Analysis of Human Phagocyte Flavocytochrome b558 by Mass Spectrometry*

The catalytic core of the phagocyte NADPH oxidase is a heterodimeric integral membrane protein (flavocytochrome b (Cyt b)) that generates superoxide and initiates a cascade of reactive oxygen species critical for the host inflammatory response. In order to facilitate structural characterization, the present study reports the first direct analysis of human phagocyte Cyt b by matrix-assisted laser desorption/ionization and nanoelectrospray mass spectrometry. Mass analysis of in-gel tryptic digest samples provided 73% total sequence coverage of the gp91phox subunit, including three of the six proposed transmembrane domains. Similar analysis of the p22phox subunit provided 72% total sequence coverage, including assignment of the hydrophobic N-terminal region and residues that are polymorphic in the human population. To initiate mass analysis of Cyt b post-translational modifications, the isolated gp91phox subunit was subject to sequential in-gel digestion with Flavobacterium meningosepticum peptide N-glycosidase F and trypsin, with matrix-assisted laser desorption/ionization and liquid chromatography-mass spectrometry/mass spectrometry used to demonstrate that Asn-132, -149, and -240 are genuinely modified by N-linked glycans in human neutrophils. Since the PLB-985 cell line represents an important model system for analysis of the NADPH oxidase, methods were developed for the purification of Cyt b from PLB-985 membrane fractions in order to confirm the appropriate modification of N-linked glycosylation sites on the recombinant gp91phox subunit. This study reports extensive sequence coverage of the integral membrane protein Cyt b by mass spectrometry and provides analytical methods that will be useful for evaluating posttranslational modifications involved in the regulation of superoxide production.

In response to infection and tissue injury, phagocytic leukocytes (monocytes, macrophages, and neutrophils) are recruited to affected areas in large numbers and provide a first line of defense in the innate immune system. Following the recruitment and activation of phagocytic cells, a complex array of oxidative (1) and nonoxidative (2) pathways are utilized in the course of a productive inflammatory response. The generation of reactive oxygen species is initiated by the phagocyte NADPH oxidase, a multisubunit enzyme complex composed of membrane (flavocytochrome b (Cyt b)) 2 and cytosolic (p40 phox , p47 phox , p67 phox , and Rac) components that assemble to generate large quantities of superoxide in activated cells (3,4). The resulting superoxide is released into the phagosomal compartment and extracellular medium, where it provides the common precursor for a host of biologically relevant reactive oxygen and nitrogen species (5). The oxidase-derived superoxide also plays a role in the activation of granule proteases (6) and may directly regulate signal transduction pathways (7). The critical importance of the phagocyte NADPH oxidase in host defense is highlighted by individuals that lack this enzymatic activity (a condition known as chronic granulomatous disease) and suffer from recurrent, life-threatening infections. Although crucial for the elimination of infectious agents, the generation of superoxide must be strictly regulated, since elevated levels of phagocytederived reactive oxygen species are believed to contribute to unwanted host tissue damage that is observed in a variety of inflammatory disease states (8).
The catalytic core of the phagocyte NADPH oxidase (Cyt b) is a heterodimeric integral membrane protein composed of a glycosylated large subunit of 570 amino acids (gp91 phox ) and a tightly associated 195 residue small subunit (p22 phox ) (4,9). The gp91 phox subunit contains all of the redox cofactors required for the transfer of electrons from NADPH to molecular oxygen (10) and represents the prototype for a homologous family of enzymes (Nox family) that generate superoxide in a variety of tissues (11). Consideration of the gp91 phox primary sequence has generated predictions for the overall structure of this subunit, including an N-terminal domain containing six transmembrane ␣-helices (12) and two heme prosthetic groups (13), and a cytosolic C-terminal domain that binds NADPH and FAD (14). The identity of the gp91 phox N-terminal half as the heme-containing domain has been experimentally demonstrated by limited proteolysis studies (15), and site-directed mutagenesis has been used to identify His-101, -115, -209, and -222 as heme-coordinating residues (13). Direct support for transmembrane topology models of gp91 phox has been provided by numerous lines of experimental evidence, including 1) the use of anti-peptide polyclonal antibodies demonstrating the cytoplasmic orientation of the N terminus (16); 2) phage display antibody epitope mapping and peptide inhibition studies supporting the assignment of the first intracellular loop region (17,18); 3) the identification of Asn-132, -149, and -240 as N-linked glycosylation sites using an in vitro transcription/translation system (19); and 4) the specific binding of antibodies directed to the second and third extracellular loop regions of gp91 phox to intact cells (20,21). The C-terminal domain of gp91 phox shows homology to the ferridoxin nucleotide reductase family of oxidoreductases (22), and models of this region have been generated by structure-based alignment programs (14,23). The ability of this domain to bind redox cofactors (NADPH and FAD) has been demonstrated by biochemical labeling experiments (22,24), whereas a variety of methods have been used to identify residues in the gp91 phox C-terminal domain that are critical for assembly of the NADPH oxidase complex (17,18,25,26). Current models localize the gp91 phox C-terminal domain to the cytoplasmic aspect of the plasma or phagosomal membrane (4,12,27).
