Crucial Role of Two Potential Cytosolic Regions of Nox2, 191TSSTKTIRRS200 and 484DESQANHFAVHHDEEKD500, on NADPH Oxidase Activation*

Assembly of cytosolic factors p67phox and p47phox with cytochrome b558 is one of the crucial keys for NADPH oxidase activation. Certain sequences of Nox2 appear to be involved in cytosolic factor interaction. The role of the D-loop 191TSSTKTIRRS200 and the C-terminal 484DESQANHFAVHHDEEKD500 of Nox2 on oxidase activity and assembly was investigated. Charged amino acids were mutated to neutral or reverse charge by directed mutagenesis to generate 21 mutants. Recombinant wild-type or mutant Nox2 were expressed in the X-CGD PLB-985 cell model. K195A/E, R198E, R199E, and RR198199QQ/AA mutations in the D-loop of Nox2 totally abolished oxidase activity. However, these D-loop mutants demonstrated normal p47phox translocation and iodonitrotetrazolium (INT) reductase activity, suggesting that charged amino acids of this region are essential for electron transfer from FAD to oxygen. Replacement of Nox2 D-loop with its homolog of Nox1, Nox3, or Nox4 was fully functional. In addition, fMLP (formylmethionylleucylphenylalanine)-activated R199Q-Nox2 and D-loopNox4-Nox2 mutants exhibited four to eight times the NADPH oxidase activity of control cells, suggesting that these mutations lead to a more efficient oxidase activation process. In contrast, the D484T and D500A/R/G mutants of the α-helical loop of Nox2 exhibited no NADPH oxidase and INT reductase activities associated with a defective p47phox membrane translocation. This suggests that the α-helical loop of the C-terminal of Nox2 is probably involved in the correct assembly of the NADPH oxidase complex occurring during activation, permitting cytosolic factor translocation and electron transfer from NADPH to FAD.

Baldridge and Gerard (1, 2) discovered the respiratory burst in which dramatic oxygen consumption occurred in neutrophils during bacteria phagocytosis. Subsequent research showed that the NADPH oxidase complex of phagocytic cells is responsible for the innate host defense by producing superoxide anions (O 2 . ) and reactive oxygen species to kill invading bacterials and fungi (3)(4)(5)(6). The importance of this system is demonstrated by chronic granulomatous disease (CGD), 1 characterized by severe and recurrent life-threatening infections, resulting in defective NADPH oxidase activity (7)(8)(9)(10). However, recent evidence indicates that the killing activity of neutrophils is mediated through activation of proteases by K ϩ flux, and the large-conductance Ca 2ϩ -activated K ϩ channel is essential for innate immunity (2,11). NADPH oxidase is a multicomponent complex, composed of a heterodimeric transmembrane protein known as flavocytochrome b 558 , cytosolic proteins p47 phox , p67 phox , and p40 phox , and two small GTPase, Rac2 and Rap1A. Cytochrome b 558 , the redox core of the respiratory burst oxidase, consists of a large glycoprotein gp91 phox or Nox2 and a small protein p22 phox (12). Nox2 is the catalytic center that transfers the electrons from intracellular NADPH to extracellular O 2 . It contains two nonidentical hemes and consensusbinding sequences for FAD and NADPH (13)(14)(15)(16). NADPH oxidase becomes activated and generates O 2 . after cytosolic factor assembly with cytochrome b 558 . In the resting state, p67 phox connects p40 phox and p47 phox by its PB1 and C-terminal SH3 domains with the PC domain of p40 phox and the proline-rich region of p47 phox , respectively (17). Upon activation, a series of protein-protein interactions occurs. The phosphorylation of p47 phox causes a conformational change of the protein, which leads to binding the membrane phosphoinositides and the proline-rich region of p22 phox with its PX domains and SH3, respectively (18 -22). p22 phox serves as a docking site for p47 phox and an accessory for p67 phox . p67 phox was shown to be involved in both assembly and activation of the oxidase complex, whereas p47 phox proceeded as a positive effector and increased the affinity of p67 phox with cytochrome b 558 (23,24). Current evidence suggests that there is a direct interaction between p67 phox and cytochrome b 558 , promoted by Rac1-p67 phox binding (25)(26). However, the precise sequence of the p67 phox and cytochrome b 558 interaction has not been fully elucidated. In PMA-stimulated phagocytes from X-CGD patients, no translocation of p47 phox , p67 phox , and p40 phox to the membranes occurs, indicating that gp91 phox is absolutely required for the binding of cytosolic components to cytochrome b 558 (27). Sequence alignment of Nox2 with members of the ferredoxin-reductase family demonstrates the presence of six transmembrane ␣-helices in the N-terminal hydrophobic region, four cytosolic regions, the 1 MGNWVAVNEGL 11 sequence, the B-loop 70 PVCRNLLSFLRGSSACCSTRIRRQLDRNLTFHK 102 , the D-loop 191 TSSTKTIRRS 200 , and a C-terminal region containing binding sites for FAD, NADPH, and cytosolic components of NADPH oxidase (12). A peptide inhibitory study, random-sequence peptide phage analysis, and the mutagenesis approach suggested that complementary binding sites of gp91 phox encompassing residues 451 FEWFADLL 458 and 559 RGVHFIF 565 presumably interact with cytosolic factors upon stimulation (19,21,(28)(29)(30)(31)(32)(33)(34)(35). Furthermore, mutations in Nox2 found in X ϩ CGD cases (C369R, G408E, E568K, and H303N/P304R) have been reported to abolish the oxidase assembly (36)(37)(38)(39)(40). In addition, Taylor et al. (28) had predicted a three-dimensional structure of Nox2 using ferredoxin-NADP ϩ reductase as template and proposed that an ␣-helical loop 491 FAVHHDEEKDVITG 504 covers the cleft in which NADPH binds. During oxidase activation, access of NADPH into the binding site could potentially be regulated by interaction of this loop with cytosolic oxidase factors. However, there were conflicting results for the role of this potential ␣-helical loop. Leusen and co-workers found that cytosolic factors p47 phox and p67 phox did not translocate to the plasma membrane of phagocytes from an X ϩ CGD case because of the D500G missense mutation located in this sequence of gp91 phox . In addition, a peptide mimicking 491-504 residues of Nox2 inhibited the translocation of p47 phox and p67 phox to the membrane (30). In contrast, in another X ϩ CGD case consisting of a ⌬488-497 gp91 phox deletion in the ␣-helical loop, translocation of p47 phox and p67 phox to the plasma membrane occurred normally (40). Recently Dinauer and co-workers have studied the role of a second cytosolic region of Nox2, the B loop, on oxidase activity (35). They found that two Arg at positions 91 and 92 were essential for NADPH oxidase activity and assembly. However, the role of the second intracytosolic polybasic loop of Nox2, the D-loop 191 TSSTK-TIRRS 200 , has never been elucidated.
