Characterization of neutrophil NADPH oxidase factors p47-phox and p67-phox from recombinant baculoviruses.

Superoxide production by phagocytic blood cells involves assembly of an active NADPH oxidase complex from components found both in membrane and cytosolic locations in resting cells. We recently cloned cDNAs encoding two cytosolic components (p47-phox and p67-phox) of the oxidase that are deficient in distinct forms of autosomal recessive chronic granulomatous disease. The precise roles of p47-phox and p67-phox were explored further using purified factors produced in large quantities using recombinant baculoviruses to infect cultured Sf9 insect cells. Neither p47-phox nor p67-phox are thought to represent the flavoprotein components of the oxidase, since neither of the purified recombinant factors contained or bound FAD. Recombinant p47-phox and p67-phox are capable of restoring the deficient cytosol from chronic granulomatous disease patient neutrophils to nearly normal levels in a cell-free reconstitution system. Both p47-phox and p67-phox, used together in the absence of neutrophil cytosol, are incapable of supporting cell free production of superoxide, confirming the involvement of other soluble factor(s) in the assembly of an active oxidase in vitro.

to be the terminal component of the system responsible for donating electrons directly to molecular oxygen (22).
The cDNAs encoding the two cytosolic components of the oxidase (p47-phox and p67-phox) affected in the majority of autosomal recessive cases of chronic granulomatous disease were recently described (14- 16). Recombinant fusion proteins encoded by these cDNAs produced in Escherichia coli were used to demonstrate partial restoration of superoxide generating activity to deficient patients' cytosols in vitro (14)(15)(16). Both of these proteins contain duplicated sequence motifs exhibiting significant homology to noncatalytic (SH3 or A box) domains in src-related proteins (15,16), although these structural relationships have provided few clues regarding their precise functional roles within the oxidase. In an attempt to understand further the roles of these soluble factors and the mechanism by which they assemble with the other components into an active oxidase complex, we have explored several recombinant DNA expression systems to produce the cytosolic factors for use in cell free oxidase reconstitution studies. In this paper we report the development of an efficient expression system using recombinant baculoviruses (23, 24) to produce large amounts of highly active recombinant p47phox and p67-phox. Both recombinant proteins were purified to homogeneity from baculovirus-infected cultured Sf9 (Spodopterafrugiperda) cells. The availability of both recombinant proteins in pure form has enabled a detailed biochemical characterization of these factors and confirmed the involvement of other cytosolic factors for cell-free reconstitution of the active NADPH oxidase.

Materials
The baculovirus transfer vector, pVL1392 (24), and wild type baculovirus strain AcMNPV were obtained from Dr. Max D. Summers (Department of Entomology, Texas Agricultural Experiment Station, Texas A & M University). The insect host cultures, S. frugiperda (Sf9) cells, were obtained from the American Type Culture Collection (Rockville, MD). Cell cultures were grown in a modified Grace's insect cell culture medium (Formula No. 86-0050AJ, Gibco Laboratories, Gaithersburg, MD).
Secondary antibodies (goat anti-rabbit and rabbit anti-goat IgG) conjugated to alkaline phosphatase, as well as the phosphatase substrates used to develop immunoblots, were obtained from Bio-Rad. ELISA reagents were from Vector Laboratories (Burlingame, CA). Reagents used in the cell-free superoxide generation assay were from previously described sources (8, 25).

Methods
Production of Recombinant Baculoviruses BV/p47 and BV/p67-Bacuioviruses carrying the cDNAs for human p47-phx and p67-phox were constructed with the transfer vector pVL1392, a derivative of pAC373 containing a mutation within the polyhedrin translation initiation site (24). The cytosolic factor cDNAs were derived from pBluescript clones 8A (p47-phox) (14) and 10 (p67-phoz) (16) and subcloned into an EcoRI site occurring downstream from the polyhedrin promoter within pVL1392. These plasmids (4 pg of pVL1392/ p47 or pVL1392/p67) were used to cotransfect Sf9 cells along with wild type baculovirus AcMNPV DNA (2 pg) by the standard calcium phosphate coprecipitation methods described previously (23, 24). Polyhedrin-deficient recombinant viruses were selected and cloned by the standard plaque assay (23) and then confirmed both by hybridization with p47-phx and p67-phox cDNA probes and immunoblotting using rabbit anti-rp47-phox and anti-rp67-phox antisera previously raised against the recombinant factor fusion proteins produced in E. coli (14,16).
