compared with lucKD cells at both 10 and 50 ng/ml. Next, we investigated the ability of p40KD PLB-985 cells to produce ROS when challenged with serumopsonized S. aureus. In these experiments we observed a 43% decrease in total ROS production in the p40 phox -deficient cells compared with the controls (Fig. 1C). We also observed in each of eight separate experiments that the p40KD cells were profoundly defective in their ability to kill opsonized microbes (Fig.  1D), indicating that endogenous p40 phox plays a physiologically important role in NADPH oxidase-dependent bacterial killing.
p40 phox Is Necessary for Optimal Superoxide Production by the NADPH Oxidase in Permeabilized PMNs-To further investigate the mechanism by which p40 phox activates the neutrophil NADPH oxidase, we utilized a permeabilized neutrophil system previously developed in our laboratory that recapitulates intracellular ROS production (26). In this system, the plasma membranes of human neutrophils are permeabilized using streptolysin-O, resulting in the production of cytosol-free neutrophil cores containing intact intracellular granules and vesicles ( Fig. 2A). Robust PMA-stimulated activation of the NADPH oxidase is obtained in these neutrophil cores when they are supplemented with purified cytosol, ATP, GTP␥S, and NADPH, as measured by luminol-dependent chemiluminescence (26). In contrast to traditional cell-free NADPH oxidase assay systems, the permeabilized core system preserves much of the complex intracellular structure of the neutrophil and permits intracellular oxidase activation on granules and vesicles, where much of the initial oxidase activation is believed to occur in intact neutrophils (33,34). Importantly, the cytosol used in the reconstitution step can be manipulated prior to re-addition, allowing removal and/or replacement of specific cytoplasmic constituents.
To specifically address the biochemical role of p40 phox in NADPH oxidase-dependent ROS production, we reconstituted PMN cores with neutrophil cytosol that was immunodepleted of p40 phox or mock-depleted with nonspecific rabbit IgG (Fig.  2B). Depletion of all of the p40 phox detectable by immunoblotting dramatically reduced the activity of the NADPH oxidase, resulting in a 72 Ϯ 0.8% decrease in the peak rate of ROS production, a kinetic delay of 7.6 Ϯ 3.6 min in the time until the peak rate was reached (Fig. 2C), and a 90 Ϯ 7.1% reduction in the total amount of ROS produced (total integrated chemiluminescence over 60 min; Fig. 2D). Immunodepletion of p40 phox also resulted in co-depletion of 43 Ϯ 13% of the p67 phox present in the cytosol, with no change in p47 phox levels (Fig. 2B). Thus, the decrease in ROS production that we observed in the p40 phox -depleted reactions could have been due either to the loss of p40 phox itself or to the partial depletion of p67 phox .
To distinguish between these possibilities, we supplemented the p40 phox -depleted reactions with purified bacterially produced recombinant p40 phox (r-p40 phox ). As shown in Fig. 2C, the observed decrease in ROS production because of p40 phox  depletion, together with the kinetic delay, was largely reversed by re-addition of r-p40 phox , suggesting that p40 phox functions as a direct positive regulator of NADPH oxidase activity in this system. In contrast to the ϳ600% increase in ROS production when r-p40 phox was added to p40 phox -depleted cytosols, we observed only minimal increases in ROS production (Ͻ20%) when r-p40 phox was added to the mock-depleted cytosols. Furthermore, addition of recombinant p67 phox to levels 20-fold higher than those in the p40 phox -depleted lysates failed to confer any increased ROS production in the absence of r-p40 phox (data not shown). p40 phox Enhances the Initial Enzymatic Activity of the NADPH Oxidase without Affecting the Apparent K M Value for NADPH-In intact neutrophils and reconstitution systems, superoxide production involves two interconnected events that overlap in time, a progressive assembly of the NADPH oxidase holoenzyme complex (cf. Refs. [35][36][37], together with the catal-ysis of NADPH oxidation by those complexes that have already assembled. The coupling of these