Probing the Role of the Carboxyl Terminus of the gp91 phox Subunit of Neutrophil Flavocytochromeb 558 using Site-directed Mutagenesis*

Site-directed mutagenesis was used to generate a series of substitutions and deletions in the carboxyl-terminal 11 residues of gp91 phox , the 91-kDa subunit of the phagocyte NADPH oxidase flavocytochrome b 558. This region encompasses 559RGVHFIF565, implicated as a contact point for the cytosolic oxidase subunit p47 phox during oxidase activation, and a carboxyl-terminal phenylalanine (Phe570), which corresponds in position to a highly conserved aromatic residue that interacts with the flavin group in the ferredoxin-NADP+ reductase flavoenzyme family, of which gp91 phox is a member. Mutant proteins were expressed in human myeloid leukemia cells which lack expression of endogenous gp91 phox due to targeted disruption of the X-linked gp91 phox gene. Although specific residues within559RGVHFIF565 had previously been identified by alanine scanning as essential for peptide inhibition of oxidase activity in a cell-free assay, comparable substitutions in the gp91 phox polypeptide had either no or only a modest effect on oxidase activity in whole cells. Replacement of nonpolar with polar or charged residues had greater effects on oxidase activity, but were also associated with decreased gp91 phox expression, suggesting that overall protein structure was perturbed. No stable gp91 phox protein was detected upon deletion of the terminal 11 amino acids. Alanine substitution or deletion of the carboxyl-terminal Phe570 in gp91 phox resulted in a 2-fold reduction in superoxide production. This contrasts with a ≈300–800-fold reduction reported for comparable mutations in pea ferredoxin-NADP+ reductase, which suggests that structural or functional differences exist between the carboxyl terminus of gp91 phox and other ferredoxin-NADP+ reductases.

Neutrophils and other phagocytic leukocytes possess an NADPH oxidase (respiratory burst oxidase) that generates large quantities of superoxide during the respiratory burst (1). Upon phagocyte activation by opsonized bacteria or other inflammatory stimuli, the active NADPH oxidase complex is rapidly assembled at the plasma membrane from cytosolic and membrane components to catalyze the transfer of electrons from NADPH to molecular oxygen. Oxidase subunits include two polypeptides, gp91 phox and p22 phox , which form a membrane-bound phagocyte flavocytochrome b 558 heterodimer, and three cytosolic proteins, p47 phox , p67 phox , and a low molecular weight GTP-binding protein, Rac (1)(2)(3). A fourth cytosolic protein, p40 phox , is present as a complex with p67 phox in resting neutrophil cytosol (4,5), but does not appear to play a direct role in superoxide production. Superoxide and its derivatives are essential for normal microbicidal activity, and genetic defects in the NADPH oxidase result in chronic granulomatous disease (CGD), 1 a syndrome characterized by life-threatening fungal and bacterial infections (1). Mutations in the X-linked gene for gp91 phox account for the majority of cases of CGD, and the p22 phox subunit is defective in an uncommon autosomal recessive subgroup of CGD (1). The remaining cases of autosomal recessive CGD result from mutations in the genes encoding p47 phox or p67 phox (1).
Current evidence suggests that flavocytochrome b 558 functions as the redox center of the NADPH oxidase, and is regulated by the cytosolic oxidase subunits p47 phox , p67 phox , and Rac, which translocate to the cell membrane upon neutrophil activation (6 -13). The p47 phox subunit appears to mediate the first steps of interaction with the flavocytochrome in assembling the active oxidase complex (6,7,14), although it is not required in vitro if high levels of p67 phox and Rac are supplied (15,16). Sites within flavocytochrome b 558 that function as contact points with p47 phox have been identified in both the gp91 phox and p22 phox subunits (2,3). The cytochrome has an NADPH-binding site and bears a flavin group that acts as the initial acceptor of a pair of electrons from NADPH (17)(18)(19)(20)(21)(22)(23), although recently p67 phox has also been reported to contain a functional NADPH-binding site (24). The subsequent one-electron transfer to molecular oxygen is mediated by a pair of heme groups in the flavocytochrome heterodimer that are embedded within the membrane (25)(26)(27)(28).
