Tripartite Chimeras Comprising Functional Domains Derived from the Cytosolic NADPH Oxidase Components p47phox, p67phox, and Rac1 Elicit Activator-independent Superoxide Production by Phagocyte Membranes

The superoxide-generating NADPH oxidase is converted to an active state by the assembly of a membrane-localized cytochrome b559 with three cytosolic components: p47phox, p67phox, and GTPase Rac1 or Rac2. Assembly involves two sets of protein-protein interactions: among cytosolic components and among cytosolic components and cytochrome b559 within its lipid habitat. We circumvented the need for interactions among cytosolic components by constructing a recombinant tripartite chimera (trimera) consisting of the Phox homology (PX) and Src homology 3 (SH3) domains of p47phox, the tetratricopeptide repeat and activation domains of p67phox, and full-length Rac1. Upon addition to phagocyte membrane, the trimera was capable of oxidase activation in vitro in the presence of an anionic amphiphile. The trimera had a higher affinity (lower EC50) for and formed a more stable complex (longer half-life) with cytochrome b559 compared with the combined individual components, full-length or truncated. Supplementation of membrane with anionic but not neutral phospholipids made activation by the trimera amphiphile-independent. Mutagenesis, truncations, and domain replacements revealed that oxidase activation by the trimera was dependent on the following interactions: 1) interaction with anionic membrane phospholipids via the poly-basic stretch at the C terminus of the Rac1 segment; 2) interaction with p22phox via Trp193 in the N-terminal SH3 domain of the p47phox segment, supplementing the electrostatic attraction; and 3) an intrachimeric bond among the p67phox and Rac1 segments complementary to their physical fusion. The PX domain of the p47phox segment and the insert domain of the Rac1 segment made only minor contributions to oxidase assembly.

Phagocytes produce reactive oxygen radicals, part of their microbicidal arsenal, by means of a tightly regulated enzyme complex commonly referred to as NADPH oxidase. At the origin of all oxygen radicals is the superoxide anion (O 2 . ), generated by the NADPH-derived one-electron reduction of molecular oxygen. The O 2 . -generating NADPH oxidase complex (briefly "oxidase") consists of a membrane-associated flavocytochrome (cytochrome b 559 ) comprising two subunits (gp91 phox and p22 phox ) and four cytosolic components (p47 phox , p67 phox , p40 phox , and small GTPase Rac1 or Rac2) (reviewed in Refs. [1][2][3]. Electron flow from NADPH to oxygen occurs along three redox stations, all of which are located on gp91 phox : the NADPH-binding site, FAD, and two non-identical hemes. It is assumed that initiation of electron flow is the consequence of a conformational change in gp91 phox induced by its interaction with p67 phox . The region in p67 phox presumed to be involved in such interaction is known as the "activation domain" and consists of residues 199 -210 (4). It has been suggested that the roles of p47 phox and Rac are to serve as carriers of p67 phox to the membrane or as membrane anchors for p67 phox to enable the correct juxtaposing of the activation domain on p67 phox to its target site on gp91 phox (4 -7). We have recently proposed that an additional function of p47 phox is to augment the stability of the assembled complex (8). The carrier/anchor for p67 phox function of p47 phox is based on two sets of interactions: a tail-to-tail interaction between a proline-rich region (PRR) 3 at the C terminus of p47 phox and the C-terminal Src homology 3 (SH3) domain of p67 phox (9) and between the two SH3 domains of p47 phox and a PRR in the p22 phox subunit of cytochrome b 559 (10,11). Further links of p47 phox to the membrane environment of cytochrome b 559 are provided by the binding of the Phox homology (PX) domain at its N terminus to specific phosphoinositides (12) and possibly also by the binding * This work was supported by the Julius Friedrich Cohnheim-Minerva Center of a C-terminal domain to gp91 phox (13). The mechanism responsible for the binding of p47 phox to p22 phox has been worked out in great detail and consists of the exchange of an intramolecular autoinhibitory bond between the SH3 tandem and a polybasic region at the C terminus of p47 phox for an intermolecular bond between the same SH3 tandem and a PRR at the C terminus of p22 phox . The "opening" of the intramolecular bond is the result of phosphorylation of critical serine residues in the polybasic region, found to occur in the course of neutrophil activation (14,15). The carrier/anchor function of Rac rests on the interaction between the pre-switch I and switch I regions of Rac1 and the tetratricopeptide repeat (TPR) domain in p67 phox at one end (16,17) and the binding of Rac via its polybasic and prenylated C terminus to negatively charged membrane phospholipids (18) and possibly to gp91 phox (19,20) at the other end. Rac has first to be liberated from its bond with the regulatory protein Rho GDP dissociation inhibitor, which keeps Rac in an inactive, GDP-bound conformation in the cytosol (reviewed in Ref. 21). There is evidence for the involvement of free (22) or membrane-associated (23) anionic phospholipids and possibly a membrane-bound guanine nucleotide exchange factor (24) in causing the dissociation of the Rac-Rho GDP dissociation inhibitor complex.
There is evidence for the need for a conformational change to occur in p67 phox to allow a productive interaction with the catalytic subunit of cytochrome b 559 , gp91 phox . This change is thought to be the consequence of the binding of the GTPbound form of Rac to p67 phox (25) or of the relief of autoinhibition by the C terminus of p67 phox containing the two SH3 domains (26). It is not clear whether the result of the conformational change in p67 phox is to augment the actual binding of p67 phox to gp91 phox (6) or to endow p67 phox with an ability to elicit electron flow in gp91 phox (27).
A conceptual and methodological advance in our understanding of oxidase activation was the development of in vitro cell-free systems. In these, phagocyte membranes or purified cytochrome b 559 preparations are exposed to total phagocyte cytosol or to purified or recombinant cytosolic components in the presence of an activator, represented by an anionic amphiphile such as arachidonate, SDS, or phosphatidic acid (28 -33). The availability of recombinant proteins for all cytosolic components made it possible to identify functionally significant domains by combining mutagenesis with assay of the mutated components in what became known as the semirecombinant cell-free system (34). Cell-free systems reproduce the essential steps in oxidase assembly as they occur in vivo, as demonstrated by the fact that anionic amphiphiles mimic the effect of phosphorylation of p47 phox , leading to the relief of autoinhibition (35). A further advance was the design of cell-free systems in which oxidase activation takes place in the absence of an amphiphilic activator. Such systems are based on C-terminal truncation of both p47 phox and p67 phox (26), on prenylation of Rac (5), or on enrichment of the phagocyte membrane with negatively charged phospholipid (37,38).
A novel approach to the study of protein-protein interactions in oxidase assembly was initiated by the design of chimeric constructs consisting of selected segments derived from two cytosolic components. The first of these was a fusion of p47 phox (residues 1-286) with p67 phox (residues 1-210) (37). This was followed by the independent description by two groups of p67 phox -(1-210 or 1-212)-Rac1 chimeras (39,40), which could be prenylated at the C terminus (41). Chimeras of cytosolic components are, in general, characterized by EC 50 values lower than those of the non-fused proteins. There is less agreement about the effect of fusion on V max values; one group reported higher activities for both p47 phox -p67 phox and p67 phox -Rac1 chimeras in a cell-free system (37,39), whereas lower activities for p67 phox -Rac1 chimeras were found by us (40). Fusion of p67 phox and Rac1 reduces the dependence of oxidase activation on p47 phox (39 -41). Finally, prenylation of p67 phox -Rac1 chimeras enables oxidase activation in the absence of amphiphile and further reduces its dependence on p47 phox (41).
In this study, we describe the design and bacterial expression, in the form of soluble proteins, of tripartite chimeras consisting of selected segments of p47 phox , p67 phox , and Rac1, which we call "trimeras." The prototype trimera, resulting from the fusion of p47 phox -(1-286), p67 phox -(1-212), and full-length Rac1 (amino acids , was found to act as a potent amphiphile-dependent oxidase activator upon addition to phagocyte membrane or purified cytochrome b 559 . Modifying the phospholipid composition of the membrane by supplementation with anionic phospholipids enabled oxidase activation by the trimera in the absence of an amphiphile. Truncations, domain replacements, and point mutations applied to critical regions in the three segments composing the prototype trimera led to the acquisition of novel information on the participation of specific sequences and residues in oxidase assembly.
