Mutagenesis of an Arginine- and Lysine-rich Domain in the gp91 phox Subunit of the Phagocyte NADPH-oxidase Flavocytochromeb 558 *

Site-directed mutagenesis was used to generate a series of mutants harboring point or multiple substitutions within the hydrophilic, polybasic domain of gp91 phox encompassed by residues 86–102, which was previously identified as a site of interaction with p47 phox during phagocyte NADPH oxidase assembly. Recombinant wild-type or mutant gp91 phox was expressed in a human myeloid leukemia cell line in which the endogenous gp91 phox gene was disrupted by gene targeting. NADPH oxidase activity was measured in a cytochrome c reduction assay following granulocytic differentiation of cells that expressed recombinant gp91 phox . Expression of a gp91 phox mutant in which amino acids 89–97 were replaced with nine alternate amino acids abolished NADPH oxidase activity. Expression of gp91 phox mutants R89T, D95A, D95R, R96A, R96E, or K102T did not significantly affect NADPH oxidase activity. However, mutations of individual or paired arginine residues at positions 91 and 92 had substantial effects on superoxide generation. The R91E/R92E mutation completely abolished both NADPH oxidase activity and membrane-translocation of the cytosolic oxidase proteins p47 phox , p67 phox , Rac1, and Rac2. The phorbol 12-myristate 13-acetate-induced rate of superoxide production was reduced by ∼75% in cells expressing R91T/R92A, R91E, or R92E gp91 phox along with an increased lag time to the maximal rates of superoxide production relative to cells expressing wild-type gp91 phox . Taken together, these results demonstrate that Arg91 and Arg92 of gp91 phox are essential for flavocytochrome b 558 function in granulocytes and suggest that these residues participate in the interaction of gp91 phox with the cytosolic oxidase proteins.

Site-directed mutagenesis was used to generate a series of mutants harboring point or multiple substitutions within the hydrophilic, polybasic domain of gp91 phox encompassed by residues 86 -102, which was previously identified as a site of interaction with p47 phox during phagocyte NADPH oxidase assembly. Recombinant wild-type or mutant gp91 phox was expressed in a human myeloid leukemia cell line in which the endogenous gp91 phox gene was disrupted by gene targeting. NADPH oxidase activity was measured in a cytochrome c reduction assay following granulocytic differentiation of cells that expressed recombinant gp91 phox . Expression of a gp91 phox mutant in which amino acids 89 -97 were replaced with nine alternate amino acids abolished NADPH oxidase activity. Expression of gp91 phox mutants R89T, D95A, D95R, R96A, R96E, or K102T did not significantly affect NADPH oxidase activity. However, mutations of individual or paired arginine residues at positions 91 and 92 had substantial effects on superoxide generation. The R91E/R92E mutation completely abolished both NADPH oxidase activity and membrane-translocation of the cytosolic oxidase proteins p47 phox , p67 phox , Rac1, and Rac2. The phorbol 12myristate 13-acetate-induced rate of superoxide production was reduced by ϳ75% in cells expressing R91T/ R92A, R91E, or R92E gp91 phox along with an increased lag time to the maximal rates of superoxide production relative to cells expressing wild-type gp91 phox . Taken together, these results demonstrate that Arg 91 and Arg 92 of gp91 phox are essential for flavocytochrome b 558 function in granulocytes and suggest that these residues participate in the interaction of gp91 phox with the cytosolic oxidase proteins.
The phagocyte NADPH oxidase (respiratory burst oxidase) is a multicomponent enzyme that catalyzes the transfer of electrons from NADPH to molecular oxygen to form superoxide, a precursor of toxic oxidants that are important for host defense against microorganisms. The redox site of the oxidase is flavocytochrome b 558 , which is a membrane-associated heterodimer composed of gp91 phox and p22 phox . NADPH oxidase activity also requires the cytosolic proteins p47 phox , p67 phox , and either of the small G-proteins Rac1 or Rac2 (reviewed in Refs. 1 and 2). The p47 phox and p67 phox subunits are present as a complex in the cytosol of resting neutrophils, along with p40 phox , the latter of which is not required for superoxide formation in cell-free NADPH oxidase assays but may stabilize p67 phox in intact cells (3). While the precise function of each of the oxidase subunits is incompletely defined, it is well documented that gp91 phox , p22 phox , p47 phox , and p67 phox are absolutely required for superoxide production in intact cells, since a deficiency in any one of these proteins results in chronic granulomatous disease (CGD), 1 a rare genetic disorder characterized by severe recurrent bacterial and fungal infections (see Ref. 4

for review).
