Point Mutations in the Proline-rich Region of p22 phox Are Dominant Inhibitors of Nox1- and Nox2-dependent Reactive Oxygen Generation*

The integral membrane protein p22 phox is an indispen-sable component of the superoxide-generating phagocyte NADPH oxidase, whose catalytic core is the mem-brane-associated gp91 phox (also known as Nox2). p22 phox associates with gp91 phox and, through its proline-rich C terminus, provides a binding site for the tandem Src homology 3 domains of the activating subunit p47 phox . Whereas p22 phox is expressed ubiquitously, its participa-tion in regulating the activity of other Nox enzymes is less clear. This study investigates the requirement of p22 phox for Nox enzyme activity and explores the role of its proline-rich region (PRR) for regulating activity. Coexpression of specific Nox catalytic subunits (Nox1, Nox2, Nox3, Nox4, or Nox5) along with their corresponding regulatory subunits (NOXO1/NOXA1 for Nox1; p47 phox / p67 phox /Rac for Nox2; NOXO1 for Nox3; no subunits for Nox4 or Nox5) resulted in marked production of reactive oxygen. Small interfering RNAs decreased endogenous p22 phox expression and inhibited reactive oxygen generation from Nox1, Nox2, Nox3, and Nox4 but not Nox5. Truncated forms of p22 phox that disrupted the PRR-inhibited reactive oxygen generation

The integral membrane protein p22 phox is an indispensable component of the superoxide-generating phagocyte NADPH oxidase, whose catalytic core is the membrane-associated gp91 phox (also known as Nox2). p22 phox associates with gp91 phox and, through its proline-rich C terminus, provides a binding site for the tandem Src homology 3 domains of the activating subunit p47 phox . Whereas p22 phox is expressed ubiquitously, its participation in regulating the activity of other Nox enzymes is less clear. This study investigates the requirement of p22 phox for Nox enzyme activity and explores the role of its proline-rich region (PRR) for regulating activity. Coexpression of specific Nox catalytic subunits (Nox1, Nox2, Nox3, Nox4, or Nox5) along with their corresponding regulatory subunits (NOXO1/NOXA1 for Nox1; p47 phox / p67 phox /Rac for Nox2; NOXO1 for Nox3; no subunits for Nox4 or Nox5) resulted in marked production of reactive oxygen. Small interfering RNAs decreased endogenous p22 phox expression and inhibited reactive oxygen generation from Nox1, Nox2, Nox3, and Nox4 but not Nox5. Truncated forms of p22 phox that disrupted the PRR-inhibited reactive oxygen generation from Nox1, Nox2, and Nox3 but not from Nox4 and Nox5. Similarly, p22 phox (P156Q), a mutation that disrupts Src homology 3 binding by the PRR, potently inhibited reactive oxygen production from Nox1 and Nox2 but not from Nox4 and Nox5. Expression of p22 phox (P156Q) inhibited NOXO1-stimulated Nox3 activity, but co-expression of NOXA1 overcame the inhibitory effect. The P157Q and P160Q mutations of p22 phox showed selective inhibition of Nox2/p47 phox /p67 phox , and selectivity was specific for the organizing subunit (p47 phox or NOXO1) rather than the Nox catalytic subunit. These studies stress the importance of p22 phox for the function of Nox1, Nox2, Nox3, and Nox4, and emphasize the key role of the PRR for regulating Nox proteins whose activity is dependent upon p47 phox or NOXO1.
Evidence over the last decades shows that reactive oxygen species (ROS), 1 including superoxide anion, hydrogen peroxide, and their reaction products, play crucial roles in diverse physiological processes including growth (1), regulation of vasodilatation (2), hormone synthesis (3), sperm capacitation (4), fertilization (5), oxygen sensing (6), bone absorption (7), and the innate immune response (8 -10). However, until the last several years, the enzymes responsible for generating these ROS were unknown, except for the phagocyte NADPH oxidase, a superoxide-generating enzyme whose catalytic moiety, flavocytochrome b 558 , is composed of a large catalytic subunit (gp91 phox or Nox2) and a smaller subunit (p22 phox ). Beginning in 1999, novel homologues of gp91 phox have been reported (11)(12)(13)(14)(15)(16)(17), and the Nox family now consists of seven members in humans, including five Nox proteins (the NAD(P)H oxidases Nox1, Nox2 (gp91 phox ), Nox3, Nox4, and Nox5) and two dual oxidases (Duox1 and Duox2). The latter contain a peroxidase homology domain in addition to the NADPH-oxidase domain. These homologues share a core structure consisting of six predicted transmembrane ␣-helixes, which bind two heme groups, plus a C-terminal flavoprotein domain containing the conserved binding sites for NADPH and FAD. In addition to the core structure of Nox, Nox5 contains an N-terminal extension containing calcium binding EF-hand motifs (for a review, see Refs. 18 and 19).