Construction of Chimeric Proteins-See supplemental "Experimental Procedures." Expression and Purification of Recombinant Proteins-All recombinant proteins used in this work, with the exception of full-length p47 phox , were expressed in and isolated from Escherichia coli BL21-CodonPlus TM (DE3)-RIL (Stratagene). Rac1 was produced in E. coli as described previously (43). Chimera 3 (p67 phox -(1-212)-full-length Rac1) was expressed and purified as described (40). Full-length p67 phox , p67 phox -(1-212), and p47 phox -(1-286) were expressed as glutathione S-transferase (GST) fusion proteins and purified by batch affinity chromatography on glutathione-agarose (Sigma), followed by thrombin cleavage (44), as described previously (45). p47 phox was prepared in baculovirus-infected Sf9 cells as described (43). The prototype trimera, the trimera lacking the Rac1 C terminus (Rac1⌬C), the trimera with positive residues 183-188 in the Rac1 segment replaced with six neutral residues (glutamines; Rac1(183Q-188Q)), and the trimera lacking the PX domain of p47 phox (p47 phox ⌬PX) were also expressed as GST fusion proteins and purified by affinity chromatography on glutathioneagarose, followed by cleavage with thrombin. The expression vector pGEX-2T (Amersham Biosciences) carrying cDNAs encoding the above-mentioned trimeras was introduced into E. coli BL21-CodonPlus TM (DE3)-RIL cells, and bacteria were induced with 0.4 mM isopropyl ␤-D-thiogalactopyranoside at 18°C for 14 -16 h. The induced cells were suspended in TMN buffer (50 mM Tris-HCl, pH 7.5, 4 mM MgCl 2 , 150 mM NaCl, and 2 mM dithiothreitol) supplemented with Complete EDTAfree protease inhibitor (Roche Applied Science). The bacteria were disrupted by exposure to lysozyme (Sigma) at a concentration of 0.5 mg/ml for 20 min at 4°C with stirring and by sonication using a 500-watt sonic disruptor (Vibra-Cell, Sonics and Materials, Inc.) at 20% amplitude for 5 min in the 50% pulse mode in ice-cooled beakers. The resulting material was supplemented with 1% Triton X-100 (Sigma) and stirred on ice for 15 min. The bacterial lysate was subjected to centrifugation at 27,000 ϫ g for 25 min at 4°C; the clear supernatant was applied to glutathione-agarose beads; and binding was allowed to proceed in the batch mode for 60 min at room temperature. The beads were washed with TMN buffer supplemented with 0.1% Triton X-100, followed by two washes with buffer supplemented with 1 M NaCl and 0.1% Triton X-100. Next, in preparation for thrombin digestion, the beads were washed two times with TMN buffer supplemented with 2.5 mM CaCl 2 and 10 mM n-octyl ␤-D-glucopyranoside. The latter reduces the aggregation of trimeric fusion proteins, which was encountered occasionally. The bound protein was separated from the glutathione-agarose beads by two sequential digestions with bovine thrombin (Sigma); each digestion was for 60 min at room temperature. After each digestion, the digested proteins were immediately supplemented with 1 mM 4-(2-aminoethyl)benzenesulfonyl fluoride hydrochloride (Roche Applied Science) to inhibit the proteolytic activity of thrombin. The trimera with residues 183-188 (KKRKRK) in the Rac1 segment replaced with residues 183-188 (RQQKRA) in Rac2 (Rac13 Rac2), the trimera lacking the Rac1 insert domain (Rac1⌬insert), the trimera containing p47 phox (W193R), the trimera containing p67 phox (R102E), and the trimera containing Rac1(A27K) were expressed as N-terminal His 6 -tagged fusion proteins and purified by metal chelate affinity chromatography. Briefly, pET-28a plasmids (Novagen) carrying the trimera genes listed above were introduced into E. coli BL21-CodonPlus TM (DE3)-RIL cells, and the bacteria were induced under conditions identical to those described for GST-fused trimeras. The induced cells were disrupted by lysozyme treatment and by sonication in buffer consisting of 50 mM sodium phosphate, pH 7.5, and 300 mM NaCl supplemented with Complete EDTA-free protease inhibitor as described for GST-fused chimeras. The bacterial lysate was supplemented with 1% Triton X-100 and 5 mM imidazole and stirred in an ice-cooled vessel for 15 min. Insoluble material was sedimented by centrifugation at 23,000 ϫ g for 25 min at 4°C; the cleared cell-free extract was applied to BD Talon TM metal affinity resin beads (Clontech); and binding was performed in the batch mode at room temperature for 60 min. The beads were washed with the sodium phosphate buffer supplemented with 5 mM imidazole, followed by two washes with the same buffer supplemented with 20 mM imidazole. The His 6tagged trimeras were eluted from the resin with the same buffer supplemented with 150 mM imidazole. All purified trimera proteins were supplemented with 30% glycerol and stored in small aliquots at Ϫ75°C.
Characterization of Recombinant Trimeras-The protein concentration of the recombinant fusion proteins was measured by the method of Bradford (46), modified for use with 96-well microplates (81), using Bio-Rad protein assay dye reagent concentrate and bovine ␥-globulin as a standard. The level of purity of the recombinant proteins was assessed by SDS-PAGE analysis, and the gels were stained with GelCode TM Blue stain reagent (Pierce). Immunoblot analysis was carried out as described (47). For the detection of p47 phox , p67 phox , and their corresponding segments in the trimera, we used goat anti-p47 phox and anti-p67 phox polyclonal antibodies (gifts from T. L. Leto, National Institutes of Health), both diluted 1:2000. For the detection of Rac1 and the Rac1 segment in the trimera, we used an affinity-purified rabbit anti-Rac1 C-terminal peptide polyclonal antibody (Santa Cruz Biotechnology, Inc.), diluted 1:2000. The secondary antibodies were affinity-purified alkaline phosphatase-conjugated anti-goat IgG (Sigma) for detection of p47 phox and p67 phox and affinity-purified alkaline phosphatase-conjugated anti-rabbit IgG (Sigma) for detection of Rac1, both diluted 1:2000. The blots were exposed to the primary and secondary antibodies for 1 h. Alkaline phosphatase activity on the blots was detected as described (48). The ability of the trimera to bind GTP was assayed using the fluorescent guanine nucleotide analog mant-GMPPNP as described (49). The identity and amount of nucleotides bound to recombinant trimera were determined by liberating the bound nucleotides from the protein as described previously (50) and identifying the nucleotides by anion exchange chromatography on a Partisil 10 SAX column as described (49).
Nucleotide Exchange-For use in oxidase activation assays, Rac1, chimera 3, and trimeras were subjected to nucleotide exchange from the native, GDP-bound form to a state in which the protein-bound nucleotide was GMPPNP or mant-GMPPNP. Nucleotide exchange reactions were performed at a free Mg 2ϩ concentration of 0.5 M (attained by addition of 12.5 mM EDTA) at a 10-fold molar excess of nucleotide over protein, followed by incubation for 30 min at 30°C. The p47 phox -p67 phox -Rac1 Chimeras Activate the NADPH Oxidase exchange was stabilized by addition of MgCl 2 to a final concentration of 25 mM.
Gel Filtration-The prototype trimera exchanged to mant-GMPPNP was loaded onto a Superose 12 10/300 GL fast protein liquid chromatography gel filtration column (Amersham Biosciences), and chromatography was performed on a Waters HPLC system with TMN buffer at flow rate of 0.2 ml/min at 4°C. Absorbance at 280 nm was measured continuously by a Jasco Model MD-1510 diode array detector. When the trimera was labeled with mant-GMPPNP, the fluorescence signal (excitation at 361 nm and emission at 440 nm) was also recorded continuously by passing the column eluate through a Jasco Model FP-750 spectrofluorometer fitted with a Jasco Model MFC-132 HPLC flow cell. Fractions (0.6 ml) were collected, and the trimera-containing fractions were identified by their ability to support cell-free oxidase activation. The column was standardized with molecular mass markers (range of 12-200 kDa).
Preparation of Macrophage Membrane Liposomes-Phagocyte membranes were prepared from guinea pig macrophages obtained by injection of mineral oil into the peritoneal cavity as described (28). The membranes were solubilized in 40 mM n-octyl ␤-D-glucopyranoside and then reconstituted into liposomes by dialysis against detergent-free buffer as described previously (51). The specific cytochrome b 559 heme content of membrane liposomes was measured by the difference spectrum of sodium dithionite-reduced minus oxidized samples (52).
Purification and Relipidation of Cytochrome b 559 -Cytochrome b 559 was purified from solubilized macrophage membranes, relipidated, and reflavinated as described (53). PC 20% was used for relipidation at a ratio of 0.2 mg of lipid (260 nmol) to 360 pmol of cytochrome b 559 heme.