two events greatly complicates a kinetic analysis of the oxidase reaction. For example, in all of our permeabilized PMN reconstitution experiments, we observed a slow steady rise in the rate of superoxide production that peaked at ϳ15 min after PMA stimulation (Fig. 2C), similar to that observed with intact neutrophils (27,38,39). In an effort to decouple these two events and measure the initial enzymatic activity of the oxidase, we made use of our previous observation that preincubation of the cores with cytosol, ATP, GTP␥S, and PMA prior to addition of NADPH results in assembly of the NADPH oxidase complex, without ROS production (27). As shown in Fig. 3A, panels i-iv, addition of NADPH to reconstituted core reactions following 0, 10, 20, or 30 min of incubation progressively accelerated the onset of peak reaction velocity and dramatically increased the peak initial rate of ROS production. No further decrease in time to peak onset or increase in initial velocity was observed beyond 30 min of preincubation (data not shown). We therefore interpret this initial spike of FIGURE 2. p40 phox is required for optimal ROS production in a permeabilized neutrophil reconstitution system. A, schematic of PMN permeabilization and NADPH oxidase reconstitution procedure. B, purified neutrophil cytosol was immunodepleted using nonspecific rabbit IgG (Mock) or anti-p40 phox antibodies and analyzed for remaining levels of p40 phox , p47 phox , and p67 phox by immunoblotting. Each lane contains 20 g of total cytosolic protein. C and D, streptolysin-O-permeabilized PMN cores (8 ϫ 10 4 /100 l) were reconstituted with mock depleted (F), p40 phox -depleted (f), or p40 phox -depleted cytosol re-supplemented with recombinant p40 phox (ࡗ), stimulated with PMA, and ROS production measured using luminol-dependent chemiluminescence. C, typical trace from n ϭ 6 experiments, where CL/min denotes the recorded number of chemiluminescence events per min. D, mean values Ϯ S.D. of total integrated chemiluminescence over 60 min, normalized to the mock-depleted controls (n ϭ 6). WT, wild type.  oxidase activity when NADPH is added at 30 min to reflect the initial velocity of the pre-assembled oxidase complex. Interestingly, the total integrated chemiluminescence measured over 90 min remained the same regardless of the duration of preincubation (Fig. 3B), indicating that total ROS production was unchanged.
This preincubation assay, with addition of NADPH at 30 min, was then used to investigate the contribution of p40 phox to the kinetics of ROS production by comparison of reactions with p40 phox -depleted cytosol, with or without r-p40 phox added prior to NADPH addition. As shown in Fig. 3C, the initial reaction velocity was much lower in the absence than in the presence of r-p40 phox ; however, the absence of p40 phox did not significantly affect the time to peak ROS production following preincubation.
p67 phox , the constitutive binding partner of p40 phox , has been implicated in enzymatic regulation of NADPH oxidase activity through direct participation in electron flow from NADPH (2,6,10,40). As such, we reasoned that the decrease in the initial rate of ROS production observed in p40 phox -depleted reactions might result from interference with this process, manifesting from an alteration in the apparent K M value of the enzyme for NADPH. Preincubation assays with NADPH added at 30 min were therefore performed both in the presence and in the absence of r-p40 phox over a broad range of NADPH concentrations (Fig. 3D). Analysis of the initial rates of reaction revealed little difference in the apparent K M value of the reaction for NADPH in the absence or presence of r-p40 phox (15 Ϯ 6.7 versus 24 Ϯ 7.7 M, respectively) despite the Ͼ500% increase in the maximal velocity that was observed when p40 phox was present.
Our measured values compare favorably with the K M value of 34 Ϯ 3.9 M for NADPH in a traditional cell-free NADPH oxidase system following stimulation by the amphiphile SDS reported by Curnutte et al. (41). These findings suggest that p40 phox enhances NADPH oxidase activity through a route other than a direct effect on the apparent K M value for NADPH.