The gp91 phox subunit of flavocytochrome b 558 is a 570-amino acid membrane glycoprotein with multiple hydrophobic domains in the amino-terminal half followed by a hydrophilic carboxyl terminus (1). The extreme carboxyl terminus of gp91 phox , which resides at the cytoplasmic face of the membrane (29,30), has been implicated as both a docking site for the cytosolic oxidase subunit p47 phox and as a participant in NADPH binding. A variety of approaches have suggested that gp91 phox residues 559 RGVHFIF 565 interact with p47 phox at an early step of oxidase assembly (14,29,(31)(32)(33)(34). Residues Ϸ400 -570 of gp91 phox also contain regions with sequence homologies to the NADP ϩ -binding site of ferredoxin-NADP ϩ reductase flavoenzymes (17)(18)(19)(20)22). An aromatic amino acid at or near the carboxyl terminus is highly conserved among members of the ferredoxin-NADP ϩ reductase flavoprotein family, including gp91 phox which has a carboxyl-terminal phenylalanine (Phe 570 ). Crystallographic analysis of the spinach ferredoxin-NADP ϩ reductase has localized the carboxyl-terminal aromatic side chain to the NADP ϩ -binding pocket, where it may interact with the isoalloxazine ring of FAD in the absence of NADP ϩ (35). Altering the aromatic character or deleting this conserved residue in the pea ferredoxin-NADP ϩ reductase leads to major impairments in catalytic efficiency of the enzyme (36).
Although the majority of missense mutations in gp91 phox identified in patients with X-linked CGD result in apparent structural instability of the protein, in rare cases, gp91 phox expression is preserved but the mutant polypeptide forms a non-functional flavocytochrome b 558 (37,38). These latter mutations have been informative in identifying residues critical for gp91 phox function. The expression and analysis of mutant gp91 phox polypeptides generated by site-directed mutagenesis has not been previously reported, at least in part due to the lack of systems in which adequate levels of functional recombinant gp91 phox can be readily expressed. To undertake a more systematic analysis of structure-function relationships in gp91 phox using this approach, we developed a human myeloid leukemia cell line that lacks endogenous gp91 phox after targeted disruption of the X-linked gp91 phox gene (39). This "X-CGD" cell line has been a valuable tool for expression of recombinant gp91 phox , which, in the case of wild-type recombinant gp91 phox , assembles with p22 phox to form a functional flavocytochrome b 558 heterodimer (39 -41).
In the present study, we have used site-directed mutagenesis to probe the role of the carboxyl terminus of gp91 phox in NADPH oxidase activity. We examined both the requirements for an intact 559 RGVHFIF 565 sequence and for a carboxylterminal aromatic residue in supporting superoxide production by intact granulocytic cells. Mutant gp91 phox cDNAs encoding polypeptides with amino acid substitutions or deletions in the carboxyl terminus were transfected into the X-CGD myeloid cell line for analysis of expression and function. The results suggest that the distal carboxyl terminus is an important determinant for gp91 phox stability, but that neither an intact 559 RGVHFIF 565 sequence nor a carboxyl-terminal aromatic residue are absolutely essential for NADPH oxidase activity in intact granulocytic cells. Alanine substitutions had either no or only a modest effect on NADPH oxidase activity. The greatest effects on superoxide production were seen with substitution of polar or charged residues for hydrophobic amino acids in 559 RGVHFIF 565 , but these were also associated with reduced expression of gp91 phox , suggesting that the overall protein structure was perturbed.

EXPERIMENTAL PROCEDURES
Plasmids-A full-length wild-type gp91 phox cDNA, extending from 12 nucleotides upstream of the initiator ATG to a SacI site in the 3Јuntranslated region (39), was cloned into the NotI site in the multiple cloning site of pBluescript II KS(ϩ) (Stratagene). This construct was used as a phagemid to produce single-stranded DNA for oligonucleotide-directed mutagenesis of specific codons. Mutations were introduced into the carboxyl-terminal region of the gp91 phox cDNA using the Sculptor in vitro mutagenesis system (Amersham), and verified by dideoxynucleotide sequencing. Deletion mutants were made by replacement of nucleotides with a premature stop codon. Mutant gp91 phox cDNAs were subcloned into the NotI site of the pEF-PGKneo mammalian expression vector (39), or a related vector, pEF-PGKpac, which contains a linked expression cassette for puromycin-N-acetyltransferase instead of neomycin phosphotransferase. The mutated gp91 phox cDNA expression constructs were resequenced to confirm the mutations and were linearized with KpnI prior to electroporation into a human myeloid leukemia cell line (see below). Preparation and other manipulations of plasmid DNAs were performed by standard protocols (42). Restriction enzymes and other reagents for molecular biology were obtained from Promega, Boehringer Mannheim, New England Biolabs, and U.S. Biochemical Corp.