The molecular mechanisms underlying the activation of Nox2 have been studied extensively. Nox2 is complexed with a small subunit p22 phox to form flavocytochrome b 558 (9). Nox2 activation requires stimulus-induced assembly of the cytochrome b 558 with the cytosolic regulatory proteins p47 phox , p67 phox , p40 phox , and the small GTPase Rac (for a review, see Refs. 9,20,and 21). In the resting state, p47 phox and p67 phox are located in the cytosol. Upon cell activation (e.g. exposure to opsonized bacteria, chemoattractants, or a protein kinase C activator, phorbol 11-myristate 12-acetate (PMA)), p47 phox becomes hyperphosphorylated in the autoinhibitory region, freeing its tandem SH3 domains to bind to a C-terminally located proline-rich region (PRR) of p22 phox and resulting in translocation from cytosol to the membrane-associated flavocytochrome b 558 (22,23). Following GTP binding, the small GTPase Rac translocates to the membrane (24). Both Rac and p47 phox provide binding sites for p67 phox , resulting in its association with Nox2 and activation of electron transfer from NADPH to FAD (25)(26)(27)(28). Another cytosolic subunit p40 phox was identified as a p67 phox -binding protein and, while not essential, facilitates the stimulus-induced translocation of p47 phox and p67 phox to the membrane (29), resulting in increased Nox2 activity.
Recently, homologues of p47 phox and p67 phox were reported by several groups and have been named Nox organizer 1 (NOXO1) and Nox activator 1 (NOXA1), respectively (30 -33). Several biochemical studies indicate that like gp91 phox , Nox1 activity is also governed by regulatory subunits, in this case NOXO1 and NOXA1, since Nox1-dependent reactive oxygen generation required co-expression of both NOXO1 and NOXA1 (30 -33). In addition, NOXO1 and Nox1 are coordinately expressed in the gastrointestinal tract, consistent with NOXO1 as a physiological regulatory subunit for Nox1 (31,34,35). Other Nox proteins show both subunit-dependent and subunit-independent regulation. Human Nox3 requires regulatory subunits, but its regulation is less discriminating than that of Nox1. Nox3 can be activated by NOXO1 alone or by the combination of p47 phox and p67 phox in a PMA-dependent manner (36). Mouse Nox3 shows similar regulation, except that it requires NOXA1 in combination with NOXO1 (37). On the other hand, Nox4 is constitutively active (14), and it is not activated by these subunits (38). Recent studies have demonstrated that the insulin receptor-or toll-like receptor 4-signaling pathway increases Nox4-derived ROS generation (39,40), but the mechanism of this regulation is not known. Nox5-dependent ROS generation is activated by calcium, consistent with its N-terminal calcium-binding domain (16,41). Similarly, a preparation of Duox1 and Duox2 partially purified from thyroid is also activated by calcium (42). Roles for additional subunits for calcium-dependent Nox proteins have not been reported.
Despite an increasing number of studies documenting the roles of regulatory subunits in Nox regulation, the role of p22 phox in Nox family members other than Nox2 remains relatively unexplored. Activity studies have been hampered by the near universal expression of p22 phox in experimental cell systems. We thus used siRNA in the present studies to inhibit expression of endogenous p22 phox and to determine the requirement of Nox1, Nox2, Nox3, Nox4, and Nox5 for this subunit. We find that Nox1, Nox2, Nox3, and Nox4 require p22 phox , but the calcium-dependent enzyme Nox5 does not. In addition, for each of these systems, we have investigated the importance of the PRR of p22 phox in regulating ROS generation. Unexpectedly, we have found that forms of p22 phox mutated in this region function as dominant inhibitors for Nox1, Nox2, and Nox3 activities but not for Nox4 or Nox5. In addition, specific mutations can discriminate between p47 phox -dependent regulation and NOXO1-dependent regulation. The present studies implicate both regulatory subunit binding-dependent and -independent functions of p22 phox in regulating ROS production by Nox enzymes.