Recently, large homologs of gp91 phox , called Nox and Duox, have been identified. They are present in various tissues of nonphagocytic cells and can generate low amounts of O 2 . , suggested to be involved in cell signaling, host defense, hypoxia response, and also as proton transport (41). The family contains seven members in humans. Nox1, Nox3, and Nox4 resemble Nox2 in that they consist of six ␣-helix transmembrane regions in the N-terminal, with two intracytosolic loops (B and D-loops) and a cytosolic C-terminal tail containing the putative binding sites for FAD and NADPH (42). A recent study indicates that p51 and p41 (homolog of p67 phox and p47 phox , respectively) regulate Nox1 to generate O 2 . (43)(44)(45)(46). This suggests that Nox proteins need some cytosolic components to be activated or regulated. Meanwhile, the molecular mechanisms of the regulation of Nox2 homologs has not been fully elucidated. The aim of this work was to investigate the role of two regions of Nox2, the intracytosolic D-loop 191 TSSTKTIRRS 200 and the C-terminal ␣-helical loop 484 DESQANHFAVHH-DEEKD 500 , on oxidase activity and assembly by means of the mutagenesis approach in the X-CGD PLB-985 cell model (37). Results suggest that these two regions are critical for NADPH oxidase activity, especially the charged amino acids Arg 195 , Arg 198 , Arg 199 , Asp 484 , and Asp 500 . The D-loop of Nox2 is probably involved in electron transfer from FAD to oxygen independently of cytosolic factor translocation, whereas the C-terminal ␣-helical loop is crucial for the assembly of oxidase and electron transfer from NADPH to FAD. Chimeric Nox2 proteins containing the D-loop of Nox1/3/4 support NADPH oxidase activity, suggesting that this region should play a similar role on Nox analog activation. Finally, two "super-mutants" of the D-loop of Nox2, with a significant increase in oxidase activity, were obtained, suggesting that the R199Q and D-loop Nox4 mutations optimize the NADPH oxidase activation process.

EXPERIMENTAL PROCEDURES
Materials-Phorbol 12-myristate 13-acetate (PMA), dimethylformamide (DMF), diisopropylfluorophosphate, and horseradish peroxidase, cytochrome c (horse heart, type VI), and latex beads were obtained from Sigma. TaqDNA polymerase, ATP, GTP␥S, and NADPH were from Roche Applied Science. Endofree Plasmid Maxi Kit was purchased from Qiagen. The Sephaglas kit and molecular weight markers for SDS-PAGE were from Amersham Biosciences. Nitrocellulose sheets for Western blotting were purchased from Bio-Rad. G418 was purchased from Invitrogen. Monoclonal antibodies 449 and 48 were kindly provided by Dr. D. Roos (Sanquin Research at CLB, Amsterdam, the Netherlands). Polyclonal antibody anti-p47 phox (rabbit antiserum) was purchased from Upstate Biotechnology, Inc. (New York, NY). Monoclonal antibody specific for gp91 phox , 7D5 was kindly provided by Dr. M. Nakamura (Nagasaki University, Nagasaki, Japan). Fetal bovine serum and RPMI 1640 were from Invitrogen.
Site-directed Mutagenesis and Expression of Recombinant gp91 phox in X-CGD PLB-985 Cell Line-Mutations were introduced into the wildtype (WT) Nox2 cDNA in pBluescript II KS(ϩ) vector using the QuikChange site-directed mutagenesis kit (Stratagene) according to the manufacturer's instructions. The sequence of WT and the mutated gp91 phox cDNA were verified by dideoxynucleotide sequencing (Genome Express, Grenoble, France). The WT or mutant Nox2 cDNA were subcloned into the mammalian expression vector pEF-PGKneo as previously described (37). The pEF-PGKneo constructs were electroporated into X-CGD PLB-985 cells in which the gp91 phox gene was disrupted by gene targeting, resulting in the absence of Nox2 expression and NADPH oxidase activity as previously described (37,38). Clones were selected by limiting dilution in 1.5 mg/ml G418.