Cell Lines-Wild-type PLB-985 cells (43), a human myeloid leukemia cell line, and a derivative line in which the X-linked gp91 phox gene has been disrupted by gene targeting (X-CGD PLB-985 cells) were maintained as described (39). X-CGD PLB-985 cells do not express gp91 phox protein and lack NADPH oxidase activity. After electroporation of expression constructs into X-CGD PLB-985 cells, clones were selected by limiting dilution in either 1.5 mg/ml G418 or 3 g/ml puromycin (39). Since up to one-third of clones selected for resistance to the linked antibiotic marker will not have a full-length transgene and therefore fail to express any transgenic gp91 phox mRNA, 2 clones were analyzed for recombinant gp91 phox expression by either Northern blot and/or immunoblot with gp91 phox -specific probes. For subsequent analysis, three to five independent clones determined to express transgenic gp91 phox were pooled to minimize potential clone-to-clone variation in recombinant gp91 phox expression and NADPH oxidase activity. PLB-985 and derivative cell lines were differentiated into granulocytes with dimethylformamide as described (39).
Analysis of gp91 phox Expression-Northern blot and immunoblot analyses were performed as described previously (39). A gp91 phox monoclonal antibody (mAb 48) (44) was kindly provided by A. Verhoeven and D. Roos (Central Laboratory of the Netherlands Blood Transfusion Service). Polyclonal gp91 phox antibodies raised against a gp91 phox peptide (residues 86 -102), a ␤-galactosidase-gp91 phox fusion protein, and a p22 phox peptide have previously been described (45)(46)(47). Scanning densitometry was employed to measure the relative intensity of the gp91 phox signals on immunoblots developed with the ECL detection system (Amersham) and performed using a Silver Scan II scanner and image 1.45 software (W. Rasband, National Institutes of Health) (40). Different exposures as well as serial dilutions of cell extracts were scanned to ensure that measurements were taken within a linear response range.
The expression of gp91 phox was also examined by flow cytometry and confocal microscopy. After washing twice with cold phosphate-buffered saline, 5 ϫ 10 5 cultured cells were resuspended in 450 l of relaxation buffer consisting of 10 mM PIPES, pH 7.3, 100 mM KCl, 3.5 mM MgCl 2 , 3 mM NaCl, and 1 mM EGTA and blocked for 30 min on ice by adding 50 l of normal goat serum (Boehringer Mannheim). The cells were then pelleted by centrifugation and resuspended in 100 l of the relaxation buffer, and mixed with 100 l of a hybridoma culture medium containing monoclonal antibody 7D5 (kindly provided by M. Nakamura, Nagasaki University) (48,49). After a 30-min incubation on ice, the cells were washed once and then further incubated for 30 min with 200 l of fluorescein isothiocyanate-conjugated goat anti-mouse IgG (Boehringer Mannheim) antibody diluted 1:100 in the relaxation buffer. The cells were then washed twice, fixed with 1% paraformaldehyde and analyzed by FACScan (Becton Dickinson, San Jose, CA). Mouse IgG1 (Sigma) was used as an isotype control. A total of 5,000 cells were analyzed. For imaging demonstration, 30,00ϳ5,000 cells from the above staining were deposited onto a glass slide by centrifugation at 450 rpm for 5 min and observed by confocal microscopy.
Assay of Superoxide Formation-A continuous cytochrome c reduction assay was used for quantitative measurement of superoxide dismutase-inhibitable superoxide formation by granulocyte-differentiated PLB-985 cell lines (39). The assay was performed at 37°C using a Thermomax microplate reader and associated SOFTMAX Version 2.02 software (Molecular Devices) on whole cells after stimulation with either phorbol myristate acetate (100 ng/ml) or formyl-methionylleucyl-phenylalanine (FMLP) (1 M).