Although the general topology and function of g91 phox is broadly understood, the architecture and biological role of the p22 phox subunit remains less clearly defined. The p22 phox subunit is required for biosynthesis of the mature Cyt b heterodimer in human phagocytes (28), and current models suggest that p22 phox contains 2-4 membrane-spanning domains (12,29,27). Concerning oxidase function, p22 phox has been shown to be phosphorylated in activated neutrophils (30), and several regions in the primary sequence appear to be involved in binding the cytosolic oxidase subunits p47 phox and p67 phox during oxidase assembly (17,31). The association of p47 phox with a polyproline-rich region of p22 phox is critical for oxidase assembly (31), and the high resolution structure determination of p47 phox tandem Src homology 3 domains complexed with synthetic peptides encompassing residues 149 -168 of p22 phox has characterized the molecular details of this interaction (32,33). Using a fluorescently labeled antibody that binds the p22 phox subunit, physiologically relevant anionic lipids (arachidonic acid and phosphatidic acid) have been shown to induce conformational changes in Cyt b that may trigger electron flow and/or binding of the regulatory subunits (34). Although numerous homologues have been identified for the gp91 phox subunit, homologues have yet to be identified for p22 phox , and recent studies have demonstrated that p22 phox forms a functional complex with several Nox homologues (35)(36)(37). Since Cyt b has yet to be isolated in quantities required for high resolution structure determination, major questions remain concerning the structural basis of superoxide production and sensitive, site-specific methods are required to directly study localized regions of this integral membrane protein.
MALDI and electrospray mass spectrometry are sensitive and highly accurate methods that have shown broad utility in protein biochemistry (38,39). In light of the fact that integral membrane proteins are often isolated in limited quantities, mass spectrometry represents a particularly valuable method for analysis of structure (40,41), conformational changes (40,42), and post-translational modifications (43,44). Since mass spectrometry will facilitate efforts at determining the structural basis of superoxide production by the NADPH oxidase complex, this study reports the first detailed mass analysis of human phagocyte Cyt b and demonstrates extensive sequence coverage of the gp91 phox and p22 phox subunits. The broad utility of these methods is further highlighted by the mass analysis of p22 phox residues that are polymorphic in the human population and the determination of glycosylation sites on the gp91 phox subunit following isolation from both neutrophils and the PLB-985 cell line.

EXPERIMENTAL PROCEDURES
Materials-Trypsin Gold was obtained from Promega; PNGase F was from New England Biolabs; and acetonitrile, isopropyl alcohol, and trifluoroacetic acid were from J.T. Baker. Dodecylmaltoside (DDM) was purchased from Anatrace; phenylmethylsulfonyl fluoride and DTT were from Calbiochem; and Protein A-Sepharose was from Amersham Biosciences. Centricon YM-100 concentrators were purchased from Millipore; ␣-cyano-4-hydroxycinnamic acid (␣-CHC) was from Aldrich; and 2,5-dihydroxybenzoic acid (DHB) and the stainless steel MALDI plate were from Bruker. The Zorbax 300SB-C18 HPLC-Chip used for LC-MS/MS analysis was purchased from Agilent, and Labsafe Gel Blue was obtained from G Biosciences. The 50 Sonic dismembrator probe sonicator and FS30 bath sonicator were obtained from Fisher. The elution peptide Ac-PQVRPI-CONH 2 used for Cyt b immunoaffinity purification was obtained from Macromolecular Resources. Econo-Pac 10 DG desalting columns (30 ϫ 10 ml) were from Bio-Rad, and fetal bovine serum was purchased from Hyclone. All other reagents were obtained from Sigma. Human neutrophil membrane fractions were generated as previously described (9), and all monoclonal antibodies were generated in house using standard hybridoma methods.
Immunoaffinity Purification of Human Neutrophil Cyt b-Immunoaffinity purification of Cyt b was conducted as previously described (45) with modifications. Following incubation of the DDM neutrophil membrane extract with the mAb 44.1 affinity matrix, the beads were poured into a disposable column and washed with 10 ml of 10 mM Hepes (pH 7.4), 100 mM KCl, 10 mM NaCl, 1 mM EDTA (Buffer A), 0.25% DDM. The beads were then washed with Buffer A, 0.1% DDM until the final 280 nm absorption of the flow-through fraction was at base line. Cyt b was eluted from the affinity matrix with Buffer A, 0.1% DDM, 1.5 mM elution peptide (Ac-PQVRPI-CONH 2 ), and the pooled Cyt b elution fraction was concentrated to ϳ0.5 ml in a Centricon YM-100 concentrator. For removal of elution peptide, the concentrated sample was passed over an Econo-Pac 10 DG desalting column (equilibrated with Buffer A, 0.1% DDM) at 4°C with 0.5-ml samples collected and analyzed for Cyt b content by absorption spectroscopy. Cyt b was then stored at Ϫ20°C prior to SDS-PAGE and in-gel digestion. The mass spectral analysis in the present study was carried out using two separate Cyt b preparations.
Batch Mode Fermentation Growth of PLB-985 Cells Expressing Cyt b-In order to obtain stable expression of recombinant Cyt b, X-CGD PLB-985 cells were transduced with the RD114psuedotyped MFGS-gp91 phox vector as previously described (46). Transduced cells were stained with mAb 7D5, and fluorescence-activated cell sorting was used to isolate a population of cells with the highest constitutive expression of Cyt b. For growth of the transduced PLB-985 cells, the initial stationary cell culture was performed in tissue culture flasks using RPMI 1640 medium (pH 7.4) supplemented with 2 g/liter NaHCO 3 , 10 mM Hepes, 100 mg/liter streptomycin sulfate, 100 kilounits/ liter penicillin G, nonessential amino acids (100-fold dilution of stock solution obtained from Sigma; catalogue number M7145), 1 mM sodium pyruvate, 5-10% fetal bovine serum, and 5% CO 2 at 37°C.
