Assembly of the human neutrophil NADPH oxidase involves binding of p67phox and flavocytochrome b to a common functional domain in p47phox.

The human neutrophil NADPH oxidase is a multi-component complex composed of membrane-bound and cytosolic proteins. During activation, cytosolic proteins p47phox, p67phox, Rac2, and possibly p40phox translocate to the plasma membrane and associate with flavocytochrome b to form the active superoxide-generating system. To further investigate the role of p67phox in this complex assembly process, experiments were performed to identify possible regions of interaction between p67phox and other NADPH oxidase proteins. Using random sequence peptide phage-display library analysis of p67phox, we identified a novel region in p47phox encompassing residues 323-332 and a previously identified SH3 binding domain encompassing p47phox residues 361-370 as p67phox binding sites. Synthetic peptides mimicking p47phox residues 323-332 inhibited the p47phox-p67phox binding interaction in an affinity binding assay; however, peptides mimicking flanking regions were inactive. Surprisingly, this same region of p47phox was found previously to represent a site of binding interaction for flavocytochrome b (DeLeo, F. R., Nauseef, W. M., Jesaitis, A. J., Burritt, J. B., Clark, R. A., and Quinn, M. T. (1995) J. Biol. Chem. 270, 26246-26251), and this observation was confirmed in the present report using two different in vitro assays that were not evaluated previously. Using affinity binding assays, we also found that p67phox and flavocytochrome b competed for binding to p47phox after activation, suggesting that prior to full NADPH oxidase assembly the 323-332 region of p47phox is associated with p67phox and at some point in the activation process is transferred to flavocytochrome b. Thus, taken together our data demonstrate that both p67phox and flavocytochrome b utilize a common binding site in p47phox, presumably at distinct stages during the activation process, and this p47phox region plays a key role in regulating NADPH oxidase assembly.

Human neutrophils play an essential role in the body's defense against invasion of pathogenic organisms and are also one of the primary cell types involved in the inflammatory response. Upon exposure to a pathogenic stimulus, the neutro-phil becomes activated and generates superoxide anion (O 2 . ) and other more potent antimicrobial agents (1,2). The generation of these oxidants by neutrophils occurs through the activation of a membrane-associated complex known as the NADPH oxidase. In resting cells, the components of the NADPH oxidase are segregated into cytosolic and membrane compartments; during activation, however, the cytosolic components translocate to the plasma membrane and assemble with membrane-bound components, resulting in the active, O 2 .
The key membrane-associated component of the NADPH oxidase is a heterodimeric flavocytochrome b which is composed of a 91-kDa glycoprotein (gp91 phox ) and a 22-kDa protein (p22 phox ) (6 -8). Flavocytochrome b has been reported to contain all of the redox components of the NADPH oxidase (7)(8)(9) and is the key catalytic component responsible for the direct transfer of electrons from NADPH to molecular oxygen. In neutrophil membranes, a low molecular weight GTP-binding protein, Rap1A, is also associated with flavocytochrome b and plays a role in NADPH oxidase regulation in vivo (10 -12). The essential cytosolic components of the NADPH oxidase are known as p47 phox , p67 phox , and a second low molecular weight GTP-binding protein, Rac (13)(14)(15)(16)(17). All three of these cytosolic proteins have been shown to translocate from the cytosol to the membrane during NADPH oxidase assembly (1,(3)(4)(5). A fourth cytosolic protein known as p40 phox has been shown to interact with p47 phox and p67 phox and may function to stabilize a 240-kDa complex of cytosolic oxidase proteins in resting cells (18,19).
