Activation of the Phagocyte NADPH Oxidase Protein p47 phox PHOSPHORYLATION CONTROLS SH3 DOMAIN-DEPENDENT BINDING TO p22 phox *

Activation of phagocyte NADPH oxidase requires interaction between p47 phox and p22 phox . p47 phox in resting phagocytes does not bind p22 phox . Phosphorylation of serines in the p47 phox C terminus enables binding to the p22 phox C terminus by inducing a conformational change in p47 phox that unmasks the SH3 A domain. We report that an arginine/lysine-rich region in the p47 phox C terminus binds the p47 phox SH3 domains expressed in tandem (SH3 AB ) but does not bind the individual N- terminal SH3 A and C-terminal SH3 B domains. Peptides matching amino acids 301–320 and 314–335 of the p47 phox arginine/lysine-rich region block the p47 phox SH3 AB /p22 phox C-terminal and p47 phox SH3 AB /p47 phox C-terminal binding and inhibit NADPH oxidase activity in vitro . Peptides with phosphoserines substituted for serines 310 and 328 do not block binding and are poor inhibitors of oxidase activity. Mutated full-length p47 phox with aspartic acid substitutions to mimic the effects of phosphorylations at serines 310 and 328 bind the p22 phox proline-rich region in contrast to wild-type p47 phox . We conclude that the p47 phox SH3 A domain-bind-ing site is blocked by an interaction between the p47 phox SH3 AB domains and the C-terminal arginine/lysine-rich region. Phosphorylation of serines in the p47 phox C terminus disrupts this interaction leading to exposure of the SH3

The catalytic core of NADPH oxidase is the integral membrane protein cytochrome b 558 that consists of equimolar amounts of gp91 phox and p22 phox subunits in a 1:1 complex (4,5). Cytochrome b 558 binds NADPH and FAD as well as two hemes (6,7). However, cytochrome b 558 is inactive in the absence of at least three cytosol proteins p47 phox , p67 phox , and Rac1 (or Rac2) that translocate to membrane and bind cytochrome b 558 coincident with stimulation of neutrophil pathways that elicit generation of superoxide (3,6,7). The functions of the cytosol components are not completely understood, but p67 phox is required to facilitate electron transfer via cytochrome b 558 (8,9). Rac1 and Rac2, members of the Rho class GTPases, exist in cytosol in the inactive GDP-bound state. Stimulation of neutrophils leads to GTP/GDP exchange, and the resulting active GTP-bound Rac binds p67 phox , a step required for NADPH oxidase activation (10 -12). Stimulation of neutrophils with phorbol myristate acetate leads to multiple phosphorylations of p47 phox serines by one or more kinases with the most heavily phosphorylated forms of p47 phox bound to membrane (13)(14)(15)(16). p47 phox can be activated in a cell-free NADPH oxidase system by in vitro phosphorylation with protein kinase C (17).
There are two SH3 1 domains located in the middle of p47 phox . The N-terminal SH3 domain (SH3 A ) binds to a proline-rich region in the p22 phox C terminus, an interaction essential for NADPH oxidase activity (18,19). The function of the C-terminal SH3 domain (SH3 B ) is less well understood. A proline-rich region conforming to a consensus SH3 domain-binding site is found at the p47 phox C terminus. Potential targets for the p47 phox proline-rich region include SH3 domains in p67 phox , p40 phox , and p47 phox (18 -26). A positively charged arginineand lysine-rich (Arg/Lys) domain is located in the p47 phox C terminus between the SH3 B domain and the C-terminal proline-rich sequence. Multiple serines are found in this Arg/Lys domain with amino acid sequences conforming to consensus phosphorylation sites recognized by several protein kinases (16). The Arg/Lys domain and C-terminal proline-rich domain are thought to regulate NADPH oxidase assembly but are not required for NADPH oxidase catalytic activity since a recombinant truncated p47 phox lacking these two domains is fully functional (27).