Membrane Enrichment with Exogenous Phospholipid-Native macrophage membranes were supplemented with a number of exogenous phospholipids to generate membrane liposomes with an artificially modified electrical charge. As a first step in this procedure, we dissolved PC 20%, PC, PA, PG, PS, PI, and DOPG at a concentration of 5 mM in a buffer also used to solubilize membranes (51) following a procedure described previously (23). The buffer consisted of 120 mM potassium/ sodium phosphate buffer, pH 7.4, 1 mM MgCl 2 , 2 mM NaN 3 , 1 mM EGTA, 1 mM dithiothreitol, 10 M FAD, 20% glycerol, and 40 mM n-octyl ␤-D-glucopyranoside. The phospholipids were added to the solubilized macrophage membranes at a constant ratio of 4 volumes of phospholipid (5 mM) to 1 volume of membrane (at a concentration equivalent to 1.2 M cytochrome b 559 heme) unless mentioned otherwise. In another set of experiments, membranes were supplemented with DOPG and either PtdIns(3,4)P 2 or PtdIns(4,5)P 2 . For this purpose, 4 volumes of DOPG (5 mM) dissolved in the above buffer containing 40 mM n-octyl ␤-D-glucopyranoside were mixed with 0.5 volumes of either PtdIns(3,4)P 2 or PtdIns(4,5)P 2 (both at a concentration of 100 M in n-octyl ␤-D-glucopyranoside-containing buffer) and 0.5 volumes of solubilized macrophage membrane at a concentration equivalent to 2.4 M cytochrome b 559 heme. All membrane/phospholipid mixtures were dialyzed against 100 -200 volumes of n-octyl ␤-D-glucopyranoside-free buffer for 18 h at 4°C to convert them into phospholipid-enriched membrane liposomes. The concentration of cytochrome b 559 in the modified liposomes was determined and was typically found to be close to 240 pmol/ml. The theoretical final concentration of exogenous phospholipids (PC, PA, PG, PS, PI, and DOPG) was considered to be 4 mM, and that of PtdIns(3,4)P 2 or PtdIns(4,5)P 2 was considered to be 10 M (representing 0.25 mol % of the total exogenous phospholipid). The total concentration of phospholipids in the native unmodified membranes was determined as described (23).
Cell-free Oxidase Activation Assay-Activation of the oxidase in vitro was assessed by measuring the NADPH-dependent O 2 . production in a semirecombinant cell-free system in the presence or absence of the amphiphilic activator lithium dodecyl sulfate (LiDS) as described (54). Trimeras exchanged to GMPPNP or in the native (GDP-bound) form were tested for their ability to support oxidase activation at various concentrations (5-300 nM). For comparison with trimeras, mixtures of the individual oxidase components p47 phox (full-length or truncated at residue 286), p67 phox (full-length or truncated at residue 212), and Rac1 (full-length and exchanged to GMPPNP) and mixtures of chimera 3 (exchanged to GMPPNP) and p47 phox in the same concentration range were assayed in parallel. Two cell-free oxidase activation systems were utilized: (a) an amphiphile-dependent system consisting of membrane liposomes (equivalent to 5 nM cytochrome b 559 heme) and the various cytosolic activators in the presence of 130 M LiDS and (b) an amphiphile-independent system consisting of membrane liposomes supplemented with exogenous phospholipid (5 nM cytochrome b 559 heme) and cytosolic activators in the absence of LiDS. The assay mixtures were incubated in 96-well microplates in a total volume of 200 l of assay buffer (55) per well with or without LiDS for 90 s at 24°C before addition of 240 M NADPH to initiate O 2 . production. This was quantified by the kinetics of cytochrome c reduction as described previously (55).
Curve Plotting-Plotting of dose-response curves and calculation of V max and EC 50 values were performed using GraphPad Prism Version 4.03.

The Rationale on Which the Design of Trimeras Was Based-
We have reported in the past that a recombinant chimeric protein consisting of p67 phox -(1-212) fused with full-length Rac1 (referred to as chimera 3) is capable of eliciting NADPH-dependent O 2 . production by preparations of phagocyte membranes in the presence of p47 phox and an anionic amphiphilic activator (25,40) and in the absence of p47 phox and amphiphile when the chimera is prenylated (25,41). We now extend these studies to the generation of a single molecule activator of the oxidase by constructing a tripartite p47 phox -p67 phox -Rac1 fusion protein in which p47 phox -(1-286) was fused to chimera 3 by a 10-amino acid spacer (Fig. 1). The basic construct, which we call the "prototype trimera," is composed of the PX domain and the two SH3 domains of p47 phox , the TPR and activation domains of p67 phox , and full-length Rac1. p47 phox was truncated at residue 286, right after the C terminus of the second SH3 domain, to generate an "open" conformation of the p47 phox segment (10,11,15,58) and also because the p47 phox -(1-286) segment is part of a p47 phox -p67 phox chimera originally described by Ebisu et al. (37). p67 phox was truncated at residue 212, right after the end of the activation domain (4), eliminating the PRR and the two SH3 domains. Rac1 was kept full-length because of the requirement for an intact polybasic domain for oxidase activation by Rac1 (18,56,57) and p67 phox -Rac1 chimeras (25,41), the possible involvement of the insert domain in oxidase activation (19,20) and in binding to anionic phospholipids (59), and for preserving the potential for prenylation. The order of the components in the trimera was dictated by our wish to conserve the original structure of chimera 3 as an integral part of the trimera, to place the C terminus of the Rac1 segment at the C terminus of the trimera, and to maintain the N-terminal position of the PX domain in p47 phox at the N terminus of the trimera. All trimera mutants were derived from the p47 phox -(1-286)-p67 phox -(1-212)-full-length Rac1 prototype and are listed in Fig. 1. Mutants were generated to establish the importance of a particular region or residue in protein-protein and protein-phospholipid interactions, essential for the assembly of a functional oxidase complex. Some of these regions or residues are involved in interactions between the trimera and the phagocyte membrane (the first six mutants listed in Fig. 1), and some are involved in intrachimeric bonds, representing interactions between cytosolic components (the last two mutants listed in Fig. 1).

Properties of the Prototype and Mutant
Trimeras-The prototype and all mutant trimeras were successfully expressed in E. coli and recovered in the soluble fraction, and the purified proteins were found to be of the expected molecular mass when analyzed by SDS-PAGE ( Fig. 2A). Thus, the molecular mass of the prototype trimera was close to the expected 80 kDa. The reduction in molecular mass of mutant trimeras p47 phox ⌬PX (lane 3) and Rac1⌬C (lane 7) is evident. The level of purity of all recombinant trimeras exceeded 90%. To assess the integrity of each segment in the tripartite fusion protein, the prototype trimera was subjected to immunoblot analysis with antibodies to p47 phox , p67 phox , and Rac1. As shown in Fig. 2B, the trimera was recognized by all three antibodies (lanes 1-3). The specificity of the antibodies is demonstrated by their ability to react with the respective individual purified components (lanes 4 -6).
To assess the size of the trimera under nondenaturing conditions, the prototype trimera was exchanged to the fluorescent nucleotide mant-GMPPNP and injected into a Superose 12 size exclusion column. The elution of the protein from the column was followed by recording the fluorescence signal in-line and by collecting the eluate into tubes (0.6 ml/tube). Fig. 2C illustrates the elution pattern of the trimera; in three independent experiments, the main fluorescent peak eluted at 61.5 min, corresponding to a molecular mass of 80 kDa, in good agreement with the SDS-PAGE data. The collected fractions were also examined for their ability to support oxidase activity by adding a sample from each fraction to membrane liposomes in the presence of LiDS as described below. As shown in Fig. 2D, oxidase-activating ability was detected in fractions eluting at a volume of 12-12.6 ml (60 -63 min) and overlapping the fluorescent peak in Fig. 2C.
The presence of a Rac1 moiety in chimera 3 was found to confer on it the properties of a bona fide small GTPase (25,40,41). The presence of a full-length Rac1 segment in the trimera was likely to have the same effect. To test this, we determined the type of nucleotide bound to the prototype trimera. We found that, in the native form, it contained only GDP (data not shown), as expected to be the case based on information on recombinant Rac produced in E. coli, which exhibits high intrinsic GTPase activity (49,60,61).