Production and Purification of Recombinunt p47-phox and p67p b x from Bacubuirus-infected Sf9 Cells-Large scale production of recombinant p47-phox and p67-phox was achieved by infecting monolayer cultures of Sf9 cells in 150-cm2 flasks at a density of 2-4 X 10' cells/ml. Typically, cells were infected with recombinant baculovirus at a multiplicity of infection >5 plaque forming units/cell and were harvested 84-90 h post-infection. Harvested cells were washed in phosphate-buffered saline by centrifugation at 400 X g for 10 min, resuspended to 25 x 10' cells/ml into 50 mM KC1,3 mM NaC1,2 mM M&12, 0.1 mM PMSF,' 0.1 mM dithiothreitol, 5 mM diisopropyl fluorophosphate, and 5 mM PIPES, pH 7.5, and incubated on ice for 30 min. All subsequent work was conducted at 4 "C. The cells were disrupted by nitrogen cavitation (400 p. s. i. for 20 min) and centrifuged at 12,000 X g for 10 min. The supernatant fractions were recentrifuged at 100,000 X g for 1 h and the high speed supernatant fractions, containing a majority of the recombinant p47-phox or p67phox, were then chromatographed on ion exchange media described below.
Crude recombinant p47-phox preparations were diluted to 2.5 volumes with distilled water and applied to a 10-ml column of CM-Sepharose Fast Flow (Pharmacia LKB Biotechnology Inc.) equilibrated with 5 mM potassium phosphate, 0.15 mM PMSF, pH 7.0 (CM-Sepharose buffer). Following a wash with three column volumes of the same buffer, proteins were eluted with a 0-0.3 M NaCl gradient (80 ml, flow rate 0.66 ml/min) in the same buffer.
crude recombinant p67-phox preparations were chromatographed on Q-Sepharose (Pharmacia). Briefly, the high speed supernatant was diluted to 2.5 volumes with distilled water and loaded onto a 10ml Q-Sepharose column equilibrated with 20 mM Tris-HC1, 0.1 mM dithiothreitol, 0.15 mM PMSF, pH 7.0 (Q-Sepharose buffer) and washed with 3 volumes of the same buffer. Proteins were eluted with a 0.1-0.3 M NaCl gradient (80 ml, flow rate 0.66 ml/h) using the same buffer. Fractions were assayed for total protein with bicinchoninic acid assay reagents (Pierce Chemical Co.) using bovine serum albumin as a protein standard. Fractions were also analyzed further by SDS-PAGE, immunoblotting, and the cell free superoxide generation assay described below. Both pure recombinant proteins were routinely stored at -80 "C, without noticeable losses of activity.
Immunochemical Studies-Peak fractions containing intact purified recombinant p47-phox or p67-phox (200 pg) were used as antigens to immunize goats with complete Freund's adjuvant. The goats were boosted three times at 2-week intervals using 200 pg of the antigen in incomplete Freund's adjuvant. Antibody titers of whole goat sera were determined by an ELISA assay using 2 pg of pure antigen/well and anti-goat avidin-biotin-alkaline phosphatase conjugates to detect the antigen bound goat IgG (Vectastain" ABC-AP, Vector Laboratories), using protocols described by the manufacturer. Immunoblotting of proteins from SDS-PAGE (10% acrylamide) gels (26) was performed as described previously (27). Electroblots were routinely probed with rabbit (14, 16) or goat anti-p47-phox or anti-pg7-phox sera at a 1:1000-fold dilution and developed with secondary antibodyalkaline phosphatase conjugates (1:1000) using p-nitroblue tetrazolium and 5-bromo-4-chloro-3-indolyl phosphate substrates (Bio-Rad) according to manufacturer's protocols. graphed on a Sephadex G-25 (Pharmacia) column at 4 "C using 100 mM KCl, 3 mM NaC1, 3.5 mM MgCl,, 0.15 mM PMSF, and 10 mM PIPES, pH 7.5. Fractions (1 ml) were collected at a flow rate of 0.33 ml/min and scanned for flavin fluorescence. Protein containing fractions were identified by total protein assay and immunoblotting with a mixture of anti-p47-phox and anti-p67-phx antibodies as described above. The protein in these fractions was precipitated by trichloroacetic acid (10%) or heat (100 "C, 5 min) denaturation, and the supernatants resulting from a 10-min (16,000 X g) centrifugation were analyzed for flavin fluorescence as described above.