The oxidase is dormant in resting phagocytes until cellular activation by inflammatory stimuli signals the assembly of flavocytochrome b 558 , p47 phox , p67 phox , and Rac into the active respiratory burst enzyme. The mechanism of activation and assembly of the oxidase is currently under investigation. According to the current paradigm (reviewed in Ref. 2), flavocytochrome b 558 serves as a docking site for p47 phox and p67 phox , which translocate to the plasma membrane as a unit after one or both of these cytosolic proteins become phosphorylated. The p47 phox subunit appears to play an important role in mediating the initial phases of oxidase assembly. In cell-free assays of superoxide production, participation of p47 phox was shown to precede that of p67 phox in formation of the active NADPH oxidase (5). In addition, p67 phox translocation does not occur in p47 phox -deficient CGD, whereas, in the genetic absence of p67 phox , translocation of p47 phox is not impaired (6). Current evidence suggests that Rac translocates to the membrane concurrently (7) with, but as a separate unit from, p47 phox and p67 phox (8); however, it is not clear if translocation of p67 phox is a prerequisite for translocation of Rac. While Rac has been shown to interact directly with p67 phox (9 -12), Rac2 translocates to flavocytochrome b 558 even in the absence of p47 phox and p67 phox as shown in the analysis of intact CGD neutrophils deficient in either p47 phox or p67 phox (13,14) or in cell-free reactions using p47 phox -or p67 phox -deficient cytosol (13).
Multiple sites within both gp91 phox and p22 phox have been proposed as binding sites for p47 phox including a hydrophilic, polybasic domain within gp91 phox between residues 86 and 102. A portion of this domain was identified as a site of interaction with p47 phox by DeLeo and colleagues using random sequence peptide phage display analysis (15). This domain is likely to reside at the intracytoplasmic face of the membrane based on current models of the topologic organization of gp91 phox (16). Peptides derived from this sequence are potent inhibitors of superoxide formation (15,17) and translocation of the cytosolic subunits p47 phox and p67 phox (17) in cell-free assays.
In the present study, we used site-directed mutagenesis of gp91 phox amino acids 86 -102 to identify key residues within this hydrophilic domain that are involved in NADPH oxidase assembly and function. Mutant derivatives of gp91 phox were expressed in a myelomonocytic cell line that lacks endogenous gp91 phox expression due to gene targeting. We found that replacement of a pair of arginine residues at positions 91 and 92 with two glutamic acid residues abolished superoxide production and inhibited the stable translocation of the cytosolic oxidase proteins p47 phox , p67 phox , Rac1, and Rac2 to the membranes of activated granulocytes. Replacement of both arginine residues with neutral rather than acidic amino acids did not abolish NADPH oxidase activity but resulted in an increased lag time to maximal rates of enzyme activity and an overall decrease in superoxide formation. We conclude that Arg 91 and Arg 92 of gp91 phox participate in a critical step in assembly of the active oxidase complex.
Cell Culture and Differentiation-Wild-type PLB-985 cells, a human myeloid leukemia cell line, and X-CGD PLB-985 cells, a derivative line in which the X-linked gp91 phox gene has been disrupted by gene targeting, were maintained as described (18). For granulocytic differentiation, cells at a starting density of 1 ϫ 10 5 cells/ml were exposed to 0.5% N,N-dimethylformamide for 6 days (18).
Site-directed Mutagenesis and Expression Plasmids-Mutations of individual or multiple amino acids in the hydrophilic, polybasic domain of gp91 phox between residues 86 and 102 were introduced into a fulllength wild-type gp91 phox cDNA cloned into the NotI site of the multiple cloning site of pBluescript II KS(ϩ) (Stratagene) using the Quik-Change mutagenesis kit (Stratagene) or the Sculptor in vitro mutagenesis system (Amersham Pharmacia Biotech) according to the manufacturer's instructions. Wild-type or mutant gp91 phox cDNAs were verified by dideoxynucleotide sequencing and were subcloned into the pEF-PGKpac mammalian expression vector (19). The mutant gp91 phox expression constructs were resequenced to confirm the mutations, and all constructs were linearized with KpnI prior to electroporation into X-CGD PLB-985 cells. All preparations and manipulations of plasmids were performed using standard protocols (20). Following electroporation, clones were selected by limiting dilution in 2 g/ml puromycin and screened for gp91 phox expression by immunoblot. To minimize any clone-to-clone variation in recombinant gp91 phox expression or NADPH oxidase activity, 4 -6 independent clones determined to express transgenic gp91 phox were pooled and used for subsequent analysis.