RESULTS
To examine the role of the carboxyl terminus of the gp91 phox subunit of flavocytochrome b 558 in NADPH oxidase activity, site-directed mutations were introduced into the gp91 phox cDNA (Fig. 1). Mutant forms of gp91 phox that were generated included a series of alanine replacements within the 559 RGVH-FIF 565 domain at positions (Arg 559 , Val 561 , Phe 563 , Ile 564 , and Phe 565 ) previously identified as crucial for oxidase inhibition by alanine-substituted peptides derived from this region (33). Substitutions with a more polar or charged amino acid were prepared at positions 561, 563, and 565. Two mutants with double substitutions (R559A/V561A and I564T/F565V) were also generated. A mutant gp91 phox with an alanine-substituted His 562 residue was also prepared, which would be predicted to have no impact on oxidase activity based on peptide inhibition studies (33). To investigate the requirement for a carboxylterminal aromatic residue for gp91 phox function, the phenylalanine at position 570 was substituted with an alanine (F570A) or deleted (F570⌬) by insertion of a premature stop codon at this position. Finally, the effect of entirely deleting the last 10 residues (⌬560 -570) by introduction of a premature stop codon at position 560 was examined.
The mutated gp91 phox cDNAs were subcloned into mammalian expression vectors under control of the EF1-␣ promoter, and the expression of mutant polypeptides evaluated by immunoblot analysis (Fig. 2) after stable transfection of a derivative of human myeloid leukemia PLB-985 cells which lack endogenous gp91 phox due to targeted disruption of the gp91 phox gene (X-CGD PLB-985 cells) (39). Note that the wild-type gp91 phox protein migrates as a diffuse band centered at Ϸ91 kDa (Fig. 2), and the heterogeity in size of immunoreactive species most likely reflects variations in glycosylation. Two related vectors were used in these studies, pEF-PGKneo and pEF-PGKpac, that differ only in the selectable marker gene upstream of the EF1-␣ promoter/gp91 phox cDNA cassette. Using the pEF-PGKneo vector, which has a linked neomycin phosphotransferase marker gene, expression of recombinant wild-type gp91 phox cDNA is 10 -20% of the endogenous gp91 phox level in granulocyte-induced wild-type PLB-985 cells (39). During the course of these studies, we observed that higher levels of expression of recombinant wild-type gp91 phox (25-50% of wild-type PLB-985) were obtainable using pEF-PGKpac, 2 for reasons which remain unclear. The majority of gp91 phox mutants reported here were expressed using the pEF-PGKneo vector, and their expression compared with X-CGD PLB-985 cell lines transfected with a pEF-PGKneo vector containing the wild-type gp91 phox cDNA (Fig. 2, upper two panels). Several mutants prepared later during the course of this study utilized pEF-PGKpac, and expression was compared with wild-type p91 phox expressed using pEF-PGKpac (Fig. 2, lower panel). One mutant, F570⌬, was expressed in both vector backgrounds, but most mutants were not re-cloned into pEF-PGKpac as it was felt that the conclusions would not be substantially altered.
A summary of the relative levels of expression of recombinant gp91 phox derivatives in granulocyte-induced PLB-985 cells is shown in Table I. Several of the gp91 phox mutants appeared to be unstable, with deletion of the carboxyl-terminal 11 amino acids having the greatest impact on protein expression. No gp91 phox was detected in immunoblots of cell extracts prepared from either undifferentiated (not shown) or granulocyte-induced cells transfected with the ⌬560 -570 construct, using either the monoclonal antibody MoAb48 ( Fig. 2A), directed against an uncharacterized epitope on gp91 phox , or two different gp91 phox polyclonal antisera (not shown). In cells expressing F563T, F565D, or FI564/F565V gp91 phox mutants, very small but detectable amounts of the gp91 phox polypeptide were present. More modest reductions in the relative level of gp91 phox expression were seen for several other mutants (V561A, V561T, V561E, F563A, F565A, and F570⌬). The relative levels of the remaining mutants were not significantly different from wild-type recombinant gp91 phox as expressed with the corresponding pEF-PGKneo or pEF-PGKpac vector. The amount of recombinant protein detected was unrelated to the abundance of the transgene-derived gp91 phox mRNAs, which were all similar (not shown). As was previously observed with expression of wild-type recombinant gp91 phox (39), expression of mutant derivatives of gp91 phox rescued expression of the cytochrome p22 phox subunit in X-CGD PLB-985 cells, in proportion to the relative abundance of the mutant gp91 phox subunit (not shown).
