Functional Analysis of NADPH Oxidase in Granulocytic Cells Expressing a ▵488-497 gp91^phox Deletion Mutant

The Δ488-497 gp91^phox cDNA, which has an in-frame deletion of 30 nucleotides,16 was generated by oligonucleotide-directed mutagenesis by using the Sculptor in vitro mutagenesis kit (Amersham, Arlington Heights, IL). The mutant cDNA was verified by dideoxynucleotide sequencing and then subcloned into the Not I site of the mammalian expression vector, pEF-PGKpac. The pEF-PGKpac vector contains a mammalian EF-1α promoter to drive constitutive gp91^phox expression and a linked expression cassette for puromycin-N acetyltransferase.18 In parallel, the vector containing the full-length human wild-type (WT) gp91^phox cDNA was also constructed. The WT or the Δ488-497 gp91^phox-containing vectors were transfected by electroporation into X-CGD PLB-985 cells.17 Clones were selected by limiting dilution in the presence of puromycin (1 μg/mL). To minimize any potential clone-to-clone variation in recombinant gp91^phox expression or NADPH oxidase activity, 3 independent clones determined to express relatively higher levels of recombinant gp91^phox were pooled and used for subsequent analysis.

X-CGD PLB-985 cells (X-CGD), transfected PLB-985 cells expressing WT, or the deletion mutant (Δ488-497) gp91^phox were maintained in RPMI 1640 medium containing 10% fetal calf serum and 2 mmol/L L-glutamine. To induce expression of endogenous NADPH oxidase subunits, cells were differentiated for 5 days by exposure to 0.5% dimethylformamide (DMF). Under these conditions, more than 80% of the cells had undergone granulocytic differentiation as determined by observation of morphological changes and nitroblue tetrazolium (NBT) test.12 

To evaluate cell surface expression of flavocytochromeb₅₅₈, immunostaining with the 7D5 monoclonal antibody was performed as described.5,18 After staining, 3,000 to 5,000 cells were deposited on glass slides by centrifugation at 450 rpm for 5 minutes and were observed by confocal microscopy. Expression of recombinant gp91^phox and p22^phox was also determined by immunoblotting as described previously.12 

Granulocyte-differentiated cells (1 × 10⁸) were collected, washed, and then resuspended in 1 mL of relaxation buffer consisting of 10 mmol/L PIPES, pH 7.3, 100 mmol/L KCl, 3.5 mmol/L MgCl₂, 3 mmol/L NaCl, and 1 mmol/L EGTA. To activate the NADPH oxidase assembly, cells were treated with either phorbol 12-myristate 13-acetate (PMA) or dimethyl sulfoxide (DMSO) vehicle control at a final concentration of 500 ng/mL for 10 minutes at 37°C. After the addition of 12 mL of cold phosphate-buffered saline (PBS) to stop the activation, cells were pelleted, resuspended in 1.5 mL of the relaxation buffer, and disrupted by sonication for 3 times at 6 seconds each at 20% power at 4°C. Subsequently, the sonicates were spun for 8 minutes at 500g, followed by 10 minutes at 2,000g. A total of 0.75 mL of the resulting supernatant was loaded on a discontinuous sucrose gradient (1.5 mL of 20% over 1.5 mL of 38%) and centrifuged at 41,000 rpm (204,275g) in an SW55 rotor (Beckman Instruments, Fullerton, CA) for 40 minutes at 4°C. After centrifugation, the 0.6 mL fraction of the top of the gradient was collected as cytosol, and a distinct band located in the interface of the 20% and 38% sucrose gradient was collected as the plasma membranes (∼0.6 mL). To remove sucrose, the membranes were mixed with 3.5 mL of cold PBS and centrifuged at 55,000 rpm (368,000g) for 30 minutes, and the resulting pellets were resuspended in 100 μL of relaxation buffer. Translocation of cytosolic oxidase components p47^phox and p67^phox to plasma membrane was detected by immunoblotting as described previously.12 Translocation assay was also performed in a cell-free system. Briefly, 100 μg of cellular membranes was mixed with 300 μg of cytosol isolated from granulocyte-differentiated cells indicated in the figure legends in 0.5 mL of relaxation buffer containing 10 μmol/L GTPγs. After the addition of 100 μmol/L sodium dodecyl sulfate (SDS), the mixture was incubated for 5 minutes at 37°C and then centrifuged for 30 minutes at 55,000 rpm (368,000g) at 4°C. The resulting pellets were collected and used for immunoblotting.