Analysis of Nox2 Protein Expression-PLB-985 cells (5 ϫ 10 5 ) were incubated with 10 g/ml of mAb 7D5 directed against Nox2 or control monoclonal IgG1 (Immunotech, Marseille, France). Then the cells were incubated with phycoerythrin-labeled goat-F(abЈ) 2 fragment antimouse-Ig (Beckman Coulter, Marseille, France). Finally, flow cytometry analysis (FACS) was performed (FACSCalibur, BD Biosciences) (48). Sorting of 7D5-positive PLB-985 cells was done on a FACSVantage Diva (BD Biosciences) instrument at a sheath pressure of 12 p.s.i. with a 70-m nozzle. Cells were collected into phosphate-buffered saline supplemented with 0.5% bovine serum albumin and routinely contained Ͼ97% of 7D5-positive cells. Viability after sorting was Ͼ95%. After centrifugation of the cell suspension, the cell pellet was resuspended in culture medium. The expression of recombinant Nox2 was also examined by Western blot using monoclonal antibody 48 developed with the ECL detection system (Amersham Pharmacia Biotech) as described previously (49). The WT or mutant Nox2 expression was further detected indirectly by cytochrome b 558 differential spectral analysis. A molecular extinction coefficient of ⌺426 nm ϭ 106 mM Ϫ1 ϫ cm Ϫ1 for the Soret band was used for calculations (50). All experiments were done in triplicate.
Cytosol and Membrane Fraction Preparation from Transfected PLB-985 Cell Lines-10 8 PLB-985 cells were treated with 3 mM diisopropylfluorophosphate for 15 min on ice and resuspended in 1 ml of phosphate-buffered saline containing 1 mM phenylmethylsulfonyl fluoride, 2 M leupeptin, 2 M pepstatin, and 10 M 1-chloro-3-tosylamido-7-amino-2-heptanone. Cells were disrupted by sonication, and the homogenate was centrifuged at 1,000 ϫ g for 15 min at 4°C. The supernatant was withdrawn and centrifuged at 152,000 ϫ g for 1 h at 4°C. This high speed supernatant was referred to as the cytosol, and the pellet consisting of crude membranes was resuspended in the same buffer, as described before (51).
Measurement of NADPH Oxidase Activity in Intact Cells-H 2 O 2 production was measured as described previously (38). For each well in a 96-well plate, 5 ϫ 10 5 granulocytic differentiated PLB-985 cells in phosphate-buffered saline with 0.9 mM Ca 2ϩ and 0.5 mM Mg 2ϩ was mixed with 20 mM glucose, 20 M luminol, and 10 units/ml horseradish peroxidase. We added 80 ng/ml PMA to initiate the reaction. Luminescence was recorded every 30 s for a total of 90 min at 37°C using a Luminoscan® luminometer (Labsystems, Helsinki, Finland). In some experiments, PMA was replaced by 4 ϫ 10 Ϫ7 M fMLP.

Detection of NADPH Oxidase Activity in a Broken
Cell System-NADPH oxidase activity in vitro was measured in a homologous BCS using previously described protocols (52). Briefly, plasma membranes obtained from transgenic PLB-985 cells and cytosol of control human neutrophils were added to a reaction mixture containing 20 mM glucose, 20 M GTP␥S, 5 mM MgCl 2 , and arachidonic acid in a final volume of 100 l. After incubation for 10 min at 25°C, the oxidase activation was initiated in the presence of 100 M cytochrome c and 150 M NADPH. The specificity of the O 2 . production was checked by adding 50 g/ml superoxide dismutase to stop the kinetic reduction. Iodonitrotetrazolium Reductase Activity-Diaphorase activity was measured under the same BCS assay conditions, except that the 100 M cytochrome c was replaced by 50 M INT (53).
Analysis of in Vivo p47 phox Translocation-In vivo p47 phox translocation was detected by confocal microscopy analysis according to (38) with little modified. The 5 ϫ 10 5 differentiated PLB-985 cells deposited on coverslips were activated with PMA-treated latex beads (3 m in diameter) at 37°C for 15 min. The cells were fixed in 4% paraformaldehyde for 10 min and permeabilized with 0.1% Triton X-100. After extensive washing, cells were incubated for 1 h at room temperature with the diluted polyclonal anti-p47 phox (Upstate Biotechnology Inc.). After washing, cells were incubated in 5% bovine serum albumin/phosphatebuffered saline buffer containing a 2-g/ml Alexa Fluor® 488 F(abЈ) 2 fragment of goat anti-rabbit IgG(HϩL) (Molecular Probes, Eugene, OR) for 1 h. Finally, PLB-985 cells were stained with 3 M TO-PRO 3 iodide for 10 min to visualize the nuclei. Cells were examined with a confocal laser scanning microscope and analyzed with Leica confocal software (Heidelberg, Germany). In some experiments, a 10 g/ml mAb 7D5 and a 4-g/ml Alexa Fluor® 546 F(abЈ) 2 fragment of goat anti-mouse IgG(HϩL) (Molecular Probes) were added to the system as primary and secondary antibodies, respectively, to detect Nox2.
