Biosynthesis of the Phagocyte NADPH Oxidase Cytochromeb 558

The NADPH oxidase cytochromeb 558 is a membrane heterodimer comprised of a glycosylated 91-kDa subunit, gp91 phox , and a nonglycosylated 22-kDa subunit, p22 phox . The role of heme in cytochrome b 558 biosynthesis was studied using succinyl acetone, an inhibitor of heme synthesis, in PLB-985 myeloid cells undergoing granulocytic differentiation. Succinyl acetone markedly reduced expression of p22 phox and the mature 91-kDa form of gp91 phox but not its 65-kDa high mannose precursor, in association with a profound reduction in NADPH oxidase activity. Expression of non-heme-containing cytosolic oxidase components was unaffected. The reduction in cytochrome b 558 expression and NADPH oxidase activity was prevented by adding exogenous heme and was reversible upon removal of succinyl acetone. Transgenic expression of gp91 phox in monkey COS-7 and murine 3T3 cells, both of which lacked endogenous p22 phox mRNA, demonstrated that p22 phox was not required for maturation of gp91 phox carbohydrate to complex oligosaccharides. However, coexpression of transgenic p22 phox increased the abundance of the mature gp91 phox glycoprotein. These results suggest that heme incorporation plays an important role in cytochromeb 558 assembly and provide further support for the concept that stability of p22 phox and the mature gp91 phox subunit is increased by heterodimer formation.

The phagocyte NADPH oxidase catalyzes the formation of superoxide (O 2 Ϫ ), the precursor to a variety of potent oxidants that are important for the host defense against invading microorganisms (1). Dormant in resting phagocytes, the oxidase is assembled rapidly upon phagocyte activation to mediate the transfer of electrons from cytosolic NADPH to molecular oxygen (1,2). It is now established that at least three cytosolic proteins (p47 phox , p67 phox , and Rac) and a membrane-associated heterodimer, cytochrome b 558 , are involved in this electron transport system, although the detailed biochemical functions of each subunit and the mechanisms of assembly remain incompletely defined (1)(2)(3)(4).
The NADPH oxidase cytochrome b 558 , expressed almost exclusively in phagocytic cells, is comprised of two integral membrane proteins, a glycosylated 91-kDa subunit (gp91 phox ) and a 22-kDa subunit (p22 phox ) and contains both flavin and heme groups (5)(6)(7)(8)(9)(10)(11). This heterodimer is hence believed to be the redox center of the oxidase, although p67 phox has recently been shown to contain an NADPH binding site that may be important for oxidase function (12). Multiple hemes, probably two, have been reported to be incorporated within cytochrome b 558 purified from human neutrophils, with one residing in gp91 phox and a second that may be shared by the two subunits (6,13,14). The heme prosthetic groups have been proposed to reside within the membrane in the relatively hydrophobic NH 2 -terminal portion of the gp91 phox polypeptide (1,2,13,15) and appear to be in a six-coordinated state with axial imidazole or imidazolate ligands supplied by histidine residues (16,17). The hydrophilic carboxyl-terminal half of gp91 phox contains regions with homology to the ferredoxin-NADPH ϩ reductase flavoenzyme family, including flavin and NADPH binding domains (7,8,18).
The physiologic importance of the NADPH oxidase is illustrated by the inherited immunodeficiency, chronic granulomatous disease (CGD), 1 which results from genetic defects in different oxidase subunits (1,19,20). Affected patients develop recurrent, severe bacterial and fungal infections caused by the deficient generation of phagocyte oxidants. Two genetic subgroups involve mutations in the cytochrome b 558 . Defects in the X-linked gene of gp91 phox account for about two-thirds of cases of CGD, whereas the gene encoding p22 phox is the site of mutations in a rare autosomal recessive form of CGD. Genetic defects in either the p47 phox or p67 phox subunit account for the remainder of autosomal recessive CGD.