Protein Extraction and Immunoblot Analysis-Cultured granulocyte-differentiated PLB-985 and derivative cell lines were harvested by centrifugation at 500 ϫ g for 10 min at 4°C and washed once with cold phosphate-buffered saline. After diisopropyl fluorophosphate treatment for 10 min on ice, whole cell extracts were made as described previously (21).
Expression of recombinant wild-type and mutant gp91 phox and endogenous p22 phox proteins were analyzed by immunoblot using monoclonal antibodies 49 and 449, respectively (22) (kindly provided by A. Verhoeven and D. Roos, Central Laboratory of the Netherlands Blood Transfusion Service) as described previously (18). Scanning densitometry was employed to measure the relative intensity of the gp91 phox signal using a Silver Scan II scanner and Image 1.60 software (W. Rasband, National Institutes of Health). The relative levels of gp91 phox expression were confirmed using two polyclonal gp91 phox antibodies (23,24).
Measurement of NADPH Oxidase Activity-A continuous cytochrome c reduction assay was used to quantitate the superoxide dismutaseinhibitable superoxide formation by granulocyte-differentiated PLB-985 cells and derivative cell lines (18). The assay was performed at 37°C using a Thermomax microplate reader (Molecular Devices) and associated SOFTMAX version 2.02 software on whole cells after stimulation with PMA (100 ng/ml) or opsonized zymosan (6 mg/ml) freshly prepared as described (25). Superoxide production was quantitated using an extinction coefficient of 21.1 mM Ϫ1 cm Ϫ1 for cytochrome c. Data were analyzed using SOFTMAX to determine V max over a 3-min interval and the time to V max , measured as the elapsed time from the start of the assay until the maximum reaction rate was obtained. Statistical analysis was performed using InStat 2.0 software. NADPH oxidase activity in selected PLB-985 derived cell lines was also analyzed in a cell-free assay. For fractionation of membrane-associated and cytosolic proteins, cells were harvested and diisopropyl fluorophosphate-treated as described above, resuspended to 1 ϫ 10 8 cells/ml in relaxation buffer (19) and sonicated at 20% power (Sonics and Materials, Inc.) for 3 ϫ 6 s at 4°C. The membrane and cytosolic fractions were prepared by sequential centrifugation (26) and were added to a standard reaction mixture containing 100 M cytochrome c, 10 M FAD, 100 M SDS in relaxation buffer with or without superoxide dismutase. After a 3-min incubation at 25°C, the reaction was initiated with 200 M NADPH and was monitored at 550 nm using a Thermomax microplate reader.
Measurement of Translocation of Cytosolic NADPH Oxidase Components-PLB-985 and derivative cell lines were harvested by centrifugation at 500 ϫ g for 10 min at 4°C, washed in phosphate-buffered saline, and resuspended at 1 ϫ 10 8 cells/ml in relaxation buffer. The NADPH oxidase was activated by stimulating the cells with 1 g/ml PMA in Me 2 SO or with Me 2 SO alone for 10 min at 37°C. Cells were then placed on ice and, following the addition of 12 ml of ice-cold phosphate-buffered saline, were centrifuged at 500 ϫ g for 10 min at 4°C. Cells were adjusted to 1 ϫ 10 8 cells/ml in relaxation buffer and disrupted by sonication for 3 ϫ 6 s at 20% power (Sonics and Materials) at 4°C. Disrupted cells were centrifuged at 500 ϫ g for 10 min at 4°C, and the supernatants were collected and centrifuged at 2,000 ϫ g for 10 min at 4°C. The supernatants in a volume of 1 ml were then layered on a discontinuous sucrose gradient (1.5 ml of 20% over 1 ml of 38%) and centrifuged at 204,000 ϫ g for 30 min. After centrifugation, the top 600 l was collected as cytosol, and a distinct band at the gradient interface was collected as the membrane fraction. The membrane fractions were mixed with 3.5 ml of cold relaxation buffer and centrifuged at 368,000 ϫ g for 30 min, and the pellets were resuspended in 100 l of relaxation buffer with protease inhibitors. The cytosol and membrane fractions were stored at Ϫ80°C until SDS-PAGE analysis. For analysis of translocation of cytosolic proteins, the membrane fractions were separated by SDS-PAGE, and the proteins were transferred to nitrocellulose and immunoblotted sequentially with polyclonal anti-p47 phox , polyclonal anti-p67 phox (27) (both kindly provided by David Lambeth, Emory University), monoclonal anti-Rac1 (Upstate Biotechnology, Inc.), polyclonal anti-Rac2 (7) (kindly provided by Gary Bokoch, The Scripps Research Institute), and polyclonal anti-Rap1a (28) (kindly provided by Mark Quinn, Montana State University) and developed with the ECL detection system (Amersham Pharmacia Biotech) as described previously (18). Integrated densitometry was employed to measure the relative intensity of the protein signal using an Eagle Eye II Still Video System and associated software (Stratagene). Statistical analysis was performed using InStat 2.0 software.