To provide evidence that the wild-type and mutant gp91 phox are successfully delivered to the plasma membrane when transgenically expressed in X-CGD PLB-985 cells, we stained the transfected and parental X-CGD cells with a monoclonal antibody, 7D5, and examined cell surface expression of flavocytochrome b 558 by flow cytometry and confocal microscopy. The 7D5 antibody reacts with an extracellular epitope on flavocytochrome b 558 (49). Since the mutant gp91 phox proteins were expressed at variable levels, we chose F570⌬ as an example of a mutant that was expressed at relatively high levels, comparable to the wild-type recombinant gp91 phox , and F563T as an example of poorly expressed gp91 phox mutant. As seen in FIG. 1. Mutagenesis of the carboxyl terminus of gp91 phox . The wild-type amino acid sequence from residues 552 through 570 is shown in the top line. The 559 RGVHFIF 565 sequence identified as a p47 phoxbinding site is shown in bold, and critical residues identified by peptide alanine scanning are marked with asterisks. Also shown in bold is the carboxyl-terminal phenylalanine residue, which is highly conserved among ferredoxin-NADP ϩ reductase family members. The gp91 phox mutants generated in this study are listed below the wild-type sequence, with point substitutions or deletions as indicated by the shaded boxes. show results obtained using the pEF-PGKneo vector for transfection, and the bottom panel shows extracts obtained from pEF-PGKpac transfectants (asterisks). Blots were probed with a gp91 phox monoclonal antibody, MoAb48. Twenty g of protein was loaded in each lane, except for wild-type PLB-985 cells, for which 5 g was loaded. In most cases, two independently prepared representative samples from mutant gp91 phox transfectants are shown. Fig. 3, the 7D5 antibody produced a complete shift in fluorescence intensity in flow cytometry of cells transfected with wildtype gp91 phox , and confocal microscopy showed immunofluorescent staining on the cell surface. Similar results were obtained for wild-type PLB-985 granulocytes (not shown), whereas no staining was seen for X-CGD PLB-985 cells (Fig. 3). Immunofluorescent staining of cells expressing recombinant F570⌬ gp91 phox was similar to that seen for cells expressing recombinant wild-type gp91 phox . However, only weak staining of F563T gp91 phox -expressing cells was seen, consistent with the results seen by immunoblotting (Fig. 2). Taken together, these observations demonstrate that the transgenic expression of wildtype and mutant gp91 phox in X-CGD PLB-985 cells results in the plasma membrane expression of flavocytochrome b 558 , at levels that correlate with the amount of gp91 phox seen in immunoblots of cell extracts.
We next examined the impact of carboxyl-terminal mutations in gp91 phox on NADPH oxidase activity in intact cells after granulocyte differentiation (Table I). A continuous cytochrome c reduction assay was used to quantitate superoxide formation after NADPH oxidase activation by phorbol myristate acetate. X-CGD PLB-985 cells expressing wild-type recombinant gp91 phox at only 21-37% of that seen in wild-type PLB-985 granulocytes exhibited NADPH oxidase activity that was not significantly different from wild-type PLB-985 granulocytes (Table I). This is consistent with our previous studies which have shown that expression of wild-type recombinant gp91 phox at Ն10% of wild-type levels fully reconstitutes NADPH oxidase activity, and that cells expressing as little as Ϸ5% of wild-type gp91 phox levels exhibit 40 -60% of wild-type activity (39 -41, 50). These observations have suggested that gp91 phox and flavocytochrome b 558 are normally present in excess and not rate-limiting for superoxide formation in wildtype neutrophils or PLB-985 granulocytes.