The PX, SH3, and PB1 Domains of p40 phox Are All Required for p40 phox Function-The p40 phox protein has three modular domains: a PX domain, an SH3 domain, and a PB1 domain. To investigate what roles these domains play in the p40 phox -dependent stimulation of the NADPH oxidase, we made single inactivating point mutations in each domain (Fig. 4A). We and others have shown previously that the p40 phox PX domain is a phosphoinositide binding domain that specifically recognizes PtdIns(3)P (19,20). We therefore made an R58Q mutation in the PX domain that is sufficient to disrupt PtdIns(3)P binding (42). Although the physiological target(s) of the SH3 domain of p40 phox remain unclear, this domain is known to recognize Prorich peptides in vitro (21). We therefore mutated Trp-207 to Arg (W207R), because this residue makes critical contacts with a synthetic Pro-peptide ligand in the p40 phox SH3 crystal structure (21), and the equivalent Trp mutations in other SH3 domains are known to disrupt their interactions with cognate ligands (43,44). We also mutated Asp-293 in the p40 phox PB1 domain to Lys (D293K), because this mutation is known to disrupt the constitutive interaction between p40 phox and p67 phox (45). Finally, we made a double mutant with both the R58Q mutation and the W207R mutation (R/W).
Recombinant wild-type and mutant p40 phox proteins were purified from bacteria (Fig. 4B), added to p40 phox -depleted cytosol, and ROS production then measured in reconstituted neutrophil cores. As shown in Fig. 4C, mutations within any of the three domains of p40 phox , as well as the double R/W mutant, dramatically reduced the ability of r-p40 phox to stimulate NADPH oxidase activity in response to PMA. The least competent mutant was the D293K PB1 domain mutant, which produced amounts of total ROS approximately equivalent to cytosol lacking p40 phox altogether. This finding is consistent with an obligatory requirement for interaction between p40 phox and p67 phox for p40 phox -dependent stimulation of the NADPH oxidase.
One postulated function of p40 phox is to facilitate the translocation of p67 phox to intracellular membranes containing cytochrome b 558 (14). If the point mutants described above reduced the amount of p67 phox that was associated with the cytochrome, this could explain why these mutants are unable to rescue ROS  production in p40 phox -depleted cores. To examine this possibility, we monitored the association of p67 phox with cores following PMA stimulation by re-isolating the cores by centrifugation and washing at the end of the reconstitution experiments (Fig. 4D). Control reactions using mock-depleted cytosol revealed that 5-10% of the total p67 phox present in the reactions translocated to the cores after PMA stimulation, consistent with what we had observed previously with unperturbed cytosol (26). These values compare favorably with the amount of p67 phox (10 -18%) reported by Quinn et al. (46) to become membrane-associated in PMA-stimulated neutrophils. Reactions with p40 phox -depleted cytosol demonstrated very little, or no, core-associated p67 phox after stimulation with PMA, whereas re-addition of wild-type r-p40 phox to p40 phox -depleted cytosol resulted in substantial translocation of p67 phox to the cores (compare Fig. 4D, lanes 2 and 3). When cores reconstituted with p40 phox -depleted cytosol were supplemented with r-p40 phox D293K, little if any p67 phox was core-associated (Fig.  4D, lane 6), demonstrating the requirement of a direct PB1/PB1 interaction with p67 phox to allow this translocation to occur. Interestingly, the R58Q PX domain mutant, the W207R SH3 domain mutant, and the R/W double mutant, none of which were able to rescue significant ROS production in p40 phox -depleted core reactions, caused core association of p67 phox to levels at least equivalent to those observed with wild-type r-p40 phox (compare Fig. 4D, lanes 4, 5, and 7 to lane 3). The observation that p40 phox -dependent translocation of p67 phox did not strictly parallel reconstitution of ROS production demonstrates that, whereas p40 phox does indeed translocate and localize p67 phox , this is not sufficient for stimulation. Additional events must occur (through at least the PX and SH3 domains) to promote NADPH oxidase activation. Addition of p40 phox to Ongoing Oxidase Reactions Results in the Instantaneous Acceleration of ROS Production-To further examine the parameters of the ability of p40 phox to stimulate NADPH oxidase activity, we investigated the effect of delayed addition of r-p40 phox to ongoing oxidase reactions. In experiments analogous to those described in Fig. 3, reactions containing permeabilized PMN cores, p40 phox -depleted cytosol, ATP, GTP␥S, and PMA, were incubated with NADPH at t ϭ 0 min and then supplemented with recombinant p40 phox protein 0, 10, 20, or 30 min later (Fig. 5). The addition of wild-type r-p40 phox at 10, 20, or 30 min resulted in the instantaneous enhancement of the rate of oxidase activity at all subsequent time points (Fig. 5A, panels i-v). In contrast, little to no enhancement in rate was observed upon addition of the R58Q or W207R r-p40 phox mutants (Fig. 5B, panels i-v and data not shown), despite the ability of these mutants to localize p67 phox to the cores (Fig. 4D). These data demonstrate that even when NADPH oxidase activity is decreasing, re-addition of p40 phox induces a rapid and significant increase in ROS production, through a mechanism not solely reliant on simple co-localization of p67 phox , requiring input from both the PX and SH3 domains.