According to the current model of NADPH oxidase assembly, p47 phox , p67 phox , and possibly p40 phox translocate en bloc to associate with flavocytochrome b during neutrophil activation (3,20). Rac appears to translocate simultaneously but independently of the other cytosolic components to associate with the membrane-bound NADPH oxidase (5,21,22). Studies of oxidase assembly in neutrophils from patients with various forms of chronic granulomatous disease suggest that p47 phox binds directly to flavocytochrome b (23), and at least six regions of flavocytochrome b have been identified as potential sites for interaction with p47 phox (24 -29). Complimentary sites of interaction present on p47 phox have recently been mapped to a cationic domain encompassing residues 323-342 (30) and to the first and possibly second SH3 domain of p47 phox (27,28). However, the process whereby p47 phox associates with p67 phox to form a complex that translocates to the membrane and the ultimate assembly of the active NADPH oxidase complex is not fully understood. Recently, there have been several reports that SH3 domain interactions mediate the formation of a p47 phox -p67 phox complex (27,28,31,32) and possibly a complex with p40 phox (18,19); however, the function of these regions in NADPH oxidase activation still remains unclear. Peptides mimicking the SH3 domain target sites are relatively weak inhibitors of NADPH oxidase complex formation and have little effect on O 2 . production in the cell-free assay (26,31,33). In addition, SH3-mediated interactions between p47 phox and p67 phox are found in resting neutrophils and in unactivated in vitro assay systems (28, 31, 34 -36), suggesting that SH3 domains mediate only part of the assembly process but are not completely responsible for the activation-induced assembly of a stable NADPH oxidase.
To gain a further understanding of the assembly of the NADPH oxidase complex and investigate the interaction of p47 phox with p67 phox , we used peptide mapping in an in vitro binding assay and random sequence peptide phage-display library analysis to define regions of p47 phox that interact with p67 phox . Our data not only confirmed previous reports that the proline-rich carboxyl-terminal domain of p47 phox (residues 361-370) is involved in p67 phox binding (28,31,32,34,36) but also demonstrated the presence of a novel binding site for p67 phox encompassing residues 323-332 of p47 phox . Surprisingly, this same region of p47 phox also binds to flavocytochrome b (30), and our present studies now provide direct evidence that p47 phox 323-332 represents a mutually exclusive binding site for both p67 phox and flavocytochrome b and that this p47 phox region plays a central role in NADPH oxidase assembly.

MATERIALS AND METHODS
Preparation and Fractionation of Neutrophils-Purified human neutrophils, isolated as described previously (37), were disrupted by N 2 cavitation, and membrane and cytosolic fractions were prepared from the cavitate by sequential centrifugation as described by Fujita et al. (37).
Production and Purification of Recombinant p67 phox , p47 phox , and Rac2-Recombinant p47 phox and p67 phox were produced in a baculovirus expression system and purified as described by Leto et al. (38). Glutathione S-transferase (GST) 1 -Rac2 was produced in Escherichia coli using a pGEX-2T expression vector (kind gift of Dr. Ulla Knaus, Scripps Research Institute, La Jolla, CA) (39), and GST was cleaved from Rac2 using thrombin (40). All recombinant proteins were found to be Ͼ95% pure using SDS-PAGE and Coomassie Blue staining, their identity was confirmed by Western blotting (data not shown), and they were found to be active in reconstituting NADPH oxidase activity in a cell-free NADPH oxidase assay system containing only recombinant cytosolic proteins (see below). In addition, GTP-binding activity of the recombinant Rac2 was confirmed using an [ 35 S]GTP␥S binding assay as described by Bokoch and Quilliam (41) (data not shown).
Peptide Synthesis-Peptide RPPGFSPFR was obtained from The American Peptide Co. (Sunnyvale, CA); peptide AYRRNSVRFL was obtained from Macromolecular Resources (Colorado State University, Fort Collins, CO); and all other peptides were synthesized by the Montana State University peptide synthesis facility. The sequence and purity of all synthetic peptides was confirmed by high performance liquid chromatography and mass spectrometry.
Protein Affinity Binding Assays-Recombinant p47 phox , p67 phox , and bovine serum albumin (control) were conjugated to CNBr Sepharose 4B at 1-2 mg of protein/ml of Sepharose following the manufacturer's instructions (Pharmacia Biotech Inc.). To determine NADPH oxidase protein binding to the conjugated Sepharose beads, 20 -50 l of beads were incubated with 5-25 g of recombinant p67 phox or p47 phox , 4 ϫ 10 6 cell equivalents of human neutrophil cytosol or 10 7 cell equivalents of human neutrophil membranes in cell-free assay buffer (10 mM sodium phosphate, 130 mM NaCl, 2 mM EGTA, 2 mM MgCl 2 , and 2.7 mM KCl (pH 7.2)) containing 10 M GTP␥S, and 100 M SDS in a final volume of 500 l). Reactions were incubated with mixing for 3 h at 4°C, pelleted in a microcentrifuge at 1000 rpm for 4 min, and then subsequently washed 4 times in 1 ml of cell-free assay buffer. Samples were then normalized to contain identical amounts of Sepharose beads and analyzed by SDS-PAGE and Western blotting to detect bound proteins. Nonspecific binding of proteins to the beads was determined using unconjugated and BSA-conjugated Sepharose beads. In all experiments, nonspecific binding was Ͻ5% of the specific binding observed under activating conditions (see Fig. 1B).