EXPERIMENTAL PROCEDURES
Cells and Reagents-Human embryonic kidney (HEK) 293 cells were grown in Dulbecco's modified Eagle's medium with 10% fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin (36). Chicken anti-human NOXO1 and NOXA1 antibodies were previously described (36). Monoclonal antibodies 44.1 and NS2 against human p22 phox were kindly provided by Dr. A. Jesaitis (Montana State University, Bozeman, MT). Polyclonal antibody R3179 was prepared against human neutrophil p22 phox and was shown previously to be specific for p22 phox (43). Monoclonal antibody E39.1, raised against a synthetic peptide representing amino acid residues 235-248 of human Nox1 was kindly provided by Jackie Papkoff (DiaDexus, Inc.). Polyclonal antibody N4-8478 was prepared by immunizing rabbit with expressed protein corresponding to amino acids 322-428 of human Nox4. For immunostaining of tissues, Alexa Fluor 488-conjugated goat anti-mouse or anti-rabbit IgG and Alexa Fluor 555-conjugated goat anti-mouse or anti-rabbit IgG were purchased from Molecular Probes, Inc. (Eugene, OR). Anti-green fluorescent protein (GFP) polyclonal, anti-␤-actin, and anti-Myc tag monoclonal antibodies were purchased from BD Biosciences, Cell Signaling, and Sigma, respectively. The cDNA clones encoding human Nox1, Nox2, Nox3, Nox4, p22 phox , p47 phox , p67 phox , NOXO1, NOXA1, and V12Rac1 (the constitutively active substitution of human Rac1), were previously described (33,36). Human colon tissue was obtained from Emory's Human Tissue Procurement and Banking Center from surgeries performed at Emory and generally represented normal tissue obtained along with resected tumors. The procurement of tissues was approved by the Institutional Review Board of Emory University, and written informed consent was obtained. Tissues were embedded in Cryomatrix. Five-micrometer sections were obtained at Ϫ22°C using a cryostat (Thermo Shandon Cryotome) and placed on glass microscope slides. Tissue slides of the human kidney cortex were purchased from BioChain. Nox5 cDNA was amplified by PCR of human spleen cDNA library (BD Biosciences) using a set of primers (5Ј-GCAT-GGATCCACCATGAACACATCTGGAGACCCA-3Ј and 5Ј-GCATA-AGCTTCTAGAAATTCTCTTGGAAAAATCTGAAGCCGAAC-3Ј), containing BamHI (italic) and HindIII (underline) sites. After digestion by each enzyme, the cDNA fragment was subcloned into the pCMV-tag 5A vector (Stratagene). The construct was sequenced in both strands using ABI model 377 sequencer and was identified as Nox5␣ (GenBank TM accession number AF317889).
To generate cDNA that encodes N-terminal GFP-fused protein of Nox1, human Nox1 cDNA was amplified using sense primer (AAAAG-AGCTCGGAAACTGGGTGGTTAACCACTGGT) and antisense primer (AAAAGTCGACTCAGAAGTTTTCTTTGTTGAAGTAGAATTG). This sense primer has been designed to delete the initial methionine of native Nox1. The PCR product contained SacI (italic type) and SalI (underlined) sites. For C-terminally Myc-tagged Nox4 protein, human Nox4 cDNA was amplified using sense primer (AAAAGCGGCCGCGG-TACCATGGCTGTGTCCTGGAGGA) and antisense primer (AAAAGT-CGACGCTGAAAGACTCTTTATTGTATTCAAATCTTGTCCCA) that contain NotI (italic type) and SalI (underlined) sites, respectively. After p22 phox Mutants as Nox Inhibitor digestion by each enzyme, the fragments including Nox1 and Nox4 cDNAs were subcloned into the pEGFP-C3 vector (BD Biosciences) and the pCMV-tag 5A vector, respectively. All constructs were verified by sequencing.
Transient Transfections-HEK293 cells were grown for 24 h in 6-well plates and allowed to reach to 50% confluence in 2 ml of culture medium. Cells were transfected with vectors carrying Nox1 through Nox5, p67 phox , p47 phox , NOXO1, NOXA1, full-length p22 phox , and mutated or truncated p22 phox cDNAs alone or in the indicated combinations using the tranfection reagent FuGENE 6 (Roche Applied Science) according to the manufacturer's instructions. After 48 h, or 72 h when co-transfected with shRNA vectors, cells were washed twice with cold Hanks' balanced salt solution and were removed from the well, after which cells were harvested by centrifuging at 500 ϫ g for 5 min and resuspending in Hanks' balanced salt solution.
Measurement of ROS-ROS was measured using luminol luminescence as previously described (33,36). Cells (4 ϫ 10 4) in Hanks' balanced salt solution with calcium and magnesium were mixed with 200 M luminol plus 0.32 units of horseradish peroxidase in a 200-l total volume in each well of a 96-well plate. Luminescence was quantified using a FluoStar TM luminometer (BMG Labtech), recording data every minute for 2 h. Luminescence increased during early time points, peaking at 30 min, and this value was reported herein. For the measurement of Nox2 activation, cells were preincubated with 200 nM PMA for 5 min at 37°C. For Nox5 activation, cells were stimulated with 1 M ionomycin or the same amount of vehicle at 10 min after beginning data collection and immediately achieved the maximum rate, which was reported at the 12-min time point.