Babior and collaborators (15,16) showed that most p47 phox serines phosphorylated in response to phorbol stimulation of neutrophils mapped to the Arg/Lys domain. To test the contributions of phosphorylations of individual serines to overall p47 phox activity, Faust et al. (28) constructed a series of p47 phox mutants with alanine substitutions for every individual p47 phox serine between residues 303 and 379. EBV-trans-formed B lymphocytes derived from a patient with the p47 phoxdeficient form of chronic granulomatous disease were transfected with the alanine-substituted p47 phox cDNA. No superoxide was generated by p47 phox -deficient EBV B cells expressing p47 phox S379A stimulated with phorbol esters. EBV B cells expressing mutated p47 phox with S303A, S304A, S310A, and S328A mutations generated less than 50% of wild-type oxidase activity. p47 phox S320A supported wild-type levels of NADPH oxidase activity, whereas S315A, S345A, S348A, and S370A mutations resulted in only moderate decreases in superoxide generation. In contrast, alanine substitutions at any serine had no effect on p47 phox activity in vitro consistent with the observation that the p47 phox C terminus containing the Arg/Lys domain is not required to reconstitute oxidase activity in the cell-free NADPH oxidase assay (27).
In this paper, we report that the Arg/Lys domain maintains p47 phox in an inactive state that does not interact spontaneously with cytochrome b 558 . Our data suggest that phosphorylation of serines within the Arg/Lys domain activates p47 phox by unmasking the p47 phox SH3 A domain enabling binding to the p22 phox C-terminal proline-rich region.
EGY48 was transformed with pJG4-5, pEG202, and pSH18-34 using lithium acetate as described (29). Expression of LexA and B42 fusion proteins in EGY48 was confirmed by immunoblot analysis (12). Prototrophy in leucine-deficient media indicated reconstitution of LexA transactivator function and transcription of the LEU2 reporter (29). Cell-free Assay for NADPH Oxidase Activity-Neutrophil membrane and cytosol fractions were prepared as described (30). Peptides were synthesized by the Biopolymer Laboratory at the University of Maryland, Baltimore. Final peptide purity was Ͼ90% estimated by high pressure liquid chromatography. Correct sequence of synthesized peptides was confirmed by mass spectrometry. The cell-free assay for superoxide generation was performed in 96-well microtiter plates with a final volume of 100 l/well. Cytosol (2 ϫ 10 6 cell equivalent) and synthetic peptides were added to individual wells prior to activation by adding a reaction mixture containing neutrophil membrane (1 ϫ 10 7 cell equivalent), FAD (1 M), NADPH (200 M), GTP␥S (10 M), and arachidonic acid (30 M) (30). Membrane was solubilized with deoxycholate (1:12 w/w). The maximal rate of superoxide formation (V max ) was obtained by calculating the maximal rate of superoxide dismutaseinhibitable reduction of ferricytochrome c monitored at 550 nm using a Molecular Devices (Menlo Park, CA) Thermomax kinetic microtiter plate reader. All experiments were repeated at least three times. IC 50 (concentration of peptide that inhibited 50% of NADPH oxidase activity) was calculated as described (30). Recombinant p47NT-SH3AB (residues 1-285) and Rac1 C189S were expressed in DH5␣ Escherichia coli as glutathione S-transferase (GST) fusion proteins, purified, and cleaved from GST with thrombin as described (31). Recombinant fulllength p47 phox and p67 phox were expressed in baculovirus-infected Sf9 cells and purified as described (32).

RESULTS
Full-length p47 phox does not bind spontaneously to cytochrome b 558 in the absence of activation despite affinity of the p47 phox N-terminal SH3 A domain for the p22 phox C-terminal protein-rich region (21,34). EGY48 expressing full-length p47 phox /B42 ad and the p22 phox C-terminal/LexA bd did not grow on leucine-deficient media (see below) in contrast to yeast expressing p47 phox SH3AB/or p47 phox SH3A/B42 ad with the p22 phox C-terminal/LexA bd (Fig. 2a). We hypothesized that access to the SH3 A domain is blocked in full-length p47 phox by another region of p47 phox that obstructs the SH3 A domain binding groove. The abilities of N-terminal and C-terminal p47 phox truncations to bind to the p47 phox SH3 domains were tested in the yeast two-hybrid assay (Fig. 2b). p47 phox truncations containing the C-terminal Arg/Lys-rich domain bound p47 phox SH3AB. p47P, which encompasses the p47 phox C-terminal proline-rich peptide, bound p47SH3A but not p47SH3AB.