For analysis of peptide biological activity, peptides were added to the binding assays prior to activation with SDS. For protein competition assays, 2 ϫ 10 7 cell equivalents of membranes or 30 g of p67 phox were added at 1-1.5 h after SDS activation. Samples were analyzed as described above.
Random Sequence Peptide Phage-Display Library Analysis-Recombinant p67 phox was biotinylated to yield a final biotin/protein ratio of approximately 20:1 (42). Three rounds of biopanning were performed as described previously (42) using 10 or 250 g of biotinylated p67 phox per biopan (both p67 phox concentrations gave similar results) and 75 l of a nonapeptide phage-display library (number J404 -3) (43). The random peptide regions of bacteriophage obtained from the third round eluate were sequenced as described previously (29,43,44).
Cell-free NADPH Oxidase Reconstitution Assay-NADPH oxidase activity was measured in a cell-free assay system as described previously (29), except recombinant cytosolic proteins were used instead of neutrophil cytosol. The standard reaction mixture containing 2 ϫ 10 6 cell equivalents of human neutrophil membranes, 350 -430 nM recombinant p67 phox and p47 phox , and 180 nM recombinant Rac2 was incubated for 2 min at 25°C followed by the addition of 100 M SDS. After a 5-or 10-min incubation, 200 M NADPH was added, and the rate of O 2 . production was monitored on a Molecular Devices THERMOmax microtiter plate reader for 15 min at 25°C as described previously (29).
To analyze NADPH oxidase assembly, cell-free NADPH oxidase assays were performed as described above but without either p47 phox or p67 phox during the 5-or 10-min incubation with SDS. Following the SDS incubation period, the missing NADPH oxidase factor was added back Ϯ500 M peptide, incubated an additional 3 min, and then NADPH was added.
Electrophoresis and Western Blotting-Proteins were separated by SDS-polyacrylamide gel electrophoresis on 5-20% polyacrylamide gradient gels and electroblotted as described previously (5). Relative amounts of binding proteins were quantified by densitometric analysis (IS-1000 Digital Imaging System, Alpha Innotech, San Leandro, CA) of the immunoreactive p47 phox , p67 phox , gp91 phox , and p22 phox . The random regions of phage recovered from p67 phox biopanning were sequenced, and the putative motifs were aligned. Residues identical to the corresponding p47 phox sequence are in bold, conservative substitutions are underlined, and residues shifted one position are represented by italics.

RESULTS
Random Sequence Peptide Phage-Display Library Analysis of p67 phox -To identify possible regions of interaction between p67 phox and other NADPH oxidase proteins, we screened a random sequence peptide phage-display library with recombinant p67 phox to affinity-select phage from the library that specifically bound to p67 phox . The predicted amino acid sequences from the random region of 113 affinity-selected bacteriophage were analyzed, and two dominant consensus motifs were identified (Table I). When compared with the amino acid sequence of p47 phox , these motifs mapped to residues 319 -332 and 361-370 in p47 phox . The strongest homology among the phage peptides was evident in those representative of the p47 phox region 319 LSQDAYRRNSVRFL 332 , which represents a putative novel site of interaction between p47 phox and p67 phox . As Table I shows, 19 phage isolates contained 3-4 residues identical to this region, and 8 other phage isolates supported this consensus. Additionally, the presence of conservative substitutions in these phage sequences compared with the p47 phox sequence suggests an even greater similarity of many of the phage peptides to this region. Interestingly, our previous screening of purified flavocytochrome b, using a similar experimental approach, demonstrated that this same region in p47 phox , residues 323-342, represented part of a flavocytochrome b binding domain (30).