The PX Domain of p40 phox Plays Both a PtdIns(3)P-dependent and a PtdIns(3)P-independent Role in the Activation of the NADPH Oxidase-To better understand the role of the PX domain of p40 phox in p40 phox -dependent ROS production, we made a variety of additional PX domain mutants. We were particularly interested in whether the p47 phox PX domain could functionally substitute for the p40 phox PX domain. The p47 phox PX domain has an almost identical structure to the p40 phox PX domain (42), but each has different lipid-binding specificities.
The p47 phox PX domain shows most specific binding to PtdIns(3,4)P 2 , whereas the p40 phox PX domain recognizes only PtdIns(3)P (19,20). We made a chimeric protein that swapped the PX domain of p40 phox for the PX domain of p47 phox (p47PX-p40 phox ). We also made this same p47PX-p40 phox fusion protein with an R43A point mutation in the lipid-binding pocket of the p47 phox PX domain, which eliminates PtdIns(3,4)P 2 binding, and is the functionally equivalent mutation to R58Q in the p40 phox PX domain. Finally, we also gener- v. iv. iii.
v. ated a PX domain truncation of p40 phox (⌬PX) (Fig. 6, A and B), which retains its ability to bind to p67 phox (supplemental Fig. 1). Consistent with the inability of the R58Q point mutant to rescue oxidase activity of p40 phox -depleted cores, the r-p40 phox ⌬PX truncation protein was also unable to rescue activity (Fig.  6C). However, in contrast to the R58Q point mutant, which was able to localize equivalent amounts of p67 phox to cores membranes as wild-type p40 phox (Fig. 4C), the ⌬PX truncation was unable to localize p67 phox to the cores (Fig. 6D). This demonstrates that co-translocation of p67 phox is PX-domain-dependent, but PtdIns(3)P-independent, whereas reconstitution of oxidase activity requires both the PX domain and PtdIns(3)P binding.
The p47PX-p40 phox chimeric mutant was able to activate the oxidase in p40 phox -depleted cores at comparable levels to wildtype p40 phox (Fig. 6C), suggesting that in the context of the fusion protein, the PX domain of p47 phox can functionally sub-stitute for that of p40 phox . However, the R43A p47PX-p40 phox chimeric fusion protein also performed equivalently to the wild-type p47PX-p40 phox fusion protein, demonstrating that, unlike the p40 phox PX domain, the p47 phox PX domain does not require contact with 3-phosphoinositides to confer activity on the system in the specific context of the p40 phox fusion protein. This idea was further reinforced by the observation that although the NADPH oxidase activity of cores reconstituted with wild-type r-p40 phox was highly sensitive to the PI3K inhibitor wortmannin, the fusion proteins were substantially less sensitive (Fig. 6C), with the R43A mutant being no more sensitive than the wild-type p47PX-p40 phox chimera. This again suggests that the p47PX-p40 phox fusion protein is activating the oxidase in a manner that is not dependent on the interaction of the fusion protein with PI3K-generated phosphoinositides. Upon examination of the ability of these p47PX-p40 phox chimeric fusion proteins to translocate p67 phox , we found that both were able to localize p67 phox in amounts equivalent to that achieved by wild-type p40 phox (Fig. 6D).