Immunostaining-Slides were fixed with 4% paraformaldehyde in phosphate-buffered saline at room temperature for 20 min. After washing with phosphate-buffered saline, tissue sections were permeabilized using 0.1% Triton X-100 at 4°C for 5 min and were blocked with 1% goat serum at room temperature for 1 h. Slides were incubated with primary antibodies (monoclonal anti-Nox1 antibody E39.1 and rabbit polyclonal anti-p22 phox antibody R3179 for colon or rabbit polyclonal anti-Nox4 antibody and monoclonal anti-p22 phox antibody 44.1 for kidney) for 2 h. After washing for 30 min, slides were incubated with the appropriate secondary antibodies (Alexa Fluor 488 goat anti-mouse IgG and Alexa Fluor 555 goat anti-rabbit IgG for the colon or Alexa Fluor 488 goat anti-rabbit IgG and Alexa Fluor 555 goat anti-mouse IgG for the kidney) for 1 h. Nuclei were stained with ToPro3 (Molecular Probes) for 20 min at room temperature, and slides were washed for 15 min. Tissues were mounted in Vectashield (Vector Laboratories), and fluorescence images were scanned with a confocal laser-scanning microscope (LSM 510 Meta; Zeiss). Controls were incubated with secondary antibodies only.

Effects of p22 phox Expression on Activation of Nox Enzymes-
Cellular ROS-generating systems that utilize Nox enzymes and regulatory proteins co-expressed in HEK293 cells have been reported previously (30 -33, 36). For example, co-expression of Nox2, p67 phox , p47 phox , and V12Rac1 resulted in high rates of PMA-dependent reactive oxygen generation. However, the reconstituted systems do not require ectopic expression of p22 phox . Fig. 1A demonstrates that untransfected HEK293 cells express p22 phox message, suggesting that endogenous expression of p22 phox might support the activation of Nox enzymes without the need for additional overexpressed protein. In addition, we investigated a number of other cell types to try to find a cell line lacking p22 phox that was capable of reconstituting Nox-dependent activities, and all but one contained endogenous p22 phox . We confirmed that human pulmonary carcinoma H292 cells (ATCC), which have been reported to lack endogenous p22 phox , 2 fail to express p22 phox , but for unknown reasons, H292 cells transfected with an expression plasmid encoding this subunit also failed to express ectopic p22 phox (data not shown). In these cells, either with or without cotransfection with p22 phox , only Nox5 was active, and no activity was seen with Nox1, Nox2, Nox3, and Nox4 when appropriate subunits were co-expressed (supplemental Fig.  S1). Therefore, Nox5 does not require p22 phox for activity, but the H292 cell type did not appear to be suitable for studies of Nox1-Nox4.

p22 phox Mutants as Nox Inhibitor
Nox1-Nox3, Nox4-dependent ROS-generating activity does not require regulatory subunits, but p22 phox expression was still needed for Nox4 activity (Fig. 2D). As shown in Fig. S2, scrambled shRNA-p22-N and -C as well as shRNA-p22-M did not change the rates of Nox3-and Nox4-dependent ROS generation.
Banfi et al. (41) showed that Nox5 activity is stimulated by calcium ion/ionomycin. In the present studies, low level basal Nox5-dependent ROS generation was seen without the addition of calcium ionophore, and this activity was markedly increased by ionomycin (Fig. 2E, hatched bars). In contrast to the other Nox enzymes, the silencing of p22 phox did not affect either the spontaneous or the ionomycin-triggered ROS generation by Nox5 (Fig. 2E).
Nox2 (gp91 phox ) protein has previously been shown to be stabilized by p22 phox , since in a form of chronic granulomatous disease lacking p22 phox , Nox2 expression is also impaired (45). Therefore, the decreased activity seen above when expression of p22 phox is impaired might be due to decreased expression of Nox subunits. To investigate whether silencing of p22 phox affects expression of Nox1 protein, we examined the expression of Nterminal GFP-fused Nox1 vector. A GFP fusion protein was uti-lized, since examination of several antibodies against Nox1 failed to reveal Nox1 expression by Western blotting. The N-terminal fusion was selected, since fusion with GFP at the N terminus resulted in full activity (supplemental Fig. S4), whereas GFP fused at the C terminus resulted in complete absence of Nox1 activity (33). When GFP-Nox1 was co-expressed with NOXO1 and NOXA1 in HEK293 cells, shRNA-p22-N and -C vectors significantly suppressed ROS generation by GFP-Nox1 (Fig. S4). The knockdown of p22 phox resulted in an ϳ50% decrease in GFP-Nox1 expression (Fig. 3A). However, for shRNA-p22-N, the decreased expression of Nox1 was insufficient to account for the nearly complete inhibition of ROS generation by this vector, suggesting that Nox1-dependent ROS generation utilizes p22 phox not only for stabilization but also for other functions. Nox4 fused at the C terminus with the Myc epitope retained ϳ15% of the ROS-generating activity compared with untagged Nox4 protein (supplemental Fig. S4). Silencing of p22 phox completely suppressed ROS generation by Myc-tagged Nox4 (Fig. S4). However, under these conditions, the expression of Nox4 protein was not significantly affected by silencing of p22 phox (Fig. 3B). Hence, Nox4 also requires p22 phox for functions other than stabilization of the Nox4 catalytic subunit.