The Prototype Trimera Is a Potent Oxidase Activator in the Presence of an Anionic Amphiphile-Both p47 phox -p67 phox and p67 phox -Rac1 chimeras supplemented with the missing cytosolic component are potent oxidase activators in a variety of cellfree systems (25,37,39,40,41). We thus first examined the ability of the tripartite fusion protein to support amphiphiledependent oxidase activation in a cell-free system. A critical factor in cell-free oxidase activation by mixtures of individual cytosolic components is optimization of the concentration of the activating amphiphile. This was studied in the past in relation to arachidonate (28), SDS (32), and LiDS (54), and in all cases, the optimal activating concentration was found to be in the 100 -150 M range.
In preliminary experiments, we found that incubation of membrane liposomes (equivalent to 5 nM cytochrome b 559 heme) with the prototype trimera at a concentration of 100 nM in the presence of 130 M LiDS resulted in oxidase activation. We next examined the effect of varying the concentration of

p47 phox -p67 phox -Rac1 Chimeras Activate the NADPH Oxidase
LiDS from 0 to 300 M on the level of oxidase activation by the above concentrations of membrane and trimera and compared this with oxidase activation by a mixture of individual fulllength and truncated cytosolic components and by a combination of chimera 3 and full-length p47 phox . In all experiments, the trimera, Rac1, and chimera 3 were subjected to nucleotide exchange to GMPPNP. As shown in Fig. 3, when the cytosolic components were present as full-length individual entities, the LiDS optimum lay between 120 and 160 M, as reported previously (54). Oxidase activation by chimera 3 combined with fulllength p47 phox exhibited a similar optimum at 100 -160 M LiDS. In contrast to these values, the trimera was most active at 40 -100 M LiDS. The LiDS dose-response curve of mixtures of C-terminally truncated p47 phox and p67 phox (truncations corresponded to those of the respective segments in the trimera) and full-length Rac1 was flat, with an optimum at 80 -140 M LiDS. This concentration optimum is in a range located between that of the trimera and that of the fulllength components. It is also apparent that peak oxidase activities elicited by the trimera were only about half of those obtained with fulllength individual cytosolic components and were similar to those obtained with individual truncated p47 phox and p67 phox . No oxidase activation was elicited by the trimera or by any of the combinations of cytosolic activators in the absence LiDS. These results show that, despite the C-terminal truncation of both p47 phox and p67 phox segments, oxidase activation by the trimera remained amphiphile-dependent, albeit the concentration of amphiphile required for maximal activation was lower than that necessary for activation by individual fulllength components and by chimera 3 and full-length p47 phox . This finding generated a dilemma as to the concentration of LiDS to be used in experiments in which oxidase activation by the trimera was to be compared with that by non-fused components. We opted for 130 M LiDS, which is closest to the concentration causing maximal activation by both fused and nonfused components.
Having established the most propitious conditions for activation, we studied the ability of the trimera to activate the oxidase under amphiphile-dependent conditions by performing dose-response experiments in which activation by the prototype trimera (either exchanged to GMPPNP or in the unexchanged, GDPbound form) was compared with that by a combination of the individual intact components p47 phox , p67 phox , and full-length Rac1-GMPPNP and by a combination of p47 phox -(1-286), p67 phox -(1-212), and full-length Rac1-GMPPNP. In the doseresponse experiments, the concentration of each of the individual components, in the combinations was increased in parallel and compared with oxidase activation by the trimera at the same concentration. As shown in Fig. 4, oxidase activation by trimera-GMPPNP exhibited a V max value 1.8 and 1.6 times lower than those achieved by full-length and truncated individual components, respectively. However, the EC 50 value measured with the trimera was 2.5 and 11 times lower than those measured with full-length and truncated individual components, respectively. The lower EC 50 value for the trimera in comparison with non-fused components is in good agreement with previous findings of lower EC 50 values for p47 phox -p67 phox . The proteins were subjected to SDS-PAGE, transferred onto a nitrocellulose membrane, and detected with the indicated specific antibodies as described under "Experimental Procedures." C and D, 2.5 nmol of prototype trimera exchanged to the fluorescent GTP analog mant-GMPPNP were subjected to gel filtration on a Superose 12 fast protein liquid chromatography column as described under "Experimental Procedures." C illustrates the in-line recording of the fluorescent signal. One major peak representing the mant-GMPPNP-conjugated trimera was detected and corresponds to a protein with a molecular mass of 80 kDa. A shoulder (at the left) and a minor peak (at the right of the major peak) represent aggregated and partially degraded trimera, respectively. D illustrates the ability of fractions (0.6 ml) collected from the column eluate to activate the oxidase. 20 l of each fraction were added to a cell-free assay (see "Experimental Procedures") consisting of native membrane (equivalent to 5 nM cytochrome b 559 heme) and the amphiphilic activator LiDS (130 M). Oxidase-activating fractions overlap the major fluorescent mant-GMPPNP-carrying peak. The results in C and D represent one characteristic experiment of three performed.
(37) and p67 phox -Rac1 (39) chimeras in comparison with those measured with mixtures of individual truncated components. The data in Fig. 4 also reveal that the dose-response curve obtained with the mixture of truncated components was sigmoidal as opposed to the dose-response curves of the trimera and full-length components, which were hyperbolic. It is also apparent in Fig. 4 that the trimera in the GDP-bound form exhibited significant activity, particularly at high concentrations. This was expressed in a V max similar to that for the trimera in the GMPPNP-bound form, but the EC 50 value was four times higher, suggesting a lower affinity for its target. The ability of the GDP-bound form of p67 phox -Rac1 chimeras to elicit oxidase activation at a significant level was described by us in the past (25,40,41).
Stability of the Trimera-Cytochrome b 559 Complex-The oxidase complex generated in vitro is unstable (reviewed in Ref. 62). The mechanism of deactivation of the assembled complex has been the subject of extensive studies, and the dominant idea is that the stability of the complex is the expression of the balance between assembly and disassembly. The stability of the assembled complex could be improved by the presence of nonhydrolyzable GTP analogs (63), by chemical cross-linkers (64), and by the fusion of p67 phox with p47 phox (37,65,66) and, to a lesser degree, with Rac1 (39,66). An important role for p47 phox in the stabilization of the oxidase complex was reported in the early nineties (64) and reemerged in a recent report (8).
We assessed the stability of the oxidase complex consisting of membrane and the prototype trimera and compared this with the stability of complexes comprising membrane and a mixture of individual cytosolic components (p47 phox , p67 phox , and Rac1) or a mixture of chimera 3 and p47 phox . In these experiments, . production was measured. Oxidase activities assessed at the various time intervals were expressed as the percentage of initial activities measured at time 0. As shown in Fig. 5, there was a marked loss in activity of mixtures consisting of membrane, p47 phox and p67 phox (full-length and truncated), and Rac1 and  p47 phox -p67 phox -Rac1 Chimeras Activate the NADPH Oxidase of membrane, chimera 3, and p47 phox within the first 20 min of incubation at a non-activating concentration of LiDS. As opposed to this, membrane/trimera mixtures remained extremely stable and lost Ͻ10% of activity during the same incubation period. Half-life calculations revealed that, upon amphiphile removal, the membrane-trimera complex remained active, with a calculated half-life of 8.6 h. In comparison, a complex generated by membrane and individual cytosolic components had a half-life of 8.5 min (full-length p47 phox and p67 phox ) and 5.5 min (truncated p47 phox and p67 phox ), and a complex formed by membrane with chimera 3 and p47 phox decayed with a half-life of 2.6 min.
In experiments to be described below, we found that exposure of purified relipidated cytochrome b 559 to the trimera in the absence of amphiphile resulted in oxidase activation. We thus examined the stability of the oxidase complex assembled from cytochrome b 559 and trimera-GMPPNP in the absence of amphiphile. In these experiments, purified cytochrome b 559 incorporated in liposomes of PC 20% (50 nM cytochrome b 559 heme) were incubated with the trimera (1 M) in the absence of LiDS for 90 s to induce assembly of the oxidase complex. Following assembly, aliquots of the mixture were diluted 10-fold in buffer lacking LiDS, resulting final concentrations of 5 nM cytochrome b 559 heme and 100 nM trimera. The mixtures were incubated at 24°C for time intervals varying from 0 to 80 min. At each time interval, NADPH (240 M) was added, and O 2 .
production was measured. Similar to the membrane-trimera complex, the purified and relipidated cytochrome b 559 -trimera complex generated in the absence of LiDS was extremely stable, with a calculated half-life of 7.8 h. These results demonstrate that linking all three cytosolic components covalently markedly improves the stability of the assembled oxidase complex in comparison with complexes assembled with non-fused components or with bipartite chimeras.