Large scale cultivation of PLB-985 cells was performed in a 7.5-liter CelliGen Plus Bioreactor with a BioFlo-3000 console (New Brunswick Scientific). This system was set up with a ring sparger, marine blade impeller and was operated in batch mode. For initial inoculation of the Bioreactor, PLB cells were grown to ϳ1 ϫ 10 6 cells/ml in a spinner flask using medium conditions described above with the exception that 2% horse serum was used in place of the fetal bovine serum. These cells were then added to the Bioreactor containing 5 liters of RPMI 1640 medium (pH 7.4) supplemented with 2 g/liter NaHCO 3 , 100 mg/liter streptomycin sulfate, 100 kilounits/liter penicillin G, 3.25 g/liter glucose, 0.54 g/liter glutamine, and 2% horse serum. For cell growth, the Bioreactor was maintained at 37°C, with a dissolved oxygen content maintained at 40% of air saturation and an agitation of 80 -100 rpm. The pH of the growth medium was maintained at 7.1-7.4 by sparging CO 2 gas into the Bioreactor. Once the PLB-985 cells reached a density of 1-1.8 ϫ 10 6 cells/ml, they were partially harvested by removing 5.5 liter of culture, and the fermentation process was continued by diluting the remaining cells with 5.5 liters of fresh medium.
For purification of Cyt b, 20 -30 ml of the PLB membrane fraction was homogenized, washed with 1 M NaCl for 5 min on ice, and then centrifuged at 100,000 ϫ g for 30 min at 4°C. The membrane pellet was homogenized in 20 ml of Buffer A, 2 mM MnCl 2 , 100 g/ml DNase, 100 g/ml RNase and incubated for 10 min on ice. The mixture was then adjusted to 2 mM phenylmethylsulfonyl fluoride and 1 l/ml P8340 prior to the addition of DDM to 0.8%, brief probe sonication (3 ϫ 5 s on setting 3), and rotation at 4°C for 30 min. This membrane extract was centrifuged at 100,000 ϫ g for 30 min at 4°C, and the supernatant fraction was diluted 7-fold in 10 mM Hepes (pH 7.4), 0.1% DDM, 1 l/ml P8340. The diluted extract was rotated with 10 ml of heparin-Sepharose for 1 h at 4°C. The heparin matrix was then poured into a column and washed with 10 mM Hepes (pH 7.4), 20 mM NaCl, 0.1% DDM until the 280-nm absorption of the flow-through fraction was at base line. Bound material was eluted from the heparin matrix with 10 mM Hepes (pH 7.4), 800 mM NaCl, 0.1% DDM, with 1-ml fractions collected and analyzed by absorption spectroscopy. Peak fractions containing heme absorption (414 nm) were pooled, diluted 5-fold into 10 mM Hepes (pH 7.4), 0.1% DDM, and added to 2 ml of mAb 44.1-Protein A-Sepharose for immunoaffinity purification as outlined above for neutrophil Cyt b.
In-gel Digestion of gp91 phox and p22 phox -Following purification, Cyt b (15-60 pmol/gel lane) was resolved by SDS-PAGE (12.5% separating gel) and then silver-stained by the modified Blum method (47) or stained with Labsafe Gel Blue according to the manufacturer's instructions. For tryptic digestion, protein bands were excised, placed in microcentrifuge tubes, and washed 3 ϫ 10 min in 300 l of 25 mM NH 4 HCO 3 , 50% acetonitrile. Washed gel slices were dried in a vacuum centrifuge and rehydrated in 75 l of 25 mM NH 4 HCO 3 , 10 mM DTT for 30 min at 56°C. For alkylation, the reduced gel slices were dehydrated with 200 l of acetonitrile and incubated with 100 l of 25 mM NH 4 HCO 3 , 55 mM iodoacetamide in the dark for 20 min at room temperature. The gel slices were subsequently washed for 10 min at room temperature with 200 l of 25 mM NH 4 HCO 3 followed by dehydration in 200 l of acetonitrile. After washing gel slices three times in the above fashion, they were dried in a vacuum centrifuge and rehydrated with 50 l of 25 mM NH 4 HCO 3 , 10% acetonitrile, 40 ng/l trypsin on ice for 30 min. Excess buffer was then removed from the rehydrated slices, and 20 -40 l of 25 mM NH 4 HCO 3 , 10% acetonitrile was added for overnight incubation at 37°C.
Following digestion, enough distilled H 2 O was added to cover the gel slices, and the samples were bath-sonicated for 15 min. The aqueous solution was collected, and gel slices were further extracted with 50% acetonitrile, 0.1% trifluoroacetic acid for 20 min at room temperature. The two fractions were then combined and dried by vacuum centrifugation. The dried digest fractions were prepared for MALDI analysis by solubilization in 5-15 l of 50% acetonitrile, 0.1% trifluoroacetic acid. For LC-MS/MS, the above samples were diluted 5-fold with 0.1% trifluoroacetic acid prior to analysis (Extraction Protocol 1).
In an alternative extraction procedure designed to enrich hydrophobic peptides, gel slices were treated as above with the following modifications. After the first (aqueous) extract was collected, the second (organic) extraction was carried out with 40% isopropyl alcohol, 40% acetonitrile, 0.1% trifluoroacetic acid, and the two resulting fractions were combined and dried by vacuum centrifugation. The dried sample was then extracted three times with 50 l of 5% acetonitrile, 0.1% trifluoroacetic acid for 10 min at room temperature (to remove hydrophilic peptides) followed by a single extraction with 15 l of 40% isopropyl alcohol, 40% acetonitrile, 0.1% trifluoroacetic acid for 10 min at room temperature (Extraction Protocol 2).