Dominant Inhibitor Effects of the C-terminal Truncations of p22 phox on Activation of Nox Enzymes-For activation of the phagocyte NADPH oxidase containing Nox2, the p22 phox subunit plays two critical roles: 1) formation of a stabilizing complex with Nox2 and 2) mediation of the assembly between flavocytochrome b 558 and cytosolic regulatory subunits (9,20,21). Based on this idea, we hypothesized that a mutant of p22 phox lacking an intact PRR would still associate with the Nox subunit but would fail to mediate interactions with regulatory subunits. As such, these mutant forms of p22 phox could compete with wild type p22 phox and would therefore function as dominant inhibitors for Nox proteins that require an organizer protein subunit such as p47 phox or NOXO1. Binding of p47 phox requires direct interaction between the PRR of p22 phox (amino acids 151-160) and the tandem SH3 domains of p47 phox (28). Consistent with the hypothesis, the C-terminal truncations p22 phox (residues 1-149) and p22 phox (residues 1-155) that disrupt the PRR inhibited ROS generation by Nox1, Nox2, and Nox3 (Fig. 4, A, B, and C, respectively). These inhibitory effects were not observed using either full-length p22 phox or p22 phox (residues 1-172) in which the PRR is intact (Figs. 4, A-C). As seen in Fig. 2, Nox4 activity requires the p22 phox subunit, but
Dominant Inhibitor Effects of the Point-mutated p22 phox (P156Q) on the Activity of Nox Enzymes-The missense mutation p22 phox (P156Q) is causally linked to chronic granulomatous disease, a condition that is characterized by impaired ability of the neurophil NADPH oxidase (Nox2) to produced microbicidal ROS (46). Previous biochemical studies demonstrated that p22 phox (P156Q) fails to bind to p47 phox (28,46,47). To examine whether mutation of the PRR would cause p22 phox to function as a dominant inhibitor, experiments similar those in Fig. 4 were performed using p22 phox (P156Q). As with truncated forms, co-expression of p22 phox (P156Q) markedly suppressed Nox1-dependent ROS generation in a dose-dependent manner (Fig. 5A), with greater than 95% inhibition at a vector concentration of 0.25 g/ml. Similarly, when this amount of p22 phox (P156Q) was expressed, Nox2 activity was also completely inhibited (Fig. 5B). Like Nox1 and Nox2, p22 phox (P156Q) abrogated NOXO1-dependent Nox3 activation (Fig.  5C). Consistent with the results shown in Fig. 4, p22 phox (P156Q) did not influence Nox4 and Nox5 activities (Figs. 5, D  and E, respectively).

p22 phox Mutants as Nox Inhibitor
Effects of Point Mutation p22 phox (P156Q) on Nox3 Activities Regulated by Activator Subunits--Unlike Nox1 and Nox2, Nox3 activity can be modestly activated by the "activator" subunit NOXA1 or p67 phox by itself in the absence of an organizer subunit (36). Unlike NOXO1-dependent activation, NOXA1-dependent Nox3 activity was not affected by the p22 phox mutant (Fig. 6A). In the presence of NOXO1 plus NOXA1, activity was significantly higher, about the same as the activity supported by NOXO1 alone. Remarkably, co-expression of NOXA1 along with NOXO1 completely rescued the dominant inhibitor effect of p22 phox (P156Q) toward Nox3 activity (Fig. 6B). Similarly, Nox3 activity regulated by p67 phox alone or in combination with p67 phox was not affected by p22 phox (P156Q) (Fig. 6, C and D). The explanation for this phenomenon is not entirely clear and will require additional investigation. Nevertheless, results in-dicate that the activator protein (NOXA1 or p67 phox ), when complexed to an organizer protein, bypasses the requirement for interaction with the PRR of p22 phox , perhaps by allowing the activator protein to bind directly to the Nox3 protein.

p22 phox Mutants as Nox Inhibitor
Nox2, two of these point mutations showed different specificity compared with that for Nox1 inhibition; p22 phox (P157Q) inhibited Nox2 activity much more effectively than Nox1 activity, and p22 phox (P160Q) functioned as a moderately strong inhibitor of Nox2 but had little effect on Nox1 (Fig. 7B).
To examine whether the differences in specificity were dictated by the catalytic subunits or by the co-expressed organizer proteins (p47 phox or NOXO1), we tested heterologous combinations in which NOXO1 and p47 phox were substituted for one another in the Nox1 and Nox2 systems (i.e. Nox1 (Nox1, p47 phox , and NOXA1) and Nox2 (Nox2, NOXO1, p67 phox , and V12Rac1)). Inspection of Fig. 7, C and D, reveals that the inhibitor specificity of different p22 phox mutants was independent of the Nox subunit and was dictated solely by which organizer subunit was expressed in the assay cells.