We examined the effects of p47 phox serines 310 and 328 on p47 phox binding to the p22 phox C terminus in the yeast twohybrid assay (Fig. 5). Mutations substituting alanines and aspartic acids at serines 310 and 328 were introduced into p47 phox cDNA. Alanine substitutions were used to assess the effects of serine loss since alanine presents a steric profile similar to serine but has neutral polarity and cannot form hydrogen bonds. Aspartic acid substitutions mimic the nega-tive charges that occur with phosphorylations of serines. In contrast to wild-type p47 phox , both p47 phox S310A/and p47 phox S310D/B42 ad interacted with p22 phox C-terminal/LexA bd in the two-hybrid assay (Fig. 5). Binding of full-length p47 phox S328A to the p22 phox C terminus was not substantially different from wild-type p47 phox . However, the S328D mutation enabled p47 phox /p22 phox C-terminal binding. The p47 phox S310D/ S328D double mutant also bound p22 phox C terminus. Substituted p47 phox /p22 phox C-terminal binding was SH3 domain-dependent since no binding was seen to p22 phox P156Q C terminus. Serine 310 and 328 substitutions did not affect binding of the p47 phox C-terminal proline-rich region to the p67 phox C-terminal SH3 B .
The effects of p47 phox Arg/Lys domain peptides-(301-320) and -(314 -335) on NADPH oxidase activity were tested in the cell-free assay (Fig. 6) using membrane and cytosol fractions isolated from resting neutrophils. Levels of NADPH oxidase activity were quantitated as maximal rates of superoxide generation (V max ) expressed as milliOD/min (Fig. 6). V max was measured with varying concentrations of wild-type and phosphorylated p47 phox Arg/Lys domain peptides. Wild-type p47 phox -(301-320)-and -(314 -335)-peptides inhibited superoxide production in a dose-dependent manner with IC 50 values among the lowest reported for peptide inhibitors of NADPH oxidase activation (Table I). Phosphorylated p47 phox Arg/Lys peptides were 7-20-fold less potent inhibitors of NADPH oxidase activity compared with wild-type peptides ( Table I).
The role of the p47 phox Arg/Lys domain in activating the NADPH oxidase was examined further in a cell-free system with recombinant p67 phox , Rac1 C189S, and p47NT-SH3AB (instead of full-length p47 phox ) used in place of neutrophil cytosol. p47NT-SH3AB spanning amino acids 1-285 is a truncated p47 phox consisting of the N terminus plus both SH3 FIG. 2. The p47 phox Arg/Lys domain binds to p47 phox SH3 AB . The abilities of p47 phox SH3 domains to bind p22 phox C terminus (a) and truncated p47 phox (b) were tested in the yeast two-hybrid binding assay as described under "Materials and Methods." Yeast expressing either p47SH3A, p47SH3B, p47SH3AB, or full-length p67 phox (p67) fused to B42 ad with either p22CT, p22*CT, p47NT, p47P, p47CT, or p47Arg/ Lys fused to LexA bd were spotted on leucine-deficient media. Growth (shown in white) indicates binding. Binding of p47 SH3A and SH3AB to p22CT is SH3 domain-dependent since no binding to p22*CT is seen (P156Q mutation eliminates affinity of the p22 phox proline-rich region for the p47 phox SH3 A domain) (21,34). Figure shown is representative of at least five similar experiments.
