|Home | About | Journals | Submit | Contact Us | Français|
Regulated generation of reactive oxygen species (ROS) is primarily accomplished by NADPH oxidases (Nox). Nox1 to Nox4 form a membrane-associated heterodimer with p22phox, creating the docking site for assembly of the activated oxidase. Signaling specificity is achieved by interaction with a complex network of cytosolic components. Nox4, an oxidase linked to cardiovascular disease, carcinogenesis, and pulmonary fibrosis, deviates from this model by displaying constitutive H2O2 production without requiring known regulators. Extensive Nox4/Nox2 chimera screening was initiated to pinpoint structural motifs essential for ROS generation and Nox subcellular localization. In summary, a matching B loop was crucial for catalytic activity of both Nox enzymes. Substitution of the carboxyl terminus was sufficient for converting Nox4 into a phorbol myristate acetate (PMA)-inducible phenotype, while Nox2-based chimeras never gained constitutive activity. Changing the Nox2 but not the Nox4 amino terminus abolished ROS generation. The unique heterodimerization of a functional Nox4/p22phox Y121H complex was dependent on the D loop. Nox4, Nox2, and functional Nox chimeras translocated to the plasma membrane. Cell surface localization of Nox4 or PMA-inducible Nox4 did not correlate with O2− generation. In contrast, Nox4 released H2O2 and promoted cell migration. Our work provides insights into Nox structure, regulation, and ROS output that will aid inhibitor design.
The family of NADPH oxidases consists of seven members termed Nox/Duox that differ in their tissue expression profiles, modes of activation, reactive oxygen species (ROS) outputs, and physiological functions. Understanding their distinguishing features is a prerequisite for rational inhibitor design and thus targeted intervention in ROS-mediated pathophysiologies (4). The coexpression of different Nox isoforms, each with potentially distinct functional profiles, in the same cell type necessitates a more discriminating approach than application of pan-Nox inhibitors. Detailed structure-function studies are necessary to identify unique regions and their impact with respect to catalytic function or localization of the enzyme. All Nox/Duox enzymes share a Nox backbone with six predicted transmembrane domains and an intracellular carboxyl-terminal domain which harbors FAD and NADPH binding sites. Nox5 and Duox1/2 enzymes contain additional structural elements such as amino terminal EF-hand motifs, a hallmark of their regulation by the intracellular calcium concentration (13, 30).
The founding member of the NADPH oxidase family, the phagocyte oxidase, consists of membrane-bound Nox2 in a complex with the smaller subunit p22phox (3). Heterodimerization of these two proteins is required for maturation and translocation of the enzyme complex to the plasma membrane or to intracellular vesicles. The Nox family members Nox1, Nox3, and Nox4 follow this paradigm (1, 14, 21, 25, 31). Heterodimer formation and association of the Nox/p22phox complex at particular cellular membranes is essential for catalytic activity, i.e., for ROS generation. Nox2, and to a lesser degree Nox1 and Nox3, remain dormant under resting conditions and rely on stimulus-dependent translocation and assembly of oxidase components such as p47phox and p67phox, or NoxO1 and NoxA1 in the case of Nox1 and Nox3 (16). These steps, together with activation and translocation of the GTPase Rac, ultimately lead to the assembled, catalytically active oxidase and to ROS generation.
Nox4 differs from the usual theme of multimeric assembly of active NADPH oxidases found in Nox1 to Nox3 (21, 22, 28, 32). Constitutive H2O2 production by Nox4 localized at perinuclear vesicles has been reported (1, 21, 28). Since NADPH oxidases catalyze the one-electron reduction of molecular oxygen to superoxide anion, the current dogma suggests that Nox4 generates intracellular superoxide. The superoxide produced will then dismutate rapidly to H2O2, diffusing from the cell into the extracellular milieu. Cytosolic proteins, which regulate the activity of Nox1 to Nox3 by binding to the carboxyl-terminal domains of Nox1 to Nox3, seem to be irrelevant for Nox4 function. The membrane-bound subunit p22phox is to date the only known protein associated with Nox1 to Nox4. Heterodimerization, translocation, and enzymatic function of these oxidases require p22phox. Recent structure-function analyses of complexes between Nox2 or Nox4 and the subunit p22phox documented specific regions and amino acid residues in p22phox necessary for complex formation and oxidase activity (35, 37). Interestingly, a p22phox mutant (p22phox Y121H) is capable of distinguishing between Nox1 to Nox3 and Nox4 by forming a functional complex only with Nox4, further suggesting unique structural features in Nox4 (35).
In this study, we expand structure-function analysis of the oxidase complex by comparing Nox4/Nox2 chimeric enzymes with respect to NADPH oxidase activity, type of reactive oxygen species produced, requirement for additional oxidase components, and detailed subcellular localization.