Digestion of gp91 phox for the Identification of Glycosylation Sites-For analysis of N-linked glycosylation sites, gel slices were processed as outlined above to the step of protease addition, at which point dried slices were rehydrated in 50 l of 25 mM NH 4 HCO 3 containing 6 units/l of PNGase F on ice for 30 min. Excess buffer was then removed and 20 -40 l of 25 mM NH 4 HCO 3 was added prior to overnight incubation at 37°C. Following the deglycosylation reaction, gel slices were washed 3 ϫ 10 min with 300 l of 25 mM NH 4 HCO 3 , dehydrated with 200 l of acetonitrile for 5 min, and briefly dried in a vacuum centrifuge. Dried gel slices were then subject to tryptic digestion and processed by Extraction Protocol 1.
MALDI Mass Spectrometry-MALDI analysis was conducted on a Bruker BiFlexIII mass spectrometer in reflectron mode with positive ion spectra averaged over 100 -600 laser shots. For MALDI, digest fractions were diluted 3-fold with either a saturated solution of ␣-CHC in 50% acetonitrile, 0.1% trifluoroacetic acid, or 25 mg/ml DHB in 50% acetonitrile, 0.1% trifluoroacetic acid. Following a brief incubation, 1 l of sample was spotted on the MALDI plate and allowed to air-dry. For MALDI analysis, the instrument was calibrated (internally or externally) using the singly protonated monoisotopic masses of bradykinin fragment 1-7, angiotensin II, and adrenocorticotropin hormone 18 -36. Theoretical tryptic digests of Cyt b were generated using the program PeptideMass (available on the World Wide Web at us.expasy.org/cgi-bin/ peptide-mass.pl.) for comparison with the observed masses. Alternatively, the observed MALDI masses were analyzed with the program Mascot (available on the World Wide Web at www.matrixscience.com/search_form_select.html) using the following parameters: 0.15 Da error, 1 potential missed cleavage, variable carbamidomethyl (Cys), variable deamidation (Asn/Gln), and variable oxidation (Met).
Nanospray LC-MS/MS-For the nanospray studies, digest fractions were analyzed using an integrated Agilent 1100 LC-ion-Trap-XCT-Ultra system fitted with an Agilent ChipCube source sprayer. Injected samples (1 l) were first trapped and desalted isocratically on a Zorbax 300 SB-C18 Precolumn (5 m, 5 ϫ 300-m inside diameter; Agilent) for 5 min with 0.1% formic acid delivered by the auxiliary pump at 0.6 l/min. The peptides were then reverse eluted from the trapping column and separated on an analytical Zorbax 43-mm-long 300SB-C18 HPLC-Chip connected inline to the mass spectrometer with a flow of 0.6 l/min. Peptides were eluted with a 5-70% acetonitrile gradient in 0.1% formic acid over a 10-min interval. Data-dependent acquisition of collision-induced dissociation MS/MS was utilized, and parent ion scans were run over the mass range m/z 400 -2,000 at 8,100 m/z s Ϫ1 . For analysis of LC-MS/MS data, Mascot searches were typically conducted using the following parameters: 1.5 Da MS error, 0.8 Da MS/MS error, 1 potential missed cleavage, variable carbamidomethyl (Cys), variable deamidation (Asn/Gln), and variable oxidation (Met). Alternatively, extracted ion chromatograms were evaluated for masses of interest with the actual MS/MS data evaluated against theoretical MS/MS spectra of Cyt b tryptic peptides generated using the program MS-Product (available on the World Wide Web at prospector.ucsf.edu/ucsfhtml4.0/msprod.htm).

RESULTS
Mass Analysis of Human Neutrophil gp91 phox -Despite the fact that mass spectrometry represents a powerful technique for protein structural characterization, integral membrane proteins can present unique difficulties (48,49), and these methods have not previously been applied for the mass analysis of Cyt b. In the present study, Cyt b was immunoaffinity-purified from human neutrophil membrane fractions and resolved by SDS-PAGE to separate the gp91 phox and p22 phox subunits for in-gel tryptic digestion (Fig. 1, lane 2). MALDI analysis of the gp91 phox in-gel digest fractions (using the matrix ␣-CHC) generated spectra that were dominated by peaks that could be confidently assigned to gp91 phox ( Fig. 2A and Table 1), demonstrating the utility of MALDI for mass analysis of this highly purified sample. Evaluation of the spec-trum shown in Fig. 2A with the program Mascot identified gp91 phox as the only significant hit derived from the human NCBI Protein Data Base and identified human gp91 phox as the first significant hit when protein sequences from all organisms were searched. In these studies, the majority of the observed masses corresponded to tryptic peptides derived from the cytosolic C-terminal domain, and 70% sequence coverage of this region was obtained by the above analysis. 3 Preliminary efforts at a variety of different digest conditions (in-gel trypsin, in-gel Glu-C, solution trypsin, and solution Glu-C) clearly demonstrated in-gel tryptic digestion to provide the best sequence coverage of gp91 phox (data not shown), and this method was used for the remainder of the present study. To prepare MALDI samples, human neutrophil Cyt b was resolved by SDS-PAGE, stained with Labsafe Gel Blue, and gp91 phox was subject to in-gel tryptic digestion. A, MALDI spectrum generated using the matrix ␣-CHC following the preparation of digest samples using Extraction Protocol 1 outlined under "Experimental Procedures." Extracted peptides from three gel slices (60 pmol of Cyt b/gel slice) were dried in a vacuum centrifuge and resuspended with 5 l of 50% acetonitrile, 0.1% trifluoroacetic acid. Peaks assigned to the gp91 phox subunit are labeled with the observed monoisotopic masses and asterisks. A mass at 3607.30 corresponding to residues 469 -499 was also observed in this spectrum but was omitted for clarity. B, MALDI spectrum generated using the matrix DHB following preparation of digest samples using Extraction Protocol 2. In these studies, extracted peptides from four gel slices (60 pmol of Cyt b/gel slice) were dried in a vacuum centrifuge prior to differential solubilization and mass analysis of the organic extract. Labeled average masses indicate peaks assigned to the gp91 phox subunit, and asterisks indicate ϳ56.5-Da adducts observed in samples analyzed with this matrix.  DECEMBER 1, 2006 • VOLUME 281 • NUMBER 48