O₂⁻ production by granulocyte-differentiated cells was measured both in whole cells and in the cell-free oxidase assay by monitoring the reduction of cytochrome c at 550 nm using a Thermomax microplate reader.12 In the assay using intact cells, PMA at a final concentration of 0.1 μg/mL was used to activate the NADPH oxidase of granulocyte-differentiated cells (a total of 2.5 × 10⁵ cells in 200 μL of volume in a well). The cell-free assays using SDS as activator were performed as described previously,19,20 using flavocytochromeb₅₅₈ partially purified from membranes of the PLB-985 cell lines and cytosol from neutrophil or granulocyte-differentiated PLB-985 cells. Membrane and cytosolic fractions were prepared by continuous centrifugation followed by cell disruption by sonication.21 Michaelis-Menton kinetics were analyzed using GraphPad Prism (San Diego, CA). For peptide inhibition assays, peptides were dissolved in assay buffer and added to the reaction mixture before the addition of SDS (100 μmol/L). Protein concentration was determined by BCA assay (Pierce, Rockford, IL). INT reductase activity was measured as described previously.20 

Flavocytochrome b₅₅₈ was partially purified from 3 × 10⁹ cell equivalents of salt-washed PLB-985 cell membranes using the method described previously for flavocytochromeb₅₅₈ purification from neutrophil membranes.22 This method uses mixed-bed (carboxylmethyl [CM], diethyl aminoethyl [DEAE] Sepharose CL-6B, amin-octyl agarose) and heparin chromatography. Relipidation and reflavination of the partially purified flavocytochromeb₅₅₈ was also performed as described previously using phosphatidylcholine (type IIS; Sigma, St Louis, MO). For the determination of FAD binding constants, portions of the partially purified cytochrome were relipidated in the absence of FAD.

Reduced minus oxidized difference spectra of detergent-solubilized membranes and partially purified flavocytochromeb₅₅₈ were recorded as described previously using a Perkin-Elmer Lamda 18 spectrophotometer (Perkin-Elmer, Norwalk, CT).13 

The dissociation constant for FAD binding by the flavocytochromeb₅₅₈ preparations was determined by equilibrium dialysis using Sialomed equilibrium dialyzers (AmiKa Corp, Columbus, MD). One-hundred–microliter aliquots (130 to 150 nmol/L in concentration) of partially purified flavocytochromeb₅₅₈ samples that had been relipidated (but not reflavinated) were placed in one side of the dialysis chamber and dialyzed against 100 μL of the same buffer containing 100 nmol/L FAD. FAD standards were made by serial dilution of a freshly prepared stock solution of FAD. The concentration of the stock solution was determined from the absorbance at 450 nm using an extinction coefficient of 11.3 mmol/L⁻¹. All FAD-containing solutions were protected from light. The concentration of flavocytochromeb₅₅₈ was determined spectrophotometrically using an extinction coefficient of 21.6 mmol/L⁻¹ at 559 nm for the reduced-minus oxidized heme (10.8 mmol/L⁻¹ per mole flavocytochrome b₅₅₈).22 Samples were dialyzed on ice for 4 hours, and the contents of each chamber were removed for FAD analysis. All samples and FAD standards were heated in a 100°C water bath for 3 minutes to release enzyme-bound FAD and centrifuged to remove denatured protein.

FAD was estimated using a modification of the method of Hinkkanen and Decker.23 The assay mixture consisted of 80 to 100 μL of sample (or FAD standard), 20 mmol/L 3, 5-dichlorobenzene sulfonic acid, 200 μmol/L 4-amino antipyrene, 4 U/mL horseradish peroxidase, 0.2 U/mL apo-D-amino acid oxidase, and 35 mmol/L D-proline in a total volume of 250 μL 100 mmol/L Tris, pH 8.6. The assay was performed at 37°C and the rate of formation of N-(4-antipyryl)-3-chloro-5-sulfonate-p-benzoquinone monoimine was observed for 60 minutes by measuring the increase in absorbance at 512 nm. The FAD concentration of the samples was calculated from a standard curve of 0 to 200 nmol/L FAD plotted against maximum rate of ΔA₅₁₂. The K_d (FAD) for each sample was calculated from the final concentrations of FAD in the sample and buffer compartments, and the concentration of flavocytochromeb₅₅₈ was added.