Together, these data suggest that the PX domain of p40 phox makes two distinct contributions to NADPH oxidase activity. The first is to translocate p67 phox in a PX domain-dependent but PtdIns(3)P-independent manner, and this can be functionally substituted by the PX domain of p47 phox . The second is a role in switching the oxidase to an activated state which for wild-type p40 phox requires PtdIns(3)P binding.
"Opening" of p40 phox Causes Enhanced p40 phox -dependent ROS Production by the NADPH Oxidase-Recently the crystal structure of full-length p40 phox was solved (24). Full-length p40 phox was found to exist in a closed state in which the N-terminal PX domain and the C-terminal PB1 domain interact with each other, allowing p67 phox binding but preventing interaction of the PX domain with PtdIns(3)P (Fig. 7A) (24,25). Mutating residues at the interface between the PX and PB1 domains was found to open full-length p40 phox and permit the PX domain to bind to PtdIns(3)P (24). Masking of the lipid-binding pocket of the PX domain in the resting state clearly has implications for this study in which we observed a PtdIns(3)P-independent role for the PX domain in translocation but a PtdIns(3)P-dependent role on oxidase activation.
To determine the effect of the open or closed state of p40 phox on the activation of the oxidase, we mutated Glu-259 of p40 phox to Ala. Glu-259 is one of the key residues responsible for maintaining p40 phox in a closed state (Fig. 7B). Mutation of Glu-259 to Ala was reported to convert p40 phox to an open form and permit the PX domain to bind to PtdIns(3)P (24). When we added the r-p40 phox E259A mutant to p40 phox -depleted cores, we observed a dramatic increase in the amount of ROS produced by the cores relative to reactions reconstituted with wildtype r-p40 phox (Fig. 7C). Phe-320 of p40 phox was found to have a similarly important role in keeping full-length p40 phox closed (24). As observed with the E259A mutant, mutation of Phe-320 to Ala in r-p40 phox also resulted in increased superoxide production when added to p40 phox -depleted cores, compared with reactions containing wild-type r-p40 phox (data not shown). Importantly, however, when we made mutations in the PX, SH3, or PB1 domains (R58Q, W207R, or D293K, respectively) in the background of an E259A mutant, we found that none of these
minus wortmannin plus wortmannin r-p40phox added r-p40phox added p40phox p22phox FIGURE 6. A chimeric p47 phox PX domain-p40 phox fusion protein activates the NADPH oxidase independently of binding to PI3K lipids. A, schematic of p40 phox chimeras and mutants studied. B, recombinant wild-type and mutant p40 phox proteins containing an N-terminal His 6 tag were produced in bacteria and batch-purified on Ni-NTA-agarose. Proteins were analyzed by SDS-PAGE and stained with Coomassie Blue. Positions of molecular weight markers are indicated. C, reconstitution reactions were performed using permeabilized PMN cores and p40 phox -depleted cytosol supplemented with the indicated recombinant p40 phox proteins in the presence (white bars) or absence (gray bars) of 100 nM wortmannin. Total integrated chemiluminescence was measured over 30 min following PMA stimulation and normalized to reactions supplemented with wild-type (wt) r-p40 phox . Means Ϯ S.D. from n ϭ 3 experiments are shown. D, differential effects of p47 phox PX domain-p40 phox chimeric proteins and p40 phox PX domain truncation on association of p67 phox with neutrophil cores. Reconstitution reactions were performed as in C in the presence of PMA for 10 min. Cores were then re-isolated and analyzed for p67 phox by SDS-PAGE and immunoblotting. An immunoblot for p22 phox is shown as a loading control. Each lane corresponds to 9 ϫ 10 5 cores.
double mutants were hyperactive (Fig. 7C), suggesting that opening p40 phox does not overcome the requirement for the PX, SH3, and PB1 domain functions.