RESULTS
A hydrophilic domain of gp91 phox encompassed by amino acids 86 -102 has been implicated as a contact point for p47 phox in the assembly of the active NADPH oxidase complex (15,17). The purpose of this study was to examine the requirements for an intact 86 -102 domain in superoxide production and to identify specific residues within this domain that were essential for normal gp91 phox function. We used site-directed mutagenesis to generate a set of gp91 phox mutants in which multiple or individual amino acid residues were replaced with alternate residues (Fig. 1). We postulated that charged amino acids within the 86 -102 sequence, which contains five basic residues and one acidic residue, might mediate electrostatic interactions with p47 phox and/or other cytosolic oxidase subunits. Hence, charged amino acid residues were replaced with neutral or oppositely charged residues, thereby altering the local electrostatic charge within this hydrophilic domain. WT or mutant cDNAs were cloned into a mammalian expression vector and transfected into X-CGD PLB-985 cells, a derivative of the myelocytic cell line PLB-985 in which the coding sequence of the gp91 phox gene was disrupted by gene targeting (18) and which, therefore, lack endogenous gp91 phox expression and NADPH oxidase activity.
We assessed the expression of recombinant wild-type or mutant gp91 phox proteins by immunoblotting whole cell extracts from PLB-985-derived granulocytes with a monoclonal antibody specific for an undefined epitope of gp91 phox (Fig. 2). The level of expression of recombinant mutant gp91 phox proteins was calculated relative to that seen for recombinant WT gp91 phox in X-CGD PLB-985 cells (Fig. 2). Similar relative levels of expression were seen when blots were probed with two different polyclonal anti-gp91 phox antibodies (data not shown). The gp91 phox mutant in which amino acids 89 -97 were replaced with nine alternate residues was expressed at only one-twentieth the level of recombinant WT gp91 phox (Fig. 2). All of the mutant gp91 phox polypeptides harboring single or double point mutations were expressed between 10 and 95% of recombinant WT gp91 phox levels (Fig. 2). The expression of p22 phox in the transfected X-CGD cell lines was rescued in proportion to the level of expression of the recombinant mutant gp91 phox (data not shown).
We next examined the effect of these gp91 phox mutations on PMA-stimulated NADPH oxidase activity in intact PLB-985derived granulocytes. Superoxide production was completely absent in cells expressing mutant gp91 phox in which residues 89 -97 had been replaced with nine alternate residues (Fig.  3A). While this mutant was only poorly expressed, previous studies have shown that expression of even small amounts of recombinant wild-type gp91 phox can reconstitute considerable oxidase activity in X-CGD neutrophils (18, 29 -31).