Analysis of Translocation of p47 phox in Vitro-In vitro p47 phox translocation to the plasma membrane was measured using a classic protocol (37). Briefly, membranes purified from PLB-985 cells (100 g) were preincubated with human neutrophil cytosol (1,000 g) for 2 min at 30°C and activated with (ϩ) or without 100 M SDS and 20 M GTP␥S for 15 min at 30°C. Membranes were collected between the 45 and 20% (w/v) sucrose layers after centrifugation (30,000 rpm ϫ 1 h in an SW41 rotor (Beckman) at 18°C) and analyzed using immunoblotting with anti-peptide polyclonal antibody directed against p47 phox (24).
Protein Determination-Protein content was estimated using the Bradford assay (54) or the Pierce® method (55).

RESULTS
To further understand the role of the D-loop ( 191 TSSTK-TIRRS 200 ) and the C-terminal ( 484 DESQANHFAVHH-DEEKD 500 ) in Nox2 upon oxidase activation, homologous sequence analysis in the ferredoxin-NADP ϩ reductase family was done (Fig. 1). The D-loop is a polybasic region, containing Lys 195 , Arg 198 , and Arg 199 , which are conserved in the ferredoxin-NADP ϩ reductase family. In the C-terminal, the charged amino acids encompassing residues 484 -500 are also highly conserved in this family, especially Asp 484 , His 490 , Asp 496 , and Asp 500 . We postulated that charged amino acids within these two domains might be important for the active conformation of Nox2 and/or for further binding with cytosolic factors. Hence charged amino acid residues were replaced with neutral or reverse-charged residues by site-directed mutagenesis to alter the local electrostatic charge within these  2B) were generated by directed mutagenesis and stably transfected in X-CGD PLB-985 cells, which lack endogenous Nox2 expression because of gene targeting (56). We also produced two mutants known to disturb cytosolic factor translocation to the plasma membrane, the D500G mutant, which had reproduced a human X ϩ CGD case (30), and the RR9192EE mutant of the B loop of Nox2 previously studied by Biberstine-Kindade et al. (35). These two mutants served as controls to validate NADPH oxidase activity and assembly experiments.
The WT and mutated Nox2 cDNA subcloned into the pEF-PGKneo vector were verified by sequence analysis and purified before transfection in X-CGD PLB-985 cells, as described under "Experimental Procedures." Nox2 expression was studied by FACS using the monoclonal antibody 7D5, and IgG1 were used as negative control to test the specificity of Nox2 binding (38). For some mutants, FACS was also used to sort a highly Nox2expressing population. This was done for K195A, RR198199QQ (data not shown), and K195E mutant cells (Fig. 3A). We detected 12-20 different clones for each mutant and to minimize clone-to-clone variation in NADPH oxidase activity, three to four independent clones of each highly expressing Nox2 mutant were used for subsequent analysis (data not shown). As ex-pected, Nox2 was not detected in X-CGD PLB-985 cells or in empty vector transfected PLB-985 cells. Expression of WT or mutant Nox2 was also assessed by semiquantitative immunoblotting of soluble extracts from differentiated WT and transfected X-CGD PLB-985 cells using the monoclonal antibody 48 (38). Similar Nox2 expression was seen in all differentiated PLB-985 cells (Fig. 3B). Identical differential spectrum characteristics of flavocytochrome b 558 were observed in all mutants compared with WT Nox2-transfected PLB cells or WT PLB-985 cells (data not shown). The amount of cytochrome b 558 was equivalent in all the transfected X-CGD PLB-985 cells, confirming results previously obtained by FACS and Western blot analysis (Table I). This demonstrated correct heme incorporation in recombinant Nox2. No differential spectra were detected in X-CGD PLB-985 cells or empty vector-transfected cells. In conclusion, high, stable, and similar amounts of recombinant WT or mutant Nox2 proteins were expressed in transfected X-CGD PLB-985 cells. This was crucial for comparing the impact of each mutation on NADPH oxidase functions in such cells.
We next measured H 2 O 2 production in transfected X-CGD PLB-985 cells stimulated by 80 ng/ml PMA using luminolamplified CL. The RR9192EE Nox2 transfected PLB-985 cells exhibited no oxidase activity, whereas its cytochrome b 558 expression was comparable to the WT Nox2-transfected PLB-985 cells (Table I). This was consistent with previously published results (35) and allowed us to validate our methodology.  (Table I). In the C-terminal tail of Nox2, changing Asp 484 to a neutral amino acid (D484T) inhibited the NADPH oxidase activity, whereas replacing it with a positive charge (D484H) had little effect (79% of control). In addition, the D496H mutation had no effect on oxidase activity. However, cells expressing H490A-Nox2 exhibited normal oxidase activity, whereas the H490D mutation destroyed most of the NADPH oxidase activity (10% of control) (Table I). In conclusion, the effect of D484 and H490 charge changes had different effects on oxidase activity depending on the type of amino acid replacement. Meanwhile, the disappearance of the negative charge supported by Asp 500 in the D500A, D500R, and D500G mutations had a definitive inhibitory effect on NADPH oxidase activity (Table  I). It should be noted that the D500G Nox2-transfected cells mimicked the phenotype of a previously described X ϩ CGD case (30).