To investigate whether the increased NADPH oxidase activity caused by the E259A mutant was because of the ability of this mutant protein to recruit more p67 phox to the core membranes than wild-type p40 phox , immunoblotting of the reconstituted PMA-stimulated cores for translocated p67 phox was performed (Fig. 7D). We observed that the E259A mutant p40 phox protein translocated equal but not larger amounts of p67 phox to the cores as the wild-type r-p40 phox protein. Furthermore, both the E259A/R58Q (PX domain) and E259A/W207R (SH3 domain) double mutants also delivered p67 phox to the cores membranes to an extent comparable with wild-type r-p40 phox , although consistent with results obtained with the p40 phox D293K (PB1 domain) single mutant (Fig. 4C), the E259A/D293K double mutant translocated very little p67 phox (Fig. 7D). These data demonstrate that, whereas opening p40 phox allows greater stimulation of the oxidase, the PX, SH3, and PB1 domains are all still required and further establish that p40 phox activates the oxidase in a manner that is not simply dependent on its ability to translocate p67 phox .

DISCUSSION
We used a combination of in cellulo RNA interference experiments and ex vivo biochemical studies to probe the function and mechanism of p40 phox function in assembly and activation of the NADPH oxidase. In cellulo studies were performed in PLB-985 cells, which can differentiate into mature neutrophils (32) that express all of the endogenous phox proteins, including p40 phox , and display robust NADPH oxidase activity. Depletion of endogenous p40 phox in these cells using lentiviral RNA interference delivery revealed a pronounced decrease in PMA-dependent ROS production and a marked defect in bacterial killing. These results are similar to those observed in bone marrow-derived neutrophils from p40 phoxϪ/Ϫ mice (12). However, in contrast to p40 phoxϪ/Ϫ neutrophils, which demonstrated an ϳ60% decrease in p67 phox levels, the levels of p67 phox in the PLB-985-derived knockdown neutrophils were unchanged, demonstrating a direct effect of endogenous p40 phox on oxidase activity in a functionally relevant cell type.
An in depth biochemical analysis of p40 phox function has been difficult until now because classical cell-free recombinant systems used for studying NADPH oxidase activity either do not show any enhancement of activity by p40 phox (16,17) or demonstrate only a slight enhancement of activity (10 -25%) if the system is modified to use crude neutrophil membranes as the source of the cytochrome (18 Arg58 Arg60 p22phox FIGURE 7. Relief of structural autoinhibition of p40 phox enhances NADPH oxidase activity but still requires PX and SH3 domain function. A, ribbon diagram of full-length p40 phox (Protein Data Bank code 2DYB). B, close up of the PX-PB1 domain interface showing some of the critical interacting residues that maintain the closed auto-inhibited conformation of p40 phox . C, reconstitution reactions were performed using permeabilized PMN cores and p40 phox -depleted cytosol supplemented with the indicated recombinant p40 phox proteins. Total integrated chemiluminescence was measured over 30 min and normalized to reactions supplemented with wild-type p40 phox . Means Ϯ S.D. from n ϭ 3 experiments are shown. D, differential effects of single and compound mutants that disrupt p40 phox autoinhibition on association of p67 phox with neutrophil cores. Reconstitution reactions were performed as in C in the presence of PMA for 10 min. Cores were then re-isolated and analyzed for p67 phox by immunoblotting. Each lane corresponds to 9 ϫ 10 5 cores. An immunoblot for p22 phox is shown as a loading control. Results are representative of three separate experiments. wt, wild type.