The second consensus sequence mapped to p47 phox 361 PQPAVPPRPS 370 and was represented by 11 phage peptides (see Table I). Previously, Leto et al. (28) and Finan et al. (31) demonstrated that the proline-rich region encompassing p47 phox residues 358 -371 of p47 phox was a site of interaction with p67phox and recently proposed that this region bound to the second SH3 domain of p67 phox in resting neutrophils (34,36). Thus, our data mapping this region of p47 phox as a p67 phox interactive site not only confirms the previous results of Leto et al. (28) and Finan et al. (31) but also serves as a positive control for our phage-display library analysis. Since this site has previously been characterized (28,31,34,36), no further analyses were performed involving this region of p47 phox .
Peptide Mapping and Affinity Binding Assays-Previously, we reported that p47 phox residues 323-342 were important for association with flavocytochrome b, and peptides mimicking this region blocked interaction of p47 phox with flavocytochrome b (30). However, based on the data presented above from the phage-display analysis of p67 phox , this same region of p47 phox also appears to be a binding site for p67 phox , suggesting that following activation, p67 phox and flavocytochrome b may compete for p47 phox through a high affinity interaction involving p47 phox residues 323-332. To evaluate this hypothesis, we analyzed synthetic peptides mimicking this region of p47 phox for their ability to block p47 phox -p67 phox binding using affinity binding assays.
As shown in Fig. 1A, little or no p67 phox binds to immobilized p47 phox prior to activation with SDS, but this binding significantly increased (up to 100-fold) after activation of the system with SDS. Similarly, p47 phox bound to immobilized p67 phox only after activation with SDS. As controls for nonspecific binding or aggregation of cytosolic proteins with the Sepharose beads, we performed similar experiments using unconjugated and BSA- conjugated Sepharose beads. The results shown in Fig. 1B demonstrate that there was essentially no nonspecific binding or precipitation of p47 phox or p67 phox from human cytosol to unconjugated or BSA-conjugated Sepharose beads under activating conditions (similar results were also observed using recombinant cytosolic proteins instead of cytosol (data not shown)).
Using the affinity binding assay described above, we analyzed the peptide identified by our phage-display screening. As shown in Fig. 1A, a peptide mimicking p47 phox residues 323 AY-RRNSVRFL 332 inhibited binding of recombinant p47 phox or p67 phox to immobilized p67 phox or p47 phox , respectively, following activation (EC 50 Х250 M using either assay condition), whereas control peptides had no effect on this binding interaction. Additionally, this peptide blocked the association of p47 phox or p67 phox present in normal human neutrophil cytosol to the immobilized complimentary cytosolic protein when added at the time of activation (Fig. 1A). In contrast, addition of 1 mM peptide 323 AYRRNSVRFL 332 to the preassembled p47 phox ⅐p67 phox complex (1000-fold molar excess with respect to p67 phox ) 1.5 h after activation failed to disrupt the complex (Table II), suggesting the formation of a high affinity interaction between p47 phox and p67 phox that cannot be disrupted by this peptide alone.
In our previous studies mapping a domain of p47 phox that interacts with flavocytochrome b (30), we found that peptides representing regions flanking and/or partially overlapping the 323-332 site of p47 phox also inhibited NADPH oxidase activity in cell-free assays and in electropermeabilized neutrophils. Therefore, because p47 phox 323-332 blocked the association of p47 phox and p67 phox , we analyzed these peptides in our affinity binding assay system. As shown in Fig. 1A, we found the peptides flanking/partially overlapping our active region, 315 SRKRLSQDAYRRNS 328 and 334 QRRRQARPGPQSPG 347 , and p67 phox carboxyl-terminal peptide 513 DCTTDLESTRR-EV 526 had no effect on the p47 phox -p67 phox association when tested up to 1 mM peptide (Fig. 1A, lanes 8 and 9, and Table II, respectively). The inability of these peptides, which partially overlap the active region (e.g. peptide 315-328 overlaps 60% of peptide 323-332) and/or have an even greater number of charged residues, to inhibit formation of the p47 phox -p67 phox complex provides further evidence for a sequence and/or conformational specificity in peptide 323-332 for interfering with the formation of the p47 phox -p67 phox complex.