To improve our knowledge of H 2 O 2 production in DMF-differentiated transfected X-CGD PLB-985 cells, PMA and fMLP stimulation were compared (Fig. 4A). All the experiments were done in triplicate and reproduced twice. Inhibitory effects of some mutations on oxidase activity, observed previously in PMA-stimulated PLB-985 cells, were also obtained in fMLPactivated mutant cells. The increasing effect on NADPH oxidase activity of the R199Q mutation and the replacement of the Nox2 D-loop with the same from Nox1/3/4, was more evident using fMLP stimulation. Indeed in the R199Q and the Dloop Nox4 mutant cells, H 2 O 2 production was 7.9 and 4.6 times higher than in the WT PLB-985 cells, respectively (Fig. 4A). The kinetics of H 2 O 2 production was not the same when WT or WT Nox2-transfected PLB-985 cells were activated by PMA or fMLP (Fig. 4B). The maximum of H 2 O 2 production (V max ) was obtained in 10 -13 min (T max ) for PMA activation, whereas for fMLP, the T max was reached in roughly 4 min (Table II). Interestingly, the V max for the D-loop Nox4 -Nox2 super-mutant cells activated with PMA was obtained in 4 -5 min versus 10 -13 min in WT PLB-985 cells. However, when these cells were stimulated with fMLP, the kinetics of H 2 O 2 production remained unchanged, whereas the V max was higher than in the WT PLB-985 cells (Fig. 4B and Table II). The same kinetic changes in H 2 O 2 production were also observed in the R199Q mutant cells (data not shown).
We also examined the in vitro NADPH oxidase activity of transfected X-CGD PLB-985 cells using a broken cell system (BCS). To compare only the effect of mutations on Nox2 of the plasma membranes purified from transfected PLB-985 cells, cytosol from human neutrophils was used as the source of NADPH oxidase cytosolic components. The in vitro oxidase activity was totally restored in X-CGD PLB-985 transfected with WT Nox2 cDNA compared with WT PLB-985 cells (Table  I). For almost all the studied mutations, the results obtained in BCS were correlated with those obtained in intact cells, although residual oxidase activity (ϳ20% of the original WT PLB-985 cells) was observed in some mutants that had no oxidase activity in vivo. In addition, the highly increased NADPH oxidase activity measured in intact R199Q and in the D-loop Nox4 -Nox2 mutant cells activated with fMLP was not reconstituted in vitro.
As demonstrated above, mutations within these two cytosolic regions of Nox2 (K195A/E, R198E, R199E, RR198199QQ, D484T, and D500A/R/G) completely abolished the NADPH ox-TABLE I Cytochrome b 558 amount and NADPH activity in WT and transfected X-CGD PLB-985 cells The cytochrome b 558 concentration was quantified using Soret band absorption of 1% Triton X-100-soluble extracts from WT, X-CGD, and mutant transfected PLB-985 cells, considering that cytochrome b 558 contains two hemes. H 2 O 2 production was measured by luminol-amplified CL for 90 min in differentiated WT and transfected X-CGD PLB-985 cells after PMA activation. RLU represents the sum of relative luminescence units measured in 90 min. NADPH oxidase activity was reconstituted in a broken cell system (BCS) with purified plasma membranes of the indicated cells (30 g) in the presence of neutrophil cytosol (300 g) and activated with GTP␥S and arachidonic acid as described under "Experimental Procedures." Values represent the mean Ϯ S.D. of triplicate determinations.    (57), PLB-985 WT or mutated cells were differentiated by 0.5% DMF for 6 days before PMA-treated latex-bead stimulation, as described under "Experimental Procedures." For each translocation experiment, NADPH oxidase activity was measured the same day. The deformation of the TO-PRO 3 iodine-labeled nuclei allowed us to ascertain that the latex beads were indeed in the cell. As seen in Fig. 5, the phagocytosis of latex beads occurred in WT PLB985 cells as in empty-vector transfected X-CGD PLB-985 cells and in X ϩ CGD PLB-985 cells (D500G mutant) independently of oxidase activity. As expected, p47 phox protein was present in cytosol from all the tested PLB-985 cells.