Effect of Point Mutation and Truncation of p22 phox on Nox1 Expression-To test whether any of the inhibitory effect of p22 phox mutants might be attributed to expression of Nox1 protein, we examined the expression of N-terminally GFPfused Nox1. As with the native Nox1, p22 phox truncations (1-149 or 1-155) and mutations (P156Q, P157Q, or P159Q) also markedly inhibited the activity of the GFP-Nox1 protein (Fig.  8). Protein expression levels of GFP-Nox1 and subunits (NOXO1 and NOXA1) were not affected by co-expression of either wild type or mutant forms of p22 phox (Fig. 8). Thus, inhibitory effects of these subunits are not due to altered expression of Nox subunits.
Colocalization of p22 phox and Nox1 or Nox4 in Human Tissues-According to previous reports, human Nox1 and Nox4 mRNAs are predominantly expressed in the colon and kidney, respectively (11,15). To investigate whether p22 phox colocalized with endogenous Nox1 and Nox4 in vivo, we performed immunofluorescence analysis of the human colon epithelium and the renal cortex. Nox1 was predominantly expressed on the apical side of human colon epithelial cells. p22 phox was expressed somewhat more broadly in the epithelium but was highly expressed with the same subcellular location as Nox1, colocalizing at the surface of the epithelial cells (Fig. 9, A and B). Strong expression of Nox4 and p22 phox was observed in the epithelial cells of kidney tubules (Fig. 9, D and E), where colocalization was seen in membranes and submembranous vesicle-like structures. Thus, coexpression results are consistent with a role for p22 phox as a physiological partner of Nox1 and Nox4.

DISCUSSION
Silencing of endogenous p22 phox expression demonstrated that p22 phox is required for the activities of Nox1-Nox4 (Fig. 2) but not the calcium-dependent Nox5. This is consistent with the idea, first developed from co-isolation of Nox2 and p22 phox , that p22 phox forms a functional complex with some Nox catalytic subunits and with recent biochemical data that used fluorescence resonance energy transfer and co-immunoprecipitation to demonstrate that p22 phox interacts with GFP fusion proteins of Nox1 and Nox4 (38). Recently, Ueno et al. (48) also demonstrated that the Myc-tagged Nox3 binds to p22 phox using co-immunoprecipitation analysis.
As shown herein, human Nox5 activity was calcium-stimulated but did not require p22 phox . Nox5 contains a unique N-terminal extension including four EF-hand motifs and a single proline/arginine-rich region that is predicted to be exposed on the cytosolic side of the membrane C-terminal to the calcium-binding region (13,16). It has been suggested that this N-terminal extension of Nox5 might function in a manner similar to p22 phox. Nevertheless, it remains unknown whether Nox5 requires another endogenously expressed regulatory subunit(s). We have identified the D. melanogaster Nox5 homologue gene DmNox (CG3896 on FlyBase; see Refs. 49 and 50), and several Nox5 homologues (termed the Rboh proteins) have been reported in plants including A. thaliana (51) and tobacco (52). According to the predicted amino acid sequences, DmNox and Rboh possess conserved NADPH-oxidase domains and Nterminal extension containing EF-hand motifs. On the other hand, an extensive BLAST genome search has failed to identify a p22 phox homologue in fly or plants. To help to interpret the relationships among human Nox family and the requirement of p22 phox for ROS-producing activity, a phylogenic tree of human Nox1-Nox5, DmNox, and plant Nox (RbohA) is shown (Fig. 10). The requirement of p22 phox and for the PRR is shown to the right of the dendrogram. According to the dendrogram, human Nox5 is closer to fly Nox (DmNox) and plant Nox (Rboh) than A-E, Nox1 or Nox2 (0.1 g/ml each) was co-expressed in HEK293 cells together with 0.1 g/ml NOXO1, NOXA1, p47 phox , and p67 phox or 0.25 g/ml of V12Rac1 in combinations as indicated. At the same time, 0.25 g/ml pcDNA3 empty vector (mock), p22 phox , p22 phox (P156Q), p22 phox (P157Q), p22 phox (P159Q), or p22 phox (P160Q) was co-expressed as indicated. Nox4 was also co-expressed in HEK293 cells along with 0.25 g/ml pcDNA3 empty vector (mock), p22 phox , p22 phox (P156Q), p22 phox (P157Q), p22 phox (P159Q), or p22 phox (P160Q). For Nox2 activation in the presence of p47 phox or NOXO1 or for Nox1 activity in the presence of p47 phox , PMA was added and preincubated for 5 min. ROS generation was measured by luminol luminescence and indicated by RLU/10 4 cells, as described under "Experimental Procedures." Western blots represented under each bar indicate protein expression of p22 phox (monoclonal antibody 44.1) and the reference ␤-actin. Values are means Ϯ S.D. (n ϭ 6). Similar results were obtained in three separate experiments. p22 phox Mutants as Nox Inhibitor 31866 other human Nox proteins. Thus, the present studies provide a rationale for why a p22 phox homologue is not present in these species.