Stable expression of recombinant Δ488-497 gp91^phox in X-CGD PLB-985 cells mimicked the phenotype originally reported for neutrophils isolated from an X-CGD patient with the same Δ488-497 deletion in gp91^phox.16 Expression of Δ488-497 gp91^phox in transfected PLB-985 cells was examined by immunoblotting (Fig 1B). The level of recombinant Δ488-497 gp91^phox was similar to that of recombinant WT gp91^phox expressed in PLB-985 cells. A marked increase in expression of p22^phoxwas seen in both transgenic Δ488-497 gp91^phox and WT gp91^phox PLB-985 cells (Fig 1B). This is consistent with previous observations indicating that coexpression of both gp91^phox and p22^phox and subsequent heterodimer formation is important for stable expression of each flavocytochrome b₅₅₈subunit.11,12,24 We have previously shown that the transgenically expressed recombinant WT gp91^phox is processed and targeted normally into the plasma membrane in promyelocytic PLB-985 cells.18 To determine whether Δ488-497 gp91^phox was expressed in the plasma membranes, transfected cells were stained with 7D5, a monoclonal antibody that interacts with an extracellular epitope of gp91^phox, and examined by confocal microscopy. As shown in Fig 1A, membrane surface staining was present in both Δ488-497 gp91^phox as well as WT gp91^phox PLB-985 cells. No positive signal above the background was obtained in X-CGD PLB-985 cells, consistent with the absence of gp91^phox in the cells (Fig 1B).

We next performed reduced minus oxidized difference spectroscopy on transgenically expressed flavocytochrome b₅₅₈partially purified from membranes isolated from Δ488-497 gp91^phox, WT gp91^phox, and X-CGD PLB-985 cells. Virtually identical spectra characteristic of flavocytochrome b₅₅₈ were seen for both the Δ488-497 gp91^phox and WT gp91^phox flavocytochrome preparations, demonstrating the normal incorporation of heme groups in the gp91^phox deletion mutant (Fig 2). As expected, X-CGD samples lacked specific absorption at 558 nm (Fig 2).

NADPH oxidase activity in Δ488-497 gp91^phoxPLB-985 cells was determined in both intact cells and in the cell-free oxidase assay. After granulocytic differentiation for 5 days to induce the expression of the endogenous p47^phox and p67^phox oxidase subunits, WT gp91^phox PLB-985 cells produced O₂⁻ after stimulation with PMA, as expected (Table 1). In contrast, Δ488-497 gp91^phox PLB-985 cells were unable to generate O₂⁻ (Table 1). To confirm that the cellular defect in the NADPH oxidase in Δ488-497 gp91^phox PLB-985 cells was related to the mutation in gp91^phox, cell-free oxidase assays were performed using combinations of cytosol and membranes prepared from WT gp91^phox and Δ488-497 gp91^phox PLB-985 cells. As shown in Table 2, membranes isolated from Δ488-497 gp91^phox PLB-985 cells failed to support O₂⁻ generation in combination with cytosol from either WT gp91^phox or Δ488-497 gp91^phox PLB-985 cells, demonstrating that absence of NADPH oxidase activity resulted from a defect in the cellular membranes containing the mutant flavocytochromeb₅₅₈.

Upon phagocyte activation, cytosolic oxidase components translocate to the plasma membrane to assemble the functional NADPH oxidase. Multiple contact points between p47^phox and the gp91^phox and p22^phox subunits of flavocytochrome b₅₅₈ have been described previously.14,25-29 Among these, a missense mutation predicting an Asp-Gly substitution at residue 500 of gp91^phox has been reported to lead to defective translocation of p47^phox and p67^phox.14 The proximity of the Δ488-497 deletion in gp91^phox to Asp500 prompted us to test whether translocation of p47^phox and p67^phox to the plasma membrane during oxidase assembly was affected by this deletion. As shown in Fig 3A, PMA-stimulated translocation of p47^phox and p67^phox to the plasma membrane in intact Δ488-497 gp91^phoxPLB-985 cells was similar to that seen for WT gp91^phox PLB-985 cells. A similar result was obtained in a cell-free oxidase reconstitution system using SDS for activation (Fig 3B). The band seen below the 68 kD marker protein is the high mannose 65-kD precursor of gp91^phox.24 We have previously shown that this species is localized in the ER, as determined by cell fractionation using a 10% to 60% continuous sucrose gradient.24 The membranes isolated by discontinuous sucrose gradient (20% and 38%) in the current experiment may be contaminated with intracellular membranes including ER, which may account for the presence of the precursor in the preparation (Figs 1B and 3B). A small amount of p47^phox but not p67^phox was seen in WT and Δ488-497 gp91^phox-transfected PLB-985 granulocytes even in the absence of PMA, as well as in X-CGD PLB-985 cells (Fig 3A and B). This likely reflects nonspecific binding of p47^phoxto membranes, which has also been observed by others.30Alternatively, an unexpected priming of PLB-985 granulocytes during culture at 37°C may cause the translocation of p47^phox to the membrane, which is flavocytochrome-independent.