GTP Agarose
Binding-GTP-agarose chromatography was performed at 20 "C using a 1-ml GTP-agarose (Pharmacia) column and the buffer system described previously (9). Proteins were loaded onto the column in 100 mM KCl, 3 mM NaC1, 3.5 mM MgC12, 0.15 mM PMSF, and 10 mM PIPES, pH 7.5, allowed to bind for 30 min, and then washed with the same buffer. The columns were then eluted with 1 mM GTP and 150 mM NaCl in the same buffer. Samples (25 pl) from the peak protein containing fractions were separated by SDS-PAGE, immunoblotted with a mixture of rabbit anti-p47-phox and anti-p67-phox, and developed as described above.
NADPH Oxidase Reconstitution Studies-Cell-free superoxide production was measured essentially as described previously (8), using 96-well multiwell plates and a Molecular Devices Thermomax microplate reader (Menlo Park, CAI. Unless otherwise noted, typical reactions (100 pl) contained variable amounts of pure recombinant p47phox or p67-phox, 10' cell equivalents of neutrophil cytosol, and 5 X 10' cell equivalents of deoxycholate-solubilized membranes (8), prepared from peripheral blood neutrophils following nitrogen cavitation and differential centrifugation procedures described previously (8). Reaction mixtures contained 75 mM potassium phosphate, pH 7, 0.2 mM acetylated cytochrome c, 4 mM M&12, 10 p M FAD, 5 pM GTPyS, and 200 p~ NADPH. The reactions were initiated by addition of 40 p~ arachidonic acid. Control reactions contained 5 pg of superoxide dismutase. Superoxide generation was calculated based on superoxide dismutase-inhibitable changes in cytochrome c absorbance observed at 551 nm (1-nm bandwidth filter), assuming absorbance changes of 21.1 mM" cm" for the reduced minus oxidized cytochrome e. The reactions were followed for 30 min after addition of arachidonic acid, with absorbance readings taken at 1-min intervals. Maximum rates of superoxide generation were calculated from a linear least squares fit of 10 consecutive 1-min data points. The reactions reached maximum rates within 5-20 min following the addition of arachidonic acid. Most determinations were based on reactions performed in triplicate.  (24). The baculovirus transfer vector (pVL1392) chosen for NADPH oxidase cytosolic factor production was designed with a mutation in the polyhedrin ATG translation start codon, allowing expression of native eukaryotic proteins which lack additional fused leader sequence encoded by the vector. The presumed ATG translation initiation sites (Kozak sequences) within both of the inserted cytosolic factor cDNAs occur close to the 5' ends of these subcloned EcoRI fragments, leaving only 25 base pairs of 5"untranslated sequence in pVL1392/p67 (16) and 7 base pairs of 5"untranslated sequence in the pVL1392/p47 construct (14). The resulting recombinant baculoviruses, used under optimum infection conditions, produced maximum amounts of these recombinant proteins 3-4 days post-infection, at a time which coincided with a rapid decline in cell viability. The recombinant proteins p47-phox and p67-phx were the predominant polypeptide bands observed in cavitated Sf9 cell lysates harvested at 88 h post-infection ( Figs. 1 and 2). Schemes devised for isolation of these recombinant cytosolic factors were based on structura1,data deduced from their cDNAs and earlier observations on the fractionation of these factors from human neutrophil cytosol (8, 29). Crude p47-phox preparations were chromatographed on carboxymethyl (CM) cellulose and eluted with a 0-0.3 M NaCl gradient in CM-Sepharose buffer (Fig. 1). Although many of the insect proteins eluted in the flow-through fractions of this column, a major protein peak composed of a relatively homogeneous protein with an apparent molecular mass of 47 kDa eluted at -125 mM NaC1. This observed behavior was consistent with the high PI of p47phox and the fractionation of p47-phox observed using crude human neutrophil cytosol (29,30). This protein peak also coincided with a peak in NADPH oxidase activity observed when mixed with solubilized neutrophil membranes, suboptimum amounts (lo5 cell eq) of crude neutrophil cytosol, and pure recombinant p67-phox (described below). No significant activity was observed in other fractions. Preliminary immunoblotting studies confirmed that this protein, which was present only in the BV/p47-infected cells, cross-reacted with p47-phox detected in normal neutrophils with rabbit antibodies raised against a recombinant p47-phon fusion protein produced in E. coli (14). A slightly smaller peptide, seen in the leading fractions (fractions 8 