Single or double point mutations in the gp91 phox 86 -102 domain also resulted in altered NADPH oxidase activity, with the most dramatic effects observed with mutations at Arg 91 and Arg 92 . PMA-stimulated superoxide production was completely abolished in cells expressing recombinant gp91 phox in which both Arg 91 and Arg 92 were replaced with acidic residues (R91E/R92E). When these same arginine residues were both substituted with neutral amino acid residues (R91T/R92A), oxidase activity was markedly decreased, although not abolished (Figs. 3A and 4A). In addition, cells expressing the gp91 phox mutant R91T/R92A displayed a more than 3-fold increase in the lag period between PMA stimulation and the onset of maximal rates of superoxide production (Figs. 3B and 4A). In earlier studies, we generated transgenic X-CGD PLB-985 cells lines expressing low levels of recombinant wild-type gp91 phox that were associated with decreased rates of superoxide production but without an increased lag time (29), 2 suggesting that delayed onset of V max is independent of low oxidase activity. Replacement of individual arginine residues at position 91 or 92 with glutamic acid (R91E or R92E) also resulted in decreased and delayed PMA-stimulated oxidase activity. However, alanine substitution of either Arg 91 or Arg 92 with alanine residues (R91A or R92A) resulted in increased oxidase activity, although only that of the R91A mutant reached statistical significance (Fig. 3, A and B). Superoxide production was not significantly affected by neutral substitutions at basic residues 89, 96, or 102 (R89T, R96A, or K102T) or at aspartic acid 95 (D95A); a charge reversal at position Asp 95 by substitution with arginine (D95R) or at position Arg 96 by substitution with glutamic acid (R96E) also had no statistically significant effect on oxidase activity (Fig. 3A).
Superoxide production in selected cell lines was also measured following stimulation with opsonized zymosan. The overall rate of superoxide production was significantly reduced in cells expressing the gp91 phox mutants R91E, R92E, or R91T/ R92A (Figs. 4B and 5A). However, unlike the PMA-induced respiratory burst, the lag time to maximum rates of superoxide production after opsonized zymosan activation was not significantly increased in these mutants (Figs. 4B and 5B). The NADPH oxidase was still completely nonfunctional in cells expressing R91E/R92E gp91 phox upon stimulation with opsonized zymosan (Figs. 4B and 5A).
Finally, we investigated whether the R91E/R92E mutation in gp91 phox , which abolished NADPH oxidase activity, affected the PMA-induced translocation of the cytosolic oxidase proteins p47 phox , p67 phox , and Rac to the cell membrane. Membranes from X-CGD PLB-985 granulocytes expressing recombinant WT gp91 phox or R91E/R92E gp91 phox as well as the parental X-CGD PLB-985 cell line were extracted following PMA stimulation; separated by SDS-PAGE; and analyzed by sequential immunoblotting with anti-p47 phox , anti-p67 phox , anti-Rac1, and anti-Rac2, (Fig. 6A). To adjust for any unequal loading of lanes, the amounts of p47 phox , p67 phox , Rac1, and Rac2 were normalized against the level of the membrane-associated small G protein Rap1a. In agreement with previous studies using normal and X-CGD neutrophils (13), PMA activation of PLB-985 granulocytes expressing recombinant WT gp91 phox resulted in membrane translocation of p47 phox , p67 phox , and Rac2, whereas translocation failed to occur or was significantly reduced in  2. Immunoblot analysis of gp91 phox expression in transfected X-CGD PLB-985 cell lines. Ten g of cell extracts prepared from granulocyte-differentiated PLB-985 cells and derivatives were separated by SDS-PAGE and analyzed for gp91 phox expression by immunoblotting with a gp91 phox monoclonal antibody. The immunoblot was analyzed by densitometry, and the relative level of gp91 phox expression was calculated as a percentage of the level of expression of recombinant wild-type gp91 phox and is shown below. At least three independent extracts from each cell type were obtained and analyzed. Data shown are from one representative blot.