Earlier studies have noted partial activation of Nox1 and Nox4 by ectopic expression of p22 phox . Takeya et al. showed that co-expression of p22 phox together with Nox1 and its other regulatory subunits (Nox1, NOXO1, and NOXA1) stimulates Nox1-dependent superoxide generation (32), but it was not clear whether p22 phox was absolutely required for activity nor whether its effect was due to stabilization of Nox1 versus regulation of its activity. Similarly, co-transfection of rat p22 phox was reported to cause a ϳ3-fold increase in ROS generation in HEK293-expressing rat Nox4-transfected cells compared with Nox4 alone (38). However, in our experiments expressing human proteins, excess expression of wild type p22 phox did not further increase human Nox enzyme activities (Fig. 4), possibly due to different sublines of these cells in use in different laboratories and/or the use of rodent versus human proteins. The failure of exogenous p22 phox expression to affect Nox-dependent activities indicates that in the HEK293 cells used in our laboratory, there is sufficient expression of endogenous p22 phox to fully support the activities of these Nox enzymes. The use of shRNA to suppress p22 phox expression demonstrates for the first time a requirement of Nox3 for p22 phox and the independence of the calcium-regulated Nox5 of p22 phox . In addition, these approaches demonstrate that Nox4-and Nox1-dependent activities are absolutely dependent upon p22 phox , since there is little or no activity in the absence of this subunit.
The requirement for p22 phox may arise either from stabilization of the catalytic Nox subunit by the smaller subunit or from a functional role of p22 phox related to catalysis or recruitment of regulatory subunits. The former phenomenon has been well documented for Nox2, since in variants of chronic granulomatous disease, the absence of either the Nox2 or the p22 phox subunit results in the absence of the other subunit, presumably due to protein degradation of the unpaired subunit (44). As seen in Fig. 3, Nox1 protein levels were decreased by silencing of p22 phox , indicating that Nox1 is also stabilized to some extent by p22 phox protein. However, significant Nox1 expression still occurs in the absence of p22 phox , and under these conditions, Nox1-dependent activity is nearly completely absent, pointing to additional roles of p22 phox in Nox1-dependent ROS generation. Regarding the latter, the role of the PRR of p22 phox in the recruitment of p47 phox via its bis-SH3 domain has been extensively studied, including by structural methods, and the use of forms of p22 phox mutated in the PRR has allowed us in the present studies to demonstrate that for Nox1, the PRR is essential for the activating function of p22 phox . In contrast to Nox1 and Nox2, Nox4 expression was not affected when p22 phox expression was blocked (Fig. 3), indicating that the effect of p22 phox on Nox4 activity was not due to subunit stabilization. Thus, p22 phox is playing some other regulatory role to maintain the activity of Nox4. Importantly, the present studies show that the PRR of p22 phox does not participate in this regulation of Nox4. This is consistent with the absence of any requirement for NOXO1, NOXA1, p47 phox , or p67 phox for Nox4-dependent ROS generation.
Herein, we demonstrate that p22 phox (P156Q) is a strong inhibitor of both Nox1-and Nox2-dependent activities. In vitro binding assays using synthetic peptides corresponding to the PRR have demonstrated that wild type but not the peptide corresponding to the P156Q mutation binds to tandem SH3 domains of NOXO1 (32) like p47 phox . Recently, Groemping et al. (53) reported the x-ray crystal structure of the complex between p22 phox and p47 phox . According to details of the contacts, these interactions involve van der Waals interactions between Pro-156 in the PRR of p22 phox and Trp-193 in the N-terminal SH3 domain of p47 phox . Similarly, the residue Pro-157 of p22 phox individually interacts with the Tyr-167, Pro-206, and Phe-209 in the N-terminal SH3 region, whereas Pro-160 of p22 phox is stacked against the Phe-209 of p47 phox . Interestingly, as shown in Fig. 7, p22 phox (P157Q) and p22 phox (P160Q) inhibited p47 phox -supported ROS generation more effectively than that supported by NOXO1. These results are consistent with structural information described above, because several residues of p47 phox that are responsible for the interaction with p22 phox are different in NOXO1. Specifically, Phe-209 of p47 phox , which interacts with Pro-157 and Pro-160 of p22 phox , is replaced by tyrosine in NOXO1. Similarly, Tyr-167 of p47 phox , the residue that interacts with Pro-157 of p22 phox , is changed to phenylalanine in NOXO1. This seems likely to explain the different inhibitory specificities when Pro-157 or Pro-160 in p22 phox is mutated.