domains but missing the Arg/Lys domain and the rest of the C terminus. Hata et al. (27) reported that a similar truncated N-terminal p47 phox was active in the cell-free assay indicating that the p47 phox Arg/Lys domain was not required absolutely for activating the oxidase in vitro. Superoxide generation in vitro using p47NT-SH3AB was equivalent with and without arachidonic acid (Fig. 7). In contrast to p47NT-SH3AB, fulllength p47 phox was completely inactive without arachidonic acid (Fig. 7b). The abilities of p47 phox Arg/Lys domain-(301-320)-and -(314 -335)-peptides to inhibit this p47NT-SH3ABcontaining cell-free system was not dependent on arachidonic acid. These results indicate that the abilities of Arg/Lys domain peptides to inhibit oxidase activity in vitro do not require the p47 phox Arg/Lys domain. This suggests that these peptides do not inhibit oxidase activity by interfering with some function of the Arg/Lys domain in full-length p47 phox . DISCUSSION NADPH oxidase function requires activation of dormant p47 phox by phosphorylation of one or more serines clustered in the Arg/Lys domain in the C terminus. Considerable insight into p47 phox function has been gained in the last few years. p47 phox is not absolutely required for NADPH oxidase activity in vitro. Normal amounts of superoxide are generated in the cell-free system using neutrophil membrane containing cytochrome b 558 , GTP-bound Rac, and very high concentrations of p67 phox (8,9). p47 phox functions in neutrophils as an adaptor protein that facilitates binding of other oxidase components to each other. p47 phox has been reported to bind p67 phox , p40 phox , and p22 phox mediated by SH3 domain-dependent interactions and to bind also to gp91 phox (18 -27, 34, 37, 40).
p47 phox does not translocate spontaneously to membrane in resting neutrophils (41), and p47 phox does not bind recombinant p22 phox C terminus in vitro without arachidonic acid (34). These observations suggest that access by the p22 phox prolinerich region to the p47 phox SH3 A domain is blocked in unphosphorylated p47 phox . The requirement for arachidonic acid or SDS in the cell-free system is obviated by using p47 phox phosphorylated in vitro by protein kinase C (17). Exogenous phos- FIG. 4. p47 phox Arg/Lys-(301-320)-and -(314 -335)-peptides block binding of p47SH3AB to p22CT by immunoblot analysis. GST-p22CT immobilized on glutathione-agarose beads was incubated with p47SH3AB with and without native and phosphorylated Arg/Lys-(301-320)-peptides (a) and -(314 -335)-peptides (b) as described under "Material and Methods." GST-p22*CT was substituted for GST-p22CT in the control lane. Immobilized proteins were separated by SDS-polyacrylamide gel electrophoresis, transferred to polyvinylidene difluoride membrane, and incubated with anti-p47 phox serum. Scrambled peptides corresponding to Arg/Lys-(301-320) and -(314 -335) were substantially less potent in their abilities to block co-precipitation of p47SH3AB by GST-p22CT compared with native sequence peptides (data not shown). Figure shown is representative of three experiments.

FIG. 5.
Mutations at p47 phox serines 310 and 328 unmask the SH3 domains and enable p47 phox /p22 phox binding by yeast twohybrid analysis. Alanine and aspartic acid substitutions for serines 310 and 328 were introduced into full-length p47 phox using site-directed mutagenesis methods as described under "Materials and Methods." Yeast expressing either native or substituted full-length p47 phox fused to B42 ad with either p22CT, p22*CT, or p67 fused to LexA bd were spotted on leucine-deficient media. Alanine and aspartic acid substitutions at serines 310 and 328 did not affect p47 phox /p67 phox binding. Binding of p47 phox S310A, p47 phox S310D, p47 phox S328D, and p47 phox S310D S328D to p22CT was SH3 domain-dependent since no binding to p22*CT was seen.  phorylation by protein kinase C and incubation with arachidonic acid both induce similar changes in p47 phox conformation as measured by changes in tryptophan fluorescence (42) and susceptibility of cysteine 378 to reaction with N-ethylmaleimide (43). Several authors (3,34,43) have speculated that conformational changes induced by phosphorylation and anionic amphiphiles arise from rearrangements of internal p47 phox binding and result in exposure of the p47 phox SH3 A domain. We hypothesized that unmasking the p47 phox SH3 A domainbinding site requires disruption of binding between the p47 phox C terminus and the two SH3 domains. We found that the p47 phox Arg/Lys domain bound p47SH3AB in a survey of potential p47 phox -binding sites using the yeast two-hybrid assay. This result was confirmed in affinity precipitation experiments where GST-p47CT co-precipitated p47SH3AB (Fig. 3). The p47 phox C terminus contains a proline-rich region in addition to the Arg/Lys domain. Several laboratories identified the p47 phox proline-rich region as the ligand for the p67 phox C-terminal SH3 domain (20,21). In the yeast two-hybrid assay, we found that the p47 phox proline-rich region bound p47SH3A but not p47SH3B or p47SH3AB. This result is in partial agreement with observations by de Mendez et al. (19) who showed that GST-p47SH3A, but not GST-p47SH3B, bound full-length p47 phox . These authors assumed that the p47 phox proline-rich region was the ligand for p47 phox SH3 A domain binding site.