and 9) of this peak, was also detected with this antibody and was thought to represent a minor proteolytic product of recombinant p47-phox. Antibodies subsequently raised in goats against the pure baculovirusderived 47-kDa band detected the same 47-kDa band in normal neutrophils (Fig. 3) but not in p47-phox-deficient neutrophils from patients with autosomal recessive CGD (not shown). The recombinant p47-phox from peak fractions migrated precisely with the same mobility as p47-phox detected in normal neutrophil cytosol (Fig. 3), confirming that the complete coding sequence for p47-phox was contained within or for the ability to augment NADPH oxidase activity (+) when mixed with 0.5 pg of pure recombinant p67-phox, neutrophil membranes (5 X lo5 cell eq), and neutrophil cytosol (lo5 cell eq) as described under "Methods." Activity is expressed as the superoxide dismutase-inhibitable change in cytochrome c absorbance (551 nm). Base-line activity observed in the same reaction lacking recombinant protein was -3 mOD/min. (1 mOD = 0.001 optical density unit.) Inset, Coomassie Blue staining of proteins ( 2 0 4 sample volumes) in crude (Cr) and purified fractions separated by SDS-PAGE (10% acrylamide).

Expression and Purification of Recombinant Cytosolic
Kozak start sequence was only 44.7 kDa (14).
Similarly high levels of recombinant p67-phox were produced from the BV/p67 baculovirus construct described above (see "Methods"). Crude insect cell lysates containing p67phox were chromatographed on Q-Sepharose (Fig. 2) using a 0.1-0.3 M NaCl gradient. Pure p67-phx eluted around 225 mM NaC1, consistent with the observed fractionation of p67p h x from whole human neutrophil cytosol on Mono Q (8).
Peak fractions complemented the activity of recombinant p47-phox, when mixed with neutrophil membranes and suboptimal amounts of whole neutrophil cytosol in the cell-free superoxide generation assay described under "Methods." The slightly smaller polypeptides observed in the trailing edge of this major peak exhibited lower specific activity, suggesting that the activity of p67-phox is susceptible to minor proteolytic degradation. The major protein band had an apparent molecular mass of 67 kDa, demonstrating the same mobility on SDS-PAGE immunoblots as p67-phx from neutrophil cytosol (Fig. 3). This recombinant protein cross-reacted with p67-phon using antibody against the recombinant p67-phox fusion protein expressed in E. coli (16) and elicited antibodies in goats which detected p67-phon in neutrophil cytosols of all subjects, except patients with p67-phox-deficient autosomal recessive CGD. These findings confirm that the original p67phox clone (pBluescript 10) contained full coding sequence for p67-phox, although the calculated molecular mass of p67phox based on the proposed Kozak start site (first ATG codon occurring within this cDNA) was only 60.9 kDa (16). If posttranslational modifications (i.e. glycosylation) account for the discrepancies between the calculated and apparent molecular masses of these factors, it would appear that the recombinant proteins were processed in similar ways in the insect expression system, since both native and recombinant proteins demonstrated similar properties both in SDS-PAGE and ion exchange chromatography.
The ability to prepare significant amounts of both cytosolic factors in pure form and raise specific antibodies against them has enabled estimation of the amounts of these factors present in whole neutrophil cytosol (Fig. 3)   of serial dilutions of these proteins within a linear range of detection capabilities, using the pure recombinant factors as references and bovine serum albumin as the protein weight standard (see "Methods"). The standardized immunoblotting protocols, using the specific whole goat sera at a dilution of 1/1000, were capable of detecting as little as 10 and 20 ng of pure recombinant p67-phon and p47-phox, respectively (Fig.  3). The native factors were detectable in less than 100,000 cell equivalents of whole neutrophil cytosol but not in the membrane fraction of resting neutrophils. Comparison of band intensities suggested that the content of p47-phox was -0.5% of the total cytosolic protein fraction (-150 ng/106 cell equivalents), whereas p67-phox represented -0.3% of the cytosolic fraction (-75 ng/106 cell equivalents) of resting neutrophils. The crude insect cell lysates were estimated as 50-100-fold enriched sources of these factors in comparison with whole neutrophil cytosol, yielding 1-2 mg of the recombinant proteins from lo8 baculovirus-infected Sf9 cells.