As mentioned above, p47 phox residues 323-332 have been shown previously to be important for association with flavocytochrome b using cell-free oxidase translocation assays (30,45). Because our affinity binding assays indicated this site also interacts with p67 phox , we tested the peptide for its ability to block p47 phox -flavocytochrome b association in the same in vitro affinity binding assay to confirm that p47 phox 323-332 was indeed blocking both interactions in the same assay system. As shown in Fig. 2, this peptide inhibited flavocytochrome b binding to p47 phox at similar or slightly higher concentrations than those required to block the p47 phox -p67 phox association (IC 50 values of ϳ400 and 200 M, respectively), whereas two control peptides (slightly smaller and larger than 323 AYRRNS-VRFL 332 ) had no inhibitory effect, indicating that the inhibition exhibited by peptide 323-332 is sequence-specific. The data shown in Fig. 2 suggest that flavocytochrome b binds to p47 phox with a slightly higher affinity than does p67 phox ; this would be expected given the multi-site binding interaction that occurs between p47 phox and both subunits of flavocytochrome b (24 -29). Finally, addition of 1 mM peptide 323-332 to assays containing complexed p47 phox and flavocytochrome b 1-1.5 h after activation failed to disrupt the p47 phox ⅐flavocytochrome b complex (Table II), confirming the formation of a stable complex that is resistant to peptide disruption (46).
Competition Assays-Based on our results, it appears that both flavocytochrome b and p67 phox compete for binding to p47 phox , possibly at residues 323-332. To evaluate this possibility, we analyzed the ability of flavocytochrome b to compete with p67 phox for binding to p47 phox using competition assays. As shown in Fig. 3, addition of p67 phox to immobilized p47 phox at an approximate 1:1 molar ratio prior to addition of flavocytochrome b (flavocytochrome b was added at 1-1.5 h after activation of the assay containing p47 phox and p67 phox ) almost completely blocked (ϳ91% inhibition) flavocytochrome b binding to p47 phox compared with controls, supporting the formation of a stable complex that renders the 323-332 binding region inaccessible as was indicated by our peptide studies. Similarly, addition of flavocytochrome b to p47 phox prior to addition of p67 phox , even at a much lower molar ratio (ϳ5 times less flavocytochrome b, assuming 3 ϫ 10 6 flavocytochrome b molecules/cell), reduced p67 phox binding to p47 phox by ϳ80% (Fig. 3). The results confirm that flavocytochrome b and p67 phox do indeed compete for binding to p47 phox .
Peptide Effects in the Cell-free NADPH Oxidase Assay-In previous studies, Nauseef et al. (45) found that p47 phox peptide 323-332 inhibited O 2 . in a cell-free NADPH reconstitution assay system and blocked assembly of the oxidase in a cell-free translocation assay. Recently, we confirmed this result and further demonstrated that this peptide directly inhibited the association of p47 phox with flavocytochrome b (30). To further evaluate the site(s) of action of this peptide, we manipulated the cell-free assay following the method of Kleinberg et al. (46). As shown in Table III, p47 phox peptide 323-332 inhibited NADPH oxidase activity in a cell-free assay containing all NADPH oxidase cytosolic components when added before or simultaneously with SDS activation (i.e. prior to NADPH oxidase assembly) but loses its ability to inhibit NADPH oxidase activity when added at Ͼ5 min after activation, indicating that Sepharose conjugated with p47 phox or p67 phox (S-p47 and S-p67, respectively) were incubated with human neutrophil cytosol (cytosol), recombinant p47 phox (rp47), recombinant p67 phox (rp67), or human neutrophil membranes (membranes), and 1 mM AYRRNSVRFL was added either before or 1.5 h after SDS activation as described under "Materials and Methods." Bound proteins were then determined by SDS-PAGE and Western blotting with the appropriate antibody as indicated. The results are displayed as a percent of control (stimulated assay with no peptide) and represent the mean Ϯ S.D. of (n) experiments. ND, experiment not performed. the site of action of this peptide becomes inaccessible after assembly of the NADPH oxidase. As we observed previously, however, there