X-CGD PLB-985 granulocytes (Fig. 6, A and B). PLB-985 granulocytes had readily detectable levels of Rac1, which upon PMA-induced activation, was also detected in the membrane fraction of cells expressing wild-type gp91 phox (Fig. 6). Similar to Rac2, Rac1 translocation was markedly decreased in X-CGD PLB-985 cells (Fig. 6). PLB-985 granulocytes expressing the mutant R91E/R92E gp91 phox resembled X-CGD cells in that PMA-induced translocation of p47 phox , p67 phox , and both Rac1 and Rac2 was absent (Figs. 6, A and B). DISCUSSION The phagocyte NADPH oxidase is a multicomponent enzyme assembled from membrane-associated and cytosolic components upon cellular activation by inflammatory stimuli. The objective of this study was to examine the role of a hydrophilic domain encompassing residues 86 -102 of gp91 phox in oxidase assembly and function. This region was previously identified by random sequence peptide phage display library analysis as a potential site of interaction with p47 phox (15). Consistent with this observation, synthetic peptides derived from residues 86 -102 have been shown to inhibit superoxide production in the cell-free superoxide assay (15,17) and in electropermeabilized neutrophils (32), and can block the translocation of p47 phox and p67 phox to flavocytochrome b 558 in a cell-free system (17). In this study, we used site-directed mutagenesis to probe the role of specific charged amino acid residues within this region of gp91 phox , hypothesizing that electrostatic interactions between charged residues in gp91 phox and p47 phox may be important for assembly of the active oxidase complex. We focused our initial analysis on the functional effect of these mutations in intact granulocytic cells since the protein-protein interactions required for oxidase assembly in intact cells appear to differ from those utilized in the cell-free oxidase assay. For example, while two carboxyl-terminal SH3 domains of the p67 phox subunit are required for superoxide production in Epstein-Barr virustransformed B cells (33), the truncation of these domains does not interfere with oxidase activity in the cell-free assay (33). Also, phosphorylation of the p47 phox subunit is required for NADPH activity in intact cells (34) but not in the cell-free system (35). In fact, p47 phox is not required for superoxide production in the cell-free system if large amounts of p67 phox are supplied (36,37). FIG. 4. Superoxide production in intact cells or in the cell-free system. The PMA-stimulated (A) or opsonized zymosan-stimulated (B) superoxide formation was measured by a continuous cytochrome c reduction assay in intact, N, N-dimethylformamide-induced X-CGD PLB-985 cells expressing wild-type, R91T/R92A, or R91E/R92E gp91 phox . The V max was measured over a 3-min interval and is indicated by a line tangent to the kinetic curve. Data are from one of 11 representative experiments using PMA and one of three representative experiments using opsonized zymosan. C, 10 g of membrane and 50 g of cytosolic fractions isolated from granulocyte-induced X-CGD PLB-985 and derivative cell lines were used in a cell-free assay of superoxide production. The superoxide dismutase-inhibitable superoxide production was measured using a cytochrome c reduction assay. Data are from one of three representative experiments. We found that replacement of amino acids 89 -97 of gp91 phox with nine alternate residues completely abolished oxidase activity. This gp91 phox mutant was expressed at only 5% of the level of recombinant WT gp91 phox , suggesting that the mutant protein had decreased stability. Hence, we cannot rule out the possibility that the absence of enzymatic activity was due to a perturbation of overall gp91 phox structure rather than specific to the replacement of residues 89 -97. Single or double substitutions within the 86 -102 domain generally had much less of an effect on the level of gp91 phox expression. Mutation of individual amino acid residues at Arg 89 , Asp 95 , Arg 96 , or Lys 102 , did not significantly affect superoxide production, even when the mutation resulted in a reversal of electrostatic charge. However, as discussed below, striking effects on oxidase activity were observed with mutations of arginine residues at positions 91 and 92, particularly when Arg 91 and Arg 92 were mutated simultaneously. It is relevant that superoxide production was not completely abolished by mutation of any single amino acid, suggesting that at least two amino acids within this gp91 phox domain must be altered to produce a nonfunctional flavocytochrome b 558 . This is consistent with the lack of reported cases of X-CGD due to point missense mutations or deletions in this region of gp91 phox (38).