Human H661 lung carcinoma cells (ATCC HTB-183) were cultivated in RPMI 1640, and Cos-p22phox cells (36), Cos-phox cells (27), and Cos-Nox4/p22phox cells were cultivated in Dulbecco modified Eagle medium (DMEM). Cos-Nox4/p22phox and Cos-Nox4 P437H/p22phox cells were generated by lentiviral transduction and sorted as described previously (20, 35). All growth media were obtained from Invitrogen, CA, and supplemented with 10% fetal calf serum (FCS).
Expression plasmids pcDNA3.0 hNox4 and pcDNA hNox2 were described previously (37). Nox chimeras were generated using PCR or the PCR-based QuikChange site-directed mutagenesis kit (Stratagene, CA) according to the manufacturer's protocol and were verified by sequencing. Amino acid sequences of chimeric proteins can be found in the supplemental material. Alignments were generated with AlignX (Informax, CA). Nonsimilar amino acids are shown in black, conservative changes are shown in blue, blocks of similar amino acids are highlighted in green, identical sequences are depicted in red with a yellow background, and weakly similar residues are in green. Loop (A to E) and transmembrane (TM 1 to 6) assignments are according to Kawahara and coworkers (13). FAD and NADPH binding sites in the C terminus were based on the work of Shiose et al. (29), although different predictions exist (7, 34). Transient transfections of H661 and Cos-p22phox cells were performed using Lipofectamine Plus (Invitrogen, CA) or FuGene6 (Roche, Switzerland) according to the manufacturer's instructions. Nox4, Nox2, and chimeras were always coexpressed with p47phox and p67phox. Experiments were performed at 48 h posttransfection.
Cells were lysed in radioimmunoprecipitation assay (RIPA) buffer [50 mM Tris(2-carboxyethyl)phosphine, 100 mM NaCl, 1% NP-40, 0.1% SDS, 0.5% deoxycholate] containing protease inhibitors (Complete mini; Roche, Switzerland). Lysates were clarified by centrifugation at 15,000 rpm for 10 min at 4°C. Proteins were separated by 11% SDS-PAGE and blotted onto nitrocellulose (Bio-Rad, CA). Membranes were blocked in Tris-buffered saline containing 1.5% bovine serum albumin (BSA) and 2% goat serum. Primary antibodies were as follows: polyclonal rabbit anti-Nox4 antibody (35), anti-Nox2 monoclonal antibody (MAb) 54.1 (6), rabbit polyclonal anti-p22phox antibody FL-195 (Santa Cruz, CA), polyclonal rabbit anti-p47phox and anti-p67phox (Millipore, MA), anti-V5 MAb (Sigma, MO), and rabbit polyclonal antiactin antibody (Sigma, MO). Secondary antibodies used were goat anti-mouse or goat anti-rabbit antibody conjugated to horseradish peroxidase (Southern Biotech, AL), followed by detection with ECL (Pierce, IL).
Cells were trypsinized, washed in phosphate-buffered saline (PBS) containing 1.5% BSA and 2% goat serum, and incubated with rabbit polyclonal anti-Nox4 antibody, anti-Nox2 MAb 7D5 (24), anti-Nox2 MAb 54.1 or anti-V5 MAb in cold fluorescence-activated cell sorting (FACS) buffer (PBS, 0.5% BSA) on ice for 30 min. For intracellular staining, cells were incubated with 0.2% saponin for 10 min on ice. After washing in FACS buffer, cells were incubated with secondary antibodies, goat anti-rabbit antibody-biotin or goat anti-mouse antibody-biotin (BD Bioscience, NJ), followed by washing and incubation with streptavidin-phycoerythrin (BD Bioscience, NJ). After the final washing step, fluorescence of cells was measured using a BD LSR II flow cytometer and FACSDiva 6.0 (BD Bioscience, NJ). Cell populations were gated for live cells. Data were analyzed with FlowJo (Tree Star). Untransfected Cos-p22phox cells served as a negative control.
Cells were grown on glass coverslips and transfected as described above. Cells were washed with PBS and fixed in 2% paraformaldehyde, followed by permeabilization with 0.1% Triton X-100 and blocking in PBS containing 1.5% BSA and 1% goat serum. Cells were stained with polyclonal rabbit anti-Nox4 antibody and anti-p22phox MAb 449 (33) or anti-Nox2 MAb 54.1 (6) and polyclonal rabbit anti-p22phox antibody FL-195 (Santa Cruz, CA). Nuclei were stained using DAPI (4′,6′-diamidino-2-phenylindole) (Sigma, MO). Secondary antibodies used were goat anti-rabbit antibody-Alexa Fluor 568 and goat anti-mouse antibody-Alexa Fluor 488 (Molecular Probes, OR). Images were taken on a Bio-Rad/Zeiss Radiance 2100 Rainbow laser scanning confocal microscope (laser 405 nm, 488 nm, and 543 nm) with a 63× oil objective lens (Plan Apo, 1.4 numerical aperture) and were processed using Zeiss LSM Examiner, Bio-Rad LaserSharp 2000 (v. 6.0), Image J (v. 1.34), and Adobe Photoshop CS3.
