Regulation of Phagocyte NADPH Oxidase by Hydrogen Peroxide through a Ca²⁺/c-Abl Signaling Pathway

Isolation of neutrophils

Human neutrophils >90% pure were isolated from heparinized venous blood by dextran sedimentation, centrifugation on Hypaque-Ficoll, and hypotonic lysis of erythrocytes as described previously [9].

Expression of the NOX2 NADPH oxidase system in K562 cells

The clonal K562 human leukemia cell line stably expressing human NOX2 (gp91^phox) has been described previously [10]. The cells were maintained in complete RPMI medium (RPMI-1640 supplemented with 10% fetal bovine serum, 100 U/ml penicillin, and 100 µg/ml streptomycin). For the expression of p47^phox and p67^phox, cells in the logarithmic stage of growth were washed once, and resuspended in complete medium to 15×10⁶ cells/ml. The cDNAs for human p67^phox and p47^phox (25 µg total) in the episomal vector pREP4 (Invitrogen, Carlsbad, CA), were added to 375 µl of the cells and 325 µl aliquots electroporated at 250 volts/960 µFd. Stable expressing cells were selected in medium containing 250 µg/ml of hygromycin for 7 days, then subsequently maintained in medium with 150 µg/ml of hygromycin. Expression of the p67^phox/p47^phox proteins in K562/NOX2 cells was confirmed by western blot analysis and by measuring superoxide generation (see below). These cells are referred to as K562/NOX2 cells. When the over-expression of a protein in addition to the phox proteins was required, we used transient transfection and pcDNA3.1 vectors containing the cDNAs of interest.

Stable expression of GFP-c-Abl fusion proteins

The expression plasmid pcDNA3.1/Zeo (+) containing cDNAs for either wild-type c-Abl (GFP-c-Abl) or kinase-dead c-Abl (GFP-KD-c-Abl) was linearized and transfected into K562 cells as described above. Stable expressing cells were selected in 250 µg/ml of zeocin for 14 days. Single-cell clones were established by limiting dilution in 96-well plates. The expression of GFP-c-Abl and GFP-KD-c-Abl in the selected clones was determined by fluorescence microscopy. These cells were then transiently transfected with pcDNA3.1 vectors encoding NOX2, p67^phox, and p47^phox.

Cell culture, inhibitors, and subcellular fractionation

Transfected K562 cells grown in complete RPMI medium, or freshly isolated neutrophils were treated as indicated in the text with inhibitors of PKC (staurosporine 1 µM, 30 min), PKCδ (rottlerin 20 µM, 3h), Src family kinases (genistein 10 µM, 30 min), c-Abl tyrosine kinase (imatinib mesylate 10 µM, overnight), SERCA (thapsigargin [TG] 100 nM, 30 min), T-type Ca²⁺ channels (mibefradil 15 µM, 1h), G-proteins (pertussis toxin [PTX] 10 µg/ml, 1h) and RhoGTPase (Clostridium difficile toxin B 1 nM, overnight). Cells were also treated, where indicated, with PMA 1µg/ml or the extracellular Ca²⁺ chelator BAPTA (50 µM). Control cells were treated with vehicle dimethyl sulfoxide or phosphate-buffered saline (PBS) plus 10 mM glucose (PBS-G). At the end of the treatment the cells were washed in PBS-G and treated with H₂O₂ (100 µM) for 10 min at 37°C. Cell lysis was carried out in buffer A (20 mM HEPES, pH 7.9; 350 mM NaCl; 0.5 mM EDTA; 0.5 mM EGTA; 1 mM MgCl₂) plus 10% glycerol, 1% Nonidet P-40, 10 mM NaF, 0.1 mM NaVO₄, 8 mM β-glycerophosphate, phosphatase inhibitor cocktail I and II (Sigma), and a protease inhibitor cocktail (Roche, Mannheim, Germany; 1 tablet/50 ml buffer). Lysates were cleared by centrifugation, and when required, the total protein extracts were centrifuged at 100,000 g for 1 h to separate crude membranes from cytosolic proteins. Protein content was estimated as described [11].