Mutants of p22 phox may find utility as tools for cell biology and possibly as therapeutic agents. Inappropriate activation of Nox2 is associated with inflammatory conditions including shock lung, arthritis, and damage to endothelium (54,55), whereas overexpression of Nox1 may be implicated in cancer, atherosclerosis, and hypertension (56,57). Nox3 is implicated FIG. 8. Effects of p22 phox mutants on ROS generation and protein expression of GFP-Nox1. 0.2 g/ml GFP-tagged Nox1 (GFP-Nox1) or EGFP-C3 empty vector (GFP) was co-expressed in HEK293 cells along with NOXO1 and NOXA1 (each 0.2 g/ml). In parallel, 0.25 g/ml pcDNA3 empty vector (mock), p22 phox , p22 phox (1-149), p22 phox (1-155), p22 phox (P156Q), p22 phox (P157Q), p22 phox (P159Q), or p22 phox (P160Q) was co-expressed in cells as indicated. ROS generation was measured by luminol luminescence and indicated by RLU/10 4 cells, as described under "Experimental Procedures." Western blots represented under each bar indicate protein expression levels of GFP-Nox1, GFP, p22 phox (monoclonal antibody NS-2), NOXO1, NOXA1, and the reference ␤-actin using corresponding antibodies. Positions for marker proteins are indicated in kDa. Values are mean Ϯ S.D. (n ϭ 6). Similar results were obtained in three separate experiments. p22 phox Mutants as Nox Inhibitor in otolith formation in the inner ear and hence plays a role in balance perception (58), whereas Nox4 has been shown to play a role in the insulin response (40). Thus, it is potentially useful to have therapeutic agents that can selectively inhibit one or a small number of Nox proteins with little or no effect on other Nox proteins. Currently, there are few if any selective inhibitors of Nox enzymes. Diphenylene iodonium has been used in a variety of Nox studies but shows broad specificity for most or all Nox proteins. Unfortunately, this compound also inhibits a fairly broad range of flavoprotein dehydrogenases and is therefore not very useful for cell biology studies or as a therapeutic agent. A cell-internalized peptide corresponding to an intracellular loop of the transmembrane region of Nox2 or adenovirus vector encoding antisense Nox2 partially inhibits angiotensin II-induced ROS generation in mouse vascular smooth muscle and the respiratory burst in human neutrophils (59,60). Recently, fungal gliotoxin from Aspergillus fumigatus has been reported as a natural Nox2 inhibitor. Its mechanism of inhibition remains unclear, although it is known to abrogate phosphorylation of p47 phox , preventing formation of an assembled, active Nox2 complex (61). In the present study, we demon-strated that the p22 phox PRR mutants did not affect Nox4 activity in contrast to their potent effect on Nox1-and Nox2dependent activities. This was despite the clear requirement of Nox4 ROS generation for p22 phox . In addition, two of the mutants showed greater efficacy against p47 phox -dependent rather than NOXO1-dependent processes and are therefore expected to have greatest efficacy in Nox2-dependent ROS generation as is seen in phagocytes and selected other cell types including lung endothelium (54). Hence, if appropriate methods are developed for delivery or expression, these mutants might find use in the treatment of inflammatory conditions or other Nox2dependent pathological conditions. FIG. 9. Colocalization of p22 phox and Nox1 or Nox4 in human tissues. The human colon and kidney tissues were fixed and immunostained as described under "Experimental Procedures." A-C, anti-Nox1 monoclonal antibody (E39.1), anti-p22 phox polyclonal antibody (R3179), Alexa Fluor 488 goat anti-mouse IgG (green), and Alexa Fluor 555 goat antirabbit IgG (red) were used for detection of Nox1 and p22 phox in the colon. The arrows in A and B indicate the apical surface of epithelial cells. D-F, anti-Nox4 rabbit polyclonal antibody and anti-p22 phox monoclonal antibody (44.1), Alexa Fluor 488 goat anti-rabbit IgG (green) and Alexa Fluor 555 goat anti-mouse IgG (red) were used for detection of Nox4 and p22 phox in the kidney. The arrows in D and E indicate kidney tubular epithelial cells. Nuclei were stained with ToPro3 (blue in A-F). For detection, tissues were viewed under a confocal laser-scanning microscope. The same procedures were performed without primary antibody as negative controls (C and F). Magnifications are ϫ28 (A), ϫ120 (B and C), ϫ20 (D), ϫ232 (E), and ϫ63 (F). In each staining image, the lower right panel shows merged images, and yellow or yellowgreen color indicates co-localization. Similar results were obtained in three separate experiments.
FIG. 10. Relationships among human Nox family members and the requirement of p22 phox for ROS-producing activity. A dendrogram was created from an alignment of amino acid sequences of human Nox1-Nox5, fruit fly Nox (DmNox), and A. thaliana Nox (RbohA) that was defined as an outgroup. Evolutionary distances from a common ancestor sequence are represented as a bar, meaning 0.1 nucleotide substitutions per site. The requirement for p22 phox and for its proline-rich region for ROS-generating activity are indicated on the right side of dendrogram. For Nox3, the requirement for the proline-rich region of p22 phox depends upon the combination of regulatory subunits that are present. p22 phox Mutants as Nox Inhibitor