No fluorescence was observed in WT PLB-985 cells when primary antibodies were omitted. The Red and Green fluorescence colors representing Nox2 and p47 phox , respectively, surrounded the phagosome membranes around the latex beads only in intact WT PLB-985 cells (Fig. 5). A yellow merged image indi-cated the co-localization of p47 phox and Nox2 in phagosome membranes. However, the D500G-Nox2 mutation disrupted p47 phox membrane translocation. Nox2 was localized in phagosome and plasma membranes, whereas p47 phox was uniformly distributed in the cytosol, confirming previously published results (30) and confirming this method. In K195A/E, R198E, R199E, and RR198199QQ mutations, as the RR198199AA mutation (data not shown) in the D-loop of Nox2, which totally inhibited NADPH oxidase activity, a strong green fluorescence was seen surrounding phagocytosed latex particles (Fig. 6A), suggesting that the p47 phox protein translocated normally to the phagosome membranes. Other mutations generated in the D-loop of Nox2 had no influence on NADPH oxidase assembly (data not shown). In contrast, in some mutants of the ␣-helical loop of the C-terminal of Nox2 (D484T, D500A, and D500R) with no oxidase activity, the p47 phox membrane translocation FIG. 5. Localization of p47 phox and Nox2 during activation of WT PLB-985 cells, empty vector or X ؉ CGD PLB-985 cells. A total of 5 ϫ 10 5 WT or transfected X-CGD PLB-985 cells was stimulated with PMA-treated latex beads for 15 min on day 6 after 0.5% DMF differentiation. Nox2 and p47 phox translocation was followed by confocal microscopy analysis as described under "Experimental Procedures." Polyclonal anti-p47 phox Ab and monoclonal 7D5 were used as primary antibodies, Alexa Fluor 488 and Alexa Fluor 546 F(abЈ) 2 fragments of goat anti-mouse IgG(HϩL) were used as second antibodies. Negative control was provided by incubating differentiated WT PLB-985 cells with phosphate-buffered saline, 5% albumin instead of primary antibodies. The D500G-Nox2 transfected X-CGD PLB-985 cells mimicked an X ϩ CGD case previously described (30). was totally disrupted, as in the RR9192EE mutant of the B loop of Nox2, previously studied by Biberstine-Kankade et al. (35). However, in the H490D-Nox2-transfected PLB-985 cells that had ϳ10% activity of original WT PLB-985 cells, a normal p47 phox translocation occurred as in WT Nox2-transfected PLB-985 cells (Fig. 6B). Similar results were obtained from in vitro translocation experiments using purified plasma membranes from transfected X-CGD PLB-985 cells after SDS and GTP␥S stimulation (Fig. 7). The in vitro p47 phox translocation was normal for K195A/E, R198E, R199E, and RR198199QQ/AA-Nox2-transfected PLB-985 cells similar to WT PLB-985 cells (data not shown) or WT Nox2 PLB-985 cells. In contrast, the p47 phox membrane translocation was defective in the D484T and in the D500A/R/G-Nox2 mutant cells. As expected, no p47 phox translocation occurred in the X-CGD PLB-985 cells (data not shown), in the empty vector transfected X-CGD PLB cells, or in the cells expressing the RR9192EE Nox2 mutant (Fig. 7A) (35).
Finally, diaphorase activity was measured in purified plasma membranes from WT or transfected mutant X-CGD PLB-985 cells (Fig. 8). Cytochrome b 558 from WT Nox2-transfected cells supported INT reductase activity to the same extent as was assessed in original WT PLB-985 cells, whereas in the empty vector-transfected X-CGD PLB-985 cells, this activity was null. While the K195A/E, R198E, R199E, and RR198199QQ mutants of the D-loop of Nox2 had a totally abolished oxidase activity, they exhibited 80% INT activity of the WT PLB-985 cells (Fig. 8). However, D484T and D500G mutations in the ␣-helical loop of the C-terminal of Nox2 led to the inhibition of NADPH oxidase and INT reductase activities. This was also observed in D500A/R mutant cells (data not shown).

DISCUSSION
The NADPH oxidase of phagocytic cells is an enzymatic complex assembled from membranous cytochrome b 558 and cytosolic components, p67 phox , p47 phox , p40 phox , and Rac2, upon cellular activation by microorganisms or/and inflammatory stimuli. The objective of this study was to examine the role of two cytosolic domains of Nox2 encompassing residues 191-200 (D-loop) and residues 484 -500 (C-terminal) of Nox2 in NADPH oxidase function and assembly (30,34). The D-loop has never been studied, whereas the C-terminal region of Nox2 was previously identified, by a predicted three-dimensional structure of Nox2, as an ␣-helical loop constituting a potential binding site for cytosolic factors (35). A point mutation D500G in this region in an X ϩ CGD case led to inhibition of cytosolic factor translocation to the plasma membrane of patient's neutrophils, confirming the previous hypothesis. However, in a second case of X ϩ CGD caused by a ⌬488 -497 gp91 phox deletion in the ␣-helical loop, p47 phox and p67 phox translocation was normal, whereas electron transfer from NADPH to FAD was disturbed (40). In this study, we used site-directed mutagenesis to probe the role of specifically charged amino acid residues within the D-loop and the C-terminal ␣-helical loop of Nox2, hypothesizing that electrostatic interactions between charged residues of Nox2 and cytosolic factors may be important for assembly of the active NADPH oxidase complex. To study the functional effect of these mutations we chose the granulocytic cell line PLB-985 as the cellular model, so as to be close to physiological events occurring in phagocytic cells.