FAD Binding-Early attempts to define the NADPH and flavin binding component of the oxidase, based both on isolation studies and affinity labeling experiments (31)(32)(33)(34)(35)(36), suggested that this protein was 65-67 kDa in size (for review see Ref. 37). Others (38) recently suggested that the 66-kDa protein labeled with an NADPH dialehyde analogue in membranes is not a component of the oxidase and that the NADPH binding site is a soluble protein, based on inactivation of cytosol with this compound. A 45-kDa polypeptide was identified by affinity labeling with diphenyleneiodonium, a potent oxidase inhibitor (39). Antibodies raised against this purified protein detected it both in cytosol and membrane fractions (40). In light of these observations, the possible binding of FAD to the recombinant p47-phox and p67-phor was examined. Neither of the pure factors, isolated from insect cell lysates by the ion exchange chromatography steps described above, exhibited detectable flavin fluorescence, characterized by an excitation wavelength maximum at 450 nm and an emission wavelength maximum at 525 nm. Furthermore the fluorescence spectra of 10 PM FAD solutions were unaffected when combined with both pure recombinant proteins (0.5 mg of each). In order to examine further whether recombinant p47-phox and p67-phox could bind exogenously added flavin, these proteins were incubated with 20 PM FAD and separated 19815 by gel filtration on Sephadex G-25 (Fig. 4). Although flavoproteins typically bind FAD with very high affinities, the methods used here should detect even moderate binding affinities, based on the sensitivity of the fluorescence assay (<lo-' M) and the amounts of recombinant proteins used. No flavin fluorescence was detected in the void volume fractions which contained all of the recombinant p47-phon and p67-phox. In order to rule out the possibility of quenching of protein bound flavin fluorescence, the proteins in these fractions were precipitated by trichloroacetic acid (10%) or heat (100 "C) denaturation, and the resulting supernatants were analyzed for free flavin. The low amount of flavin fluorescence detected (< 0.05 mol of flavin/mol of protein) suggested that neither p47-phox nor p67-phox are flavoprotein components of the oxidase. Consistent with the absence of detectable flavin in these recombinant preparations, neither protein contains the FAD and nucleotide binding consensus sequence elements that are shared by a number of flavoproteins including ferrodoxin NADP+ reductase, cytochrome P450 reductase, and cytochrome b5 reductase (41). The oxidase activity supported by these pure cytosolic factors was not significantly affected in the absence of exogenous FAD, although significant endogenous flavin fluorescence was detected in the membranes used in these reactions (data not shown). Although direct binding of flavin to these pure active recombinant factors was not observed, the possibility that both factors represent only a part of a flavoprotein complex involving other proteins could not be ruled out.
GTP-agarose Binding-Work from several laboratories has supported the involvement of GTP-sensitive components in NADPH oxidase activation (9,29,(42)(43)(44)(45)(46)(47). GTP and GTP analogues were noted to enhance oxidase activity assayed in vitro (29,(42)(43)(44)(45)(46)(47), and attempts to identify the GTP binding component in cytosol revealed that soluble oxidase components from resting cells were retained on GTP-agarose and could be eluted by 1 mM GTP and high salt (9). Although the bound fraction contained a large number of protein species, antibodies raised against this preparation were specific for p47-phox, p67-phox, and one other minor peptide (9). In order to address the question of whether p47-phox or p67-phon interact directly with GTP-agarose and represent the putative GTP-sensitive oxidase components, the pure baculovirusderived recombinant factors were reexamined under the same conditions of GTP-agarose chromatography reported earlier  (9). In control experiments (Fig. 50) both of the native factors from whole neutrophil cytosol were efficiently bound and eluted with 1 mM GTP and high salt (0.15 M NaCl), consistent with the earlier observations. This binding was affected by ionic strength, in that the inclusion of 0.15 M NaCl to the loading buffer inhibited complete binding of p47-phox and p67-phox, and the omission of salt from the GTP containing elution buffer resulted in significantly decreased recoveries. Pure recombinant p47-phox was also retained completely on GTP-agarose and was subsequently eluted with the same buffer containing 1 mM GTP plus 0.15 M NaCl (Fig. 5A), similar to that observed