we took advantage of a streptolysin-O-permeabilized neutrophil assay system, as described previously by our laboratory, that maintains the ultrastructural features of neutrophils such as granules, compartments, etc. and allows intracellular oxidase activation in a PI3K-dependent manner, mirroring the behavior of intact neutrophils (26). This permeabilized neutrophil cores system permits biochemical manipulation and analysis, and it allowed a series of structure-based mutations in p40 phox to be studied in detail. Immunodepletion of p40 phox from this system resulted in a profound decrease in PMA-stimulated intracellular ROS production that could be rescued by complementation with recombinant wild-type p40 phox but not by p40 phox containing mutations in either the PX, PB1, or SH3 domains, illustrating the p40 phox dependence of this system and indicating that all three domains are required for its biochemical activity. p40 phox is constitutively bound to p67 phox through PB1-PB1 domain-mediated interactions, and p67 phox has been shown to both bind to NADPH (40) and to facilitate electron flow from NADPH to FAD in the cytochrome (6,10). As such, we investigated whether p40 phox might function as a molecular "chaperone" for p67 phox to facilitate interactions of NADPH with the oxidase. However, kinetic analysis of the reconstituted neutrophil cores revealed a similar K M value for NADPH regardless of the presence or absence of p40 phox , indicating that p40 phox functions to stimulate the oxidase through its PX, PB1, and SH3 domains in some other manner. Even after the system has been assembled and activated in the absence of p40 phox , the addition of p40 phox still increases oxidase activity. We suggest this effect could be through either allosteric regulation of the pre-assembled holoenzyme complex or by increased recruitment of p67 phox to the region of the cytochrome, perhaps by enhancing the affinity of p67 phox for p47 phox during oxidase reconstitution as proposed by Cross (18). Indeed, we observed that p40 phoxdependent translocation of p67 phox was prevented by disruption of the PB1/PB1 domain interaction (D293K p40 phox ). However, contrary to the hypothesis that p40 phox -dependent recruitment of p67 phox would require the PX domain to interact with PtdIns(3)P-containing membranes, the R58Q p40 phox mutant was fully competent to translocate p67 phox compared with wild-type, as was an SH3 domain mutant (W207R p40 phox ). These data show that p40 phox -dependent translocation of p67 phox is necessary but not sufficient for oxidase activation in this system.
Very recently, the x-ray crystal structure of full-length p40 phox was solved, revealing a closed conformation via an intramolecular association of the PB1 domain with the PX domain that permits p40 phox -p67 phox binding but prohibits the PX domain from binding to PtdIns(3)P-containing membranes. Because mutations that abrogate PtdIns(3)P binding by the PX domain have significant impact on p40 phox function, yet the full-length protein is in a closed conformation such that the PX domain cannot contact PtdIns(3)P, then at some point during physiological activation of the oxidase, p40 phox must transition to an open state. This suggests that p40 phox may itself be subject to allosteric regulation, perhaps via one or more interdomain interactions.
To investigate this, we first individually mutated key residues responsible for the intramolecular PX-PB1 domain interaction that have been shown to relieve the masking of the PX domain, E259A and F320A (24,25). These mutants displayed ϳ2.5-fold more activity than wild-type p40 phox in our system, without increasing the amount of p67 phox translocation, suggesting that opening of p40 phox may be a limiting event in its mode of action. Furthermore, this result indicates that the stimulatory effect of p40 phox on NADPH oxidase activity is functionally auto-inhibited in the resting state and relieved upon disruption of the PX/PB1 interaction.
One function of the p40 phox SH3 domain and/or the PX domain might be to assist in conversion of p40 phox from its closed to its open conformation during oxidase activation (i.e. breathing movements in p40 phox could permit transient association of the PX domain with PtdIns(3)P to prevent re-closure and stabilize the open form). Individual point mutations in the three modular domains (R58Q, W207R, and D293K) were therefore generated in the background of the constitutively open E259A mutant. All three double mutants reversed the ability of the open form to activate the oxidase. These data demonstrate that although p40 phox is opened during activation, none of the three domains are solely responsible for this event, such that constitutively opening the protein can circumvent the requirement for any of them. Instead both the PX domain and the SH3 domain directly participate in triggering and/or maintaining oxidase activity.