Enrichment of the Phagocyte Membrane with Exogenous Anionic Phospholipid or Its Replacement by Cytochrome b 559
Relipidated with Anionic Phospholipid Makes Oxidase Activation by the Trimera Amphiphile-independent-Assembly of the oxidase complex in the canonical cell-free system requires an anionic amphiphile (28 -34). The dominant opinion is that the principal role of amphiphile is to disrupt the autoinhibitory intramolecular bond in p47 phox (35). There is, however, also evidence for an effect of amphiphile on cytochrome b 559 (67). C-terminal truncation of p47 phox and p67 phox was reported to free oxidase activation from the need for an activating amphiphile by removing the C-terminal partner in the intramolecular bond (26). In view of the fact that, in the prototype trimera, both p47 phox and p67 phox segments are C-terminally truncated, we examined the ability of the trimera to activate the oxidase in the absence of amphiphile. As shown in Fig. 6A, the prototype trimera did not elicit oxidase activity by native membranes in the absence of amphiphile; as expected, a mixture of the three individual components or chimera 3 combined with p47 phox was also inactive. Surprisingly, when phagocyte membrane liposomes were replaced with liposomes of purified cytochrome b 559 relipidated with a partially pure preparation of PC (PC 20%, which consists of 14 -23% PC and many other lipids and is derived from soybean), the trimera behaved as a potent, dosedependent oxidase activator in the absence of amphiphile (Fig.  6B). The final concentration of PC 20%, which served for the relipidation of cytochrome b 559 , in the cell-free assay was 3.5 M. The combined individual cytosolic components or chimera 3 combined with p47 phox was incapable of eliciting oxidase activation by cytochrome b 559 liposomes under the same conditions (Fig. 6B). Addition of LiDS (130 M) to the assay containing the trimera and cytochrome b 559 relipidated with PC 20% did not augment O 2 . production above the level found in the absence of amphiphile (data not shown).
To elucidate the mechanism responsible for the difference in amphiphile dependence between liposomes of native membrane and of purified cytochrome b 559 relipidated with PC 20%, we initiated experiments in which the native phospholipid composition of phagocyte membranes was modified by the incorporation of exogenous phospholipids. Supplementation of membrane with PC 20% resulted in a final concentration of exogenous phospholipid in the cell-free assay of 80 M, with . generation measured at incubation time zero, which was considered as 100%. The results represent the means Ϯ S.E. of 3-11 experiments for each combination of cytosolic components. p47 phox -p67 phox -Rac1 Chimeras Activate the NADPH Oxidase JULY 27, 2007 • VOLUME 282 • NUMBER 30 endogenous membrane lipids contributing 12 M. As evident in Fig. 6C, the membrane supplemented with PC 20% became responsive to activation by the trimera in the absence of amphiphile. Only minor activation was induced by a mixture of the three full-length cytosolic components at a high concentration of components, whereas the combination of chimera 3 and p47 phox failed to activate. To ascertain that the ability of the trimera to elicit amphiphile-independent oxidase activation was the result of enrichment of the native membrane with the particular preparation of partially pure PC (PC 20%), we performed a series of control experiments in which phagocyte membrane was enriched with an identical amount of a highly purified (99%) preparation of the neutral phospholipid PC. The final concentration of PC (99%) in the cell-free assay was, as before, 80 M, with endogenous membrane lipids contributing 12 M. As shown in Fig. 6D, neither the trimera nor a combination of individual components or of chimera 3 and p47 phox was capable of eliciting amphiphile-independent oxidase activation. We hypothesized that the reason for this difference lies with the fact that a major component of PC 20% preparations is anionic phosphatidylinositols, and it thus appears likely that an increase in the proportion of anionic phospholipids in the membrane relieves oxidase activation by the trimera of its dependence on exogenous anionic amphiphile.

Oxidase Activation by the Trimera in the Absence of Amphiphile Is Absolutely Dependent on a Negative Membrane Charge and Is Influenced by the Nature of the Anionic Phospholipid Responsible for This
Charge-The findings described above raised the question of the mechanism by which membraneembedded anionic phospholipid enables amphiphile-independent oxidase activation. Because PC 20% is a poorly defined mixture of PC with other lipids rich in anionic phosphatidylinositols, it was essential to study the effect of supplementing the membrane with well characterized purified phospholipids. We thus enriched phagocyte membranes with four different, highly purified, anionic phospholipids: PA (98% pure), PG (99% pure), PS (98% pure), and PI (98% pure). The characteristics of these phospholipids are described under "Experimental Procedures." We compared the ability of the four types of phospholipid-supplemented membrane to serve as targets for amphiphile-independent oxidase activation by the trimera, by a combination of full-length individual cytosolic components, and by chimera 3 combined with p47 phox . As before, the final concentration of anionic phospholipid in the cell-free assay was 80 M, with endogenous membrane lipids contributing 12 M. As shown in Fig. 7, there were considerable differences between the effects of the various anionic phospholipids. In all cases, with the exception of membranes supplemented with PI, maximal activation was induced by a mixture of individual p47 phox , p67 phox , and Rac1, followed by the trimera, with the combination of chimera 3 and p47 phox being the least effective. When activation by the trimera was used as a criterion for comparing the effectiveness of the various phospholipids, on the basis of kinetic characteristics, we found that the V max values were quite similar, with the exception of the membrane supplemented with PA, which exhibited a higher V max value (Fig. 7, table). EC 50 values were, however, markedly different, following the order PA Ͻ PG Ͻ PS Ͻ PI, indicating that PA was the most effective and PI the least effective phospholipid. It should be noted that supplementation of macrophage membrane with exogenous anionic phospholipids did not influence the spectral characteristics of cytochrome b 559 in the membrane (data not shown).
Amphiphile-independent Oxidase Activation in Phagocyte Membranes Enriched with Synthetic PG by the Trimera and Individual Cytosolic Components-For an in-depth study of the effect of membrane enrichment on responsiveness to the p47 phox -p67 phox -Rac1 Chimeras Activate the NADPH Oxidase trimera, we used a well characterized and highly purified anionic phospholipid. Because supplementation of phagocyte membrane with PA led to a low level of spontaneous oxidase activation in the absence of cytosolic components (Fig. 7A), we chose the next most potent anionic phospholipid, PG. Whereas the experiments illustrated in Fig. 7B were performed with ␤-oleoyl-␥-palmitoyl-PG, we now chose a preparation of ␤,␥-dioleoyl-PG (DOPG) of Ͼ99% purity. The final concentration of DOPG incorporated in membrane liposomes in the cell-free assay was kept at 80 M. As shown in Fig. 8, we set up experiments similar to those in Fig. 4, but this time, oxidase activation was performed in the absence of amphiphile. It is evident that membrane enrich-ment with DOPG enabled amphiphile-independent oxidase activation by trimera-GMPPNP and by a combination of full-length p47 phox , p67 phox , and Rac1-GMPPNP, confirming the results obtained with ␤-oleoyl-␥-palmitoyl-PG (Fig. 7B). DOPG-supplemented membranes also responded to a mixture of p47 phox - (1-286), p67 phox -(1-212), and full-length Rac1-GMPPNP by amphiphile-independent oxidase activation. This response was similar to that obtained with unmodified membranes in the presence of LiDS (Fig. 4), as evident in the sigmoidal dose-response curve as opposed to the hyperbolic curve found with the trimera and with full-length individual components. The V max value for oxidase activation by trimera-GMPPNP was 1.8 times lower than that achieved with individual fulllength components, but within the range of that measured with truncated components. However, the EC 50 value measured with the trimera was 3 and 8 times lower than the values measured with individual full-length components and individual truncated components, respectively. The trimera in the GDP-bound form exhibited significant activity, particularly at high concentrations. In general, the kinetic characteristics of oxidase activation by the trimera with DOPG-supplemented membranes in the absence of an exogenous amphiphile were similar to those found with unmodified membranes in the presence of amphiphile, suggesting that negatively charged phospholipid incorporated into the phagocyte membrane functions as a substitute for a soluble anionic amphiphilic activator.