Superoxide assay in whole cells

Superoxide generation was measured using a luminol-based chemiluminescence assay (Diogenes, National Diagnostics). Cells were collected by centrifugation, washed once in PBS, resuspended at 5×10⁶/ml in PBS-G, and kept on ice until assayed. For the assay, 100 µl of the luminol reagent were mixed with 0.25 to 0.5×10⁶ cells and incubated at 37°C for 2–4 min. Superoxide generation was stimulated by the addition of PMA 1 µg/ml in PBS-G, H₂O₂ 0–500 µM, or glucose oxidase (0.25 mU/ml) in the presence of glucose (5 mM) or the addition of formyl-methionyl-leucyl-phenylalanine (fMLF; 100 nM). Chemiluminescence was measured every 30–60 s using a Turner Designs 20/20 luminometer and a 5 s integration time.

Broken-cell NADPH oxidase assay

Neutrophils were disrupted by sonication in buffer B (100 mM KCI, 3 mM NaCI, 3.5 mM MgCI₂ 1.25 mM EGTA, 10 mM PMSF, 10 mM NaF, 0.1 mM, NaVO₄, 8 mM β-glycerophosphate, phosphatase inhibitor cocktail I and II [Sigma], and a protease inhibitor cocktail [Roche, Mannheim, Germany; 1 tablet/50 ml buffer]). Lysates were cleared by centrifugation. Crude membranes were separated from cytosolic proteins by centrifugation at 100,000 g for 1 h. Protein content was estimated as described [11]. Mixtures containing PBS, pH 7.5, 2 mM MgCl₂, 20 mM GTPγS, 200 µM ATP, 10 nM arachidonic acid/µg of protein and 5 µg of membrane protein mixed with 45 µg of cytosolic proteins in a final volume of 100 µl were used to reconstitute NADPH oxidase activity. Superoxide production was initiated with the addition of 200 µM NADPH and determined by measuring the SOD-inhibitable reduction of cytochrome C (100 µM) and quantitated using Δε 550 = 21,000 M⁻¹ cm⁻¹ [12] blanked against identical wells containing 400 U/ml SOD. Average rates of superoxide generation were calculated from the linear region of increase in absorbance at 550 nm and were expressed as nmol of O₂^•/min/mg of membrane protein.

Rac activation assay (GST-PAK-CRIB pull-down assay)

Cell lysates prepared as above in buffer A were pre-cleared with GST-bound GSH-Sepharose, then incubated at 4°C for 30 min with 10 µg of GST-Cdc42/Rac interactive binding (CRIB) region (amino acids 54–165) of PAK1 (p21-activated kinase 1) protein (GST-PAK-CRIB) immobilized on glutathione-Sepharose beads [13]. The bead pellet was thrice-washed with the lysis buffer and resuspended in 40 µl of SDS-PAGE sample buffer. The samples were separated on 12% SDS-PAGE gels, transferred to a nitrocellulose membrane, and blotted with a monoclonal antibody to Rac1/2 (Upstate Biotechnology). Parallel immunoblots were carried out on 20 µg of total cell lysate to quantitate the total amount of Rac1/2 protein. The antigen-antibody complexes were visualized by enhanced chemiluminescence (ECL, Amersham Corp.).

Immunoblotting

Protein extracts (20–50 µg) were separated on 4–20% SDS polyacrylamide gradient gels and transferred to polyvinylidene difluoride membranes. The membranes were then incubated with antibodies against c-Abl (Sigma, cat # A5844), phosphotyrosine-245 c-Abl (Cell Signaling, cat # 2861), PKCδ (Santa Cruz Biotechnology, cat # sc937), phosphotyrosine-311 PKCδ (Cell Signaling, cat # 2055), p47^phox or p67^phox (William M. Nauseef, University of Iowa), or NOX2 [14]. Actin (Sigma) was used as a loading control. The antigen-antibody complexes were visualized by enhanced chemiluminescence (ECL, Amersham Corp.).

Live cell imaging

Transfected K562 cells were washed twice in PBS-G and plated on poly-L-lysine-coated-glass coverslips for 60 min (3×10⁵ cells/ml) in complete RPMI medium without phenol red. The cells were loaded with 6 µM fluo-4-AM and 300 nM dihydroethidium (DHE; Molecular Probes). Ca²⁺ influx and superoxide production were recorded using a confocal microscope (LSM 510; Carl Zeiss MicroImaging, Inc.) with a 63X/1.4 NA plan-apochromat objective. Fluo-4 and DHE fluorescence were excited with dual laser at 488/543 nm, attenuated to avoid photobleaching and saturation. Detection was through a 545 nm long-pass dichroic mirror and a band-pass filter at 500–530 nm for fluo-4 fluorescence and LP560 for DHE fluorescence. Image acquisition was performed with the Zeiss LSM510 Software 3.2 SP2.