JOURNAL OF BIOLOGICAL CHEMISTRY 37049
Although the high mass accuracy of MALDI allows for confident interpretation of spectra generated from highly purified digest samples (such as the gel-purified gp91 phox subunit), the use of LC-MS/MS has the advantage of providing peptide mass and sequence information. In order to obtain support for the above MALDI assignments and provide a complementary method of mass analysis, gp91 phox in-gel digest samples were also analyzed by nanospray LC-MS/MS. As shown in Table 1, the nanospray results correlated well with MALDI analysis and identified a significant number of tryptic peptides derived from the gp91 phox C-terminal domain (49% sequence coverage of this region). Mascot analysis of LC-MS/MS spectra resulted in gp91 phox as the most significant hit in the NCBI human protein data base, with the common contaminant keratin also identified as a significant hit. In light of the difficulties associated with the mass analysis of hydrophobic transmembrane (TM) domains, it serves to note that masses assigned to gp91 phox TM helices 2 and 6 were consistently observed in the above MALDI and LC-MS/MS studies.
Despite the good sequence coverage obtained for gp91 phox , a large tryptic peptide from the cytoplasmic domain (residues 385-421) and four membrane-spanning domains were notably absent from all spectra. In attempts to facilitate mass analysis of these difficult peptides, an extraction procedure was developed for the crude fractionation of in-gel digests. For this procedure, dried digest samples were sequentially solubilized with 5% acetonitrile, 0.1% trifluoroacetic acid (for more polar peptides), followed by 40% isopropyl alcohol, 40% acetonitrile, 0.1% trifluoroacetic acid (for more hydrophobic peptides; Extraction Protocol 2 under "Experimental Procedures"). When material from the high organic fraction was analyzed by MALDI using the matrix DHB, the resulting spectra contained two clearly defined peaks that were not observed in previous MALDI or LC-MS/MS studies (Fig. 2B). Importantly, these masses could be assigned to the gp91 phox TM domain 5 (3301.2 Da; residues 200 -226) and to residues 385-421 from the cytoplasmic domain (3758.6 and 3775.0 Da with Met oxidation), significantly increasing the overall sequence coverage. It is important to note that use of the MALDI matrix DHB proved critical for the detection of these large tryptic peptides, since the corresponding masses were not observed with ␣-CHC.
Taken together, this MALDI and LC-MS/MS analysis provided extensive sequence coverage of the gp91 phox C-terminal domain (87% total coverage) and demonstrates the utility of these methods for the mass analysis of pmol levels of the gelpurified subunit. These studies also allowed for the assignment of proposed TM domains 2, 5, and 6, which are modeled to comprise roughly half of the gp91 phox residues residing within the membrane bilayer. This represents the first detailed mass analysis of human phagocyte gp91 phox .
Mass Analysis of Human Neutrophil p22 phox -For mass analysis of p22 phox , MALDI and LC-MS/MS were conducted following SDS-PAGE and in-gel digestion as outlined above for the gp91 phox subunit. MALDI analysis of in-gel tryptic digests (using the matrix ␣-CHC) generated spectra that were dominated by peaks that could be assigned to p22 phox (Fig. 3A and Table 2) and provided 44% total sequence coverage. LC-MS/MS analysis of p22 phox digest fractions corresponded well with MALDI assignments and provided 33% total sequence coverage ( Table 2). Of interest, Mascot analysis of LC-MS/MS data (using the NCBI human protein data base) also led to the identification peptides that were assigned to the small G-protein Rap1A (22% total sequence coverage), which represented the only significant identification aside from keratin and p22 phox (data not shown). It should be emphasized that peptide masses matching the Rap1A sequence were of relatively low abundance in the LC-MS/MS extracted ion chromatograms FIGURE 3. MALDI analysis of human neutrophil p22 phox . To prepare MALDI samples, human neutrophil Cyt b was resolved by SDS-PAGE, stained with Labsafe Gel Blue, and p22 phox was subject to in-gel tryptic digestion. A, MALDI spectrum generated using the matrix ␣-CHC following the preparation of digest samples as described in the legend to Fig. 2A. The labeled peaks indicate monoisotopic masses assigned to the p22 phox subunit. B, MALDI spectrum generated using the matrix DHB following the preparation of digest samples as described in the legend to Fig. 2B. The labeled peaks indicate average masses assigned to the p22 phox subunit, and asterisks indicate ϳ56.5-Da adducts observed in samples analyzed with this matrix. and were not detected by MALDI, indicating that Rap1A represented a minor component of the gel-purified p22 phox subunit. Nevertheless, the LC-MS/MS analysis provided an independent line of evidence supporting the association of Cyt b and Rap1A (50).