Cos-p22phox cells were trypsinized, washed in cold FACS buffer, and incubated with primary antibodies for 30 min on ice. The cells were then fixed in 2% paraformaldehyde in PBS on ice and washed, and free aldehydes were quenched with 50 mM glycine in PBS prior to incubation with gold-tagged secondary antibodies. Secondary antibodies used were goat anti-rabbit IgG(H+L) 12-nm gold and goat anti-mouse IgG(H+L) 12-nm gold (Jackson ImmunoResearch, West Grove PA). Cells were subsequently washed in PBS and fixed in 2.5% glutaraldehyde in 0.1 M cacodylate buffer. Following postfixation in 1% osmium tetroxide, the pelleted cells were dehydrated in a graded ethanol series, transitioned in propylene oxide, and embedded in Epon/Araldite (Electron Microscopy Sciences, Hatfield, PA). Following an initial polymerization to form a small resin button at the bottom of the microcentrifuge tube, the resin pellets were removed from the microcentrifuge tubes, trimmed, reembedded in liquid resin in a regular flat embedding mold (Ted Pella Inc, Redding, CA), and oriented such that transverse sections could be prepared through the depth of the original cell pellet to show the history of sedimentation. Thin sections (60 nm) were cut with a diamond knife (Diatome, Hatfield, PA), mounted on copper slot grids coated with Parlodion, and subsequently stained with uranyl acetate and lead citrate for examination on a Philips CM100 electron microscope (FEI, Hillsbrough, OR). Images were documented using a Megaview III charge-coupled device (CCD) camera (Olympus Soft Imaging Solutions, Lakewood, CO). Images were then further processed in Adobe Photoshop CS3.
To measure H2O2 production the homovanillic acid (HVA) assay was performed as described previously (21). In short, cells were washed with Hanks balanced salt solution (HBSS) and incubated at 37°C for 1 h in HVA solution (100 mM homovanillic acid and 4 U/ml horseradish peroxidase in HBSS with Ca2+ and Mg2+). The reaction was terminated by adding stop buffer (0.1 M glycine-0.1 M NaOH [pH 12] and 25 mM EDTA in PBS). Fluorescence was read on a Biotek Synergy HT (320-nm excitation, 420-nm emission). Fluorescence readings were converted into nmol H2O2 based on the H2O2 standard curve. Superoxide generation was measured by cytochrome c assay as described previously (35). Cells were incubated for 1 h at 37°C in HBSS containing 1 mg/ml cytochrome c (Sigma, MO). Absorbance was measured at 550 nm on a Biotek Synergy HT. When indicated, cells were stimulated with phorbol myristate acetate (PMA) at a concentration of 1 μg/ml. The background absorbance of the cytochrome c solution was subtracted from all measurements.
Transwell migration experiments were performed on semipermeable membranes (8 μm) (Costar Transwell Clear; Corning, Corning, NY) precoated with 0.05% rat tail collagen (Sigma, MO). Cells were suspended in 0.5% FCS-containing DMEM and seeded in triplicate with 1 × 105 cells per membrane; when indicated, cells were stimulated with PMA at a concentration of 1 μg/ml. After 30 min of attachment, directional migration was induced by addition of 10% FCS to the bottom chamber. Cells were allowed to migrate for 5 h at 37°C. Cells were washed with PBS and fixed in 4% paraformaldehyde (PFA). After removal of cells from the upper face of the membrane, membranes were incubated for 1 h with 20 μg/ml DAPI. Nuclei on the bottom membrane were counted in at least 20 random fields per membrane with Image-Pro Plus 7.0 (Media Cybernetics, Inc., MD). Migration rates were analyzed using a two-tailed Student's t test. A P value of <0.05 was considered statistically significant. Significance levels are indicated as follows: *, P < 0.05; **, P < 0.01; and ***, P < 0.001.
All experiments were performed at least three times each in triplicate (error bars indicate standard deviations [SD]; n = 3). Shown are representative examples of at least three independent experiments.