Relatively little is known about the biogenesis of the cytochrome b 558 and the role of heme incorporation in this process. In virtually all patients with cytochrome b 558 mutations, neutrophils lack both gp91 phox and p22 phox polypeptides, regardless of which subunit is affected by the mutation (20,21). This observation has suggested that formation of the gp91 phox -p22 phox heterodimer is important for stable expression of each subunit within the neutrophil. Porter and co-workers (22) have identified a partially processed 65-kDa precursor of gp91 phox in B cell lines from p22 phox -deficient CGD patients, where it may be more stable compared with neutrophils or more abundant because of an increased content of endoplasmic reticulum. This 65-kDa gp91 phox intermediate has high mannose carbohydrate side chains, indicating that the initial steps of oligosaccharide addition and processing within the endoplasmic reticulum occur in the absence of p22 phox . Expression of the mature 91-kDa form of gp91 phox with fully processed oligosaccharide side chains can be restored in p22 phox -deficient CGD B cell lines by expression of recombinant p22 phox (22,23). The high mannose gp91 phox precursor has also been detected in B cell lines from four unrelated X-CGD patients who otherwise lacked mature gp91 phox (24). The patients all had missense mutations or an in-frame deletion of the coding sequence, and it was postulated that these mutations impaired normal biosynthetic processing of gp91 phox either by interfering with incorporation of redox cofactors or by disrupting the association with p22 phox .
The heme prosthetic groups have themselves been postulated to play a role in heterodimer formation and stability of cytochrome b 558 . The hemes are tightly bound to the cytochrome b 558 heterodimer, and the detectable spectrum is always accompanied by the stable expression of both subunits (2,19,20,25). When heme synthesis was inhibited in myeloid leukemia HL-60 cells induced to undergo granulocytic differentiation, cells failed to show the normal increase in NADPH oxidase activity (26). The typical cytochrome b 558 spectrum was absent, and indirect immunofluorescence microscopy showed markedly decreased expression of p22 phox , although gp91 phox expression appeared normal. The identification of CGD patients with undetectable cytochrome b 558 who have point mutations in candidate heme-binding histidine residues within gp91 phox or p22 phox has also been taken as indirect evidence for a role of heme incorporation in cytochrome b 558 expression (20). However, three of the four of the aforementioned mutations involve His 3 Arg substitutions within hydrophobic domains, which could also have a nonspecific effect on protein stability.
In the current study, we have investigated further the biosynthesis of cytochrome b 558 , examining the role of both heme incorporation and heterodimer formation in the maturation and stability of gp91 phox and p22 phox subunits. We found that succinyl acetone (SA), an inhibitor of heme biosynthesis, reversibly reduced the expression of both p22 phox and the mature 91-kDa form of gp91 phox but not the 65-kDa precursor of gp91 phox , in association with a profound reduction in NADPH oxidase activity in cultured PLB-985 myeloid cells undergoing granulocytic differentiation. Expression of the cytosolic oxidase components p47 phox , p67 phox , and Rac2 was unaffected. Transgenic expression of gp91 phox in monkey COS-7 and murine 3T3 cells, both of which lack endogenous p22 phox mRNA, demonstrated that the presence of the p22 phox polypeptide was not required for maturation of gp91 phox oligosaccharide side chains, although coexpression of transgenic p22 phox increased the abundance of the mature gp91 phox glycoprotein. Taken together, these data confirm that the incorporation of heme plays an important role in the assembly of the cytochrome b 558 and suggest that the stability of phagocyte p22 phox and the mature gp91 phox subunit is dependent on both heme incorporation and heterodimer formation.
Cell Culture and Differentiation-Human myeloid leukemia PLB-985 (27) cells (obtained from P. Newburger, University of Massachusetts) and HL-60 (28) cells (obtained from ATCC CCL 240) were maintained in RPMI 1640-glutamine (Life Technologies, Inc.) supplemented with 10% fetal calf serum (Sigma), 50 units/ml penicillin, and 50 g/ml streptomycin (Life Technologies, Inc.). For granulocytic differentiation, cells at a starting density of 1 ϫ 10 5 cells/ml were exposed to 0.5% DMF in CO 2 -independent medium (Life Technologies, Inc.) in the absence of fetal calf serum but supplemented with Nutridoma-SP (Boehringer Mannheim) and 2 mM L-glutamine. The medium was changed once at day 3 during the period of differentiation. Under these conditions, at least 85% of the cells had undergone granulocytic differentiation by 4 -6 days based on acquisition of respiratory burst oxidase activity as deter-mined using the nitro blue tetrazolium test (29).