All mutants within the 559 RGVHFIF 565 with detectable gp91 phox expression were capable of supporting at least a small amount of superoxide generation (Table I). The lag time between stimulation of cells with phorbol myristate acetate and the onset of superoxide generation (Յ2 min) was the same for cells expressing mutant gp91 phox derivatives as for wild-type gp91 phox (not shown). Alanine substitutions at Arg 559 , His 562 , or F563A did not affect NADPH oxidase activity significantly, as was also seen for the double R559A/V561A substitution. Point substitutions at Val 561 , Ile 564 , and Phe 565 resulted in a 2-5-fold reduction in the rate of superoxide formation relative to wild-type PLB-985 cells; the greatest reduction seen in the V561E mutant, which was only Ϸ40% as active as the V561T mutant despite similar levels of gp91 phox expression. Superoxide formation observed with the F565D gp91 phox derivative, which was expressed at very low levels, was 20-fold less than wild-type. In cells transfected with the F563T or I564T/F565V mutants, in which expression was also very low, NADPH oxidase activity appeared to be even further reduced but was difficult to measure reliably using the cytochrome c assay. Note that the above comparisons do not take into account variations in recombinant gp91 phox expression levels, and therefore the relative function of gp91 phox mutants expressed in reduced amounts may be an underestimate. All of the above gp91 phox mutants supported the reduction of nitro blue tetrazolium dye, with the intensity of resultant formazan staining proportional to enzyme activity measured by the cytochrome c reduction assay. However, NADPH oxidase activity was undetectable by either the cytochrome c or nitro blue tetrazolium assay in cells transfected with the ⌬560 -570 mutant, which did not express detectable Ϸ91 kDa gp91 phox protein. Superoxide generation in response to fMLP was studied for several mutants (R559A, V561A, V561E, V561T, and F565A) (not shown). The kinetics of superoxide formation were unaffected, and the rank order of the relative magnitude of the response corresponded to that seen for phorbol myristate acetate.
a p Ͻ 0.05 (wild-type PLB-985 versus transfected X-CGD PLB-985 expressing recombinant gp91 phox ), using the unpaired t test. b p Ͻ 0.05 (wild-type recombinant gp91 phox versus mutant gp91 phox expressed using the same vector), using the unpaired t test.
Mutant derivatives of gp91 phox in which the final carboxylterminal residue, Phe 570 , was substituted with an alanine or deleted entirely (F570⌬) were studied to test the hypothesis that an aromatic amino acid at this position is critical for NADPH oxidase enzymatic activity, as observed for pea ferredoxin-NADP ϩ reductase (36), another member of the ferredoxin-NADP ϩ reductase family. Although the F570A mutant was expressed at levels comparable to wild-type recombinant gp91 phox , the rate of superoxide formation was reduced by Ϸ2-fold (Table I). Deletion of the carboxyl-terminal residue (F570⌬) resulted in a Ϸ4-fold reduction in expression of recombinant gp91 phox and an almost 10-fold reduction in NADPH oxidase activity compared with wild-type cells (Table I). When the relative level of recombinant F570⌬ protein was increased by using pEF-PGKpac for expression, an almost 2-fold reduction in NADPH-oxidase activity relative to wild-type cells was still observed (Table I). DISCUSSION The phagocyte flavocytochrome b 558 is a plasma membraneassociated heterodimer that is the electron transfer center of the superoxide-generating NADPH oxidase. The objective of this study was to use site-directed mutagenesis to probe the role of the distal carboxyl terminus of the gp91 phox subunit, which contains domains implicated in the binding of NADPH and in a critical interaction with p47 phox during assembly of the active oxidase complex. This is the first reported study using site-directed mutagenesis to investigate structure-function relationships in gp91 phox , and utilized an X-CGD myeloid cell line for expression of mutant gp91 phox polypeptides. For functional analysis, we specifically focused on analysis of NADPH oxidase activity in intact granulocytic cells, since the protein-protein interactions required for oxidase assembly and function in the cell-free assay do not always reflect those relevant in whole cells. For example, phosphorylation of p47 phox is required for NADPH oxidase activity in neutrophils (12,51,52), but not in the cell-free system (53). As another example, the p67 phox subunit contains two SRC homology 3-binding domains that are essential for oxidase function in whole cells, but are dispensable in the cell-free system (54).