Chronic granulomatous disease (CGD) phenotypes and mutational studies of the phagocyte oxidase Nox2 suggested that the two intracellular loops of the Nox domain are important for superoxide generation (13, 19). To assess how complete exchange of these loops would alter the unique regulation of Nox4 versus Nox2, chimeric constructs were prepared by swapping the amino acid residues between the second and third transmembrane domains (B loop) or between the fourth and fifth transmembrane domains (D loop) at positions with sequence overlap in at least two identical amino acids (Fig. 1A and B; see Fig. S1 in the supplemental material). In general, chimeras featuring a Nox4 backbone belong to the “b” series, while corresponding Nox2 chimeras are in the “a” series. Chimeras were expressed in a Cos7 cell line containing human p22phox (Cos-p22phox), and cells were analyzed for H2O2 production, protein expression, and intracellular localization of Nox. Wild-type Nox2, in contrast to wild-type Nox4, requires the presence and stimulus-induced assembly of oxidase components for activity. Hence, Cos-p22phox cells expressing Nox chimeras were always cotransfected with the cytosolic components p47phox and p67phox. Transfected cells were assayed for ROS generation while adherent to provide optimal conditions for Nox4 activity. Both procedures show no adverse or synergistic effects on the activity of wild-type Nox4 or Nox2 (21). Chimeras with B-loop substitution lost their ability to produce ROS, either as a constitutively active enzyme (Nox4 13b) or as a PMA-activated oxidase (Nox2 13a). In contrast, exchange of the D loop did not alter Nox4 or Nox2 activity (Nox4 12b and Nox2 12a) (Fig. (Fig.1B).1B). Immunoblotting with specific anti-Nox4 or anti-Nox2 antibodies revealed similar expression and maturation of wild-type and chimeric proteins (Fig. (Fig.1C).1C). Using antibodies recognizing extracellular domains in Nox backbones (24, 35), Nox2, Nox4, and all chimeras were detected on the cell surface of nonpermeabilized Cos-p22phox cells (Fig. (Fig.1D).1D). This observation was confirmed by immunofluorescence staining (Fig. (Fig.1E).1E). Although transiently expressed wild-type Nox4 and Nox4 chimeras were also found in intracellular structures reminiscent of the endoplasmic reticulum (ER) or perinuclear vesicles, the enzymes clearly lined the plasma membrane. Nox2 and Nox2-based chimeras were distributed between the plasma membrane and the ER, although maturation and translocation to the cell surface were somewhat reduced in the Nox2 13a B-loop chimera.
The C-terminal cytoplasmic domain of Nox2 contains binding sites for the oxidase components p47phox, p67phox, and the GTPase Rac, which are essential for assembly and activation of a functional Nox2-based oxidase (12, 17, 18). When exogenously expressed, Nox4 activity is not dependent on these proteins (21). Carboxyl-terminal Nox4/Nox2 exchange chimeras were prepared, which included switching of the complete C terminus or of shorter fragments (Fig. (Fig.2;2; see Fig. S2 in the supplemental material). All of the Nox4/Nox2 (“b”) and Nox2/Nox4 (“a”) chimeras were expressed at similar protein levels and translocated to the cell surface as shown by immunoblotting, immunofluorescence, and flow cytometry (Fig. (Fig.2).2). HVA assays were performed to detect constitutive or inducible activation of the chimeras in direct comparison with the wild-type proteins. Replacement of the complete C terminus of Nox4 with the Nox2 sequence resulted in a Nox chimera (Nox4 3b) that displayed the hallmarks of Nox2. This chimera lost the capacity for constitutive ROS generation and acquired a PMA-inducible phenotype. Thus, the intracellular region following the sixth transmembrane domain contains all the binding sites required for assembly of a functionally active Nox2-like oxidase (Fig. (Fig.2A).2A). Interestingly, swapping sequences 12 amino acids further downstream (Nox4 1b) decreased ROS generation by approximately 30%, while PMA-mediated ROS generation was still retained. In contrast, the corresponding C-terminal substitutions in Nox2 (Nox2 3a and 1a) remained catalytically inactive (Fig. (Fig.3E3E and data not shown; see Fig. S4 in the supplemental material), indicating that the Nox4 C terminus alone cannot transform Nox2 into a constitutively active oxidase.
Although exchange of the complete C terminus after the last transmembrane domain was well tolerated in Nox4, swapping shorter sequences located after the second NADPH binding site led to catalytically inactive Nox4 chimeras (Fig. 2A and B). These chimeras still formed heterodimers with p22phox and localized to the cell surface (Fig. 2C and E). Nox4 tolerated the substitution of amino acid residues located between the second NADPH domain starting with the consensus WFADLL and terminating at a putative additional NADPH domain ending with GRP, although ROS generation in this case was reduced (Nox4 10b) (Fig. (Fig.2;2; see Fig. S2 in the supplemental material). Removal of the penultimate C-terminal serine in Nox4 was inconsequential (Nox4 20b) (Fig. (Fig.2A).2A). Analysis of Nox2 chimeras featuring shorter C-terminal Nox4 sequences showed that the last 50 five amino acid residues in Nox2 do not contain information vital for inducible Nox2-dependent ROS production (Nox2 11a) (Fig. (Fig.2D;2D; see Fig. S2 in the supplemental material).
We then wished to determine if the catalytic function of Nox isoforms is dependent on the interaction between the cytosolic C terminus and one of the intracellular loops. The PMA-induced activity of the Nox4 3b chimera indicated that the combination of a Nox4 B loop with a Nox2 C terminus was well tolerated. Additional substitution of the B or D loop in Nox4 3b (Nox4 4b and 5b) did not alter the inducible Nox2-like phenotype (Fig. 3A to D; see Fig. S3 in the supplemental material), although a Nox4 backbone incorporating the Nox2 C terminus and Nox2 B loop (Nox4 4b) displayed somewhat reduced ROS production. Chimeras 3b, 4b, and 5b maintained maturation and were expressed at the cell surface of Cos7 cells. Reverse chimeras, constructed by integrating a Nox4 B loop or a Nox4 D loop together with the Nox4 C terminus into a Nox2 backbone (Nox2 4a and 5a) did not produce ROS (Fig. 3E and F; see Fig. S4 in the supplemental material). Thus, matching the B loop and the C terminus of Nox4 did not rescue enzymatic activity when the Nox2 backbone was present.