is a variable lag time after activation where the peptide is partially inhibitory, presumably due to the presence of unassembled, partially assembled, and/or unstable oxidase complexes. At longer time points after activation, fully assembled oxidase complexes are resistant to peptide effects. To further investigate the site of action of p47 phox peptide 323-332, peptide was added to assays deficient in either p47 phox or p67 phox following activation with SDS and simultaneously with the missing cytosolic factor, either p47 phox or p67 phox , at 5 and 10 min after SDS addition. As shown in Table  III, addition of peptide concurrently with p67 phox at 5 and 10 min after SDS activation of a mixture containing only p47 phox and membranes had no inhibitory effect on O 2 . production. In contrast, addition of peptide simultaneously with p47 phox at 5 and 10 min after activation of a mixture containing only p67 phox and membranes inhibited O 2 . production by ϳ78.7 and 91.7%, respectively (Table III). This level of inhibition is similar or even better than that observed if the peptide is added before activation of the assay containing membranes and both p47 phox and p67 phox (Table III). Interestingly, when the activation mixture contained only p47 phox and membranes, and p67 phox (with or without peptide) was added after activation, O 2 . production was significantly higher than in control values (Table III). This phenomenon was previously observed by Kleinberg et al. (46) although it was not discussed in their report. In the context of our present studies, however, this result suggests that p67 phox may be serving a regulatory function via its binding to p47 phox 323-332.

DISCUSSION
The nature of the p47 phox -p67 phox interaction and its function in human neutrophils is not well understood. Activation in neutrophils and in vitro using a cell-free assay system appears to cause a conformational change or unfolding of an active domain(s) in p47 phox , allowing it to associate with or form a higher affinity association with p67 phox and then translocate en bloc to the membrane where both stably associate with the active NADPH complex. Several recent reports indicate that these p67 phox -p47 phox complexes may be mediated by SH3 domain interactions (27, 28, 31, 32, 34 -36) as p47 phox and p67 phox FIG. 2. p47 phox peptide 323 AYRRNSVRFL 332 inhibits association of p47 phox with flavocytochrome b. Sepharose conjugated with p47 phox was incubated with human neutrophil membranes (f) or recombinant p67 phox (E) as described under "Materials and Methods." Western blots stained with anti-p22 phox antibody (f) or anti-p67 phox antibody (E) are shown in the upper panel (blotting for gp91 phox using anti-gp91 phox antibodies gave the same results as those shown for p22 phox (data not shown)). In all blots, lanes 1 and 2 represent control, SDS-stimulated and -unstimulated assays, respectively; lanes 3-7, represent assays containing 100, 250, 500, 750, and 1000 M AYRRNS-VRFL, respectively; lanes 8 and 9 are assays containing 1000 M control peptides RPPGFSPFR and AVEGGMKPVKLLVGC, respectively, and lane 10 is a control lane of normal human neutrophil cytosol (E) or neutrophil membranes (f). The densitometric analysis of lanes 3-7 from both blots is shown in the lower panel (the symbols are the same as described above for the blots). Prestained molecular weight standards were used on all gels (Std), and the molecular masses are as indicated. The results are expressed as a percent of control assays (no peptide) and represent the mean Ϯ S.D. of three separate experiments. each have two SH3 domains and two proline-rich SH3 targets (14,15,27,28,31,32,47). It has recently been reported that the carboxyl-terminal regions of both proteins were necessary for the formation of a p67 phox -p47 phox complex (28,31,34,36) in resting cells and that the amino-terminal SH3 domain of p47 phox interacts preferentially with p22 phox (35). However, the requirement for these interactions in initiating O 2 . production is uncertain as deletion of both p67 phox SH3 domains has no effect on in vitro O 2 . reconstitution (32,33), indicating that SH3-mediated interactions are only partially responsible for assembly of the oxidase components and activation of the NADPH oxidase.