SA is a specific inhibitor of the enzyme 5-aminolevulinic acid dehydratase that catalyzes the formation of porphobilinogen from 5-aminolevulinate in heme biosynthesis (30). In experiments using SA, PLB-985 or HL-60 cells were differentiated with DMF in the presence of 10 g/ml SA as described previously (26). Cell growth and viability were similar in cells differentiated in either the presence and absence of SA. After a 6-day incubation, control DMF-induced PLB-985 cells increased to 235 Ϯ 17% (n ϭ 3) of the original cell number and were 93% viable based on trypan blue exclusion. SA treated, DMF-induced PLB-985 cells multiplied to 246 Ϯ 13% (n ϭ 3) of the initial inoculum and were 92% viable. Both SA-treated and untreated DMF-induced cells had similar morphology by cytospin and light microscopy. For SA reversal studies, PLB-985 cells cultured for 3 days in the presence of DMF with SA were collected by centrifugation at 500 ϫ g for 5 min at room temperature and washed once with PBS, after which the cells were resuspended in fresh medium containing DMF, but no SA, and cultured further for 1, 2, or 3 days before harvest. In a separate experiment, DMF-induced differentiation of PLB-985 cells was carried out in the presence of 5 M hemin.
NIH 3T3 mouse fibroblasts and monkey kidney COS-7 cells were obtained from the ATCC (CRL-1658 and CRL-1651, respectively). COS-7 cells were grown in Dulbecco's modified Eagle's low glucose medium (Life Technologies, Inc.) supplemented with 10% fetal calf serum (Sigma), 50 units/ml penicillin, and 50 g/ml streptomycin (Life Technologies, Inc.). 3T3 cells were cultured under the same conditions except newborn calf serum (Summit Biotechnology) was supplemented instead of fetal calf serum. Both of these cell lines are adherent and were scraped gently at about 90% of confluence for passage or harvesting.
Transfection of 3T3 and COS-7 Cells with gp91 phox and p22 phox cDNAs-For expression of recombinant gp91 phox in non-phagocytic cell lines, a full-length human gp91 phox cDNA was cloned into the NotI site of pEF-PGKpac, which is similar to the gp91 phox -pEF-PGKneo mammalian expression vector (29) but contains a linked expression cassette for puromycin N-acetyltransferase instead of neomycin phosphotransferase. For expression of the p22 phox , a full-length cDNA for human p22 phox (23,31) was cloned into the NotI site of pEF-PGKneo. DOTAP (Boehringer Mannheim) was used to mediate transfection of plasmid expression vectors essentially as recommended by the manufacturer. Briefly, 10 g of DNA in 100 l of sterile PBS was gently mixed with 100 l of DOTAP mixture consisting of 30 l of DOTAP reagent and 70 l of PBS. After a 10-min incubation at room temperature, the transfection mixture was mixed with 5 ml of fresh medium and added to the cells growing at Ϸ60% confluence in a 100 ϫ 20-mm culture dish. The cells were transfected overnight, and the clones were selected by limiting dilution in the presence of the corresponding antibiotic (1.8 mg/ml G418 and 1 g/ml puromycin). To generate cell lines expressing both recombinant gp91 phox and p22 phox , p22 phox pEF-PGKneo was transfected into clones that had been transfected previously with gp91 phox -pEF-PGKpac.