We hypothesized that regions in the last 55 amino acids of the C terminus may interact with intracellular loops if the majority of the FAD and NADPH domains would be conserved. Matching the B loop and the last 55 residues of Nox2 (Nox4 30b) could not rescue the catalytic activity of Nox4 11b (Fig. (Fig.4A;4A; see Fig. S5 in the supplemental material), although the chimeras showed similar expression on the cell surface (Fig. 4B and C). Interestingly, these studies indicated that a sequence comprised of the last 22 amino acid residues is critical for Nox4 activity (Nox4 41b). Similar substitutions were performed on Nox2 11a (see also Fig. Fig.2D),2D), the only Nox2 chimera harboring C-terminal Nox4 amino acid residues, which was capable of retaining Nox2 activity. Additional B- or D-loop substitutions in this chimera indicated that the Nox2 B loop is critically involved in ROS production (Nox2 18a) (Fig. (Fig.4D;4D; see Fig. S5 in the supplemental material). Maturation and cell surface expression of Nox2 18a were decreased, but the chimera clearly translocated to the plasma membrane (Fig. 4E and F). In conclusion, the B loop of Nox4 is fully capable of interacting with the Nox2 C terminus, forming a PMA-activated Nox2-type oxidase on a Nox4 backbone, while similar substitutions in the Nox2 backbone render Nox2 inactive.
Since we were not able to recover constitutive ROS generation by Nox4 with the aforementioned chimeras, new sets of chimeras were prepared, which featured a Nox2 backbone with complete substitution of all intracellular Nox2 sequences by Nox4 sequences. To permit easier detection of these chimeras using commercially available antibodies, N termini were tagged with a V5 epitope. Addition of this tag to Nox4 did not alter expression, localization, or constitutive activity of Nox4 (Fig. (Fig.5).5). On the other hand, introduction of the V5 tag into wild-type Nox2 or replacement of the first 10 amino acids of Nox2 with Nox4 residues (Nox2 00a) decreased PMA-mediated ROS generation (Fig. (Fig.5A),5A), although expression and membrane translocation were maintained (Fig. 5B and C). We also added V5 tags to the N termini of Nox4 backbone chimeras 3b and 5b. Both V5-3b and V5-5b displayed impaired ROS production upon PMA stimulation compared to the untagged proteins (Fig. (Fig.5D),5D), while expression levels, maturation, and cell surface localization remained unchanged (Fig. 5E and F). This result suggests that the inducible catalytic activity of chimeras harboring a Nox4 backbone depends on the N-terminal Nox2 sequence. We conclude that alterations of the Nox2 N terminus either interfere with stimulus-dependent oxidase assembly or hinder dynamic folding required for ROS generation.
Replacement of all or only a subset of intracellular Nox2 amino acid residues with Nox4 residues (Fig. (Fig.6A;6A; see Fig. S6 in the supplemental material) revealed that these chimeras remain nonfunctional (Fig. 6A and C) and show highly diminished maturation and cell surface expression (Fig. 6B and C). A matching substitution of all the intracellular regions in a Nox2 backbone with Nox4 sequence did not rescue Nox function (Nox2 17aI). Thus, the ability of Nox4 to produce ROS constitutively not only depends on the B loop or interactions between the B loop and FAD/NADPH sites but also involves undefined amino acids or structural elements incorporated into the transmembrane domains or in the extracellular regions.