Although reasonable sequence coverage of p22 phox was obtained in the above MALDI and LC-MS/MS studies (44% total sequence coverage), three large, hydrophobic tryptic peptides (residues 1-32, 33-56, and 91-127) were notably absent in all spectra. For attempts to detect the missing peptides, in-gel digest fractions were dried and prepared using Extraction Protocol 2 as described for gp91 phox . When the resulting organic extracts were analyzed using the matrix DHB, peptide masses corresponding the p22 phox residues 2-32 (3199.3 and 3215.6 Da with Met oxidation) and 33-56 (2888.1 Da) could be assigned from the resulting MALDI spectra (Fig. 3B). The detection of residues 2-32 confirms proteolytic processing of the N-terminal Met residue in the mature protein (15). Similar to gp91 phox , the MALDI matrix DHB proved critical for mass analysis, since the above peaks were not observed with ␣-CHC. Using these combined approaches for the preparation of tryptic digest samples, MALDI provided 72% sequence coverage of the p22 phox subunit, including detection of the hydrophobic N-terminal region and 90% of the proposed cytoplasmic domain according to the topology model in Fig. 7 (27).
Mass Analysis of Polymorphisms in the p22 phox Subunit-DNA sequencing studies have revealed a number of naturally occurring p22 phox amino acid polymorphisms in the human population (Lys/Thr-60, His/Tyr-72, Glu/Lys-135, and Ala/ Val-174) that appear to have no effect on host defense (51,52). Although sequencing studies to date suggest that Thr-60 and Lys-135 represent rare polymorphisms, the amino acid variants at residues 72 and 174 have been shown to occur with high frequency in the human population.
The monoisotopic masses at 1184.58 and 1437.72 shown in Fig. 3A were consistently observed in p22 phox tryptic digest fractions by MALDI and could be assigned to residues 128 -137 and 128 -139, respectively, assuming the Glu-135 variant. LC-MS/MS analysis supported the assignment of both MALDI masses (Table 2), directly confirming the presence of Glu-135 in our human donor population. Close manual inspection of the MALDI and LC-MS/MS data provided no indication of the Lys-135 variant in p22 phox digest fractions (data not shown).
Since the Ala-174 and Val-174 p22 phox variants appear to be prevalent in the human population, mass analysis of the p22 phox subunit derived from a pool of human donors would be expected to provide clear evidence for this polymorphism. As shown in Fig. 3A, monoisotopic masses were observed at 3011.73 and 3039.77 Da that could be assigned to p22 phox residues 165-195 assuming Ala-174 (theoretical monoisotopic mass of 3011.53 Da) and Val-174 (theoretical monoisotopic mass of 3039.56 Da). Since MALDI masses assigned to the Val-194 polymorphism were consistently of low abundance, LC-MS/MS was used to provide support for the above assignments. Fig. 4A shows LC-MS/MS extracted ion chromatograms (generated from triply charged parent ions) indicating the relative ion intensities of p22 phox residues 165-195 containing either Ala-174 (curve b) or Val-174 (curve c). Fig. 4B shows the MS/MS data allowing for confident identification of the lower intensity Val-174 variant. The MS/MS spectra observed for both variants were highly similar and were dominated by b and b 2ϩ ions due to two Lys residues at the N terminus and the lack of a Lys/Arg residue at the C terminus of this tryptic peptide. Since the single amino acid substitution at position 174 will have little effect on the ionization potential of these peptides, the above mass analysis indicates a preferential usage of Ala-174 in the donor pool used for this study. To our knowledge, this provides the first direct analysis of naturally occurring p22 phox polymorphisms in the human population at the level of the isolated protein subunit.
Identification of N-Linked Glycosylation Sites on Human Neutrophil gp91 phox -The human gp91 phox subunit contains five consensus motifs for the attachment of N-linked glycans (NX(S/T)) (19), with current topology models localizing three of these sites (Asn-132, -149, and -240) to extracellular loop regions (27). To directly demonstrate that the above residues are genuinely glycosylated in human neutrophils in vivo, mass spectrometry was used to analyze the gp91 phox subunit following deglycosylation and proteolytic digestion.