In addition to SH3-mediated interactions, increasing evidence indicates there are other important functional sites of interaction between NADPH oxidase components, and these interactions are characterized by significantly higher affinities than those of the SH3 interactive domains (24,25,29,30,45). Phosphorylation also plays a role in activation of the NADPH oxidase in vivo (3, 48 -51) and may initiate a conformational change within p47 phox . There are eight or nine p47 phox phosphorylation sites within residues 303-379, including Ser-320 and Ser-328 which are in close proximity to the site of interaction identified in this report (50,52,53), and it has been proposed that phosphorylation may serve to neutralize the p47 phox cationic domain encompassing residues 314 -347, thus allowing it to interact with the membrane or target protein.
In an effort to further understand the interaction between p47 phox and p67 phox , we have utilized several approaches to identify and characterize potential sites of interaction of these proteins. Our present studies demonstrate the presence of a novel site in p47 phox (residues 323-332) that binds p67 phox after activation of the NADPH oxidase. This site was mapped using random sequence phage-display library analysis of p67 phox and peptide mapping using protein affinity binding assays and in vitro NADPH oxidase assays. These experiments demonstrated that a peptide mimicking p47 phox residues 323-332 completely blocked p47 phox -p67 phox association, whereas peptides representing flanking regions had no inhibitory effect. Thus, in addition to the interaction between the second SH3 domain of p67 phox and the proline-rich carboxyl-tail of p47 phox , which is present in resting cells, a further interaction between these proteins at p47 phox residues 323-332 is induced during activation.
Previously, we found that p47 phox residues 323-342 represented a site of binding interaction with flavocytochrome b (30), and this observation was confirmed in the present report using two different in vitro assays that were not evaluated previ-ously. Thus, taken together our data suggest that both p67 phox and flavocytochrome b utilize a common binding site in p47 phox , presumably at distinct stages of the activation process. Using affinity binding assays, we confirmed that p67 phox and flavocytochrome b do indeed compete for binding to p47 phox and further demonstrated that p47 phox 323-332 bound with high affinity to flavocytochrome b after activation and became inaccessible to peptide. Prior to assembly, however, this region of p47 phox appears to be associated with p67 phox and at some point in the activation process is transferred to flavocytochrome b.
Previously, Leto et al. (28) reported that a GST-fusion protein containing p47 phox residues 280 -338 did not bind to recombinant p67 phox , and this region of p47 phox encompasses the region we have reported here to associate with p67 phox . One explanation for this disparity is that GST-fusion proteins of this region of p47 phox do not accurately reflect conformational characteristics of the native protein and, therefore, are not capable of mediating a binding interaction between these molecules. In addition, Leto et al. (28) tested their interactions using nonactivating or resting cell conditions. In contrast, we observed specific interactions mediated by p47 phox 323-332 only after SDS activation, and none of the activation-induced conformational changes and/or charge neutralization in p47 phox and p67 phox would have been present in the assays of Leto et al. (28). Thus, studies using only nonactivated binding assays (28,34) or yeast two-hybrid assays (19,31,36) seem to provide accurate information about binding interactions occurring between the NADPH oxidase proteins in the resting cell but may fail to detect activation-induced interactive sites. In support of this conclusion, de Mendez et al. (34) recently reported that the interaction between the second SH3 domain of p67 phox and p47 phox residues 358 -171 occurs in the resting cell and, using phage-display library analysis, we confirmed their results (see Table I). However, we also report the presence of a novel, higher affinity binding site for p67 phox in p47 phox that is only utilized under activating conditions.