Conflicting data regarding the intracellular localization of Nox4 exist. Nox4 has been detected on membranes of perinuclear vesicles (10, 21, 28), in focal adhesions (11), in the nucleus (15), or on the plasma membrane (35). These differences might be due to different cell types and/or the specificity of the antibodies used. Recently, we reported localization of Nox4 at the plasma membrane of H661 lung epithelial cells, which was accompanied by H2O2 generation in lieu of detectable O2− production (35). The anti-Nox4 antibody validated in this previous study recognizes an extracellular Nox4 epitope. The data presented here for transfected Cos-p22phox cells demonstrate again plasma membrane localization of Nox4, this time accompanied by electron microscopy (EM) and flow cytometry studies (Fig. (Fig.7).7). The preparation of constitutively active V5-Nox4 and the PMA-inducible Nox4 3b chimera (Fig. 7A and B) permits detailed analysis of Nox4 versus Nox2 localization in epithelial cells and allowed us to address how localization may affect the output of a particular ROS. Staining of Nox4-expressing cells with anti-Nox4 antibody or with anti-V5 antibody revealed cell surface localization of Nox4, V5-Nox4, and the Nox4 3b chimera (Fig. (Fig.7C).7C). The V5-tagged N terminus of Nox4 resides intracellularly, which is similar to published observations using tagged Nox2 (37), and can be detected only after cell permeabilization (Fig. (Fig.7C;7C; see Fig. S7 in the supplemental material). Cell surface-expressed Nox2 was stained by the monoclonal antibody 7D5, which recognizes a complex extracellular epitope, but not with an antibody directed against an epitope located in the intracellular C terminus (MAb 54.1). Nox2 expression was detected with MAb 54.1 only upon cell permeabilization (see Fig. S7 in the supplemental material). Equal expression and maturation of Nox4, Nox2, and chimeras were confirmed by immunoblotting (Fig. (Fig.7B).7B). Nox-expressing Cos-p22phox cells were also examined by electron microscopy. Cells expressing Nox4 and Nox4 3b showed immunogold staining on the surface of intact cells and in coated pits, invaginations, and vesicles located near the plasma membrane. Gold particles were also detected using cells transfected with Nox2 (Fig. (Fig.7D).7D). Specificity of the primary antibodies as well as the immunogold-conjugated secondary antibodies was confirmed by staining nontransfected cells lacking Nox or cells where primary antibodies were omitted, respectively (data not shown). In conclusion, Nox4, Nox2, and the Nox4 3b chimeras localized to the outer plasma membrane in comparable fashion.
Since Nox4 3b displayed inducible ROS generation, reminiscent of that for Nox2, we analyzed which ROS was being produced by Nox2, Nox4, and Nox4 3b. Adherent cells expressing those oxidases in combination with p47phox and p67phox proteins were probed for H2O2 generation and for superoxide production using the HVA assay and cytochrome c reduction, respectively. Nox2 generated, as expected, both reactive species upon PMA stimulation (Fig. (Fig.7A).7A). In contrast, we could not detect superoxide in cells expressing Nox4 or Nox4 3b in the presence or absence of PMA (Fig. (Fig.7A,7A, lower panel).
We hypothesized that the initial ROS output by Nox4 (i.e., H2O2) versus Nox2 (i.e., O2−, dismutated product H2O2) may contribute to divergent cellular functions despite similar cell surface localization of both enzymes. We created Cos-p22phox-based cell lines stably expressing wild-type Nox4 or inactive Nox4 P437H and compared them with the reconstituted Nox2 model cell line Cos-phox (27). H2O2 production by cells expressing wild-type Nox4 was comparable to H2O2 output generated by spontaneous superoxide dismutation of PMA-stimulated Cos-phox cells, while p22phox-expressing control cells and cells expressing dominant negative Nox4 remained catalytically inactive (Fig. (Fig.8A).8A). HVA assays performed over a prolonged time period indicated that similar levels of H2O2 production were maintained in Nox4- and Nox2-expressing cells for at least 6 h (data not shown). In contrast to that by Nox2, H2O2 generation by Nox4 is not affected by PMA treatment (21). Localization of Nox4, Nox4 P437H, and Nox2 at the cell surface of these cell lines was comparable (Fig. (Fig.8B).8B). Boyden chamber assays showed a 2-fold, PMA-independent increase in cell migration when active Nox4 was present (Fig. (Fig.8C).8C). Control cells expressing p22phox or the inactive Nox4 P437H/p22phox complex displayed comparable, PMA-independent migration rates. Cos-phox cells harboring the reconstituted Nox2 oxidase migrated considerably less when PMA was added and superoxide production was initiated.
p22phox forms a heterodimer with the “Nox” domain of Nox1 to Nox4, which is an essential requirement for NADPH oxidase maturation and activity. A point mutation in the last transmembrane domain of p22phox (p22phox Y121H) supports Nox4-mediated ROS generation, while abolishing the oxidase activity of Nox1 to Nox3 (26, 35), indicating that formation of the Nox4/p22phox complex differs from the association of Nox1 to Nox3 with p22phox. Exchange of the D loop in Nox4 or Nox2 did not affect ROS generation in cells harboring wild-type p22phox (see 12b and 12a in Fig. Fig.1).1). For comparison, the catalytically fully functional chimera Nox4 12b was coexpressed with wild-type p22phox and p22phox Y121H in p22phox-deficient H661 cells. Although overall expression levels were similar, the Nox4 D-loop substitution chimera failed to generate ROS and did not translocate to the cell surface when coexpressed with p22phox Y121H (Fig. (Fig.99 A and C, upper panel). This suggests that the Nox4 D loop is required for complex formation with the p22phox mutant and cannot be replaced with matching Nox2-derived sequence. As shown earlier, wild-type Nox2 does not mature or translocate to the cell surface in p22phox Y121H-expressing cells (Fig. 9B and C) (35). Replacing the D loop in Nox2 (Nox2 12a) with a Nox4 D loop did not rescue Nox2 maturation or localization defects (Fig. 9B and C, middle panel), implying that the Nox4 D loop is only part of the structural elements required in bridging dimerization of Nox4 with the p22phox Y121H mutant. To identify these additional structural features, Nox4 chimeras with C-terminal and loop replacement (Nox4 3b, 4b, and 5b) were tested for Nox protein maturation and subcellular localization in H661 cells expressing wild-type or mutant p22phox (Fig. 9B and C, lower panel). These experiments indicate that the C terminus, or matching substitution of the C terminus and B loop, will not affect interaction of Nox4 chimeras with p22phox Y121H. In contrast, replacement of the D loop, either alone or in combination with the C terminus, will abolish Nox4 translocation to the plasma membrane, highlighting the importance of the Nox4 D loop in the distinct folding and interaction between Nox4 and p22phox.
