Binding of FAD to Cytochrome b558 Is Facilitated during Activation of the Phagocyte NADPH Oxidase, Leading to Superoxide Production*

The superoxide-producing phagocyte NADPH oxidase can be reconstituted in a cell-free system. The activity of NADPH oxidase is dependent on FAD, but the physiological status of FAD in the oxidase is not fully elucidated. To clarify the role of FAD in NADPH oxidase, FAD-free full-length recombinant p47phox, p67phox, p40phox, and Rac were prepared, and the activity was reconstituted with these proteins and purified cytochrome b558 (cyt b558) with different amounts of FAD. A remarkably high activity, over 100 μmol/s/μmol heme, was obtained in the oxidase with purified cyt b558, ternary complex (p47-p67-p40phox), and Rac. From titration with FAD of the activity of NADPH oxidase reconstituted with purified FAD-devoid cyt b, the dissociation constant Kd of FAD in cyt b558 of reconstituted oxidase was estimated as nearly 1 nm. We also examined addition of FAD on the assembly process in reconstituted oxidase. The activity was remarkably enhanced when FAD was present during assembly process, and the efficacy of incorporating FAD into the vacant FAD site in purified cyt b558 increased, compared when FAD was added after assembly processes. The absorption spectra of reconstituted oxidase under anaerobiosis showed that incorporation of FAD into cyt b558 recovered electron flow from NADPH to heme. From both Kd values of FAD and the amount of incorporated FAD in cyt b558 of reconstituted oxidase, in combination with spectra, we propose the model in which the Kd values of FAD in cyt b558 is changeable after activation and FAD binding works as a switch to regulate electron transfer in NADPH oxidase.

The phagocyte NADPH oxidase catalyzes the generation of superoxide anions (O 2 . ) in response to invading microorganisms. Superoxide anions are precursors of a variety of reactive oxygen species that are utilized in killing bacterial and fungal pathogens (1,2). The physiological significance of the phagocyte NADPH oxidase in host defense is illustrated by the severe recurrent bacterial and fungal infections that occur in patients with chronic granulomatous disease whose phagocytes are unable to generate O 2 . (3,4). This NADPH oxidase complex consists of membrane-bound flavocytochrome b 558 , a heterodimer composed of gp91 phox and p22 phox , four cytosolic proteins, p40 phox , p47 phox , p67 phox , and the small GTPase Rac. In the resting cells, p40 phox , p47 phox , and p67 phox exist as a heterotrimeric complex that contains one copy of each protein (5). Activation of the oxidase is initiated by phosphorylation, which might induce the conformational changes by modulating intraand intermolecular interactions in the p47-p67-p40 phox complex. With these interactions, the activated oxidase is formed via assembly of the cytosolic regulatory proteins with cyt b 558 . 1 Activation of phagocyte NADPH oxidase can be mimicked in a cell-free system reconstituted with cyt b 558 , p47 phox , p67 phox , p40 phox , and Rac. These cytosolic proteins bind to cyt b 558 to form an active oxidase complex, which can generate O 2 . in the presence of NADPH. The activated NADPH oxidase is highly labile because of the dissociation of each subunit of the active complex (6). Recently, interactions among each subunit protein have been extensively studied; protein-protein interactions mediated by Src homolog 3 (SH3), tetratricopeptide repeat domain, and switch I have been reported (7)(8)(9). In the cytoplasm of resting cells, p40 phox , p47 phox , and p67 phox exist as a tight complex that can be purified by gel chromatography with an apparent molecular mass of 250 -300 kDa (10 -11). Sedimentation equilibrium and dynamic light-scattering experiments disclosed that the p47-p67-p40 phox complex contains one copy of each protein, and the apparent high molecular weight of this complex, as estimated by gel filtration studies, is because of an extended, nonglobular shape (5). gp91 phox in cyt b 558 represents the only catalytic component in NADPH oxidase, containing both redox centers, FAD, and two nonidentical hemes. Many studies of the heme in cyt b 558 have been reported as follows: (i) two hemes are located in gp91 phox (12); (ii) two nonidentical hemes with midpoint redox potentials of Ϫ265 and Ϫ225 mV (13); (iii) the heme has a low spin six coordination site (14); and (iv) the low spin state of the heme is essential for O 2 . generation in reconstituted NADPH oxidase (15). In contrast to the extensive studies on the heme, a relatively small amount of information on FAD (16 -18) has * This work was supported by a grant-in-aid for scientific research on priority areas and National Project on Protein Structural and Functional Analyses from the Ministry of Education, Culture, Sports, Science and Technology, Japan. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ Both authors contributed equally to this work. ‡ ‡ To whom correspondence should be addressed: School of Health Sciences, Sapporo Medical University, South-1, West-17, Chuo-ku, Sapporo, Hokkaido 060-8556, Japan. Tel.: 81-11-611-2111; Fax: 81-11-612-3617; E-mail: hfujii@sapmed.ac.jp. 1 The abbreviations used are: cyt b 558 , cytochrome b 558 ; HTG, N-heptyl-␤-thioglucoside; TC, ternary complex of recombinant cytosolic proteins p47-p67-p40 phox ; GTP␥S, guanosine 5Ј-(␥-thio)triphosphate; PBS, phosphate-buffered saline; Ni-NTA, nickel-nitrilotriacetic acid; AEBSF, 4-(2-aminoethyl)-benzenesulfonyl fluoride hydrochloride; SH3, Src homolog 3; gp, glycoprotein. been obtained, including such as the contents of flavins (FAD and FMN) in the oxidase and reflavination in flavocytochrome b 558 . One plausible model is that FAD is involved in O 2 . generation in NADPH oxidase and that exogenous FAD can restore O 2 . generating activities in the cell-free system reconstructed with FAD-depleted purified cyt b 558 and cytosolic proteins. Detailed information on the behavior of FAD molecules at the FAD-binding site of the NADPH oxidase and its physiological status is requisite to elucidate the electron transfer reactions among the two redox centers, but the behavior of FAD in intact cells is still unclear. Therefore, a simplified in vitro experiment excluding unfavorable side reactions and thermal instability of participating proteins is desired and required.