For mass analysis, gp91 phox was subject to in-gel digestion with PNGase F (to liberate N-linked glycans), followed by digestion of the same gel slice with trypsin. MALDI analysis of the resulting samples revealed three relatively intense peaks with masses corresponding to the proposed gp91 phox glycosylation sites (1242.60 Da, residues 148 -157 containing Asn-149; 1836.93 Da, residues 131-147 containing Asn-132; and 1986.07 Da, residues 230 -247 containing Asn-240), assuming a single Asn/Gln deamidation (ϩ0.98 Da) for each peptide and carbamidomethylation of residues 230 -247 (Fig. 5A). Importantly, digestion with PNGase F results in deamidation of Asn residues that contain N-linked glycans (53), consistent with the above MALDI data. In addition, the above masses were strikingly absent from MALDI spectra of in-gel tryptic digests in the absence of PNGase F treatment (see Fig. 2A) and do not correspond to tryptic peptides derived from PNGase F (which is  DECEMBER 1, 2006 • VOLUME 281 • NUMBER 48

Analysis of Flavocytochrome b by Mass Spectrometry
added to the gel slice and becomes a contaminant prior to tryptic digestion). In order to confirm the identification of gp91 phox glycosylation sites, the PNGase F/trypsin digests were also analyzed by LC-MS/MS. Fig. 5B shows a representative MS/MS spectrum providing peptide sequence information that confidently identifies the tryptic fragment containing Asn-132 (residues 131-147). Importantly, this peptide was assigned to have a single deamidated Asn/Gln residue (observed in all b ions containing Asn-132) and was not detected by LC-MS/MS in the absence of PNGase F treatment. Similar results were obtained for tryptic peptides containing residues 148 -157 and 230 -247 (data not shown). Taken together, these results provide the first direct experimental data demonstrating that Asn-132, -149, and -240 are modified by N-linked glycans on gp91 phox in human neutro-  To prepare samples for mass analysis, isolated Cyt b was resolved by SDS-PAGE, and the gp91 phox subunit was subject to in-gel digestion with PNGase F, followed by in-gel tryptic digestion. A, MALDI spectrum of human neutrophil gp91 phox generated using the matrix ␣-CHC following the preparation of digest samples using Extraction Protocol 1. In these studies, extracted peptides from a single gel slice (25 pmol of Cyt) were dried with a vacuum centrifuge prior to resuspension with 5 l of 50% acetonitrile, 0.1% trifluoroacetic acid. Peaks assigned to gp91 phox (C) or PNGase F (P) are indicated, whereas peaks corresponding to gp91 phox glycosylation sites are labeled with the observed monoisotopic mass and an asterisk. B, MS/MS spectrum confirming the identification of Asn-132 as a gp91 phox glycosylation site. In these studies, MALDI samples outlined in Fig. 5A were diluted 5-fold with 0.1% trifluoroacetic acid prior to LC-MS/MS. The positive ion mass spectrum for the tryptic peptide corresponding to gp91 phox residues 131-147 is shown with b and y ions indicated. The asterisks indicate additional masses that were assigned to this peptide, and hyphens indicate sequence information obtained from the MS/MS analysis.
phils. Detection of these extracellular loop region peptides also increased sequence coverage of the gp91 phox subunit to 73% total.
Isolation of Cyt b from PLB-985 Cells and Identification of N-Linked Glycosylation Sites-The PLB-985 myeloid leukemia cell line (54) represents an important model system for investigation of the NADPH oxidase complex and has been used for the functional analysis of Cyt b (13,26,55). Although these cells grow extremely well as a suspension culture, this system has yet to be exploited for isolation of Cyt b at levels required for structure analysis. To initiate the use of recombinant Cyt b for mass spectrometry, methods were developed for purification of Cyt b from PLB-985 membrane fractions to confirm the appropriate modification of N-linked glycosylation sites on the gp91 phox subunit. In these studies, a virally transduced PLB-985 cell line expressing significant levels of Cyt b in the absence of differentiation (expression levels were ϳ5-10-fold less than human neutrophils; data not shown) was cultivated using a batch mode fermentation scheme to generate starting membrane fractions for Cyt b purification. Following extraction of the Bioreactorderived PLB-985 membranes, highly purified Cyt b could be generated at levels of ϳ300 pmol of Cyt b/1 ϫ 10 10 cell equivalents using heparin-Sepharose and a mAb 44.1 affinity matrix (Fig. 1, lane 3). Semiquantitative Western blotting indicated that neutrophil and PLB-985 Cyt b are isolated with roughly similar efficiencies by the above method (data not shown).
For the determination of N-linked glycosylation sites, the PLB-derived gp91 phox subunit was resolved by SDS-PAGE for in-gel digestion with either trypsin or PNGase F/trypsin as outlined above for neutrophil gp91 phox . Fig. 6 shows a representative MALDI spectrum collected for the PNGase F/trypsin-di-gested PLB-985 gp91 phox subunit. Similar to the data shown in Fig. 5A, this spectrum was dominated by peaks that could be assigned to gp91 phox and PNGase F and contains masses that correspond to the proposed gp91 phox glycosylation sites. LC-MS/MS confirmed the above MALDI assignments and demonstrated that the appropriate gp91 phox glycosylation sites are genuinely modified in PLB-985 cells (data not shown).

DISCUSSION
The generation of superoxide by the phagocyte NADPH oxidase initiates a complex series of biological reactions that participate in diverse physiological processes ranging from the elimination of pathogenic organisms to the propagation of detrimental, systemic inflammatory cascades (56). To advance long term efforts at defining the molecular mechanisms of superoxide production by this enzyme complex, the present study reports the first detailed characterization of human neutrophil Cyt b by mass spectrometry.
Following immunoaffinity purification and in-gel tryptic digestion, extensive sequence coverage of the Cyt b heterodimer was obtained by a combination of MALDI and nanospray LC-MS/MS, with the most prominent gaps in coverage observed for the hydrophobic transmembrane domains. The particularly high sequence coverage for regions modeled to reside on the cytoplasmic aspect of the membrane (Fig. 7) shows the potential utility of mass spectrometry for identifying post-translational modifications on Cyt b. Along these lines, multiple protein kinases have been implicated in the regulation of the NADPH oxidase complex following phagocyte activation, and although the majority of studies have described phosphorylation of the oxidase cytosolic subunits (4), the p22 phox subunit of Cyt b is also phosphorylated in a manner that correlates with oxidase activation in human neutrophils (30). The determination of Cyt b phosphorylation sites by mass spectrometry would provide an important guideline for site-directed mutagenesis studies aimed at testing the functional effects of this modification.