Effect of Deletions and Sequence Replacements Focused on the C Terminus of the Rac1 Segment in the Prototype Trimera on
Oxidase-activating Ability-The realization of the seminal importance of membrane phospholipid charge in oxidase activation prompted the construction of deletion and domain replacement mutants focused on a region likely to have a major effect on the electrostatic interaction of the trimera with anionic membrane phospholipids. This region is the polybasic stretch at the C terminus of Rac1, shown in several earlier studies to mediate charge-based interaction with the membrane (18, 23, 25, 41, 56, 57, 68, 69). p47 phox -p67 phox -Rac1 Chimeras Activate the NADPH Oxidase JULY 27, 2007 • VOLUME 282 • NUMBER 30

JOURNAL OF BIOLOGICAL CHEMISTRY 22131
We subjected the prototype trimera to the following changes: replacement of residues 183-188 (KKRKRK) in Rac1 with residues 183-188 (RQQKRA) in Rac2 (Rac13 Rac2), deletion of the C terminus of Rac1 (residues 179 -192; Rac1⌬C), and replacement of the positive residues 183-188 in Rac1 with six neutral residues (glutamines; Rac1(183Q-188Q)). These modifications were expected to result in a reduction in the positive charge at the C terminus of the trimera. In mutant Rac13 Rac2, the number of positive residues was reduced from six to three, whereas in the two other mutants (Rac1⌬C and Rac1(183Q-188Q)), the number of clustered positive residues was reduced from six to zero. The mutants were assayed in two types of cell-free assays: (a) on native membrane in the presence of amphiphile and (b) on DOPG-supplemented membrane in the absence of amphiphile. As shown in Fig. 9, mutants Rac1⌬C and Rac1(183Q-188Q) completely lost their oxidase-activating ability, as expressed in marked decreases in V max and the inability to calculate EC 50 values, whether assayed on native membranes (panel A) or on DOPG-supplemented membranes (panel B). The Rac13 Rac2 trimera expressed a moderate reduction in V max but a marked increase in EC 50 on both native and DOPG-supplemented membranes. The pronounced changes in EC 50 apparent with all three mutant trimeras demonstrate the centrality of the Rac1 segmentassociated polybasic C terminus in determining the affinity of the trimera for its most likely membrane target, anionic phospholipids. The ability of the prototype trimera (exchanged or not to GMPPNP) to activate the oxidase was compared with that of combinations of full-length and truncated cytosolic components, the latter corresponding to the length of the p47 phox and p67 phox segments in the trimera. The assay mixtures consisted of liposomes of membrane supplemented with DOPG at a concentration equivalent to 5 nM cytochrome b 559 heme, to which were added one of the following combinations of cytosolic activators (at concentrations varying from 0 to 300 nM): (a) trimera exchanged to GMPPNP (E); (b) trimera not subjected to nucleotide exchange (bound nucleotide is GDP; U); (c) full-length p47 phox , full-length p67 phox , and Rac1 exchanged to GMPPNP (Ⅺ); and (d) p47 phox -(1-286), p67 phox -(1-212), and Rac1 exchanged to GMPPNP (छ). No anionic amphiphile was added. O 2 . production was initiated by addition of NADPH (240 M) and measured as described under "Experimental Procedures." The results represent the means Ϯ S.E. of 3-10 experiments for each combination of cytosolic components. The table displays a comparison of the kinetic data (V max and EC 50 ) derived from the displayed curves. FIGURE 9. Effect of deletions and sequence replacements focused on the C terminus of the Rac1 segment of the prototype trimera on oxidaseactivating ability. The ability of the prototype trimera (E) and the mutant trimeras Rac13 Rac2 (‚), Rac1⌬C (Ⅺ), and Rac1(183Q-188Q) (छ) to activate the oxidase was tested in two cell-free systems: amphiphile-dependent, in which native membrane liposomes were used in the presence of LiDS (130 M) (A); and amphiphile-independent, in which membrane liposomes supplemented with DOPG were used in the absence of LiDS (B). In both systems, the concentration of the membrane was equivalent to 5 nM cytochrome b 559 heme, and the concentration of the trimeras was varied from 0 to 300 nM. All trimeras were exchanged to GMPPNP. Following incubation for 90 s in the presence or absence of LiDS, O 2 . production was initiated by addition of NADPH (240 M) and measured as described under "Experimental Procedures." The table displays a comparison of the kinetic data (V max and EC 50 ) derived from the displayed curves. nd, not determinable. The results represent the means Ϯ S.E. of 3-10 experiments for each combination of membrane preparation and trimera (prototype or mutant).

p47 phox -p67 phox -Rac1 Chimeras Activate the NADPH Oxidase
Effect of Deleting the PX Domain of the p47 phox Segment and the Insert Domain of the Rac1 Segment on the Oxidaseactivating Ability of the Prototype Trimera-We next examined the role of two additional regions possessing a net positive charge on the oxidase-activating ability of the trimera. These regions are the PX domain in the p47 phox segment (residues 4 -125 in the p47 phox monomer) and the insert domain in the Rac1 segment (residues 124 -135 in the Rac1 monomer) of the trimera. An additional reason for addressing the role of the Rac1 insert domain was the proposal that this region mediates the direct interaction of Rac with gp91 phox (19,20). We thus introduced the following two changes in the structure of the prototype trimera: deletion of residues 1-150 in the p47 phox segment, which includes the PX domain (p47 phox ⌬PX), and deletion of residues 124 -135 in the Rac1 segment (Rac1⌬insert). The mutants were assayed on unmodified membrane in the presence of amphiphile and on DOPG-supplemented membrane in the absence of amphiphile. As shown in Fig. 10  (A and B), removal of either the PX or Rac1 insert domain did not significantly affect the V max for oxidase activation by the mutant trimeras and caused only a minor increase in EC 50 values in both amphiphile-dependent and -independent systems. The only exception was a more pronounced increase in EC 50 upon deletion of the insert domain in the Rac1 segment when assayed on native membrane in the presence of LiDS.
We reasoned that our failure to detect anything more than a minor effect of removing the PX and Rac1 insert domains on the ability of the trimera to activate the oxidase might be due to the fact that the membrane lacked phospholipids known to serve as specific targets for the deleted domains. Thus, the PX domain of p47 phox possesses a high affinity for PtdIns(3,4)P 2 (12), and the insert domain of Rac1 binds preferentially to PtdIns(3,4)P 2 and phosphatidylinositol 3,4,5-trisphosphate (59). We consequently prepared two more forms of phospholipid-enriched membrane liposomes: the first contained, in addition to DOPG, PtdIns(3,4)P 2 , and the second contained DOPG and PtdIns(4,5)P 2 and served as a negative control. Both phosphoinositides were present at a final concentration of 10 M; in a distinct set of experiments, this concentration was found to be optimal in promoting the dissociation of Rac1-Rho GDP dissociation inhibitor complexes by negatively charged liposomes and in enhancing amphiphile-independent oxidase activation FIGURE 10. Effect of deleting the PX domain of the p47 phox segment and the insert domain of the Rac1 segment on the oxidase-activating ability of the prototype trimera. The ability of the prototype trimera (E) and the mutant trimeras p47 phox ⌬PX (e) and Rac1⌬insert (छ) to activate the oxidase was tested in two cell-free systems: amphiphile-dependent, in which native membrane liposomes were used in the presence of LiDS (130 M; A); and amphiphile-independent, in which membrane liposomes were supplemented with DOPG (B), DOPG and PtdIns(3,4)P 2 (C), and DOPG and PtdIns(4,5)P 2 (D). In all situations, the concentration of the membrane was equivalent to 5 nM cytochrome b 559 heme, and the concentration of the trimeras (wild-type and mutant) was varied from 0 to 300 nM. All trimeras were exchanged to GMPPNP. Following incubation for 90 s in the presence or absence of LiDS, O 2 . production was initiated by addition of NADPH (240 M). Enrichment of membranes with anionic phospholipid (DOPG), supplemented or not with phosphoinositides, and measurement of O 2 . production were performed as described under "Experimental Procedures." The table displays a comparison of the kinetic data (V max and EC 50 ) derived from the displayed curves. The results represent the means Ϯ S.E. of 3 to 14 experiments for each combination of membrane preparation and trimera (prototype or mutant). by such complexes. 4 The prototype trimera and mutants p47 phox ⌬PX and Rac1⌬insert were tested for the ability to activate the oxidase in membranes enriched either in DOPG and PtdIns(3,4)P 2 or in DOPG and PtdIns(4,5)P 2 in the absence of amphiphile. It was expected that the specific affinity of PX domains and of the insert domain of Rac1 for PtdIns(3,4)P 2 would be expressed in enhanced oxidase activation by the prototype trimera acting on membrane enriched in DOPG and PtdIns(3,4)P 2 and in the lack of such enhancement by the deletion mutants. The experimental results did not fulfill this prediction. Neither the V max nor EC 50 values measured with the prototype chimera acting on membrane supplemented with DOPG and PtdIns(3,4)P 2 (Fig. 10C) were different from those measured with control membrane supplemented with only DOPG (Fig. 10B) or membrane supplemented with DOPG and PtdIns(4,5)P 2 (Fig. 10D). Also, the kinetic characteristics of the PX and Rac1 insert domain deletion mutants assayed on membrane supplemented with DOPG and PtdIns(3,4)P 2 were undistinguishable from those derived by assay of the mutants on membrane supplemented with DOPG only or on membrane supplemented with DOPG and PtdIns(4,5)P 2 .