In the present work we describe the reconstitution of NADPH oxidase activity in a cell-free system with FAD-depleted purified cyt b 558 and recombinant FAD-free cytosolic regulatory proteins, i.e. the ternary complex (p47-p67-p40 phox ) and Rac. We report the effects of exogenously added FAD on the reconstituted activity of NADPH oxidase, and we show that enhancement of the activity by FAD is dependent on the assembly process of the reconstituted oxidase. Furthermore, we show that electron transfer from NADPH to heme under anaerobiosis is restored after incorporating exogenous FAD into FAD-depleted cyt b 558 . These analyses provide the direct evidence that the electron flow from NADPH to heme via FAD observed is directly related to the O 2 . -generating reaction in phagocyte NADPH oxidase.
Purification of Cyt b 558 -Neutrophils were obtained from pig blood, as described previously (19), and then treated with 2 mM diisopropyl fluorophosphate for 20 min on ice. Cyt b 558 was solubilized from the membrane fraction with HTG and was purified by using DEAE-Sepharose, CM-Sepharose, and heparin-Sepharose as described previously (14,20). The specific content of heme in the purified cyt b 558 preparation was 10.6 -12.4 nmol per mg of protein.
The full-length p47 phox was constructed by PCR from the plasmid of human p47 phox (7,21), using 5Ј-GGGAATTCATGGGGGACACCT-TCATCCGTC-3Ј as the forward primer and 5Ј-GGGTCGACCGACG-GCAGACGCCAGCTTC-3Ј as the reverse primer. The PCR product was digested with EcoRI and SalI, gel-purified, and ligated into EcoRI-and SalI-digested pGEX-His. The plasmids of full-length p47 phox with an amino-terminal glutathione S-transferase-tagged and a carboxyl-terminal hexahistidine-tagged (His-tagged) construct were transformed in Escherichia coli BL21(DE3) and were overexpressed. The cells were disrupted by sonication at 4°C in phosphate-buffered saline (PBS), pH 7.4, and 0.1 mM AEBSF. The protein was applied on a glutathione-Sepharose 4B column equilibrated with PBS buffer, pH 7.4. The bound protein was eluted with 25 mM Tris buffer, pH 8.0, and 25 mM reduced glutathione. The amino-terminal glutathione S-transferase tag in pro-teins were then removed by incubation with PreScission protease for 12 h at 4°C, and the digested proteins were dialyzed against 2 liters of 25 mM Tris buffer, pH 7.8, and 500 mM NaCl. The dialyzed proteins with carboxyl-terminal His tag were applied on a Ni-NTA column equilibrated with 25 mM Tris buffer, pH 7.8, 500 mM NaCl, and 5 mM imidazole. The bound protein was eluted with 25 mM Tris buffer, pH 7.8, 500 mM NaCl, and 250 mM imidazole. Fractions containing proteins were purified on a Superdex 200 gel filtration column and eluted with 25 mM Tris buffer, pH 7.4, and 150 mM NaCl. The protein was concentrated by Centriprep YM-30 to about 10 mg/ml.
The full-length p67 phox was amplified by PCR from the plasmid of human p67 phox (7,21), using 5Ј-GGGGATCCATGTCCCTGGTGGAGG-CCATC-3Ј as the forward primer and 5Ј-GGGTCGACCGACTTCTCTC-CGAGTGCTTTC-3Ј as the reverse primer. The PCR product was digested with BamHI and SalI, gel-purified, and ligated into BamHI-and SalI-digested pGEX-His. The plasmids of the full-length p67 phox with amino-terminal glutathione S-transferase-tagged and carboxyl-terminal His-tagged constructs were transformed and overexpressed, as shown in the preparation for the full-length p47 phox . From obtained cells, the full-length p67 phox was similarly purified, as shown above.