Measurement of NADPH Oxidase Activity-Superoxide production by SA-treated and untreated differentiated PLB-985 cells was measured in a quantitative kinetic assay based on the reduction of cytochrome c after stimulation of cells with 0.1 g/ml phorbol myristate acetate. The assay was performed at 37°C using a Thermomax microplate reader and associated SOFTMAX Version 2.02 software (Molecular Devices) as reported previously (29,32). The nitro blue tetrazolium test on PLB-985 and HL-60 cells was performed as described previously to assay O 2 Ϫ formation in individual cells by monitoring the intensity of formazan granules deposited from the reduction of nitro blue tetrazolium by O 2 Ϫ (29). Cell Extraction and Subcellular Fractionation-SA-treated and untreated differentiated PLB-985 and HL-60 cells were harvested by centrifugation at 500 ϫ g for 5 min at 4°C and washed once with cold PBS. Cells were adjusted to 5 ϫ 10 7 /ml and extracted in a Triton buffer (20 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1 mM EDTA, and 1% Triton X-100) containing protease inhibitors (2 mM phenylmethylsulfonyl fluoride, 20 g/ml chymostatin, 10 M leupeptin) with gentle shaking for 30 min at 4°C. Samples were then spun at 10,000 ϫ g for 3 min at 4°C, and the supernatant was collected and used as cell extract. Membrane fractions from PLB-985 cells differentiated in the presence or absence of SA were prepared as described previously (33), except that cells at 2.5 ϫ 10 7 /ml in a volume of 0.5-1 ml were disrupted in 1.5 ml microtubes by sonication (Sonics and Materials, Inc., Danbury, CT) three times at 20% power for 5 s each. Protein concentration of cell extracts and membrane fractions was determined by BCA assay (Pierce). Whole cell and membrane extracts were made from 3T3 and COS-7 cell lines following essentially the same protocols described for PLB-985 cells except that cells were harvested by gentle scraping.
Digestion of glycoprotein N-linked oligosaccharides with endo-␤-Nacetylglucosaminidase H or peptide N-glycosidase F was performed on membrane fractions under conditions suggested by the manufacturer. Briefly, a total of 10 g of membrane protein was treated with 500 units of enzyme for 2 h at 37°C, then separated on SDS-polyacrylamide gel (12%), and electroblotted onto nitrocellulose membrane and probed with monoclonal antibodies to gp91 phox and p22 phox . Control membrane samples were treated similarly except that PBS was added instead of enzyme.
Northern Blot Analysis-Total cellular RNA from 3T3 cells, COS-7 cells, and SA-treated and untreated PLB-985 cells differentiated for indicated times was isolated as reported previously (39). After separation on a formaldehyde agarose gel (1%), RNA samples were transferred to Magnacharge nylon membrane (Micron Separations Inc., Westboro, MA) and hybridized with random primer-labeled human gp91 phox , p22 phox , and ␤-actin full-length cDNAs according to the protocols recommended by the manufacturer (29).
Flow Cytometric Analysis of CD11b (MAC-1) Expression-PLB-985 cells differentiated for 5 days in the presence or absence of SA (10 g/ml) were collected by centrifugation and washed once with PBS, then stained with fluorescein isothiocyanate-conjugated mouse antihuman CD11b monoclonal antibodies as recommended by the manufacturer. The expression of CD11b on the cell surface was measured by FACScan (Becton Dickinson, San Jose, CA). Mouse IgG2b was used as an isotype control. A total of 2 ϫ 10 5 cells were analyzed.

Induction of Superoxide-generating Activity during Differentiation of PLB-985 Cells Was Impaired by Succinylacetone
Treatment-Human myeloid leukemia PLB-985 cells are arrested at the early promyelocytic stage and can be induced to differentiate into granulocytic forms with DMF. During differentiation, expression of NADPH oxidase subunits is induced, resulting in a marked increase in superoxide-generating activity, similar to what has been described for HL-60 myeloid leukemia cells (40). As shown in Fig. 1A, NADPH oxidase activity in DMF-induced control PLB-985 cells, as measured by the cytochrome c reduction assay, appeared at day 2 and reached a maximum level at day 5. In the presence of the heme synthesis inhibitor SA, DMF-induced cells exhibited only very low levels of superoxide production (Fig. 1A), with Ϸ5% activity detected at day 5 relative to non-SA-treated controls. The nitro blue tetrazolium test was also used to monitor O 2 Ϫ production in SA-treated and untreated PLB-985 and HL-60 cells during DMF-induced differentiation. In contrast to DMF-induced control cells, where the majority contained numerous dark purple formazan deposits 30 min after phorbol myristate acetate stimulation, SA-treated cells contained only a few formazan deposits (data not shown).