Statistical analysis

Data are presented as the mean ± SEM of the values and were normalized to controls. Statistical analysis was performed using the Dunnett multiple comparisons test to adjust for multiple testing when comparing several means against the mean for a common control sample. A value of P < 0.05 was accepted as significant.

H₂O₂ induces superoxide generation by NOX2 NADPH oxidase

Addition of 100 µM H₂O₂ to freshly isolated neutrophils induced a rapid and significant increase in chemiluminescence (via the Diogenes assay) that reached maximum values within 2 to 5 min and persisted at high levels for 20 to 40 min (Fig. 1A). Very little chemiluminescence was observed in the absence of H₂O₂ (Fig. 1A), cells, or Diogenes reagent (data not shown). Furthermore, the induction of chemiluminescence by H₂O₂ was completely abrogated by the prior addition of superoxide dismutase (400 U/ml) or diphenylene iodonium (10 µM), but not by azide (1 mM) (data not shown). Thus, H₂O₂ induced a marked and sustained production of superoxide by neutrophils. The level of superoxide generated was directly dependent on the number of neutrophils (Fig. 1B) and the concentration of H₂O₂ (Fig. 1C) and was largely blocked by the addition of catalase (Fig. 1D). A combination of glucose oxidase 25 mU/ml and 5 mM glucose mimicked the effect observed with H₂O₂ (Fig. 1E), albeit with slower kinetics (Fig. 1F). Using this system, a period of about thirty min was required to reach the level of superoxide produced with 100 µM H₂O₂ in 5–10 min. Analogous to H₂O₂, the effect was dependent on the concentration of glucose oxidase (Fig. 1F) and was inhibited by catalase (Fig. 1F). In comparison with fMLF, H₂O₂ was less potent in stimulating peak superoxide generation, but its effect was more sustained (data not shown). Furthermore, H₂O₂ pre-treatment increased superoxide production stimulated by PMA (Fig. 1G). These results show that not only does H₂O₂ serve as a primary activator of NOX2, but it also acts synergistically with PMA to enhance superoxide production.

Since neutrophils are short-lived cells and therefore only modestly amenable to molecular interventions in vitro, further studies of the mechanisms underlying the regulation of NOX2 by H₂O₂ were carried out using the human K562 cell line. Untransfected K562 cells express Rac and the p22^phox component of NADPH oxidase, but not NOX2, p67^phox, p47^phox or p40^phox. Therefore, to study the NOX2 system, the essential components NOX2, p67^phox, and p47^phox were transfected into K562 cells. H₂O₂ induced a substantial increase in superoxide production in these K562/NOX2 cells, with maximum activity observed 10–20 min after addition of 100 µM H₂O₂ (Fig. 2A). To confirm that the chemiluminescence detected was due to the activity of the NOX2 system, we compared superoxide generation in these cells with K562 cells expressing only the p67^phox and p47^phox cytosolic factors. Under these conditions there was little or no induction of superoxide generation by either H₂O₂ or PMA (Fig. 2A and 2D, respectively). The activation of NOX2 was dose-dependent from 10 to 500 µM H₂O₂ (Fig. 2B), over which range there was no effect on cell viability as determined by trypan blue exclusion (data not shown). The response to H₂O₂ was abrogated by the addition of catalase (Fig. 2C). Pre-incubation of K562/NOX2 cells with H₂O₂ enhanced their response to PMA (Fig. 2D and 2E). Relative to the effect of PMA alone, H₂O₂ pre-incubation resulted in increases in both the rate (6- to 10-fold) and total amount (3-fold) of superoxide produced. This effect gradually decreased and was undetectable after 1 h (Fig. 2D).

Role of Ca²⁺ influx in NOX2 activation by H₂O₂

Based on our previous work on H₂O₂-induced NOX5-mediated superoxide generation [5], we investigated whether Ca²⁺ influx is involved in H₂O₂-induced NOX2 activity. The induction of superoxide generation in K562/NOX2 cells by H₂O₂ was abrogated by treatment with either the Ca²⁺ chelator BAPTA or the T-type Ca²⁺-channel blocker mibefradil (Fig. 3A). In contrast, depletion of intracellular Ca²⁺ stores by thapsigargin had no significant effect on H₂O₂-induced superoxide generation, whereas it increased somewhat the basal level of superoxide production. The synergistic effect of H₂O₂ on PMA-stimulated superoxide generation was significantly reduced by pre-treatment with either BAPTA or mibefradil, but not thapsigargin (Fig. 3B). These results suggest that H₂O₂ activation of NOX2 as well as its synergistic effect on PMA stimulation requires an increase in cytosolic Ca²⁺ derived largely by influx from the extracellular pool.