The full-length p40 phox were amplified by PCR from the plasmid of human p40 phox (22), using 5Ј-CCCCATGGCTGTGGCCCAGCAGCTG-C-3Ј as the forward primer and 5Ј-CCGAATTCATTATGGCATCGTGT-TGTAGACCCTGTAGTTG-3Ј as the reverse primer. The PCR product was digested with NcoI and EcoRI, gel-purified, and ligated into BamHI-and SalI-digested pProEX HTb (pPro-p40). The plasmid pT-Trx was a generous gift from Dr. S. Ishii (Laboratory of Molecular Genetics, The Institute of Physical and Chemical Research, RIKEN) (23). These plasmids of pPro-p40 and pT-Trx were co-transformed in E. coli BL21(DE3) and were overexpressed. The cells were disrupted by sonication at 4°C in 25 mM Tris buffer, pH 7.8, and 500 mM NaCl. The protein was applied on a Ni-NTA column equilibrated with 25 mM Tris buffer, pH 7.8, 500 mM NaCl, and 5 mM imidazole. The bound protein was eluted with 25 mM Tris buffer, pH 7.8, 500 mM NaCl, and 250 mM imidazole. Fractions containing proteins were purified on a Superdex 75 gel filtration column and eluted with 25 mM Tris buffer, pH 8.0, 150 mM NaCl. The amino-terminal His tag of p40 phox was removed by incubation with TEV protease for 12 h at 4°C. Further purification was carried out on a Superdex 75 gel filtration column eluted with 25 mM Tris buffer, pH 8.5, and 150 mM NaCl. The protein was concentrated by Centriprep YM-30 to around 10 mg/ml.
The DNA encoding Rac2 constitutively active form was obtained, as described previously (8), and was subcloned into pProEX HTb (pPro-Rac2). The plasmid pProRac2 was transformed in E. coli BL21(DE3) and was overexpressed. The cells were disrupted by sonication at 4°C in PBS including 0.1 mM AEBSF, pH 7.4. The protein was applied on a Ni-NTA column equilibrated with PBS and 5 mM imidazole, pH 7.4. After washing with a 20-fold column volume of PBS and 10-fold column volume of 25 mM Tris buffer, pH 7.4, including 5 mM imidazole, the bound protein was eluted with 25 mM Tris buffer, pH 7.4, 150 mM NaCl, and 250 mM imidazole. Fractions containing proteins were purified on a Superdex 75 gel filtration column and eluted with 25 mM Tris buffer, pH 8.0, 150 mM NaCl.
Preparation of Binary and Ternary Cytosolic Complexes-To obtain binary complexes of p47-p67 phox and p67-p40 phox , p67 phox was mixed with a 2-fold molar excess of p47 phox or p40 phox . After the mixtures had been incubated for more than 10 min at 4°C, the binary complexes were purified from excess uncomplexed protein (p47 phox or p40 phox ) by gel filtration on a Superdex 200 column using a buffer containing 50 mM Tris buffer, pH 7.4, and 150 mM NaCl. Fractions containing the binary complexes were concentrated by Centriprep YM-50 to around 10 mg protein/ml. Ternary complex of p47-p67-p40 phox was obtained by mixing the binary complex p47-p67 phox with a 2-fold molar excess of p40 phox . After the mixtures had been incubated for more than 10 min at 4°C, purification of ternary complex from excess uncomplexed p40 phox was performed by size exclusion chromatography on a Superdex 200 column with a buffer containing 25 mM Tris buffer, pH 7.4, 150 mM NaCl. The ternary complex was concentrated by Centriprep YM-50 to around 10 mg of protein/ml.
Preparation of Truncated Binary Complex-Fusion protein between truncated p67 phox -(1-242) and truncated p47 phox -(151-286) (p67-p47 phox truncated binary complex) was constructed by a two-step PCR technique. The first step was amplification of the truncated p67 phox gene and the truncated p47 phox gene. The DNA fragment encoding p67 phox -(1-242) was amplified from the full-length p67 phox gene by PCR using primers as follows: 5Ј-primer (primer A) was 5Ј-GGCCATATGTCCCT-GGTGGAGGCCAATCAGC-3Ј, including NdeI digestion site at the 5Ј-end, and 3Ј-primer (primer B) was 5Ј-CAGGATGATGGGGCCGGTGA-TGTCCCCTTCCAGAGCCCTGAAGATCTC-3Ј, including 24 bases at the 5Ј-end of p47 phox -(151-286). The DNA fragment encoding p47 phox -(151-286) was amplified from the full-length p47 phox gene with PCR using primers as follows: 5Ј-primer (primer C) was 5Ј-GACATCACCG-GCCCCATCATCCTG-3Ј, and 3Ј-primer (primer D) was 5Ј-GCGCGAA-TTCATTAGTCTTGCCCCGACTTTTGCAGGTA-3Ј, including the stop codon TAA and the EcoRI site. Two PCR products had 24 overlapping base pairs (5Ј-end of p47 phox -(151-286)). For fusion of p67 phox -(1-242) gene and p47 phox -(151-286) gene as the second step, five cycles of denaturation, annealing, and extension were carried out at 98°C (15 s) and 74°C (30 s). This fused gene was amplified by an additional 25 thermal cycles under the same conditions using primer A and D. The fused gene was then subcloned into the NdeI-EcoRI site of pET28a, and this plasmid was transformed in E. coli BL21(DE3), in which the truncated binary complex was overexpressed. The cells were disrupted by sonication at 4°C in PBS, pH 7.4, and 0.1 mM AEBSF. The protein was applied on a Ni-NTA column equilibrated with PBS, pH 7.4. The bound protein was eluted with 25 mM Tris buffer, pH 7.4, 150 mM NaCl, and 250 mM imidazole. Fractions containing proteins were purified on a Superdex 75 gel filtration column.