The signaling specificity of NADPH oxidases is most likely based on a few distinct amino acid changes and/or subtle structural differences. Even though NADPH oxidases display a similar overall structure in the “Nox” domain, featuring six transmembrane segments, coordination of heme binding histidines, and largely conserved FAD and NADPH binding sites in an extended intracellular carboxyl terminus, homology or identity in amino acid residues of different Nox family members is often well below 50%. Sequence modifications and minor changes in the lengths of loops spanning transmembrane domains, as well as extensions of the amino terminus, might be the key to regulatory and functional differences between individual Nox enzymes. As Nox2 represents the prototypic inducible Nox and Nox4 the only constitutively active Nox, these proteins were chosen for comparison. To address the importance of the cytosol-exposed Nox domains, a large-scale chimera screening approach was used, with up to 50 Nox2-Nox4 and Nox4-Nox2 chimeras. Selected chimeras, showing catalytic activity and/or proper cellular localization, were tested in more detail and are summarized in this report.
In general, ROS generation by Nox4 and Nox2 was dependent on their distinct B loops. The B loop has been recognized as an important structural element for Nox2 function (5, 13, 19). Interestingly, the previously identified amino acid residues involved in electron flow from NADPH across the membrane are conserved in Nox2 and Nox4 (arginines 73, 80, 91, and 92; leucine 94; and aspartic acid 95 in Nox2). This suggests that discrete amino acids in the B loop are required for Nox enzyme-specific catalytic activity, since Nox4 B-loop chimeras showed maturation and proper cell surface localization. In the case of Nox2 B-loop chimeras (13a and 18a), however, maturation and cell surface expression were reduced and ROS production was abolished. A scenario where the C terminus connects with the B loop during enzyme activation, thus providing specificity for electron flow from one domain to the other, has been suggested. Our data, generated from combination chimeras on a Nox4 or Nox2 backbone, do not support this view. A Nox4 chimera with a matching Nox2 C terminus and B loop showed maturation and cell surface expression similar to those of chimeras containing either the Nox2 C terminus alone or a combination of the Nox2 C terminus and matching D loop. All three constructs were catalytically active upon PMA stimulation, but ROS generation by the Nox4-based chimera containing both the Nox2 C terminus and B loop was substantially reduced. Substitution of the C terminus was sufficient to convert constitutively active Nox4 into PMA-inducible Nox4. This implies that the Nox2 C terminus contains all the essential binding sites for p47phox, p67phox, and Rac. Alternatively, a potential p47phox binding site in the Nox2 B loop can be mimicked by the Nox4 B loop, at least in the context of a Nox4 backbone (8). The analogous exchange of the Nox4 C terminus into the Nox2 backbone was catalytically inactive, and additional substitutions did not rescue the phenotype. Furthermore, a Nox2 backbone chimera substituted with all of the intracellular Nox4 sequences lost most of its ability to generate ROS. Thus, we have not been able to perform substitutions that will convert an inducible Nox2 into a constitutively active oxidase. Recently, Helmcke and coworkers described a Nox1 chimera with complete C-terminal Nox4 sequence substitution. This chimera was constitutively active when expressed in HEK293 cells and analyzed by chemiluminescence in suspension conditions (10). We have only been able to detect ROS generated by Nox1, but not by Nox4, in suspended epithelial cells (21, 35) and focused in this study only on Nox2/Nox4 chimeras.
Our data illustrate the localization of Nox4 to the plasma membrane of Cos-p22phox cells. Previously, we reported Nox4 localization on the cell surface of H661 cells and on perinuclear vesicles in HEK293 cells. We have extended our studies using flow cytometry and EM analysis of Cos-22phox cells expressing wild-type and epitope-tagged Nox4, Nox2, and mutant Nox4 3b. Wild-type and mutant Nox enzymes were located at the plasma membrane and invaginations resembling coated pits. Additional experiments indicated the intracellular localization of the N termini of both Nox2 and Nox4. Constitutive H2O2 generation and plasma membrane localization of N-terminally V5-tagged Nox4 were comparable to those of wild-type Nox4. In contrast, chimeric replacement of the Nox2 N terminus with the Nox4 sequence or N-terminal sequence additions in the form of a V5 epitope tag reduced Nox2-mediated ROS generation but did not alter localization of these oxidases. V5-tagged, inducible Nox4 backbone chimeras (Nox4 3b, 4b, and 5b) and Nox2 backbone mutants with altered N termini showed similar impairment in ROS generation. Unlike that of Nox4, the catalytic activity of Nox2 or of PMA-inducible Nox4 backbone chimeras relied on the N terminus, while Nox maturation and localization were not affected. This feature deviates from recent observations where the Nox N terminus was implicated in determining intracellular localization (10).