To provide additional evidence that gp91^phoxresidues 488-497 are not essential for oxidase assembly, we synthesized a peptide corresponding to residues 488-497 and tested its ability to inhibit O₂⁻ production in the cell-free oxidase assay using membrane and cytosol isolated from normal neutrophils. As shown in Fig 4, peptide 488-497 inhibited oxidase activity only at high concentrations (IC50, ∼500 μmol/L), whereas a peptide derived from residues 86-102 of gp91^phox (containing a probable p47^phox binding motif28,29) had an IC50 of 2 μmol/L (Fig 4). The 488-497 peptide was also much less potent at inhibiting O₂⁻ production compared with other peptides derived from gp91^phox domains proposed as binding sites for cytosolic oxidase components, including peptide 491-504 containing Asp500 (IC50, 10 μmol/L),14peptide 559-570 (IC50, 28 μmol/L),31 and peptide 550-569 (IC50, 4 μmol/L).32 Consistent with our results, Kanegasaki’s group has reported that IC50 of peptide 484-502 was greater than 300 μmol/L.32 

We have previously shown that electrons are transferred to INT primarily from the flavin center of gp91^phox.4 We have used this method to show that the proximal electron transfer pathway (from NADPH to flavin) is normal in a mutant form of gp91^phox in which there is an amino acid substitution affecting one of the heme redox potentials.13 It was therefore of interest to see if the proximal electron transport pathway of the Δ488-497 mutant was functional. Partially purified flavocytochrome b₅₅₈from membranes of WT gp91^phox PLB-985 cells was able to support INT reductase activity as efficiently as flavocytochrome b₅₅₈ purified from neutrophils (not shown). In contrast, flavocytochrome b₅₅₈ from Δ488-497 gp91^phox PLB-985 cells was incapable of INT reductase activity, suggesting that either flavin or NADPH binding is affected in this mutant. As expected, the equivalent purification fraction from X-CGD PLB-985 cells also had no activity (Fig 5).

To evaluate the capacity of the Δ488-497 gp91^phoxmutant to bind FAD, we measured the affinity of the partially purified flavocytochrome b₅₅₈ preparations for FAD by equilibrium dialysis as described in the Materials and Methods. Both WT gp91^phox and Δ488-497 gp91^phox flavocytochrome preparations had virtually identical affinities for FAD of approximately 66 nmol/L (Table 3). These values are consistent with the literature values of 20 to 85 nmol/L.33-35 An equivalent volume of the X-CGD PLB fraction eluted from the heparin column that corresponded to the peak fractions of the Δ488-497 gp91^phox and WT gp91^phoxflavocytochrome b₅₅₈ showed no ability to bind FAD. In all cases, the recovery of FAD from each equilibrium dialysis experiment (sample + buffer) was 100% ± 4%.

Residues 488-497 have been postulated to lie near the NADPH binding domain of gp91^phox.36 To address whether the Δ488-497 mutation alters the affinity of flavocytochromeb₅₅₈ for NADPH, we measured the K_m for NADPH on mutant flavocytochrome partially purified from membranes of Δ488-497 gp91^phox PLB-985 cells. The NADPH oxidase has the ability to use NADPH or NADH as substrate, although the K_m for NADPH is approximately 10-fold lower (∼40 μmol/L).37-39 The K_m for NADPH of the enzyme in the cell-free system using neutrophil membranes was found to be 57.1 ± 1.8 μmol/L (n = 22), and the K_m of purified neutrophil flavocytochrome b₅₅₈ was 27.5 ± 1.3 μmol/L (n = 17; A. Cross, unpublished data). The K_m of the flavocytochrome purified from membranes of WT gp91^phox PLB-985 cells was found to be 30.4 ± 1.5 μmol/L (not shown). No activity was evident in the cytochrome purified from the Δ488-497 gp91^phox PLB-985 cells. Increasing the substrate concentration to as high as 4.9 mmol/L NADPH did not induce any O₂⁻ formation from the Δ488-497 mutant and neither did the addition of 4.9 mmol/L NADH (Fig 6). This suggests that either the Δ488-497 gp91^phox cannot bind the substrate NADPH or it cannot support electron transfer from NADPH to FAD.