Reconstitution of NADPH Oxidase and Assay for O 2 . Generation-The
. generating activity of reconstituted NADPH oxidase complex was determined by the superoxide dismutase-inhibitable reduction of cytochrome c, as described previously (19). Reconstitution of Flavin-depleted Cyt b 558 with Native FAD and Assay of Noncovalently bound FAD-Purified FAD-depleted cyt b 558 (200 pmol) was mixed with TC (80 nmol) and Rac2 (40 nmol) in the presence or absence of 2 nmol of excess FAD and was incubated with sodium myristate (600 nmol) for 5 min at 25°C. After the above treatment, the mixture was passed through a PD-10 column (Amersham Biosciences) equilibrated with 50 mM phosphate buffer, pH 7.4, 0.6% HTG, and 10% glycerol to separate FAD-reconstituted cyt b 558 from excess free flavin. FAD-reconstituted cyt b 558 fractions were pooled and concentrated by freeze-drying and then boiled at 100°C for 15 min. After centrifugation at 15,000 rpm for 15 min, FAD in boiled extract of purified cyt b 558 fractions was quantified either by fluorometry with a Wallac 1420 ARVOsx (24,25) or the chemiluminescence method (26). Calibration was performed with standard solutions of FAD.
Measurement of Electron Flux through Redox Centers under Anaerobic Conditions-The electron fluxes through two redox centers, FAD and heme, in cyt b 558 were examined. The electron flux from NADPH to heme via FAD in reflavinated cyt b 558 was measured by following the absorbance at 558 nm with Unisoku Biospectrophotometer US-401. The extent of reduction in cyt b 558 was calculated by using a millimolar extinction coefficient of 21.6 at 558 -540 nm (27). Spectral changes of cyt b 558 were observed continuously over the 400 -600 nm range. Strictly anaerobic conditions were achieved by including glucose/glucose oxidase (40 units/ml) in an airtight cuvette (19).

Ternary Complex Used in This Study-
The soluble fulllength proteins of p47 phox , p67 phox , and p40 phox were obtained in relatively high yields (Ͼ20 mg of protein/liter), as described under "Experimental Procedures." Fig. 1A shows the gel filtration chromatogram of purified proteins; binary (p47-p67 phox ) and ternary (p47-p67-p40 phox ) complexes were fractionated without contamination of monomer components. The molecular mass of purified ternary complex was calibrated by the analytical gel filtration, and the obtained value was around 250 kDa, which was similar to that in the previous report (5). Fig. 1B shows silver-stained SDS-PAGE analysis of purified proteins. The purity of each sample is demonstrated by SDS-PAGE; binary (p47-p67 phox ), and ternary (p47-p67-p40 phox ) complexes are shown in Fig. 1B.
Reconstituted Activity of NADPH Oxidase-Cyt b 558 was purified from neutrophil membranes without any modification of the heme environment and was used for evaluation of O 2 . generating activity in the cell-free system. Heme and FAD associated with the cytochrome preparations at each purification step were measured and are summarized in Table I. Crude membrane contained the heme and FAD of cyt b 558 in a molar ratio of about 2.5:1. During the purification, the specific content of FAD decreased, whereas that of the heme increased. These purified cyt b 558 exhibited high O 2 . generating activity in the cell-free system with native cytosol, which shows that cyt b 558 used in this study was highly purified without any irreversible protein denaturation. The O 2 . generating activity of the oxidase was reconstituted in the cell-free system with recombinant, full-length FAD-free cytosolic proteins (Rac2, p40 phox , p47 phox , and p67 phox ) and purified cyt b 558 . Incubation of the cytosolic component individually and in combination or the TC with purified cyt b 558 was examined, and the purified cyt b 558 showed cytosolic proteindependent superoxide production. In the cell-free reconstituted system, the activity was totally dependent on the addition of both FAD and anionic amphiphile activators (such as myristic acid or arachidonic acid). Replacement of FAD with FMN abolished the stimulation of superoxide production. The remarkably high O 2 . generating activity of the reconstituted oxidase, over 100 mol of O 2 . /s/mol of heme, was obtainable in the system with TC. The reconstituted oxidase activity with TC (p47-p67-p40 phox ) was compared with the binary complex p47-p67 phox . The activity with TC was 9.9 Ϯ 2.8% (p Ͻ 0.05) higher than with p47/p67 phox , showing that p40 phox is not essential for reconstitution of the activity. Previous findings (28,29) were also confirmed in the cell-free reconstituted system with the recombinant TC. In the case of the other binary complexes, p47-p40 phox or p67-p40 phox , any superoxide generating activity was not detected under similar reaction conditions. Stability of the Reconstituted NADPH Oxidase Activity-The purified ternary complex, p47-p67-p40 phox , exists at a tight complex of 1:1:1 stoichiometry in solutions (5) and is stable for at least 24 h without any dissociation of each component, as judged from the gel filtration experiment of purified ternary complex. Fig. 2 shows the stability of the oxidase activity reconstituted with TC at 25°C. When the reconstituted oxidase was left for 60 min at 25°C, the activity of the oxidase with TC was about 90% of its initial level, and its stability was remarkably elongated compared with that in the native cytosol. The half-lives of the activity with TC and the native cytosol were about 10 min and Ͼ12 h, respectively. Because the lability of the oxidase activity is caused by the dissociation of each protein in the oxidase complex (6), the long stability of the reconstituted activity with TC might be attributable to the stable structure of the reconstituted oxidase complex.