As comparable plasma membrane localization of Nox2, Nox4, and inducible Nox4 chimeras was detected, one would predict that these oxidases generate superoxide emitted to the extracellular milieu. Superoxide and H2O2 production by adherent cells expressing Nox2, Nox4, or Nox4 3b was compared under unstimulated and PMA-stimulated conditions. The absence of measurable superoxide in adherent Cos-p22phox cells expressing Nox4 or chimera Nox4 3b cannot be attributed to their intracellular localization followed by dismutation into diffusible H2O2. In conclusion, Nox4 releases only H2O2 to the extracellular milieu or, when localized on intracellular vesicle membranes, presumably into the lumen, a feature that has also been observed with dual oxidases (Duox) (2, 9). This particular characteristic of Nox4 seems to reside in the Nox4 backbone, extracellular loops, or a combination of both. Substitution of the C terminus of Nox4 did not result in measurable amounts of superoxide. Similarly, exchanging the N terminus of Nox2 did not alter the ability of Nox2 to produce superoxide. Although superoxide will dismutate spontaneously into H2O2, one can envision that immediate release of O2− by Nox2 versus H2O2 by Nox4 exerts a specialized, localized, and distinctive signal to cellular pathways. ROS have been previously implicated in cell motility (20, 23), and thus migration rates of Cos cells stably expressing Nox2, Nox4, or catalytically inactive Nox4 were compared. While Nox4 activity doubled migration rates, PMA-stimulated Nox2 activity inhibited cell motility. This effect was not due to constitutive Nox4 activity versus a short burst of Nox2 activity, since continuous ROS generation was maintained for the 5- to 6-h observation time for both oxidases. Therefore, the ROS output and possibly unique proximity or association of Nox4 to proteins regulating cell motility play a role in signaling specificity of Nox4 versus Nox2.
Previously, we proposed alternative complex formation of the Nox4/p22phox complex versus the Nox1 to 3/p22phox complex. Our studies provided evidence that expression of a mutant p22phox Y121H subunit discriminates between Nox1 to Nox3 and Nox4. In transient-expression studies only the catalytic activity of Nox4 could be maintained on a p22phox Y121H background (35). Mice harboring mutant CYBA (p22phox Y121H) display deficiencies in Nox2 and Nox3 activity, although their phenotype with respect to Nox1 and Nox4 function remains unknown (26). Preparation of Nox4-Nox2 chimeras permitted us to analyze requirements for localization and interaction of these enzymes with the p22phox Y121H mutant. Coexpression of almost all of the Nox chimeras with wild-type p22phox caused translocation to the cell surface, as analyzed by flow cytometry, immunofluorescence, and electron microscopy. The expression of selected catalytically active chimeras with p22phox Y121H in p22phox-deficient H661 cells revealed the unique role of the Nox4 D loop. As soon as this loop was replaced with the Nox2 D loop, complex formation of the Nox4 chimera with mutant p22phox did not occur. Nox4 D-loop chimeras paired with p22phox Y121H remained in the endoplasmic reticulum and lost their ability to generate H2O2. The same chimeras localized properly when coexpressed with wild-type p22phox and showed catalytic activity. These observations are consistent with our earlier hypothesis of a distinct structural arrangement of the Nox4/p22phox complex, which may have regulatory and/or functional relevance and may provide a suitable target for intervention.
Constitutive activity of Nox4 remains a unique feature of this particular enzyme among the Nox family of oxidases. We were unable to construct a Nox2 chimera that possessed this ability, whereas turning Nox4 into an inducible oxidase was accomplished by substituting the C terminus. Nox4 seems to be locked into an active configuration while at the same time being able to tolerate mutations in the p22phox subunit that are not tolerated by other Nox enzymes (35). Whether this is accomplished with the aid of a yet-to-be-identified cofactor or is an intrinsic attribute of the Nox4 sequence remains elusive. Evaluation of chimeras has added to our insight; however, a definitive understanding of the basis for functional differences between Nox enzymes may come only with solving the structure of the Nox/p22phox complex.
We thank Mary Dinauer for providing Cos-p22phox and Cos-phox cells.
This work was supported by Public Health Service grants AI024838 and AI077042 (to U.G.K. and J.S.F.) and by Centers for Disease Control grant CI000095 (to U.G.K.).
Published ahead of print on 7 December 2009.
†Supplemental material for this article may be found at http://mcb.asm.org/.