Effect of Point Mutations Focused on Regions in the Prototype Trimera Involved in Interaction with the p22 phox Subunit of Cytochrome b 559 and in Intrachimeric Interaction between the p67 phox and Rac1 Segments on Oxidase-activating Ability-Trp 193 in p47 phox is an essential residue in the interaction between the SH3 domains in p47 phox and the PRR in p22 phox (15,58,70). To influence the interaction of the trimera with p22 phox , we generated a mutant in which Trp 193 in the p47 phox segment was replaced with Arg. The p47 phox (W193R) trimera was found to be incapable of activating the oxidase on native membrane in the presence of amphiphile (Fig. 11A) and exhibited only weak oxidase-activating ability when assayed on DOPG-supplemented membrane in its absence (Fig. 11B). The functional impairment was most evident in the unquantifiable or markedly increased EC 50 values, a reflection of reduced affinity for its target protein in the membrane, p22 phox . The lesser effect of the W193R mutation on amphiphile-independent activation of DOPG-supplemented membranes suggests that, in the presence of a charge-related bond, the dependence on the p47 phox -p22 phox interaction is less pronounced. These results indicate that, although interaction modules connecting the trimera to the phospholipid milieu of cytochrome b 559 based on protein-lipid electrostatic attraction are essential for oxidase activation, there is an additional requirement for protein-protein interaction involving the N-terminal SH3 domain of the p47 phox segment of the trimera and the PRR of the p22 phox subunit of cytochrome b 559 .
Intrachimeric interactions between the p67 phox and Rac1 segments were shown by us to be essential for oxidase activation by chimera 3 (8,25,40,41), and it was of interest to find out whether this also applied to the trimera. Mutating Arg 102 in the p67 phox moiety or Ala 27 in the Rac1 moiety was found to reduce oxidase activation by chimera 3 (25). We designed mutants p67 phox (R102E) and Rac1(A27K) to prevent an intrachimeric FIGURE 11. Effect of point mutations focused on regions in the prototype trimera involved in interaction with the p22 phox subunit of cytochrome b 559 and in intrachimeric interaction between the p67 phox and Rac1 segments on oxidase-activating ability. The ability of the prototype trimera (E) and the mutant trimeras p47 phox (W193R) (‚), p67 phox (R102E) (छ), and Rac1(A27K) (ƒ) to activate the oxidase was tested in two cell-free systems: amphiphile-dependent, in which native membrane liposomes were used in the presence of LiDS (130 M) (A); and amphiphile-independent, in which membrane liposomes supplemented with DOPG were used in the absence of LiDS (B). In both systems, the concentration of the membrane was equivalent to 5 nM cytochrome b 559 heme, and the concentration of the trimeras was varied from 0 to 300 nM. All trimeras were exchanged to GMPPNP. Following incubation for 90 s in the presence or absence of LiDS, O 2 . production was initiated by addition of NADPH (240 M) and measured as described under "Experimental Procedures." The table displays a comparison of the kinetic data (V max and EC 50 ) derived from the displayed curves. nd, not determinable. The results represent the means Ϯ S.E. of three to nine experiments for each combination of membrane preparation and trimera (prototype or mutant). p47 phox -p67 phox -Rac1 Chimeras Activate the NADPH Oxidase interaction between the p67 phox and Rac1 segments of the trimera. The Rac1(A27K) mutant was found to be defective in its ability to activate the oxidase on both native and DOPG-supplemented membranes in the presence and absence of amphiphile, respectively, with a more pronounced impairment being apparent with native membrane (Fig. 11). The impairment was most evident in the marked increases in EC 50 values, reflecting the reduced affinity for the most likely target, cytochrome b 559 . The counterpart mutant, p67 phox (R102E), exhibited only a minor reduction in oxidase-activating ability on both types of membranes, but the existence of some functional impairment was suggested by a significant increase in EC 50 on native membrane and a lesser increase on DOPG-supplemented membrane (Fig. 11). These results demonstrate that, despite the physical fusion between the p67 phox and Rac1 segments in the trimera, an intrachimeric interaction between the TPR domain of the p67 phox segment and the pre-switch I region of the Rac1 segment is required for oxidase activation. The mild effect of mutation R102E in the p67 phox segment as opposed to the pronounced effect of mutation A27K in the Rac1 segment remains unexplained, although a similar, although less marked, difference in the effects of mutating these two residues was described in chimera 3 (25). Arg 102 was shown to make direct hydrogenbonding interactions with four residues in Rac1, but not with Ala 27 (17). It is possible that, because of steric hindrance, such bonds cannot be established in the trimera and that other residues in the TPR domain are involved in interaction with the Rac1 segment.

DISCUSSION
In this study, we have described the design, construction at the DNA level, and successful bacterial expression of a soluble fusion protein consisting of functionally important segments derived from the three cytosolic components essential for activating the O 2 . -generating oxidase in vitro. The segments were chosen based on the accumulated information concerning the involvement of specific domains in p47 phox , p67 phox , and Rac in oxidase assembly (reviewed in Refs. 1-3) and on previous work with p47 phox -p67 phox (37) and p67 phox -Rac1 (Refs. 39, 40, 41, and 25; reviewed in Ref. 8) chimeras. We hypothesized that physically fusing crucial parts of the three cytosolic components would bypass the need for protein-protein interaction between p47 phox and p67 phox and between p67 phox and Rac, resulting in a single protein, the interaction of which with cytochrome b 559 and its membrane environment can be studied. p47 phox -p67 phox and p67 phox -Rac1 chimeras fulfill this goal only partially because they require supplementation with Rac1 or p47 phox , respectively, for optimal functioning. The trimera was designed to incorporate in its structure the prototype p67 phox -Rac1 chimera, described by us in the past and known as chimera 3 (40). A spacer was introduced between the p47 phox and p67 phox segments to ensure a more flexible conformation and to mimic the spacer present in the p47 phox -p67 phox chimera described by Ebisu et al. (37). The trimera possesses several advantageous characteristics: (a) the p47 phox segment (and possibly the p67 phox segment, too) is free of an autoinhibitory region; (b) the presence of Rac makes the molecule act as a bona fide small GTPase, with a GTP/GDP switch; (c) the C terminus has the potential to be prenylated; and (d) a single molecule activator should be ideal for high throughput screening of pharmacological agents affecting oxidase function. The prototype trimera in the GMPPNP-bound form was found to be a potent activator of the oxidase in native phagocyte membrane in the absence of any additional component, with the exception of an anionic amphiphile. The dependence on amphiphile seemed surprising, at first, because of the absence of the autoinhibitory region in the p47 phox segment and because of the finding that p47 phox -p67 phox chimeras supplemented with Rac1-GTP are active with a cytochrome b 559 preparation in the absence of amphiphile (37). The explanation for the latter finding lies with the fact that the particular cytochrome b 559 preparation was relipidated with a mixture rich in anionic phospholipids, a procedure found by us in the experiments described in this study to enable amphiphile-independent oxidase activation by the trimera. The persistence of a requirement for amphiphile despite the "built-into-the-trimera" truncation of the p47 phox segment provides some support for the idea of a direct effect of amphiphile on cytochrome b 559 (67) and is in agreement with the requirement for a lower concentration of LiDS for trimera-elicited oxidase activation compared with that required for activation by components including full-length p47 phox .