Titration with FAD of Superoxide Generating Activity of the Reconstituted NADPH Oxidase-In earlier experiments using the cell-free system, native cytosolic fractions were used with either purified or crude cyt b 558 . There have been several reports on the titration of reconstituted oxidase activity with FAD, but an accurate stoichiometric analysis of FAD for activity was quite difficult because of the presence of free FAD or loosely bound FAD to the cytosolic proteins. Therefore, in this study, a titration study of reconstituted activity was carried out, employing the FAD-free p47-p67-p40 phox complex and Rac and purified cyt b 558 with different FAD contents. Fig. 3 shows the effect of exogenously added FAD on the production of superoxide by purified cyt b 558 in the presence of different con-centrations of FAD-free TC. The purified cyt b 558 lost most of its FAD, compared with that of partially purified cyt b 558 , and the content of heme and FAD was at a molar ratio of about 1:0.01-1:0.03 (94 -98% of FAD was lost), assuming that the ratio of heme to FAD was 2:1. The reconstituted oxidase activity was enhanced remarkably by the addition of exogenous FAD and reached a plateau when the ratio of FAD to heme exceeded 2.0. The minimal amount of exogenous FAD required to manifest optimal activity did not vary with increasing amounts of TC. Replacement of FAD with FMN abolished the stimulation of superoxide production. These results indicate that the enhancement of the activity by the addition of FAD depends on the ratio of added FAD to heme present in cyt b 558 but not on that of added FAD to TC.
Effects  a Superoxide generating activity was evaluated in a cell-free system reconstituted with native cytosol and membrane fractions or purified cyt b 558 at 25°C. 4A in the presence and absence of FAD. In the reconstituted oxidase system with purified cyt b 558 devoid of FAD, the activity was enhanced remarkably by the addition of FAD, but the time of addition of FAD to the system was reflected in the resultant superoxide generating activity. When FAD was added to the mixture of cytosolic proteins and purified cyt b 558 before treatment of such a mixture with anionic amphiphiles, i.e. FAD was present during the assembly process, the oxidase activity increased nearly 10 times compared with the activity without FAD. However, the activity increased only 1.4 times when the same amount of FAD was added after the assembly process, which suggests that FAD was not fully incorporated into the reconstituted oxidase, probably because of the steric hindrance at the FAD-binding site of cyt b 558 in the activated oxidase complex.
Determination of FAD Incorporated into Cyt b 558 of Reconstituted NADPH Oxidase-In order to confirm the hypothesis shown above, the amount of FAD incorporated into the purified cyt b 558 was determined before and after cell-free activation, i.e. the treatment with anionic amphiphile, in the presence of exogenous FAD with or without TC. Table II shows the ratio of FAD to heme in purified cyt b 558 , together with the resultant activity of the reconstituted oxidase. In these measurements, not only free FAD but also FAD weakly bound to cyt b 558 and/or TC were separated from cyt b 558 by the gel filtration technique. Thus, only FAD molecules, which were tightly bound to cyt b 558 , were evaluated (Table II). Before incubation with exogenous FAD, the ratios of FAD to heme in purified cyt b 558 used in this study were in the range of 0.01-0.05:1. The amount of incorporated FAD did not increase merely by the incubation with FAD, unless treated with anionic amphiphiles in the presence of TC, which indicates that incubation of cyt b 558 with FAD is not sufficient to incorporate FAD into the FAD-binding site in cyt b 558 . On the contrary, when exogenous FAD was added before the assembly process, the amount of FAD incorporated into cyt b 558 was remarkably enhanced, compared with the results when FAD was added after the assembly process. A similar experiment was carried out in the absence of TC. Even in the absence of TC, exogenous FAD molecules were incorporated into cyt b 558 by the treatment with amphiphiles, but the amount of incorporated FAD was only one-third that in the presence of TC. These results, shown in Table II, strongly indicate that the treatment of the oxidase by the anionic amphiphile is requisite for the incorporation of exogenous FAD into FAD-depleted cyt b 558 . During the treatment with anionic amphiphiles, intra-and inter-molecular conformational change took place, which might have induced the structural change in the FAD-binding site in cyt b 558 and the protein-protein interaction of cyt b 558 and TC, followed by the tight binding of FAD to cyt b 558 .
In order to test this view, the dissociation constant (K d ) of FAD in cyt b 558 of reconstituted oxidase was estimated. Under the assumption that the superoxide generating activity is proportional to the amount of FAD incorporated cyt b 558 -TC complex, the reconstituted activity was plotted as functions of added FAD (Fig. 5). The K d value of FAD in cyt b 558 of the reconstituted oxidase was estimated from Fig. 5 as 0.94 nM, which indicates that FAD molecules were strongly bound to cyt b 558 after the treatment with amphiphiles in the presence of TC. It is important to study whether the tight binding of FAD to purified cyt b 558 observed in this cell-free reconstitution system has physiological relevance. Therefore, unbound and bound FAD in activated neutrophils was compared with resting cells (Table III). Table III shows that the total amount of FAD and heme in membranes did not change upon activation of cells with phorbol 12-myristate 13-acetate, but a distinct increase in the amount of bound FAD was observed, indicating that FAD binds to cyt b 558 tightly in the activated NADPH oxidase. The cell-free activation system contained purified cyt b 558 , Rac2 (800 nM), and TC (F, 250 nM) or native cytosol (OE, 0.2 mg) and was treated with an optimal amount of sodium myristate (16 M). After the above treatment of the cell-free system for 5 min, the mixture was left at 25°C for a given time. Then the mixture was supplemented with NADPH (0.1 mM), and its activity was measured, as described under "Experimental Procedures." Activities were expressed as the percentages of initial activities. Results are means Ϯ S.D. from three experiments.