We found that the D-loop and the C-terminal ␣-helical loop of Nox2 played a crucial role in oxidase activity. We demonstrated that the positive charge of Lys 195 was essential, because reversal or cancellation of it inhibited the NADPH oxidase activity (Table I). In addition, reversal of the electrostatic charge of Arg 198 or Arg 199 induced more inhibitory effects on NADPH oxidase activity than its cancellation. In contrast, when these two arginines were mutated simultaneously by opposite residues (RR198199EE), more than half of the oxidase activity could be measured in intact cells and in an in vitro reconstituted oxidase system (BCS). However, their replacement by two neutral residues (RR198199QQ) or two alanines (RR198199AA, data not shown) inhibited the oxidase activity (Table I). These data suggest that the total electrostatic charge at positions 198 and 199 is essential for NADPH oxidase activity. Interestingly, replacing Arg 199 with a glutamine as in Nox3, increased oxidase activity in vivo, pointing out that this amino acid could play a specific role in oxidase activation process. Lys 195 of the D-loop of Nox2 is highly conserved in Nox2 proteins from different species, including animals and plants, but it is highly variant in the human Nox family (Fig.  1), suggesting it has a special function in Nox2. In contrast, Arg 198 is absolutely conserved in all the members of the Nox family, whereas Arg 199 is replaced by neutral amino acids in Nox3 and Nox4. Within the past 4 years, Nox2 homologs have been identified in various tissues that have similar features to Nox2 and can generate modest superoxide (41,42). However, although their structural arrangement is similar to Nox2, molecular mechanisms involved in reactive oxygen species production and activity regulation is unclear. We found that the three mutants in which the D-loop of Nox2 was replaced by the same region of Nox1, Nox3, or Nox4 exhibited high NADPH oxidase activity (Table I), although the electrostatic charge in this region was changed (Fig. 2B). This showed that the D-loop of Nox analogs is functional for the Nox2 activation process, sug-gesting that this loop should have a similar function in the Nox family. Negative-charged amino acids Asp 484 , Asp 496 , and Asp 500 localized in the ␣-helical loop of the C-terminal of Nox2 are also essential for NADPH oxidase activity. Cancellation of the negative charge of Asp 484 by replacement with a neutral amino acid (Thr) abolishes the NADPH oxidase activity in intact transfected PLB-985 cells as in an in vitro reconstituted oxidase system (BCS). However, a basic residue change (His) at this position can restore most of the oxidase activity, suggesting that a charged amino acid (Asp or His) at position 484 is essential for NADPH oxidase activity. In contrast, reversal of the charge at position 490 (H490D) significantly affects superoxide production. The most important acidic amino acid in this C-terminal region is probably Asp 500 . Indeed all the mutations (D500A/R/G) of Nox2 inhibited in vivo and in vitro oxidase activity (Table I). In addition, the existence of two X ϩ CGD cases attributable to mutations in this region highlight the importance of this C-terminal domain on oxidase functioning (30,58). Two other Asp 500 mutations (D500Y/N) were reported in a CYBB data base (bioinf.uta.fi/CYBBbase/CYBBbasebrowser.html). Negative-charged amino acids, Asp 484 , Asp 496 , and Asp 500 , in the ␣-helical loop of the C-terminal of Nox2 are highly conserved in the ferredoxin-NADP ϩ reductase family, however the basic amino acid His 490 is present in Nox1, Nox3, and in Nox2 proteins from different species (Fig. 1). We should emphasize that any introduced mutation had no effect on the Nox2 maturation and stability because in all the transfected PLB-985 cells mutant Nox2 expression was normal (Fig. 3).
Interestingly, our data show the difference in the NADPH oxidase activity measured in some mutant-transfected X-CGD PLB-985 cells depending on the different type of stimulus employed to activate the enzyme. The slight H 2 O 2 overproduction observed in PMA-stimulated mutant PLB-985 cells expressing the R199Q and the D-loop Nox4 Nox2 proteins (super-mutant cells) was particularly exacerbated when these cells were activated with fMLP (Fig. 4A). In addition, the activation kinetics of the NADPH oxidase in WT PLB-985 cells, WT Nox2 PLB-985 cells, and D-loop Nox4 -Nox2 mutant PLB-985 cells activated with fMLP was more rapid and transient than with PMA. However, the H 2 O 2 production kinetics was not changed in all the fMLP-activated PLB-985 cells, whereas this production was more rapid in PMA-activated D-loop Nox4 -Nox2 mutant PLB-985 cells ( Fig. 4B and Table II) and in R199Q mutant cells only (data not shown). It is known that the kinetics of oxidase activation is not the same when human neutrophils are stimulated by fMLP or by PMA, because the signaling cascades triggered by these two agents are different. The high production of H 2 O 2 in the D-loop Nox4 and the R199Q-Nox2 mutant PLB-985 cells activated by PMA or fMLP may be due to a conformational change of the mutated Nox2 in the assembly of the NADPH oxidase complex, promoting a more efficient electron transfer to reduce molecular oxygen. If we admit that the H 2 O 2 production is the result of an equilibrium between an active and inactive state of the NADPH oxidase, we can speculate that the mutations lead to a conformational change in favor of the active state of the enzyme. The difference in the activation level of the oxidase complex in the super-mutant PLB-985 cells stimulated by either PMA or fMLP can be due to different phosphorylation states of the cytosolic factors induced by these stimuli (8), leading to a more or less efficient interaction between them and the mutated D-loop Nox4 and R199Q-Nox2 proteins. This hypothesis is supported by the fact that the oxidase activity of these two super-mutants in the reconstituted system (BCS), where no phosphorylation occurs, is not higher than in the WT-PLB-985 cells (Table I). The same difference between oxidase activity measured in intact PLB-985 FIG. 8. Diaphorase activity in a cell-free system assay using cellular membranes isolated from transfected X-CGD PLB-985 cells. Diaphorase activity was determined in the same conditions as described above, except that cytochrome c was replaced with INT. The data represent mean Ϯ S.D. of three separate experiments. cells and the in vitro reconstituted oxidase activity was previously observed in the study of the R91T/R92A mutant of the B loop of Nox2. Indeed the different kinetics of O 2 . production in this mutant activated by opsonized zymosan and PMA could be explained by different phosphorylation states of p47 phox (35). In conclusion, the phosphorylation state of the cytosolic factors occurring in vivo may be essential to generate a "super-efficient" oxidase complex, including the D-loop Nox4 or the R199Q-Nox2 proteins. However, it does not mean that the D-loop of Nox2 is directly involved in cytosolic factor binding. We should note that only mutations in the D-loop produce super-mutant PLB-985 cells, pointing out a crucial role of this region on NADPH oxidase activation. In addition, only two X-CGD mutations in the D-loop of Nox2 were described. The first case was a deletion in a codon corresponding to Arg 199 leading to the creation of a premature stop codon at position 213 (59). The second one was a X-CGD case originated from a S193F missense mutation, suggesting that this region is possibly involved in the conformational stability of cytochrome b 558 (60).