Effect of Succinyl Acetone on Expression of NADPH Oxidase Subunits-We hypothesized that the reduced NADPH oxidase activity in SA-treated PLB-985 and HL-60 cells was associated with the deficient expression of the oxidase cytochrome b 558 due to the inhibition of heme synthesis. The expression of NADPH oxidase subunits during DMF-induced granulocytic differentiation was examined in cell extracts by immunoblot analysis. The abundance of both the gp91 phox and p22 phox subunits of cytochrome b 558 increased during differentiation in non-SA-treated control PLB-985 cells, which were only detected at extremely low levels in SA-treated cells (Fig. 1B).
Similar results were also obtained when HL-60 cells were differentiated with DMF in the presence of SA (data not shown). Interestingly, a Ϸ65-kDa protein immunoreactive with the gp91 phox monoclonal antibody was detected at a similar levels in both SA-treated and untreated differentiated cells, increasing modestly with differentiation (Fig. 1B). This 65-kDa species, which is more abundant in PLB-985 cells grown in serumfree medium compared with those grown with fetal calf serum, 2 appeared to be equivalent with the 65-kDa high mannose precursor of gp91 phox first described in B cell lines (22). To verify this point, cellular membranes prepared from the PLB-985 cells differentiated in the absence or presence of SA for 5 days were digested with the enzyme endo-␤-N-acetylglucosaminidase H, which only removes the high mannose form of N-linked carbohydrate, and peptide N-glycosidase F, which removes all N-linked carbohydrates. As seen in Fig. 2 gp91 phox protein seen after peptide N-glycosidase F digestion of membranes prepared from control PLB-985 cells (Fig. 2). Therefore, the 65-kDa form of gp91 phox seen in both control and SA-treated PLB-985 granulocytes appears to be identical to the high mannose precursor of gp91 phox reported previously (22).
In contrast to the expression of gp91 phox and p22 phox subunits, the expression of soluble NADPH oxidase components during DMF-induced granulocytic differentiation was unaffected by SA treatment. The p47 phox subunit was detected in small amounts at day 0 and increased during differentiation to reach a plateau at day 3 in control PLB-985 cells (Fig. 1C). The p67 phox subunit was readily detected by day 2, consistent with the onset of measurable oxidase activity, and reached a plateau at day 4. A very similar increase in expression of p47 phox and p67 phox was observed in cells induced to differentiate with DMF in the presence of SA (Fig. 1C). No differences in the pattern and the relative amounts of Rac2 expression were seen between the SA-treated and untreated control cells (data not shown).
Differentiation of PLB-985 cells in the presence of SA also did not affect the cell surface expression of the ␤2 integrin, CD11b/CD18 (MAC-1), a glycosylated plasma membrane heterodimer that does not contain heme. As studied by flow cytometry of unpermeabilized cells using an antibody directed against CD11b, no differences in MAC-1 expression were observed in PLB-985 cells differentiated with DMF for 5 days in the presence of SA compared with DMF-induced cells not exposed to SA (data not shown).
Although very little of the mature form of gp91 phox was detected in PLB-985 cells differentiated in the presence of SA, the presence of its 65-kDa high mannose precursor suggested that transcription and translation of the gp91 phox mRNA were not altered by SA treatment. Northern blot analysis for the expression of gp91 phox as well as p22 phox mRNAs was performed using their full-length cDNAs as probes and showed no differences between PLB-985 cells differentiated in the absence or presence of SA (Fig. 3). Taken together, these observations suggest that the decrease in the abundance of both p22 phox and the mature form of gp91 phox in SA-treated cells is post-translational and related to deficient heme incorporation and heterodimer formation in the absence of heme synthesis.
The Inhibition of Cytochrome b 558 Expression and NADPH Oxidase Activity by SA Is Reversible-To confirm that the marked decrease in NADPH oxidase activity and expression of the p22 phox and mature gp91 phox polypeptides in SA-treated PLB-985 granulocytes specifically result from an impairment in heme synthesis, we supplied exogenous hemin at the onset of DMF-induced differentiation of PLB-985 cells in the presence and absence of SA. NADPH oxidase activity and cytochrome b 558 expression were examined at day 5 (not shown). The in-tensity of formazan staining in the nitro blue tetrazolium test for PLB-985 granulocytes differentiated in the presence of SA and exogenous hemin was similar to control PLB-985 granulocytes. In addition, supplying exogenous heme also prevented the decreased expression of mature gp91 phox and p22 phox subunits of cytochrome b 558 as determined by immunoblot analysis. Addition of exogenous heme to PLB-985 cells differentiated in the absence of SA did not significantly alter NADPH oxidase activity or cytochrome b 558 expression.