Next, we examined whether the conformational change in cyt b 558 occurring during the treatment of the oxidase proteins with amphiphiles affects the incorporation of FAD into cyt b 558 by using fused protein between truncated p67 phox -(1-242) and truncated p47 phox -(151-286). Because this fused protein has no polybasic region, the SH3 domain in p47 phox of the fused protein can bind to the proline-rich region in p22 phox of cyt b 558 without the addition of anionic amphiphile (7,30), which results in the activation of the oxidase and the generation of superoxide. The results for the oxidase reconstituted with the fused protein and purified cyt b 558 are shown in Fig. 4B. The oxidase reconstituted with the fused protein can generate superoxide without anionic amphiphiles, but the amount of superoxide generated is greatly enhanced by the addition of anionic amphiphiles, suggesting that the conformational change in cyt b 558 by the treatment with anionic amphiphiles activator plays an important role in incorporation of FAD into cyt b 558 . It should be noted that in both cases of TC and the fused protein the reconstituted oxidase activity is not enhanced by FAD added after the treatment of the oxidase with anionic amphiphiles, i.e. after assembly processes. From the results in Table  II showing that the efficacy of incorporating FAD into cyt b 558 depends not only on the treatment with amphiphiles but also on the coexistence of TC or the fused protein, it appears that cytosolic proteins (or TC) play an important role in retaining FAD at the FAD-binding site of cyt b 558 , such as by masking the FAD-binding site of cyt b 558 in the activated oxidase.
Electron Transfer Reaction in FAD-incorporated NADPH Oxidase-The electron transfer reactions in reconstituted NADPH oxidase were compared under both aerobic and anaerobic conditions. As shown in Figs. 2-5, under aerobic conditions the superoxide generating activity of reconstituted NADPH oxidase with cyt b 558 and TC was remarkably enhanced by the incorporation of FAD to cyt b 558 . Electron transfer reactions in the reconstituted NADPH oxidase were also examined under anaerobic conditions (Fig. 6). Electron transfer from NADPH to TABLE II Reconstituted activity of NADPH oxidase in a cell-free system The cell-free reaction system contained purified cyt b 558 (200 pmol), TC (80 nmol), and Rac2 (40 nmol) and was treated with optimal amounts of sodium myristate. FAD was added into the above mixture before or after treatment with anionic amphiphiles. After free FAD was removed from the reconstituted oxidase by gel filtration, bound FAD was released from the reconstituted oxidase by its denaturation at 98°C for 5 min, and the amount of FAD tightly bound to cyt b 558 was determined by fluorometry. The data are expressed as means Ϯ S.D. from five experiments.
a "Before" indicates that FAD was added to the mixture of purified cyt b 558 and cytosolic proteins before treatment with amphiphiles. "After" indicates that FAD was added to the above mixture after treatment with amphiphiles, i.e. after assembly of the proteins. b Nondetectable. c Purified cyt b 558 was mixed with exogenous FAD without any treatment with anionic amphiphiles.

FIG. 4. Effect of the addition of FAD on assembly processes of the reconstituted NADPH oxidase.
A, the cell-free mixture containing purified cyt b 558 (1 pmol), TC (250 nM), and Rac2 (800 nM) was treated with an optimal amount of sodium myristate. The reconstituted NADPH oxidase activity was determined with or without FAD (10 pmol). In this experiment, FAD was added either before or after the assembly process. Graphs and error bars represent the means Ϯ S.D. from three experiments. B, the same experiment was carried out in the reconstituted system with the fused protein (500 nM) instead of TC. In this reconstituted system, NADPH oxidase activities were measured with or without anionic amphiphile activator (sodium myristate). Other experimental conditions were the same as in A. 5. Estimation of the K d of FAD in purified cyt b 558 of reconstituted NADPH oxidase. The cell-free reconstitution system contained purified cyt b 558 (1 pmol), TC (250 nM), Rac2 (800 nM), and increasing amounts of FAD, and the reaction mixture was treated with an optimal amount of sodium myristate. The reconstituted activities were evaluated in a similar way to that in Fig. 3, and these values were plotted as a function of FAD concentration. The K d value of FAD in purified cyt b 558 of activated oxidase is obtained by fitting reconstituted activities to the one-site saturation curve. heme in purified cyt b 558 was measured by the reduction of heme in cyt b 558 under anaerobiosis. When purified cyt b 558 was reconstituted with TC in the absence of FAD, no reduction of heme in cyt b 558 took place even in the presence of NADPH. The electron transfer path might have been broken by the removal of FAD from cyt b 558 . When purified cyt b 558 was reconstituted with TC in the presence of FAD, however, a reduction of heme in reconstituted cyt b 558 occurred because of the addition of NADPH. Fig. 6 shows the difference spectrum of native FAD-reconstituted cyt b 558 obtained by subtracting the oxidized spectrum from the reduced spectrum with NADPH, and the kinetics of the reduction of the heme in cyt b 558 are shown in the inset. The obtained spectrum is a typical cyt b 558 reduced minus oxidized difference spectrum, showing the location of the ␣ band (558 -559 nm) and Soret band (426 -427 nm) peaks. The appearance of reduced heme in FAD-reconstituted cyt b 558 by NADPH strongly suggests that the electron transfer ability of the reconstituted oxidase is recovered after incorporation of FAD into FAD-depleted purified cyt b 558 . The inset in Fig. 6 demonstrates that the heme in FAD-reconstituted purified cyt b 558 is reduced in a time-dependent manner by NADPH in anaerobiosis. Although the reduction rate of heme is slower than the superoxide-generating rate of the reconstituted NADPH oxidase (19,31), the results indicate that electrons provided from NADPH are transferred via the same electron transfer chain in NADPH oxidase under aerobic and anaerobic conditions.