with whole cytosol. In contrast, pure recombinant p67-phox did not bind to this column and was found entirely in the flow through fractions (Fig. 5B). When the recombinant proteins were combined and preincubated (20 min, 37 "C) and subjected to the same chromatographic procedures, their behavior was similar to that observed with whole cytosol, binding completely and eluting subsequently in the GTP high salt wash (Fig. 5C), suggesting that recombinant p67-phox is retained by an association with p47-phox. Whether the observed binding reflects a specific interaction between p47-phox and GTP is unclear, although the high PI of p47-phox and the inhibitory effects observed at high ionic strength may explain GTP-agarose binding in simple electrostatic terms. Neither protein contains GTP binding consensus sequence elements common to a number of GTP binding Flow-through fractions represent proteins eluted in the washing buffer. The GTP eluate was obtained from a step containing 1 mM GTP and 150 mM NaCl in the same buffer. Eluted fractions were analyzed by SDS-PAGE on 10% acrylamide gels and immunoblotted using a mixture of rabbit anti-p47-phox and anti-p67-phox as described under "Methods." proteins, including EF tu, ras, and the Ga subunits of heterotrimeric G proteins (48). If all of the soluble oxidase components exist as a stable complex in resting cell cytosol, the observed behavior of the two recombinant factors may account for binding of all active components in whole cytosol. However, these findings do not rule out the possibility that other soluble components may interact directly and specifically with GTP or GTP-agarose; indeed recent work (29) has shown that other soluble components may translocate to membranes from resting cell cytosol in the presence of GTP, independent of p47-phon and p67-phon. The finding that pure recombinant p47-phox and p67-phox interact together with the GTP-agarose matrix supports the notion that p47-phox and p67-phox can interact as a complex in resting cytosol (49), even prior to activation. Whether the other factors are part of this complex or bind to GTP-agarose independent of p47-phox and p67-phox is unclear at this time.
Reconstitution Studies-Early work in several laboratories (50-53) demonstrating reconstitution of an active neutrophil NADPH oxidase in a fully soluble cell-free system has shown that the active enzyme is assembled from membrane bound and several soluble or cytosolic components. This assay system has allowed the characterization of the pure baculovirus derived cytosolic factors in supporting superoxide production in vitro (Figs. 6 and 7). In these studies the activity of membrane preparations proved to be the most variable component of the cell-free system, therefore the reactions compared were conducted within a few hours using the same membrane preparation throughout, in order to maximize reproducibility and examine the variables affecting the activities of the cytosolic components. Optimized control reactions, which contained lo6 cell eq of normal cytosol and 5 X lo5 cell eq of membranes (-1 pg), produced 0.7-1.5 nmol of superoxide/min (Figs. 6 and 7). The factor deficient cytosols used to directly test the activities of recombinant p47-phox and p67phox were prepared at comparable concentrations (lo6 cell equivalents/reaction) and were chosen from autosomal recessive CGD patients (patients D. C. and J. H.) who were studied previously by complementation with other patient cytosols (8) and recombinant bacterial fusion proteins (14, 16). The kinetics of cell free superoxide generation were compared between normal cytosol and CGD cytosols which were restored by the complementary recombinant proteins (Fig. 6). Addition of less than 1 pg of the appropriate factors to these deficient cytosols (IO6 cell eq) was sufficient to restore cell free oxidase activity to nearly normal levels in both cases (Fig. 7). The kinetic profiles exhibited by these restored CGD cytosols exhibited the same properties as normal reactions (Fig. 6). Immediately following the addition of arachidonic acid, the final component used to activate these reactions, the rate of superoxide production showed a characteristic lag phase, followed by a linear phase (steady state), which was sustained for 10-20 min. The superoxide dismutase-insensitive component of cytochrome c reduction was minor, typically representing 10-15% of the total change in cytochrome c absorbance. The yield of oxidase in this cell-free system was expressed in terms of the maximum "steady state" rate of superoxide production observed over the linear portion of this kinetic time course, consistent with the interpretations of others (54).