Whether the decreased expression in p22 phox and the mature 91-kDa form of gp91 phox seen with SA exposure was reversible was also examined. PLB-985 cells were treated with SA for 3 days from the beginning of DMF-induced differentiation, then SA was removed, and the cells were cultured further for an additional 3 days. As shown in Fig. 4A, NADPH oxidase activity rapidly increased beginning 1 day after removal of SA, and after an additional 2 days in culture it reached 93% of the activity seen in control cells differentiated in the absence of SA. In parallel, the expression of the mature 91-kDa form of gp91 phox and p22 phox increased to levels similar to those observed in non-SA-treated controls (Fig. 4B). These observations are again consistent with a role of heme in formation of the gp91 phox -p22 phox heterodimer, resulting in the increased expression of mature gp91 phox as well as its partner, p22 phox . As expected, the differentiation-dependent increase in the levels of the cytosolic components p47 phox and p67 phox was unaffected by the presence or absence of SA (Fig. 4C).
Maturation of N-Linked Oligosaccharides in gp91 phox Does Not Require Heterodimer Formation with p22 phox -To investigate further the role of gp91 phox -p22 phox heterodimer formation in maturation of the N-linked oligosaccharide side chains of gp91 phox and stability of the two cytochrome b 558 subunits, transgenic expression of 22 phox and gp91 phox was studied in two non-phagocytic cell lines, NIH 3T3 murine fibroblasts and monkey kidney COS-7 cells. As described below, the results show that processing of gp91 phox carbohydrate side chains from high mannose to complex oligosaccharides is not dependent on an association of gp91 phox with p22 phox . However, coexpression of both p22 phox and gp91 phox appeared to increase the stability of the mature 91-kDa form of gp91 phox .
Neither gp91 phox nor p22 phox mRNA expression was detected by Northern blot analysis of the parental 3T3 or COS-7 cell lines (not shown). After stable transfection of 3T3 cells with an expression vector containing the gp91 phox cDNA, immunoblots of cell extracts were probed with a gp91 phox monoclonal antibody. A prominent Ϸ65-kDa band was seen (Fig. 5A), which was sensitive to endo-␤-N-acetylglucosaminidase H (Fig. 5C) and thus appears to correspond to the high mannose gp91 phox intermediate. Smaller amounts of more slowly migrating  3. Expression of gp91 phox and p22 phox mRNAs in PLB-985 granulocytes. Total cellular RNAs were isolated from PLB-985 cells differentiated by DMF for 5 days in the absence (ϪSA) or presence (ϩSA) of SA and analyzed by Northern blot analysis using human gp91 phox (left) and p22 phox cDNAs (right) as hybridization probes. The blots were also probed with ␤-actin cDNA (bottom) as a control for sample loading. A total of 10 g of RNA was loaded in each lane. gp91 phox species were also detected ( Fig. 5A) which were resistant to endo-␤-N-acetylglucosaminidase H (Fig. 5C), indicating that these contain mature, fully processed N-linked oligosaccharides. Subsequent transfection of 3T3-gp91 phox clones with a second transgene harboring the p22 phox cDNA resulted in a marked increase in the abundance of the larger gp91 phox species with mature oligosaccharide side chains (Fig. 5, A and C). A similar result was seen for transgenic COS-7 cell lines expressing the gp91 phox cDNA in the absence or presence of p22 phox , except that in COS-7 cells, a Ϸ58-kDa form of gp91 phox which corresponds in size to the core gp91 phox polypeptide was the prominent species detected in absence of p22 phox expression (Fig. 5, B and C). Membranes prepared from either 3T3 or COS-7 cells that coexpressed both gp91 phox and p22 phox supported superoxide production when mixed with neutrophil cytosol in the cell free NADPH oxidase assay, indicating that the two subunits formed a functional cytochrome b 558 ; membranes expressing either subunit alone were not active in this assay. 