Previous studies on X-CGD patients with deletions involving the carboxyl terminus of gp91 phox have suggested that this domain is required for the stability of this protein. Based on computer modeling analysis, the extreme carboxyl terminus of gp91 phox has been proposed to be buried within the interior of the protein, where it interacts with the flavin group (22). A complete lack of gp91 phox and flavocytochrome b 558 has been reported in X-CGD when the last 40 -50 amino acids of gp91 phox are absent (45,55), and a 5-fold reduction in gp91 phox expression was observed with deletion of the last six residues of gp91 phox (56). In this study, we found that many of the gp91 phox carboxyl-terminal mutants were expressed at significantly reduced amounts in the X-CGD PLB-985 cells compared with wild-type recombinant gp91 phox . Although expression of alanine-substituted derivatives were generally comparable to wild-type recombinant gp91 phox , decreased amounts of gp91 phox were observed when apolar residues within the 559 RGVH-FIF 565 sequence were replaced with polar or charged residues. This was particularly marked with polar substitutions at positions Phe 563 and Phe 565 . A mutant gp91 phox with the double replacement I564T/F565V was also expressed at very low levels. Removal of the last residue of gp91 phox resulted in a modest decrease in gp91 phox expression, and no recombinant gp91 phox was detected at all when the last 11 residues were deleted. These observations support the concept that the carboxyl terminus of gp91 phox is involved in maintaining a stable conformation of gp91 phox and/or in heterodimer formation with FIG. 3. Cell surface expression of wild-type and mutant gp91 phox in transgenic X-CGD PLB-985 cell lines. Flow cytometry and confocal microscopy were performed using the 7D5 monoclonal antibody to stain X-CGD PLB-985 cells and derivative lines obtained after stable transfection with pEF-PGKpac vectors for expression of either wild-type (WT) gp91 phox or the F570⌬ or F563T mutants, as indicated. A, cells were labeled with 7D5 antibody (solid lines) or a mouse IgG1 as an isotype control (dotted lines) and analyzed by flow cytometry. B, representative images obtained using confocal microscopy after cell staining with the 7D5 antibody (ϫ 730). Insets show results obtained after cells were stained with the mouse IgG1 (ϫ 340); no cells are visible due to a lack of staining. p22 phox , which plays an important role in the intracellular stability of each cytochrome subunit (1).
There has been great interest in characterizing the specific contact sites in NADPH oxidase subunits that mediate assembly of the active oxidase complex and regulate electron transfer through flavocytochrome b 558 (2,3). A number of studies have suggested that interactions between p47 phox and the 559 RGVH-FIF 565 domain encompassed within the distal carboxyl terminus of gp91 phox play a key role in this process. An antibody directed against the carboxyl-terminal 13 amino acids of gp91 phox inhibits superoxide formation in the cell-free assay (29). Synthetic peptides derived from this region inhibit NADPH oxidase activity in permeabilized neutrophils (29,34) and in the cell-free oxidase assay (29,31,57). A peptide corresponding to gp91 phox residues 559 RGVHFIF 565 is the minimum sequence capable of oxidase inhibition (IC 50 ϭ 38 M) (33), which can be abolished by preincubation of neutrophil membranes with p47 phox -containing neutrophil cytosol (14). When incubated with neutrophil cytosol, a similar carboxyl-terminal gp91 phox peptide can be cross-linked to p47 phox (31). Finally, a strategy using random sequence peptide bacteriophage display libraries of identified peptides with homologies to RGVHFIF as ligands of recombinant p47 phox (32).