DISCUSSION
The superoxide-generating phagocyte NADPH oxidase is dormant in resting cells and becomes active upon cell stimulation by pathogens or amphiphiles in vivo. The switching from a resting to an activated state in the oxidase system is precisely controlled by several different mechanisms, for example by translocation of the cytosolic proteins to the membrane and the conformational change in cytosolic proteins induced by phosphorylation reaction during the activation process. In resting cells, three cytosolic proteins, p47 phox , p67 phox , and p40 phox , exist as a tight complex, and recently these proteins were found to form a tight complex with 1:1:1 stoichiometry in vitro (5). In the present study, we prepared the ternary cytosolic protein complex, p47-p67-p40 phox (TC), together with several binary complexes, such as p47-p67 phox , p47-p40 phox , and p67-p40 phox , from each full-length recombinant cytosolic protein. By using these binary and ternary complexes, the reconstituted activity of NADPH oxidase was examined in the presence of purified cyt b 558 . The reconstituted activity with TC was higher and more stable than that with any binary complexes or the native cytosol.
Anionic amphiphiles, such as arachidonic acid (32) and myristic acid (33), can activate the phagocyte NADPH oxidase not only in intact cells but also in a cell-free system via assembly of the cytosolic regulatory proteins p47 phox , p67 phox , and Rac with the membrane-associated cyt b 558 consisting of p22 phox and gp91 phox . It is evident that the amphiphiles interact with each protein and induce conformational changes before assembly of the oxidase complex; anionic amphiphiles interact with p47 phox containing two SH3 domains, which specifically bind to a proline-rich region in the cytoplasmic tail of p22 phox . The induced p47 phox -p22 phox interaction likely plays a crucial role in the activation of the NADPH oxidase. Besides these interactions, anionic amphiphiles also interact with other oxidase proteins, which appears to render the protein into a conformation capable of oxidase activity. The interaction between amphiphiles and proteins is dependent on their hydrophobic-hydrophilic FIG. 6. Reduced minus oxidized difference spectra of native FAD-reconstituted cyt b 558 . Spectrum A, difference absorption spectrum of purified FAD-depleted cyt b 558 . Purified cyt b 558 (100 pmol) and TC (2000 pmol) were treated with sodium myristate in 50 mM phosphate, pH 7.0, without exogenous FAD. After 5 min of incubation, NADPH (0.2 mM) was added to the reconstituted mixture under anaerobic conditions using glucose-glucose oxidase as described under "Experimental Procedures." 10 min after addition of NADPH to the reconstructed mixture, a different absorption spectrum was calculated by subtracting the oxidized absorption spectrum (before addition of NADPH) from reduced spectrum (after addition of NADPH). Spectrum B, the same as A except that the reconstituted mixture with the purified cyt b 558 and TC was examined in the presence of FAD (1000 pmol). After reconstitution of the oxidase system, it was passed through a Sephadex G-25 column to separate FAD-reconstituted cyt b 558 from excess free FAD. Kinetics of the reduction of heme in FAD-reconstituted cyt b 558 are shown in the inset. The difference spectra were recorded over the indicated times, and the extent of reduction of heme was assayed by measuring the extent of absorption at 558 nm relative to the trough at 540 nm. Total reduction was assayed by the addition of sodium dithionite. balance, and common amphiphiles tend to bind not only to cytosolic proteins but also to membrane-bound cyt b 558 . Anionic amphiphiles can bind to cyt b 558 similar to other cytosolic proteins, and thus the binding of anionic amphiphiles induces a structural change in cyt b 558 that leads to the appearance of high spin heme, as evidenced by low temperature EPR studies (15,34). From these results, it is likely that not only the heme environment but also the flavin-binding site in cyt b 558 is modulated by the treatment of cyt b 558 with anionic amphiphiles in the cell-free superoxide-generating system. In the present study, a K d of nearly 1 nM was calculated for the binding of FAD to the purified FAD-depleted cyt b 558 , which shows that FAD is tightly bound to cyt b 558 of activated NADPH oxidase. If the K d value of FAD in cyt b 558 of resting intact cells is the same as that in the activated oxidase and is in the nanomolar range, FAD should be tightly bound to cyt b 558 in intact cells. It is very important to understand the FAD-binding status in the NADPH oxidase in physiological conditions. There are at least two models to explain our results on the K d of FAD. For model 1, the K d value of FAD in cyt b 558 is not different between the resting and activated states. Therefore, purified cyt b 558 lost most of its FAD because of the change in its charge, lipid environment, etc. during the purification procedures. For model 2, the K d value of FAD in cyt b 558 is decreased in the activated oxidase during the activation process, compared with that in the resting oxidase. In the resting status, FAD is in equilibrium between being bound and unbound to cyt b 558 . However, once the oxidase is activated, the structure of the FAD-binding site is modified, resulting in the tight binding of FAD to cyt b 558 . Cytosolic proteins also help maintain FAD at the FAD-binding site of activated oxidase.