A gp91 phox mutant in which a pair of arginine residues at positions 91 and 92 was substituted with two glutamic acid residues was nonfunctional in both intact cells and in the cell-free oxidase assay. In addition, we found that the cytosolic oxidase subunits p47 phox , p67 phox , Rac1, and Rac2 failed to translocate to membranes of cells expressing the R91E/R92E gp91 phox mutant, suggesting that these arginine residues play an essential role in assembly of the active oxidase complex. Taken together with previous peptide studies that identified residues 85-93 within gp91 phox as a binding site for p47 phox (15,17), these data provide strong support of an important functional role for this polybasic region of gp91 phox in the translocation of p47 phox . Most reports agree that translocation of p47 phox is a prerequisite for p67 phox translocation (5, 6, 13); therefore, the failure of p67 phox to translocate to the membrane of gp91 phox R91E/R92E is probably secondary to the inhibition of p47 phox translocation. The sequences within p47 phox that are involved in the proposed interaction with this polybasic gp91 phox domain are unknown. Binding of p47 phox to peptides X-CGD PLB-985 granulocytes expressing recombinant wild-type or mutant gp91 phox and untransfected granulocytes were stimulated with PMA dissolved in Me 2 SO or with Me 2 SO alone for 10 min. Membranes were isolated, and 10 g of membrane proteins were separated by SDS-PAGE. Translocation of cytosolic oxidase proteins was assessed by sequentially immunoblotting with anti-p47 phox , anti-p67 phox , anti-Rac1, and anti-Rac2. Blots were also reprobed with antibodies against Rap1a. A, representative immunoblots from one of seven independent experiments. B, the Western blots were analyzed by densitometry, and the relative level of membrane-associated p47 phox , p67 phox , Rac1, and Rac2 were normalized against the level of expression of Rap1a. Data are the means Ϯ S.E. of seven experiments, except for Rac1, which is the mean Ϯ S.E. of five experiments. Data were analyzed using a Friedman nonparametric repeated measures test followed by Dunn's multiple comparison test. An asterisk indicates p Ͻ 0.05 versus no PMA stimulation.
corresponding to gp91 phox residues 85-93 was detected in the absence of p47 phox phosphorylation, which normally occurs with activation of oxidase assembly in intact phagocytes (15). A domain encompassing p47 phox residues 323-332 appears to bind gp91 phox (35,39,40) in a phosphorylation-independent fashion (35). However, this p47 phox sequence contains multiple basic residues, so an interaction with the polybasic 86 -102 domain of gp91 phox would seem unlikely.
Hence, it appears that the interaction of flavocytochrome b 558 with p47 phox is quite complex and involves multiple sites of interaction. Other putative p47 phox binding sites within gp91 phox include residues 27-46 (17) and 434 -457 (15,17), a domain surrounding Asp 500 in gp91 phox (41), and a seven-amino acid sequence in the extreme carboxyl-terminal tail of gp91 phox (15,(42)(43)(44). A proline-rich motif in the carboxyl terminus of p22 phox between residues 149 and 162 has also been shown to be essential for assembly of the oxidase complex and appears to function as a binding site for the p47 phox SH3 domains located between residues 151 and 214 and residues 225 and 284 (44 -46). Less is known about potential interactions between flavocytochrome b 558 and p67 phox . An activation domain within p67 phox has recently been identified that may function as a point of contact with either gp91 phox or p22 phox (47).
In the current study, we found that the membrane association of both Rac1 and Rac2 was markedly decreased in X-CGD PLB-985 cells and in cells expressing the mutant R91E/R92E gp91 phox . Deficient Rac2 translocation in the absence of flavocytochrome b 558 confirms previous results in X-CGD neutrophils (13). To our knowledge, ours is the first report that PMAinduced translocation of Rac1 is also dependent upon expression of flavocytochrome b 558 . While either Rac1 or Rac2 is required for maximal oxidase activity in the cell-free assay (48 -50), the relative roles of Rac1 versus Rac2 in neutrophil superoxide production are uncertain. In human neutrophils, Rac2 accounts for Ͼ95% of the Rac species present (49,51). However, Rac1 was identified as the GTP-binding protein necessary for oxidase activity in guinea pig phagocytes (48) and appears to be as abundant as Rac2 in murine neutrophils 3 and in human PLB-985 granulocytes (Fig. 6). Furthermore, NADPH oxidase activity is not abolished in neutrophils of Rac2-negative mice generated after targeted disruption of the rac2 gene, indicating that Rac2 does not play an irreplaceable role in the function of this enzyme (52).