The role of Ca²⁺ in the activation of NOX2 by H₂O₂ was also investigated by confocal imaging using the Ca²⁺-sensitive probe fluo-4 and the superoxide-sensitive probe dihydroethidium (DHE) (Fig. 3C). Only K562 cells expressing the complete NOX2 system demonstrated superoxide formation (red fluorescence) following treatment with H₂O₂, whereas Ca²⁺ influx (green fluorescence) was induced by H₂O₂ both in K562 cells expressing the complete NOX2 system and in those expressing p47^phox and p67^phox only. Notably, Ca²⁺ influx in response to H₂O₂ treatment was accentuated in cells expressing the complete NOX2 system, suggesting a positive feedback effect of NOX2 products on H₂O₂ signaling, as we have described for NOX5 [5].

Role of c-Abl in NOX2 activation by H₂O₂

To investigate the role of c-Abl in H₂O₂-NOX2 regulation, we first treated K562/NOX2 cells with imatinib mesylate (10 µM, overnight), an inhibitor of Abl tyrosine kinase. This agent completely blocked NOX2 stimulation by H₂O₂ (Fig. 4A) and significantly reduced the effect of H₂O₂ on PMA-stimulated superoxide production (Fig. 4B). Since imatinib is not totally specific for c-Abl tyrosine kinase, we also used stable K562 cell lines over-expressing either the GFP-tagged wild-type c-Abl (GFP-c-Abl), or the GFP-tagged kinase-dead c-Abl (GFP-KD-c-Abl), and transiently transfected with the NOX2 system components (gp91^phox/p67^phox/p47^phox). Over-expression of the enzymatically active GFP-c-Abl significantly enhanced both the basal and H₂O₂-induced activity of NOX2, whereas over-expression of the dominant-negative GFP-KD-c-Abl markedly reduced both basal superoxide production and the response to H₂O₂ (Fig. 4C). Similar effects were observed for the effect of H₂O₂ on PMA-stimulated NOX2 activity (Fig. 4D). These results show that c-Abl is a key intermediate in the NOX2-activating effects of H₂O₂.

Role of PKCδ in NOX2 activation by H₂O₂

PKC is an important mediator of neutrophil NOX2 activation. Since both Ca²⁺ influx and c-Abl can induce PKC activation, we analyzed the effect of PKC inhibitors, namely staurosporine, a broad inhibitor of all PKC subtypes, and rottlerin, which is specific for PKCδ. A member of the novel (n)PKC subfamily of isoforms requiring diacyglycerol (DAG), but not Ca²⁺, PKCδ participates in NOX2 activation in response to fMLF or zymosan by phosphorylating p47^phox [15, 16]. PKCδ and c-Abl also cross-phosphorylate and activate each other [17]. We observed in K562/NOX2 cells that both H₂O₂ induction of NOX2 activity and its effect on PMA activation were abrogated by staurosporine (data not shown). When the cells were pre-treated with rottlerin, a significant reduction in stimulation of NOX2 activity by H₂O₂ was observed (Fig. 5A). In addition, rottlerin interfered with both the effect of H₂O₂ on PMA stimulation of NOX2 and the activation of NOX2 by PMA alone (Fig. 5B). Western blot analysis of PKCδ demonstrated that H₂O₂ induced phosphorylation of tyrosine³¹¹, a reaction that was blocked by rottlerin, BAPTA, or imatinib (Fig. 5C). PMA, used here as a positive control, predictably induced phosphorylation of PKCδ, but no additional phosphorylation was observed upon subsequent exposure to H₂O₂ (Fig. 5C). This observation correlates with marked superoxide production stimulated by PMA, but no further increment upon subsequent addition of H₂O₂. These results altogether suggest that PKCδ activation plays a major role in the regulation of NOX2 by H₂O₂ and that it lies downstream of Ca²⁺ entry and c-Abl activation.