Synthetic Arg/Lys domain peptides-(301-320) and -(314 -335) inhibited co-precipitation of p47SH3AB by GST-p47CT in contrast to p47 phox proline-rich peptide-(358 -385). These two Arg/Lys domain peptides were potent inhibitors of NADPH oxidase activity in vitro (Table I) in contrast to the p47 phox proline-rich peptide (data not shown). The p47 phox Arg/Lys domain is the critical determinant for p47 phox C-terminal binding to the p47 phox SH3 AB domains. The p47 phox proline-rich region failed to bind p47 phox SH3 AB in the yeast two-hybrid experiments, and the p47 phox proline-rich region peptide did not block p47 phox C-terminal/SH3 AB binding nor inhibit superoxide production. This result is not surprising; if phosphorylation toggles the p47 phox activation switch, then the Arg/Lys domain that encompasses most of the phosphorylated p47 phox serines should be intimately involved in p47 phox internal binding.
In this study, the effects of phosphorylation were examined directly using synthetic p47 phox Arg/Lys domain peptides where serines 310 and 328 were replaced by phosphoserines. Phosphoserine-substituted peptides were uniformly less able than wild-type Arg/Lys-(301-320)-and -(314 -335)-peptides to block co-precipitation of p47SHAB by GST-p22CT and GST-p47CT and to inhibit cell-free superoxide generation. Phosphoserines cannot be substituted directly for p47 phox serines 310 and 328 in the yeast two-hybrid binding experiments. Instead, we examined the effects of alanine and aspartic acid substitutions at serines 310 and 328 in full-length p47 phox mutants. Analysis of these substitutions in full-length p47 phox overcomes problems with using truncated proteins in binding assays where some aspects of biological specificity may be lost (3). The SH3 A domains in p47 phox S310D and p47 phox S328D both bound p22 phox proline-rich regions in the two-hybrid assay suggesting that introducing a single negative charge at either of these two positions is sufficient to disrupt p47 phox Arg/Lys domain binding to the SH3 AB domains. Alanine substitution at serines 310 and 328 yielded different results. p47 phox S328A did not bind to the SH3 AB domains indicating that loss of the serine 328 hydroxyl group was not sufficient to disrupt Arg/Lys domain binding to the SH3 AB domains. In contrast, p47 phox S310A interacted with the SH3 AB domains in the two-hybrid assay. This suggests that the serine 310 hydroxyl side group may participate directly in an interaction that stabilizes Arg/Lys domain to the SH3 AB domain.