2 For both 3T3 (not shown) and COS-7 cells transfected with a p22 phox transgene (Fig. 5B), the p22 phox polypeptide was expressed even in the absence of gp91 phox and did not increase markedly in abundance with coexpression of recombinant gp91 phox . Hence, the relative stability of the "free" p22 phox polypeptide appears to differ from PLB-985 and primary neutrophils, where the genetic absence of gp91 phox is associated with a marked reduction in p22 phox protein expression (21,29). DISCUSSION It has now been well established that a membrane-bound cytochrome b 558 and three cytosolic proteins, p47 phox , p67 phox , and Rac, are required for superoxide-generating activity of the phagocyte NADPH oxidase (1)(2)(3)(4). In this study, acquisition of NADPH oxidase activity during granulocytic differentiation of PLB-985 cells was closely correlated with the increased expression of cytochrome b 558 , p47 phox , and p67 phox , similar to what has been described previously for differentiating HL-60 myeloid cells (40). The presence of SA, an inhibitor of heme biosynthesis, during PLB-985 differentiation produced a marked decrease in the expression of 22 phox and mature gp91 phox , the two subunits of cytochrome b 558 , and in NADPH oxidase activity. This effect could be prevented by the addition of exogenous heme, was reversible upon removal of SA and did not affect the expression of other NADPH oxidase subunits or the ␤ 2 integrin MAC-1 during granulocytic differentiation. Therefore, we conclude that the effect of SA on NADPH oxidase activity is a direct result of the disruption of heme synthesis and a concomitant decrease in cytochrome b 558 expression. In the only other report in which the effect of SA on phagocyte NADPH oxidase activity was studied, NADPH oxidase activity in differentiating HL-60 myeloid leukemia cells was also found to be reduced (26). Normal expression of gp91 phox but markedly decreased p22 phox expression was observed along with an increase in the relative level of p47 phox (26). It is unclear why these latter observations differ from our results, which showed that inhibition of heme synthesis by SA resulted in a decrease in both mature gp91 phox and p22 phox without affecting expression of the non-heme-containing oxidase subunits p47 phox and p67 phox during granulocytic differentiation of PLB-985 and HL-60 cells. One explanation might be the different methods used to assess oxidase subunit expression. In the report by Henderson and co-workers (26), indirect immunofluorescence microscopy was used to monitor expression of the oxidase subunits, whereas we used immunoblotting of cell and membrane extracts. We found that although the mature 91-kDa form of gp91 phox was virtually undetectable, expression of the 65-kDa high mannose precursor of gp91 phox was unaffected by SA treatment. Hence, it is possible that the immunoreactive gp91 phox species detected by immunofluorescence represents the high mannose gp91 phox intermediate.
The NADPH oxidase cytochrome b 558 has been proposed to contain two heme groups, which mediate the final step in the transfer of electrons from NADPH to molecular oxygen to generate O 2 Ϫ (6,13,14). Our results suggest that incorporation of heme plays an additional important role in the formation of the gp91 phox -p22 phox heterodimer. In this regard, it is noteworthy that Rotrosen and co-workers (41), in producing recombinant cytochrome b 558 in Sf9 insect cells coinfected with baculovirus vectors for p22 phox and gp91 phox expression, reported that heme supplementation was required for optimal expression of cytochrome b 558 . In SA-treated PLB-985 cells, we observed that the reduced expression of both the mature 91-kDa form of gp91 phox with N-linked complex carbohydrates and the p22 phox polypeptide was caused by post-transcriptional mechanisms in that the expression of gp91 phox and p22 phox mRNAs was not altered. In addition, SA-treated PLB-985 granulocytes expressed the Ϸ65-kDa high mannose gp91 phox intermediate at levels similar to those seen in non-SA-treated cells, indicating that translation of at least the gp91 phox mRNA was not disrupted by inhibition of heme synthesis.