Role of Rac in NOX2 activation by H₂O₂

Since Rac is required for agonist-stimulated NOX2 activation and a connection between c-Abl and Rac activation has been reported [18], we analyzed the potential role of Rac activation in NOX2 regulation by H₂O₂. In K562/NOX2 cells, H₂O₂ exposure resulted in activation of Rac, as assessed by the GST-PAK pull-down assay (Fig. 6A). Pre-treatment of the cells with the Rac inhibitor toxin B of C. difficile, BAPTA, imatinib, or rottlerin abrogated or reduced the effect of H₂O₂ on Rac activation. PMA-induced Rac activation was not affected by the subsequent addition of H₂O₂ (Fig. 6A). Pre-treatment with toxin B significantly reduced the production of superoxide stimulated by H₂O₂ (Fig. 6B). In cells pre-treated with the toxin, the effect of H₂O₂ on PMA-stimulated superoxide production was significantly reduced, and the stimulation by PMA itself was reduced by approximately 20% (Fig. 6C). These results suggest a common pathway for PMA and H₂O₂ in the activation of Rac through at least PKCδ and demonstrate that the regulation of NOX2 by H₂O₂ requires activation of Rac downstream of Ca²⁺ influx and c-Abl activation.

Similar H₂O₂-NOX2 regulation pathways in neutrophils and K562/NOX2 cells

As for K562/NOX2 cells, H₂O₂-induced superoxide production by neutrophils was significantly reduced by BAPTA, mibefradil, or staurosporine, but not by thapsigargin (Fig. 7A). The synergistic effect of H₂O₂ pre-treatment for superoxide stimulation by PMA was also observed in these cells (see Fig. 1E).

Western blot analysis performed on neutrophil crude membranes showed that H₂O₂ induced the translocation to the membrane of the cytosolic factors p67^phox and p47^phox (Fig. 7B), a key feature of classical agonist-stimulated NOX2 activation. H₂O₂ also induced Rac activation in neutrophils (Fig. 7C). Furthermore, induction of superoxide production by H₂O₂ correlated with an increase in phosphorylation of both c-Abl (Fig. 7D) and PKCδ (Fig. 7E). These observations suggest that the signaling pathways involved in H₂O₂-NOX2 regulation in neutrophils are very similar to those demonstrated in K562/NOX2 cells.

Broken cell system analysis of effects of H₂O₂ pre-treatment

Reconstitution of NOX2 activity in the broken cell system was examined using different combinations of membranes and cytosol isolated from neutrophils pre-treated or not with 100 µM H₂O₂ (Fig. 7F). The rate of superoxide generation in the broken cell assay performed with membranes plus cytosol isolated from H₂O₂-treated cells, was much higher than in assays performed with membranes plus cytosol isolated from untreated control cells (p<0.05). Moreover, the level of superoxide produced by H₂O₂-treated cytosol plus untreated membranes tended to be increased relative to untreated cytosol plus untreated membranes and similar to the amount produced by untreated cytosol plus H₂O₂-treated membranes, although neither of these combinations generated as much activity as when both cytosol and membranes were derived from H₂O₂-treated cells (Fig. 7F). These observations may reflect both the presence of activated co-factors in H₂O₂-treated cytosol and translocated cytosolic factors in the plasma membranes isolated from H₂O₂-treated cells. The fact that superoxide generation was highest with the combination of treated membranes plus treated cytosol suggests that gp91^phox might itself be a final target of H₂O₂, particularly since the increased superoxide production was not attributable to an increase in NOX2 content of the membranes from H₂O₂-treated cells.

Lack of a priming effect of H₂O₂ on fMLF-stimulated superoxide production

Since H₂O₂ increases superoxide production induced by PMA, we investigated whether H₂O₂ could act as a priming factor for receptor-mediated superoxide production. We found that pre-incubation for 10 min with 10 nM to 100 µM H₂O₂ did not prime the cells for activation by 100 nM fMLF (Fig. 8A). On the contrary, a decrease in superoxide production was observed, suggesting that common signaling intermediates may be involved in H₂O₂ and fMLF-mediated superoxide production. Since signaling initiated by fMLF binding to its receptor proceeds through PTX-sensitive G proteins, which can be activated by H₂O₂ via cysteine oxidation [19], we investigated whether these G proteins might be involved in the effect of H₂O₂ on superoxide production. We observed a significant decrease in H₂O₂-induced superoxide production by PMNs pre-treated with PTX, an inhibitor of Gi/o proteins (Fig. 8B). A similar effect of PTX was observed in K562/NOX2 cells (data not shown). These observations point to at least one common proximal target of H₂O₂ and fMLF, namely a PTX-sensitive G protein.