The abilities of Arg/Lys peptides-(301-320) and -(314 -335) to block co-precipitation of p47SH3AB by GST-p22CT was surprising since the p47 phox Arg/Lys domain does not bind to p22 phox (data not shown) and does not bind to the binding groove in the p47 phox SH3 A domain (Fig. 2b). p47SH3AB is equivalent to p47SH3A in its ability to bind the p22 phox prolinerich region (Fig. 2a) suggesting that p47SH3AB exists normally in an open conformation where access to the p47 phox SH3 A domain is unimpeded. Masking the SH3 A domain within full-length p47 phox involves adoption of a new conformation by the two p47 phox SH3 domains in concert such that the binding groove of the SH3 A domain is no longer accessible. Binding to the Arg/Lys domain locks p47 phox SH3 AB in this closed conformation (Fig. 8). Stimulation of phagocytes leads to phosphorylation of serines within the p47 phox C terminus and disruption of this internal interaction. Phosphorylated p47 phox then relaxes into the open conformation characteristic of p47SH3AB with unmasking of the SH3 A domain, binding to p22 phox , and activation of the NADPH oxidase.
Internal binding of the p47 phox Arg/Lys domain to the p47 phox SH3 AB domains appears to be sufficient to block spontaneous activation of the NADPH oxidase in the cell-free system. The  Fig. 1) in a 100-l final volume. NADPH oxidase activity was quantitated as described under "Materials and Methods" with (a) and without (b) 30 M arachidonic acid. Levels of superoxide generation by p47NT-SH3AB without peptides (f) were increased modestly with arachidonic acid compared with no arachidonic acid. Molar equivalent amounts of full-length p47 phox were completely inactive in the absence of arachidonic acid (छ) but yielded levels of superoxide similar to p47NT-SH3AB with arachidonic acid (data not shown). Each point is the mean of duplicate wells and data shown are representative of four experiments. requirement for arachidonic acid as the exogenous activator was bypassed when truncated N-terminal p47 phox lacking the Arg/Lys and proline-rich domains was substituted for fulllength p47 phox (Fig. 7). Inhibition of superoxide generation by p47 phox Arg/Lys-(301-320) and -(314 -335) was also not dependent on arachidonic acid activation since both peptides inhibited oxidase activity in the cell-free system without arachidonic acid. p47 phox SH3 AB domain binding to the Arg/Lys domain is not dependent on other loci within the p47 phox C terminus (such as the p47 phox proline-rich region) since Arg/Lys peptides-(301-320) and -(314 -335) inhibited oxidase activity with p47NT-SH3AB.
An alternative model for function of the p47 phox Arg/Lys domain is that phosphorylation of one or more serines creates a ligand recognized by a binding site on another Phox protein. This would be analogous to how phosphotyrosines are recognized by SH2 domains. Our findings that phosphoserine-containing Arg/Lys peptides are uniformly less potent than native peptides in binding and cell-free oxidase activity experiments argue against this model. A more complex model was presented by DeLeo et al. (36,37) who hypothesized that the p47 phox Arg/Lys domain bound to specific sites in both p67 phox and to cytochrome b 558 at different stages of NADPH oxidase assembly. They found several peptides with sequences similar to Arg/Lys domain amino acids compassing serine 328 using purified cytochrome b 558 and recombinant p67 phox as baits in screens of a phage display library. p47 phox Arg/Lys peptides inhibited p47 phox phosphorylation, translocation of p47 phox to membrane in vitro, and superoxide generation in vitro and in permeabilized neutrophils. These experimental findings are consistent with our findings. The fact that we and others (27) can reconstitute NADPH oxidase activity in vitro using p47NT-SH3AB lacking the Arg/Lys domain indicates that the Arg/Lys domain is not required absolutely for activity in the cell-free system. Arg/Lys peptides inhibited oxidase activity in the cellfree system with p47NT-SH3AB indicating that the peptides are not acting as competitive inhibitors of the native p47 phox Arg/Lys domain. Rather, Arg/Lys domain peptides mimic the p47 phox Arg/Lys domain in vitro. In our model (Fig. 8), the p47 phox Arg/Lys domain functions normally to inhibit activation of the oxidase. Therefore, peptides that mimic activity of the Arg/Lys domain also inhibit the oxidase. We anticipate that the situation in cells will be more complicated and that transfection studies using mutated full-length p47 phox will prove useful in furthering understanding of p47 phox function.