Reconstitution of the oxidase was explored over a range in concentrations of recombinant proteins used in these reactions (Fig. 7). The maximum rate of superoxide production observed showed an approximately linear dependence on the amount of protein added over a narrow concentration range (<400 ng/100 ul reaction). At higher concentrations (up to 3 added to 100-pl reactions containing 0.5 X 10' cell eq membrane and various whole neutrophil cytosol preparations described in Fig. 6. A , p67-phox-deficient cytosol (lo6 cell eq; B, p47-phox-deficient cytosol (lo6 cell eq); C, normal cytosol, lo6 cell eq (A), 1 0 6 cell equivalents (V), or no added cytosol (X). Oxidase activity is derived from maximum rates (10 consecutive 1-min data points) of superoxide dismutase-sensitive cytochrome c reduction observed over the reaction time course described in Fig. 6. The data, shown here as the mean of triplicate reactions, showed less than 15% variance at each concentration.
pg/reaction) a plateau in oxidase yield was observed. Maximum restoration of the p47-phox-or p67-phox-deficient cytosols occurred in the range of 0.4-1 pg/lOO-pl reaction of the complementary recombinant proteins, approaching the levels of activities seen with normal cytosol reactions. The addition of recombinant p47-phox-to the p67-phox-deficient cytosol or recombinant p67-phox-to the p47-phox-deficient cytosol had no affect in restoring oxidase activity, confirming the distinct, yet essential, roles played by both oxidase factors. Addition of recombinant p47-phox and p67-phx to the control normal cytosol reaction resulted in only slightly enhanced levels of superoxide production (Fig. 7C). In contrast, it was noted previously that addition of excess whole cytosol decreased the yield of active oxidase, suggesting the presence of oxidase inhibitors or competing factors in crude cytosol (8, 53). Estimates of the specific activities of both recombinant factors, relative to the factors present in normal whole neutrophil cytosol (-75-150 ng/106 cells) suggested that a significant fraction of both recombinant proteins was active. The affinity of the reconstituted oxidase for NADPH was consistent with earlier reports (31,34,36,54,551, with an apparent K,,, of -35 M (not shown). The activity of both recombinant factors was maintained after prolonged storage at -70 "C, although recombinant p67-phox was considerably less stable than p47-phox at 4 "C. Mild heating (42 "C, 7.5 min) of pure recombinant p67-phx causes a dramatic irreversible loss in its activity. p67-phx is also sensitive to sulfhydryl (N-ethylmaleimide) and amino group modifying reagents (succinimidyl esters), whereas recombinant p47-phox is quite resistant to such treatments (data not shown). Although the reason for the remarkably higher activities exhibited by the purified baculovirus derived factors, compared with recombinant p47p h x and p67-phx fusion proteins produced in E. coli, (14, 16) was not directly addressed, several factors could account for these differences. These include the presence of oxidase inhibitors in the crude E. coli lysates (14, 16), improper protein folding in the prokaryotic expression system, inhibitory effects of the amino-terminal leader sequences in the bacterial fusion proteins, or undetermined differences in post-translational processing occurring in the two expression systems. Although both recombinant factors demonstrated high specific activities in restoring autosomal recessive CGD cytosols, the two factors together were not sufficient in replacing whole cytosol from normal resting neutrophils. Insignificant amounts of superoxide were detected if whole cytosol was omitted from reactions containing normal membranes and both recombinant factors at concentrations that exceeded those adequate in restoring CGD cytosols (Fig. 7C). The participation of other cytosolic components in cell-free superoxide generation was clearly demonstrated with addition of small amounts of whole normal cytosol. The activity of a reaction using lo5 cell eq of whole cytosol (-10% of cytosol required in an optimized reaction) was significantly augmented by addition of both of the recombinant factors (Fig.  7C). Excessive amounts of both rp47-phox and rp67-phox (2-3 pg each) supported superoxide production to levels approaching the activity of lo6 cell eq of whole cytosol (not shown), suggesting that the other complementary component(s) may be present in excess relative to p47-phox and p67-phox. The fraction of these soluble factors that actually assemble into the active oxidase and the stoichiometry of the complex assembled in vitro are unknown at this time, although recent reports (56, 57) have shown that only a minor portion of p47-phox and p67-phox translocate and assemble into the active oxidase in intact cells. Whether components in the active complex readily exchange or turnover in vitro and whether excess amounts of the exogenous factors can drive assemblv of the other comDonents through mass action I are issues currently under investigation. Preliminary experiments aimed at identifying the protein(s) which complement the activity of the recombinant cytosolic factors have indicated that a protein with properties similar to NCF3 (8), ul (58), or SOC 1 (29) may be the only additional component that is necessary for reconstitution in vitro (59). The availability of these recombinant factors in pure and highly active form should facilitate the identification of these complementary oxidase components and enable further studies on the kinetics and mechanisms of oxidase assembly.