Flavin contents and the ratio of FAD to heme in crude membranes from resting neutrophils have been reported (18, 26,[35][36]. It is generally accepted that the ratio of heme to FAD ranges from 2:1 to 3:1, and in our laboratory the value of 2.5:1 was obtained, as shown in Table I. One should note that this ratio is obtained in the crude membranes and that this ratio does not indicate the ratio in cyt b 558 . Segal et al. (35) reported that the FAD content of crude membranes from cells of patients with X-linked chronic granulomatous disease lacking cyt b 558 was one-quarter that of normal subjects and also that these FAD molecules did not belong to NADPH oxidase, which exists even in uninduced HL60 cells. Yoshida et al. (37) also published similar results, although they claimed that onethird of all FAD in crude membranes of phagocytes is not related to NADPH oxidase. Several FAD proteins had been reported as possible candidates responsible for NADPH oxidase, but these candidates were negated when cyt b 558 was found to be a flavocytochrome (35,38). Therefore, one-third or one-quarter of FAD in phagocytes might be derived from flavoproteins that are not related to NADPH oxidase. Thus, according to the above considerations, the ratio of heme to FAD in NADPH oxidase of crude membranes becomes 3.5:1-4:1. Furthermore, by using the same chemiluminescence method as Yoshida et al. (26), the ratio of free and bound FAD in crude membranes was found to be 6:4, suggesting that more than half of FAD in crude membranes is free and unbound, and more than half of the FAD sites in cyt b 558 seems to be unoccupied. As shown in Table III, the amount of FAD bound to the oxidase in stimulated neutrophils increased, compared with the resting cells, although the amount of heme in the oxidase was constant regardless of the stimulation of cells. All these facts strongly suggest that the FAD-binding sites in cyt b 558 of resting cells are not fully occupied with FAD; FAD is in equilibrium between being unbound and bound to cyt b 558 because of the weak binding capability of FAD in cyt b 558 in the resting state. Fig. 7 depicts a schematic illustration of a possible model for the FAD-binding site of cyt b 558 in both the resting and activated states of NADPH oxidase. In the resting state, some of the FAD molecules are loosely bound to cyt b 558 , i.e. the equilibrium between being unbound and bound to the FAD-binding site of cyt b 558 . Once the NADPH oxidase is treated with anionic amphiphile activators, the structure of the FAD-binding site of cyt b 558 is modified so as to fit FAD tightly into its FAD-binding site, and free unbound FAD thus binds to cyt b 558 tightly. Furthermore, this occupation of the FAD-binding site in cyt b 558 is followed by the association of the cytosolic proteins. This masking and protection of the FAD-binding site by the cytosolic protein complex is important for several reasons. By making this structure, bound FAD is not easily released from cyt b 558 . Therefore, the reconstituted oxidase activity is very stable even though the active oxidase is left at room temperature for many hours, as shown in Fig. 2. Furthermore, by masking the FAD site with the cytosolic protein complex, the FAD site might be protected from chemicals that are commonly used as inhibitors, such as cyanide, carbon monoxide, and pyridine (36).
From both the K d value of FAD and the ratio of heme to FAD in purified cyt b 558 obtained in this study, we derived the model in which K d of FAD is changeable after activation of the NADPH oxidase (Fig. 7). The change in the K d value of FAD is physiologically important in terms of avoidance of accidental FIG. 7. A schematic illustration of a possible model for the FAD-binding site of cyt b 558 in both the resting and activated states of NADPH oxidase. In the resting state, FAD is loosely bound to cyt b 558 , and FAD is in equilibrium between being bound and unbound to cyt b 558 . When the oxidase is treated with anionic amphiphile activators, the FAD-binding site in cyt b 558 is slightly distorted, which induces fitting of FAD molecules into the FAD-binding site of cyt b 558 . After the tight binding of FAD to cyt b 558 , the FAD-binding site is masked with cytosolic protein complex; therefore, FAD is not easily released from the activated oxidase. Furthermore, common inhibitors of electron transfer reactions are unable to gain access to the redox centers, FAD and heme, in cyt b 558 . production of reactive oxygen species, because in the resting state it is unlikely that electrons will flow into the electron transfer chain, where FAD is loosely or improperly incorporated into cyt b 558 . Further studies will be necessary to examine the mechanism by which FAD acts as the switch for activating NADPH oxidase.