To elucidate the reason why the NADPH oxidase activity was abolished in some mutant PLB-985 cells, assembly of the oxidase complex was studied. Translocation of p47 phox was followed by confocal microscopy in intact differentiated PLB-985 cells during latex-bead activation (Figs. 5 and 6). Experiments were conducted in all the mutant PLB-985 cells, and the data obtained were confirmed by the in vitro translocation experiment (Fig. 7). We previously described the same type of approach using opsonized zymosan particles (38). Using latex beads opsonized with human IgG or serum AB, we found a significant background in nonphagocytosed particles due to unspecific recognition by the fluorescent secondary antibody. We decided to use the latex particle method coated with PMA, as previously described to purify phagosome membranes (61,62). Although K195A, K195E, R198E, R199E, and RR198199-QQ/AA mutations in the D-loop of Nox2 led to total oxidase inhibition, the p47 phox translocation process was normal, suggesting that basic amino acids of this region do not participate in the direct binding of oxidase cytosolic factors (Fig. 6A). This suggests that translocation of p47 phox and oxidase activation are two dissociated processes. However, mutant PLB-985 cells of the C-terminal of Nox2 (D484T, D500A/R/G mutations) with no oxidase activity had a p47 phox translocation defect (Fig. 6B). This result is consistent with the hypothesis that when the oxidase is in a resting state, the potential ␣-helical loop of 20 amino acid residues (Asp 484 -Gly 504 ) lies over the NADPH binding cleft. During oxidase activation, this loop is thought to move aside to allow NADPH to reach its binding site and deliver electrons to FAD. In this activated conformation, this loop is able to bind cytosolic factors. Yet we cannot speculate on the chronology of such events on the oxidase activation process. This result is also consistent with the assembly defect observed in an X ϩ CGD patient's neutrophils with a D500G missense mutation in Nox2 (30). However, it was reported that the deletion of Nox2 residues 488 -497 did not affect translocation of p47 phox and p67 phox but only electron transfer from NADPH to FAD (40). Meanwhile, in this last case, deletion of the small region encompassing residues 488 -497 conserves the acidic amino acids Asp 484 and Asp 500 , which seems to be essential to maintain the oxidase activity and the complex assembly. Asp 500 of Nox2 is probably a crucial charged amino acid because whatever mutation was introduced, the oxidase activity and the p47 phox translocation were inhibited.
To determine whether the inhibition of oxidase activity and/or the defect of complex assembly in mutant PLB-985 cells were associated with the electron transfer process from NADPH to FAD, we measured the INT reductase activity in purified plasma membranes from all the mutant transfected X-CGD PLB-985 cells (Fig. 8). This activity was abolished only in mutant PLB-985 cells of the ␣-helical loop 484 DESQAN-HFAVHHDEEKD 500 of Nox2, whereas a p47 phox translocation defect was also demonstrated. This suggests that the assembly of the NADPH oxidase complex is closely related to the electron transfer process from NADPH to FAD. Asp 500 is possibly involved in the direct binding with cytosolic factors, resulting in the liberation of the NADPH binding site, as proposed by Taylor et al. (28). In addition, this C-terminal region of Nox2 is near the potential binding site for the adenine unit of NADPH ( 504 GLKQ 507 ). However, the electron transfer process can be dissociated from the assembly of the enzyme. Indeed in a study of an X ϩ CGD case, which was induced by a mutation that converted Gln 507 -Lys 508 -thr 509 into His-Ile-Trp-Ala in a region near the potential binding site of NADPH, the translocation of both p47 phox and p67 phox to the membrane fractions of the patient's neutrophils was normal, indicating that this region is important for a correct electron transfer but not for the assembly of the oxidase complex (63). In addition, in the X ϩ CGD case resulting from a deletion of residues 488 -497 in Nox2 reproduced in the X-CGD PLB-985 cells, cytosolic factor translocation occurred normally, whereas electron transfer from NADPH to FAD was defective (40). In the mutants of the D-loop (K195A/E, R198E, R199E, and RR198199QQ/AA), although the assembly of the oxidase complex and the electron transfer from NADPH to FAD were normal, the NADPH oxidase activity was totally abolished. The D-loop is localized close to the V transmembranous passage in the N-terminal part of Nox2, which has been proposed as being involved in heme binding. We speculate that this region might be involved in electron transfer from the FAD to the hemes. This hypothesis is currently under investigation in our laboratory.
Finally, we can conclude that charged amino acids of the D-loop (Lys 195 , Arg 198 , and Arg 199 ) and of the C-terminal region encompassing residues 484 -500 (Asp 484 , His 490 , and Asp 500 ) are essential to maintain the NADPH oxidase activity in phagocytic cells. The D-loop of Nox1, -2, -3, and -4 is probably an important functional domain for the active conformational structure of Nox2, leading the electron transfer from FAD to oxygen, whereas the potential ␣-helical loop of the C-terminal tail of Nox2 is involved in the binding of cytosolic factors and in the electron transfer process from NADPH to FAD. The study of the relationship between these two events will allow us to better understand the molecular mechanisms of oxidase assembly.