A mutant lacking the PX domain altogether (⌬PXp40 phox ) was incapable of rescuing the p40 phox -dependent stimulation of the system, indicating that the presence of a PX domain is essential for the stimulatory function of p40 phox . Surprisingly, however, we found that swapping the p40 phox PX domain with the PX domain of p47 phox resulted in production of equivalent amounts of ROS by the reconstituted oxidase, despite having very different phosphoinositide binding specificities (19). An R43A version of this p47PXp40 phox fusion protein, rendering the PX domain of p47 phox incapable of binding 3-phosphoinositides, displayed the same level of stimulatory capacity as the phosphoinositide-binding-competent version. Furthermore, compared with wild-type p40 phox , the p47PX fusion proteins were much less sensitive to inhibition by wortmannin, underscoring the differential reliance of the two PX domains on interaction with 3-phosphoinositides, suggested by the R43A mutant chimera, for their positive influence on oxidase activity. We interpret the residual effect of wortmannin on inhibition of the p47PX chimeras to other PI3K-dependent processes necessary for oxidase activation, independent of signaling through p40 phox . These data indicate a bifurcation in function of the two PX domains in the context of the rest of the p40 phox protein; the wild-type PX domain of p40 phox requires phosphoinositide binding to confer activity on the system, whereas the PX domain of p47 phox does not. The p47 phox PX domain has also been reported to bind to phosphatidic acid in vitro through a separate lipid binding pocket (47). We therefore cannot exclude the possibility that phosphatidic acid binding to the p47 phox PX domain in the p47PXp40 phox fusion might be involved in the translocation and oxidase activation process.
Superimposing the structures of the two PX domains shows that the critical residues responsible for interaction of the p40 PX domain with the PB1 domain are not conserved in the PX domain of p47 phox , suggesting that the p47PXp40 phox constructs are in an open conformation. However, the p47PXp40 phox construct displays equivalent activity as wildtype p40 phox , whereas the p40 phox E259A and F320A open mutants are 2.5-fold more active. This observation again suggests that PtdIns(3)P-binding to the p40 phox PX domain has some type of specific and direct stimulatory effect on its ability to activate the enzyme.
When the ability of the p40 phox constructs to translocate p67 phox was examined, we found a stark lack of positive correlation between the ability of specific p40 phox mutants to support oxidase activation and their ability to translocate p67 phox . As expected, mutants unable to interact with p67 phox did not translocate p67 phox to the cores, but of the battery of other mutants (many of which had little or no activity), only ⌬PXp40 phox failed to localize p67 phox . This disparity between the ability to activate the oxidase and the ability to localize p67 phox suggests the following: (a) p40 phox -dependent translocation of p67 phox is required but is not sufficient to activate the oxidase; (b) this localization is PX domain-dependent but is independent of phosphoinositide binding of the PX domain; and (c) localization is independent of ligand binding by the SH3 domain. We do not know if the observed translocation of p67 phox to the cores reflects its interaction with the membranebound cytochrome b 558 or with the membrane-associated cytoskeleton. p40 phox has been reported to bind to both coronin and moesin (48,49), two actin-binding proteins associated with the cytoskeleton. Intriguingly, the interaction of p40 phox with moesin appears to involve the PX domain (49). Thus, it is tempting to speculate that this PX domain/moesin interaction is responsible for the p67 phox translocation events that we observe, although additional experiments will be necessary to definitively support or refute this hypothesis.
Taken together, our data would be consistent with the following model for the role of p40 phox in the activation of the neutrophil NADPH oxidase (Fig. 8); in resting neutrophils, p40 phox exists in an autoinhibited conformation. Upon neutrophil activation, p40 phox is involved in the localization of p67 phox to the cytochrome in a manner that requires the PX domain but not its lipid-binding capacity. This localization is not sufficient to activate the oxidase, because mutants in the PX and SH3 domains are unable to activate the oxidase although they are competent to localize p67 phox . Binding of PtdIns(3)P by the PX domain, and engagement of the SH3 domain, then functions as a molecular "switch" that converts the holoenzyme from its inactive to its active form. Although molecular details of this activation process remain obscure, it might involve stimulating the ability of p67 phox to act in a catalytic manner with multiple cytochrome molecules, as proposed by Cross et al. (16). This model provides a succinct mechanism for how the PX domain together with PtdIns(3)P could be involved in continual assembly of the oxidase holoenzyme during phagocytosis followed by an acute burst of activation following phagosomal closure, because production of PtdIns(3)P appears to occur only following phagosomal closure via activation of a specific PI3K isoform, Vps34 (50). Definitive proof of this model, however, will likely require detailed x-ray structural studies of the holoenzyme complex in the presence and absence of PI3Kderived lipid products.