In the present study, the issue of Cyt b microheterogeneity was highlighted by the direct analysis of p22 phox polymorphisms in our human donor population. In p22 phox digest samples, the Glu-135 variant was exclusively observed by both MALDI and LC-MS/MS, consistent with DNA sequencing studies that to date have detected only a single individual in the human population encoding Lys-135 (51). Residue 135 occupies a central position in the core epitope region mapped for mAb NS2 (27), and it will be important to assess the effect of the Lys-135 substitution on immunoblot analysis of p22 phox . The disruption of NS2 binding would make this mAb a valuable diagnostic agent for the Lys-135 polymorphism. Both the Ala-174 and Val-174 variants of p22 phox were identified in this study, which is consistent with the frequent occurrence of this polymorphism at the nucleotide level (51,52). Concerning the Lys/Thr-60 polymorphism, residue 60 occurs in a polybasic region of p22 phox ( 58 KRKK 61 ), and complete tryptic digestion would result in amino acids and low molecular weight peptides that would not be readily identified by mass analysis. Although both variants of the His/Tyr-72 polymorphism would be anticipated in our donor population, masses that could be assigned FIGURE 6. Identification of PLB-985 gp91 phox N-linked glycosylation sites by MALDI mass spectrometry. Following isolation of Cyt b from PLB-985 membrane fractions, gp91 phox was prepared for MALDI analysis as described in the legend to Fig. 5A with the exception that extracted peptides were derived from three gel slices (15 pmol of Cyt b/lane). Peaks assigned to gp91 phox (C) or PNGase F (P) are indicated, whereas peaks corresponding to gp91 phox glycosylation sites are labeled with the observed monoisotopic mass and an asterisk. In each case, the assignment of MALDI masses corresponding to the respective glycosylation sites was confirmed by nanospray LC-MS/MS. DECEMBER 1, 2006 • VOLUME 281 • NUMBER 48 to either variant were not confidently observed in any MALDI or LC-MS/MS spectra. The presence of Tyr-72 has been shown to result in somewhat diminished superoxide production in both human neutrophils (57) and vascular cells (58), and it will be of interest to optimize mass analysis to detect this region. Importantly, the ability to directly evaluate relative protein expression levels in individuals that are heterozygous for p22 phox polymorphisms will facilitate studies designed to determine the effects of Cyt b microheterogeneity on oxidase function.

Analysis of Flavocytochrome b by Mass Spectrometry
Although extensive sequence coverage of the Cyt b heterodimer was obtained in the present study, the complete characterization of any integral membrane protein requires successful mass analysis of TM domains. Due to the relative lack of Lys/Arg residues, TM ␣-helices are typically present as relatively large, hydrophobic peptides in tryptic digest samples and have been shown to present difficulties for mass analysis (48,49). In the present study, the proposed gp91 phox TM domains 2 and 6 were readily detected by both MALDI and LC-MS/MS, and the detection of these fragments was undoubtedly facilitated by the presence of trypsin cleavage sites within the membrane-spanning region. A MALDI mass corresponding to the gp91 phox TM domain 5 could only be assigned using DHB as the matrix, in conjunction with tryptic peptides that were extracted using a strong organic solvent system. This differential extraction protocol also allowed for the mass analysis of p22 phox residues 2-55, which are modeled as either two TM domains (29) or the N-terminal cytoplasmic domain (27). Current efforts are focused on successful mass analysis of all Cyt b transmembrane domains. The availability of increased levels of purified Cyt b will facilitate mass analysis of difficult peptides (such as TM domains), and a major goal of our group is the development of an economically feasible, large scale isolation method for Cyt b. As a preliminary step, this study describes a PLB-985 expression system that allowed for purification of Cyt b at levels that were suitable for mass spectrometry. A similar purification procedure has been used by our group for the isolation of Cyt b following detergent solubilization of intact neutrophils (45) and may represent a robust protocol applicable to other recombinant expression systems.
Since the glycosylation pattern gp91 phox had yet to be determined in primary phagocytes, mass analysis was used in this study to directly demonstrate that Asn-132, -149, and -240 are modified by N-linked glycans on human neutrophil gp91 phox . Similar studies were also carried out for PLB-985 gp91 phox and demonstrated the appropriate modification of glycosylation sites in this model cell line. This analysis supports current gp91 phox structure models and directly confirms the topology of extracellular loops 2 and 3. Localization of these loops to the extracellular side of the plasma membrane was previously suggested by primary sequence analysis (59,60), identification of glycosylation sites in an in vitro transcription/translation system (19) and phage display analysis indicating that extracellular loops 2 and 3 form a complex epitope recognized by mAb 7D5 (20). In addition to the identification of the gp91 phox glycosylation sites, tryptic peptides containing Asn-96 and -430 (which represent potential N-linked glycosylation sites) were detected by MALDI in the absence of deglycosylation. These results support the above transcription/translation studies (19) and are consistent with topology models localizing these residues to the cell cytoplasm (16,27). Since the biological function of the gp91 phox carbohydrate groups remains to be determined, it is of interest to note that the murine gp91 phox subunit (which appears to be glycosylated at only one site on the first extracellular loop) can form a functional complex with human p22 phox in PLB-985 cells (59). Interestingly, heterodimer formation is not required for the addition of complex carbohydrates to recombinant gp91 phox (61), although it remains to be determined which Asn residues are actually modified in these expression systems. Although beyond the scope of the present study, it will be of interest to attempt a detailed characterization of Cyt b glycans by mass spectrometry, since absolutely no information exists concerning the structures of these species. Current efforts are focused on further optimizing the mass analysis of Cyt b to identify potential post-translational modifications that serve to regulate superoxide generation.