From our experiments, we cannot distinguish whether decreased Rac1/2 translocation in X-CGD PLB-985 granulocytes and in R91E/R92E gp91 phox -expressing granulocytes is due to a direct effect on the interaction of Rac with flavocytochrome b 558 or an indirect effect resulting from deficient translocation of another oxidase component. While specific associations between p67 phox and either Rac1 or Rac2 have been identified (9,10,12), it is less clear whether Rac1 or Rac2 interacts directly with flavocytochrome b 558 . In p47 phox -or p67 phox -deficient neutrophils from CGD patients, PMA-induced translocation of Rac2 was normal (13,14,53), suggesting that the membrane association of Rac2 is independent of either of these two oxidase proteins. Therefore, the dependence of Rac2 translocation on the presence of flavocytochrome b 558 (13) implies a direct interaction between Rac2 and the flavocytochrome. In contrast to Rac2, Dusi et al. (14,53) found that Rac1 translocation did not occur in either p47 phox -or p67 phox -deficient CGD neutrophils. Taken together, these data indicate that while stable membrane association of both Rac1 and Rac2 is dependent on expression of flavocytochrome b 558 , these two GTPases associate with phagocyte membranes by different mechanisms. Our own data are consistent with a model of oxidase assembly whereby Arg 91 and Arg 92 in gp91 phox are required for translocation of p47 phox , which associates directly with flavocytochrome b 558 , and, indirectly, for translocation of p67 phox (which binds to p47 phox ) and Rac1 (which binds to p67 phox ). Decreased Rac2 translocation may reflect an indirect effect of the gp91 phox R91E/R92E mutation on a putative Rac2-flavocytochrome b 558 interaction.
Other mutations in gp91 phox Arg 91 and/or Arg 92 that eliminated a net positive charge at these two positions (rather than the double charge reversal resulting from the R91E/R92E substitution) had more subtle effects on oxidase activity. Replacement of both arginines at positions 91 and 92 with neutral residues or replacement of either Arg 91 or Arg 92 with an acidic residue resulted in a substantial decrease in superoxide production following stimulation with PMA or opsonized zymosan and an increased lag time for maximal PMA-induced enzyme activity. However, alanine substitution of either Arg 91 or Arg 92 resulted in an enhanced oxidase activity, although this was statistically significant only for the R91A gp91 phox mutant.
Hence, it appears that alterations in the local electrostatic charge at residues 91 and 92 of gp91 phox can modulate oxidase function. Since the 86 -102 domain of gp91 phox has been implicated as a binding site for p47 phox , it is noteworthy that Kleinberg and co-workers found that the lag time between oxidase activation and the onset of superoxide production in the cellfree system was decreased if the neutrophil membranes were preincubated with p47 phox -containing cytosol (5) or pure recombinant p47 phox (54) but not with purified Rac (54). This result suggested that p47 phox forms an early activation intermediate with the membrane. The lag in the onset of the maximal rates of superoxide formation for the R91E, R92E, and R91T/R92A gp91 phox mutants following stimulation with PMA could thus reflect a decreased affinity between p47 phox and the mutant gp91 phox .
That no alteration in the kinetics of superoxide formation by the R91T/R92A gp91 phox mutant cells was seen in the cell-free system is not surprising given the discrepancies between the cell-free and whole cell oxidase assays highlighted above. However, it is less clear why there was no delay in the onset of V max following stimulation of gp91 phox R91T/R92A phagocytes with opsonized zymosan. We speculate that this observation reflects differences in the kinetics of activation of p47 phox translocation induced by phorbol ester versus opsonized zymosan. While PMA directly activates protein kinase C, phagocytosis of opsonized zymosan activates tyrosine kinases, leading to activation of phosphatidylinositol-specific phospholipase C and the formation of diacylglycerol and inositol 1,4,5-trisphosphate (55), which, respectively, activate protein kinase C and increase intracellular free Ca 2ϩ . However, inhibition of protein kinase C with staurosporine or 1,5-isoquinolinesulfonyl-2-methylpiperazine inhibits PMA-stimulated NADPH oxidase activity and p47 phox phosphorylation, whereas these inhibitors do not affect oxidase activity and cause only a partial inhibition of p47 phox phosphorylation in response to opsonized zymosan (56). This observation suggests that phagocytosis of opsonized zymosan phosphorylates p47 phox and activates the oxidase by a protein kinase C-independent pathway. Differences in PMA-and opsonized zymosan-induced activation were also observed in recent studies of Downey and colleagues, who found that PD098059, a potent and selective inhibitor of mitogen-activated protein kinase, significantly inhibited the respiratory burst induced by opsonized zymosan but not of that induced by PMA (57). Taken together, these data indicate that the signaling pathways utilized by PMA differ from those involved in opsonized zymosanmediated neutrophil activation. It is plausible that these dif-ferences result in alterations in the kinetics of p47 phox phosphorylation and/or other factors that regulate translocation. These, in turn, could either directly or indirectly affect the affinity of p47 phox for the Arg 91 /Arg 92 -containing gp91 phox domain and the rate of superoxide production.