One goal of the current study was to further examine the role of the 559 RGVHFIF 565 domain of gp91 phox with regards to NADPH oxidase activity in intact granulocytic cells. Specific residues within this sequence (Arg 559 , Val 561 , Phe 563 , Ile 564 , and Phe 565 ) have been identified by alanine scanning as important for the function of this domain, based on the relative ability of alanine-substituted peptides to inhibit oxidase activity in a cell-free assay (33). However, we found that comparable alanine substitutions in the gp91 phox polypeptide resulted in either no or only a modest reduction in NADPH oxidase activity when the mutant gp91 phox was expressed in whole cells. The elimination of the positive charge at position 559 (Arg 559 ), which had been proposed to mediate a critical ionic interaction between gp91 phox and another oxidase subunit (33), had no significant impact on gp91 phox expression or NADPH oxidase activity. Replacement of nonpolar residues with polar or charged residues in 559 RGVHFIF 565 resulted in relatively greater reductions in superoxide generation, but these mutations also were associated with apparent decreased stability of the gp91 phox polypeptide, suggesting that overall protein structure was perturbed. Hence, it is not possible to rule out longrange effects on other domains within the gp91 phox as the primary cause of decreased NADPH oxidase activity seen with alterations in the nonpolar character of these residues. It would also be of considerable interest to examine p47 phox translocation in gp91 phox mutants in cases where superoxide formation appeared to be affected, which would provide information as to whether enzyme assembly or catalytic activity was impaired as a result of the gp91 phox mutation. However, as noted above, mutations in gp91 phox in which NADPH oxidase activity was affected also resulted in an apparent reduction in the stability of gp91 phox . Because of the low level of expression of these gp91 phox mutants, quantitation of p47 phox translocation could not be determined reliably.
There are a number of possible explanations for the observed discrepancy between the 559 RGVHFIF 565 peptide inhibition data and the results obtained in this study. A kinetic analysis of oxidase inhibition by 559 RGVHFIF 565 has suggested that this peptide acts as a non-competitive inhibitor (57), which might explain why many of the mutations in the gp91 phox polypeptide did not have a profound effect. Alternatively, additional sites of interaction between p47 phox and gp91 phox may obviate the requirement for an intact 559 RGVHFIF 565 domain for oxidase assembly in intact cells. A proline-rich region in the cytoplasmic carboxyl terminus of p22 phox is a target for SRC homology 3 domains within p47 phox (47, 58 -62). A P156H point mutation in this domain, identified in a CGD patient homozygous for a mutant p22 phox gene, abolishes translocation of p47 phox and NADPH oxidase activity (47,58,59,61). This interaction between p47 phox and the cytochrome thus appears to be critical for oxidase assembly in intact granulocytes. A D500G mutation in gp91 phox , identified in a patient with X-linked CGD, is also associated with virtually absent oxidase activity and deficient translocation of p47 phox (63). Residues 77-93 in gp91 phox is another domain that appears to interact with p47 phox , as peptides homologous to this sequence bind to recombinant p47 phox and inhibit oxidase function in the cell-free assay (32). Finally, it is also possible that single or multiple substitutions other than those studied within the 559 RGVHFIF 565 sequence are required to fully inhibit oxidase assembly in whole cells. However, we found that the ability to analyze the impact of more drastic alterations was limited by a lack of mutant protein stability.
The presence of an aromatic amino acid as the final or penultimate residue is highly conserved among ferredoxin-NADP ϩ reductase family members, including gp91 phox , which contains a carboxyl-terminal phenylalanine (Phe 570 ). The aromatic ring has been proposed to maintain enzyme structure in the absence of NADP ϩ by acting as a pseudosubstrate (18,22,35). X-ray diffraction analysis of spinach ferredoxin-NADP ϩ reductase indicates that the carboxyl-terminal 19 residues, which lie just distal to the NADP ϩ -binding domain, form an ␣-helix/␤-strand region that extends into the FAD site, so that the terminal tyrosine residue interacts closely with the flavin ring (35). Altering the aromatic character of the corresponding residue in gp91 phox , Phe 570 , or deleting it entirely resulted in a Ϸ2-fold reduction in the rate of superoxide generation relative to cells expressing similar levels of wild-type recombinant gp91 phox . Alanine substitution of the other aromatic residues (Phe 563 and Phe 565 ) in the carboxyl terminus of gp91 phox also resulted in, at most, a 3-fold reduction in NADPH activity. These data contrast to the Ϸ300 -850-fold reduction in catalytic activity observed when non-aromatic replacements of the carboxyl-terminal tyrosine residue or its deletion were created in spinach ferredoxin-NADP ϩ reductase (36). Our results suggest that there are structural and/or functional differences between the extreme carboxyl-terminal domain of gp91 phox and the corresponding aromatic residues in other members of the ferredoxin-NADP ϩ reductase family.