Over the past 10 years, several groups have tried to identify the NADPH binding component of the oxidase complex by labeling with³²P- or ³H-labeled NADPH analogues. However, different results have been reported by different groups, including a 66-kD cytosolic protein,40 an approximately 32-kD cytosolic protein,41 a 52-kD membrane-associated protein,42 and p67^phox,9therefore leaving the issue still uncertain. By comparison with known NADPH binding regions of members of Ferredoxin-NADP⁺Reductase family, Segal’s group,6 Rotrosen et al,8 and Sumimoto et al7 have proposed that the apparent NADPH binding pocket resides in the carboxyl terminal portion of gp91^phox. This postulation has subsequently been strongly supported by the experiments showing that reflavinated and relipidated membrane fractions isolated from normal neutrophils are capable of supporting O₂⁻ generation in the absence of any cytosolic proteins.43,44 

We attempted to analyze the ability of the Δ488-497 mutant flavocytochrome b₅₅₈ to bind NADPH by affinity labeling using the photoaffinity label [4-N-(4-azido-2-nitrophenyl) aminobutyryl] NAD[³²P].45 Despite using a number of different experimental conditions, it was not possible to convincingly, or reproducibly, label WT gp91^phox from either neutrophils or WT gp91^phox PLB-985 cell membranes. Therefore, we were not able to unambiguously determine if the Δ488-497 mutant can bind NADPH.

Taylor et al36 have predicted a 3-dimensional structure of gp91^phox using Ferredoxin-NADP⁺Reductase as a template and proposed that an α-helical loop composed of residues 484-503 lies over the NADPH binding cleft. During oxidase activation, access of NADPH into the binding site could potentially be regulated by interactions of this loop with cytosolic oxidase components.36 Although the impaired translocation of cytosolic oxidase subunits reported for a gp91^phoxmutant with Asp500Gly substitution indirectly supports this hypothesis,14 we found no evidence that gp91^phox residues 488-497 are required for interactions with cytosolic oxidase components during oxidase assembly. Deletion of these residues did not affect translocation of p47^phox and p67^phox, and the peptide corresponding to residues 488-497 was a weak inhibitor of superoxide production in the cell-free oxidase assay. It therefore is likely that the 488-497 deletion in gp91^phox either leads to a failure of NADPH to bind or a defect in the subsequent transfer of electrons to FAD. Either would be consistent with the observed absence of INT reductase activity for the Δ488-497 gp91^phox flavocytochrome b₅₅₈. The data presented here further imply that simply removing the loop is insufficient to expose a functional NADPH binding site.

The majority of patients with CGD present in early childhood with severe, recurrent bacterial and fungal infections. However, the CGD patient with the Δ488-497 gp91^phox mutation was in good health until 69 years of age, when he developed an infection due to Burkholderia cepacia.16 Patient neutrophils were reported to have trace amounts of O₂⁻ production, which may have accounted for this milder phenotype, although we could not confirm the presence of residual O₂⁻-generating activity in Δ488-497 gp91^phox PLB-985 cells. Also, note that the grandson of this patient died at 5 years of age due toBurkholderia cepacia pneumonia,16 suggesting that other factors influencing host defense may have accounted for the late age of presentation in the index patient.

The authors thank Drs Dirk Roos and Arthur J. Verhoeven (Central Laboratory of The Netherlands Blood Transfusion Service, Amsterdam, The Netherlands) for kindly providing anti-gp91^phox and anti-p22^phoxmonoclonal antibodies 48 and 449, respectively. In addition, Dr David Lambeth (Emory University, Atlanta, GA) provided polyclonal anti-p47^phox and Dr Paul Heyworth (The Scripps Research Institute, La Jolla, CA) provided polyclonal anti-p67^phox antibodies. The anti-flavocytochrome b₅₅₈ monoclonal antibody 7D5 was a generous gift from Dr Michio Nakamura (Nagasaki University, Nagasaki, Japan).