The trimera forms an exceedingly stable complex with cytochrome b 559 , in marked contrast to the lability of complexes formed with three individual components, whether full-length or truncated, or with chimera 3 combined with p47 phox . Stable oxidase complexes were generated in the past by chemical cross-linking (64) or with mixtures of a p47 phox -p67 phox chimera and Rac1-GTP (37,65,66). By using a methodology similar to that used by us, complexes with half-lives of 3.5 and 4.2 h were obtained with p47 phox -p67 phox chimeras supplemented with Rac1-GTP activated in the presence and absence of amphiphile, respectively (66). The stability of the complexes was increased to half-lives of 4.5 and 6.6 h, respectively, by treatment with a chemical cross-linker. Ours appears to be the first description of a stable oxidase complex comprising a single cytosolic activator molecule without the need for artificial cross-linking. The half-lives achieved with the trimera (ϳ8 h) exceeded those obtained with bipartite chimeras combined with a third component, even when the latter combinations were treated with a cross-linker. This result supports the contention that, under physiological conditions, the stability of the catalytically active assembled complex is limited and that there is a continuous exchange of cytochrome b 559 -bound cytosolic components for fresh components translocating from the cytosol (Refs. 71 and 72; reviewed in Ref. 62). In the case of the trimera, the p47 phox and Rac1 segments bind to the membrane and possibly to cytochrome b 559 , and, by being fused to the p67 phox segment, ensure a long-lasting association of p67 phox with gp91 phox , which stabilizes an "activated" conformation in gp91 phox .
When compared with the kinetic features of activation by individual components, those of oxidase activation by the trimera are lower EC 50 values, suggesting a higher affinity for cytochrome b 559 , together with a seemingly paradoxical lowering of V max values, indicating less efficient activation. It is pos-p47 phox -p67 phox -Rac1 Chimeras Activate the NADPH Oxidase sible that the high stability of the bond between the trimera and cytochrome b 559 , as opposed to the continuous recruitment dynamics seen with individual cytosolic components, is not ideal for maximal oxidase activity and might offer an explanation for the lower V max values.
In the course of work meant to test the ability of the trimera to activate purified cytochrome b 559 , we used cytochrome b 559 preparations relipidated with non-purified PC (PC 20%) as described previously (53). This led us to the finding that cytochrome b 559 relipidated with a low purity preparation of PC responded to the trimera by NADPH-dependent O 2 . production in the absence of LiDS. The finding that native membrane and membrane supplemented with a neutral phospholipid such as pure PC did not respond to the trimera, whereas membrane supplemented with PC 20% (rich in anionic phosphatidylinositols) did, points to a role for negatively charged phospholipids. The more vigorous response of cytochrome b 559 liposomes compared with that of supplemented membrane is most likely related to the presence in the membrane, but not in relipidated cytochrome b 559 , of endogenous lipid, which is predominantly neutral (73,74). The difficulty of drawing definite conclusions from work with uncharacterized phospholipids motivated the performance of experiments with highly purified anionic phospholipids. These experiments (summarized in Fig. 7) clearly establish that modifying the lipid composition of the phagocyte membrane, leading to an increase in negative charge, causes a major change in the response of the oxidase to cytosolic activators. This is expressed in an ability to be activated by the trimera in the absence of amphiphile, but also allows activation by a mixture of individual cytosolic components and by chimera 3 combined with p47 phox .
We propose that this represents a novel mechanism of oxidase assembly and activation in vitro, which is likely to have its equivalent in vivo. Whereas in the canonical cell-free system, the anionic amphiphilic activator is free in solution, in this system, the anionic phospholipid is in the membrane. The main role of membrane-bound anionic phospholipid is likely to be the provision of an electrostatic anchor to the principal positively charged region in the trimera, namely the polybasic stretch at the C terminus of the Rac1 segment (Fig. 12).
The capacity of membrane-incorporated phospholipid to enable amphiphile-independent oxidase activation was described before for p47 phox -p67 phox chimeras supplemented with Rac (37) and for mixtures of individual cytosolic components comprising p47 phox mutants in which the autoinhibition of SH3 domains was relieved (38). Both groups reported that, under the same conditions, no amphiphile-independent activation was seen with full-length non-mutated components. We found that membrane enriched with certain anionic phospholipids also responds to full-length cytosolic components in the absence of amphiphile. This is most likely the result of the high concentration of anionic phospholipid achieved when purified phospholipids are used for membrane enrichment. Indeed, the final concentration of anionic phospholipid present in the reaction (80 M) is similar to that of soluble anionic amphiphile in the canonical cell-free system.
The theoretical pI of the prototype trimera is 6.94, and thus, charge-mediated interactions must be centered on particular regions in the trimera. We have demonstrated that the principal, if not the only such, region is the polybasic C terminus of the Rac1 segment. Binding of Rac1 to anionic phospholipids exhibits little specificity; interaction of Rac1 with PA, PG, PS, and PI and with a number of phosphoinositides was described (23,59,68,69), although a preference for PS over PG and PI was reported recently (75).
We found no good evidence for a major role of the PX domain in the p47 phox segment and the insert domain in the Rac1 segment in oxidase activation by the trimera. Domain deletion studies indicated only minor decreases in the affinity for the membrane, reflected by increases in EC 50 values. No specificity for PtdIns(3,4)P 2 could be shown, leading to the conclusion that binding of the PX and Rac1 insert domains to phospholipids is based exclusively on charge. A similar lack of specificity for phosphoinositides was found in a recent study on amphiphile-independent oxidase activation by individual cytosolic components (38).
The design of the trimera also allowed us to examine whether interaction with p22 phox forms part of the activation process. The marked inhibitory effect of the W193R mutation on the activity of the trimera was surprising in light of the many occasions in which oxidase activation by individual components (5,8,76,77) or by p67 phox -Rac1 chimeras (25,41) was found to be p47 phox -independent. A survey of the situations in which oxidase activation took place in the absence of p47 phox reveals that this requires the presence of an additional membrane localization signal such as prenylation of the C terminus of Rac (5,25,41) or supplementation of membranes with anionic phospholipid. Indeed, we found that supplementation of membrane with certain anionic phospholipids made amphiphile-independent oxidase activation by p67 phox and Rac1 or by chimera 3 in the absence of p47 phox possible (in the following order of efficiency: PA Ͼ PG Ͼ PS Ͼ PI) (data not shown).
Finally, we found that, despite the physical fusion between segments within the trimera, there is a requirement for intrachimeric protein-protein interaction between the p67 phox and Rac1 segments, similar to that described in the bipartite p67 phox -Rac1 chimeras (25,40,41). Fig. 12 presents an idealized rendering of the overall structure of the prototype trimera in the GMPPNP-bound form and of its sites of contact with membrane phospholipids and cytochrome b 559 . The approximate locations of the residues, the mutation of which causes functional changes, are indicated, as well as the intrachimeric interaction and the hypothetical point of contact between the activation domain in the p67 phox segment and the cytosolic tail of gp91 phox .
Trimera-elicited oxidase activation provides support for a model in which the process of oxidase assembly is seen as a two-stage process. In the first stage, p47 phox and Rac-GTP (acting as "organizers") establish nonspecific, electrostatic, and hydrophobic bonds with negatively charged membrane phospholipid facing the cytosol. Before, simultaneously with, or following binding to the membrane, p47 phox and Rac-GTP interact with p67 phox (which lacks intrinsic membrane tropism). In the second stage, protein-protein bonds are established between p67 phox (acting as "activator") and gp91 phox , the end p47 phox -p67 phox -Rac1 Chimeras Activate the NADPH Oxidase result of which is the induction of a conformational change in gp91 phox expressed in the promotion of electron flow from NADPH to oxygen and the generation of O 2 . . There are good arguments for the contention that the binding of p67 phox to gp91 phox and/or the induction of a conformational change in gp91 phox is the consequence of a Rac-GTP-induced conformational change in p67 phox , which is propagated to gp91 phox . A parallel pathway assisting the binding of p67 phox to gp91 phox , but not leading to the induction of a conformational change in gp91 phox , involves the binding of p47 phox to p22 phox . Our results are also in agreement with a model in which oxidase activation is initiated by the generation of anionic phospholipid "patches" facing the cytosol and providing a novel microenvironment for cytochrome b 559 . In the experiments described in this study, the concentration of anionic phospholipid in supplemented membrane liposomes was 4 mM, and its distribution was probably random. It is unlikely that such a concentration is ever achieved in membranes in vivo, and it is thus likely that lipids of higher negative charge are involved and concentrated in microdomains. The obvious candidates for such a function are phosphoinositides, as exemplified by the fact that PtdIns(4,5)P 2 has a valence of Ϫ4 compared with Ϫ1 for PS (78). Phosphatidylinositol 3,4,5-trisphosphate and PtdIns(4,5)P 2 were indeed found to target small GTPases, possessing a C-terminal polybasic cluster, to the plasma membrane (79). Another possible candidate is PA, generated by phospholipase D, by virtue of it being the membrane supplement supporting the most vigorous response to the trimera and individual cytosolic components. Both phosphoinositides (reviewed in Ref. 80) and PA (36) were proposed as intermediates in the signal transduction path leading from membrane receptors to oxidase assembly.