The role of the p47 phox -p67 phox association in regulating NADPH oxidase assembly and activation is currently unknown. Using the cell-free NADPH oxidase assay, de Mendez et al. (32) and Leusen et al. (33) found that the interaction between the carboxyl-terminal proline-rich region of p47 phox and the second SH3 domain of p67 phox was not required for activity in vitro but that this interaction was necessary for in vivo oxidase activity (32). Similarly, our peptide analysis in the cell-free NADPH oxidase assay suggests that the association of p47 phox with p67 phox via p47 phox residues 323-332 is also not TABLE III Interaction of p47 phox peptide 323-332 with NADPH oxidase components during assembly in vitro Cell-free NADPH oxidase assays were performed as indicated using human neutrophil plasma membranes and recombinant cytosolic proteins as described under "Materials and Methods," and 500 M peptide was added either before or after SDS activation. In assays containing only p67 phox or p47 phox in the initial incubation mixture, the complimentary cytosolic factor was added after the indicated SDS incubation time Ϯ peptide (buffer replaced peptide in control assays). required for assembly in vitro, and preassembly of p47 phox with flavocytochrome b (which sequesters residues 323-332 from p67 phox binding) had no effect on oxidase activity when p67 phox was subsequently added to the assay (see Table III). Thus, the association of p67 phox at this site must have a regulatory role that is only evident in vivo. One explanation, as modeled in Fig.  4, is that p67 phox acts as an escort for facilitating the translocation of activated p47 phox . As such, p67 phox binds to activated p47 phox residues 323-332 during translocation and, therefore, might conceal this cationic region from interacting nonspecifically via charge-charge interactions with other cellular proteins. Thus, in resting neutrophil cytosol, p47 phox , p67 phox , and p40 phox reside in a large complex mediated by SH3 domain interactions, including an interaction of the p67 phox carboxylterminal SH3 domain with the p47 phox proline-rich region (represented by box A in Fig. 4), as reported previously (18 -20, 28, 31, 34). p47 phox residues 323-332 are located within a highly cationic domain that also contains multiple phosphorylation sites (50,52), and, during activation, phosphorylation and subsequent conformational change in p47 phox would then expose this cationic p47 phox domain and possibly other SH3 domains, which would bind to p67 phox (represented by box B in Fig. 4) and be sequestered from nonspecific interactions during translocation to the site of NADPH oxidase assembly. This translocation process seems to be mediated both by increased p47 phox phosphorylation (49,50) and low affinity SH3 interactions between p47 phox and other NADPH oxidase components (28,31,35,36) (represented by box C in Fig. 4). At the site of assembly, p67 phox would then be switched off of this active site, possibly by the action of small GTP-binding proteins (Rac or Rap1A) or some other stimulus, exposing it for gp91 phox binding (represented by box D in Fig. 4) at a larger domain of p47 phox encompassing residues 323-342 and overlapping the previously occupied p67 phox binding site. Complementary binding sites in gp91 phox encompassing residues 85-93, 450 -457, 491-504, and 559 -565 presumably mediate this interaction with p47 phox (25,29,46). In addition to escorting p47 phox , p67 phox must play some other functional role in the NADPH oxidase, as p47 phox , Rac, and flavocytochrome b alone cannot support O 2 . production in vivo and require the addition of p67 phox even after translocation and assembly of the other essential NADPH oxidase components (in which case p47 phox residues 323-332 would be inaccessible for p67 phox binding). Such a distinct regulatory role for p67 phox is supported by the significantly higher levels of O 2 .
observed in cell-free NADPH oxidase assays where p67 phox was added after assembly of p47 phox with flavocytochrome b. Additionally, Tamura et al. (54) found that chemically cross-linked NADPH oxidase complexes were stabilized by p47 phox but were unstable if they contained p67 phox in the absence of p47 phox . Finally, Cross and Curnutte (55) recently reported that p47 phox and p67 phox played distinct roles in controlling electron flow from NADPH to oxygen, and our present report supports their observations and provides further clues to explain how these proteins carry out their distinct functional roles.
FIG. 4. Model of NADPH oxidase activation and assembly. In this schematic representation of the NADPH oxidase assembly process, the box designated A represents an SH3 interaction between the carboxyl-terminal SH3 domain of p67 phox and the carboxyl-terminal proline-rich region of p47 phox ; box B represents the interaction between the p47 phox cationic region encompassing residues 323-332 and p67 phox ; box C represents an interaction between a p47 phox SH3 domain and the proline-rich carboxyl terminus of p22 phox ; and box D represents gp91 phox binding to a larger domain of p47 phox encompassing residues 323-342 and overlapping the previously occupied p67 phox binding site. The box with SH3 and the ϩ indicate the tandem SH3 domains (residues 151-285) and the cationic domain (residues 314 -342) of p47 phox , respectively. Rac and Rap1A have been arbitrarily placed in this model, as the sites of interaction between these proteins and flavocytochrome b are currently unknown. See text for further details.