Overall, these data are consistent with a model of cytochrome b 558 biosynthesis in which the heme prosthetic groups play a requisite role in the interaction of the gp91 phox and p22 phox polypeptides and their subsequent stable expression and electron transport function in granulocytic cells. We propose that incorporation of heme promotes heterodimer formation either indirectly by facilitating proper folding of the 22 phox and gp91 phox polypeptides or by acting as a direct dimerization agent. We further propose that in the absence of heme incorporation and heterodimer formation, p22 phox and the mature gp91 phox polypeptide are unstable in PLB-985 granulocytes, although the high mannose 65-kDa gp91 phox precursor can still be detected. As discussed below, we have also shown that heterodimer formation is not required for maturation of gp91 phox carbohydrate to fully processed, endo-␤-N-acetylglucosaminidase H-resistant forms. Thus, the reduced expression of the mature 91-kDa form of gp91 phox with inhibition of heme synthesis may not be caused directly by impaired intracellular processing of the 65-kDa high mannose intermediate but instead may reflect the instability of mature gp91 phox in granulocytic cells in the absence of heterodimerization. Additional issues that remain to be clarified include whether heme incorporation occurs cotranslationally, as has been suggested for globin chain biosynthesis (42), or post-translationally, as has been shown for mitochondrial cytochrome c (43) and myeloperoxidase (44 -46). The compartment in which the formation of p22 phox -gp91 phox complexes normally occurs as the newly synthesized cytochrome b 558 subunits are transported through the endoplasmic reticulum and Golgi to the plasma membrane is also as yet unknown. Intracellular protein processing has been shown in other instances to be influenced by multimer assembly or by the incorporation of prosthetic groups. The formation of heterooligomers can affect protein turnover, and rapid degradation of unassembled subunits has been observed for the acetylcholine receptor and the T cell antigen receptor (47,48). The insertion of prosthetic groups into apoprotein precursors can also play a role in peptide stability and post-translational processing. Newly synthesized chlorophyll apoproteins require chlorophyll for stable accumulation and maturation (49), and complete translocation of cytochrome c across the outer mitochondrial membrane is closely coupled to attachment of heme (43). In neutrophils, the incorporation of heme has been shown to play an important role in the intracellular processing of myeloperoxidase, a hemoprotein whose mature form is located in azurophilic granules. In this case, the heme group is inserted posttranslationally in an apoprotein precursor of myeloperoxidase in the endoplasmic reticulum; but if heme synthesis is inhibited by SA, the apoprotein fails to undergo transport to the lysosome and further proteolytic processing and is instead degraded (44 -46, 50). Hence, it has been proposed that heme incorporation induces conformational changes in the apoprotein which is otherwise not processed correctly (44). Heme insertion and subsequent apoprotein processing have also been shown to be defective in a mutant form of myeloperoxidase with a R569W substitution due to a point mutation in the myeloperoxidase gene, which has been identified as a common cause of human myeloperoxidase deficiency (51).
Studies on COS-7 and 3T3 cells transfected with gp91 phox and/or p22 phox cDNAs indicate that the maturation of N-linked high mannose carbohydrate residues of gp91 phox to complex oligosaccharides, a marker for transport through the Golgi compartment, does not require the formation of gp91 phox -p22 phox heterodimers, in contrast to what has been suggested previously based on studies in B cell lines (22,24). The unassembled p22 phox and mature gp91 phox polypeptides appear to be more stable in these non-phagocytic cells relative to PLB-985 and B cell lines, perhaps because of differences in the proteolytic environment. However, both for COS-7 and 3T3 cells, coexpression of both p22 phox and gp91 phox appeared to enhance stability of the mature 91-kDa form of gp91 phox , consistent with what has been described previously in p22 phox -deficient CGD B cell lines (22,23). These data are also consistent with the original observations on neutrophil cytochrome b 558 expression in CGD, where genetic absence of either the gp91 phox or p22 phox cytochrome subunit results in virtually absent expression of the other cytochrome b 558 subunit (21).
In conclusion, the studies reported here provide indirect but compelling support for an essential role for heme incorporation in the assembly of the cytochrome b 558 heterodimer and provide further support for the concept that the stability of p22 phox and the mature gp91 phox subunit in granulocytic cells is increased by heterodimer formation.