Neutrophils from p40 ^phox−/− mice exhibit severe defects in NADPH oxidase regulation and oxidant-dependent bacterial killing

p40^phox−/− mice were produced by standard gene-targeting strategies (see Materials and methods and Fig. S1 A, which is available). Two independent embryonic stem cell lines were used to derive two independent strains of p40^phox−/− mice. These two strains could not be distinguished on the basis of any of the experiments presented, and, where appropriate, data is compiled from several animals of each strain. p40^phox−/−, p40^phox+/−, and p40^phox+/+ animals were produced from the breeding of p40^phox+/− heterozygotes in the expected Mendelian ratios and appeared healthy and fertile when kept under barrier conditions. p40^phox−/− mice had normal organ weights (Fig. S2) and normal blood cell counts (Fig. 1 A).

We used a mouse monoclonal antibody to p40^phox to investigate the tissue distribution of p40^phox protein. Amongst the tissues examined, we could only find strong evidence for immuno-detectable protein that was absent in equivalent p40^phox−/− samples (Fig. 1 B, arrowheads) in both bone marrow–derived neutrophils (BMNs) and in the spleen, suggesting that p40^phox expression is highly restricted. We could not detect expression of p40^phox in BMNs from p40^phox−/− mice using three different monoclonal and polyclonal antibodies raised against p40^phox (Fig. 1 C and unpublished data), indicating successful targeting of the p40^phox gene. Expression of p47^phox was unaffected by the absence of p40^phox, but expression of p67^phox was significantly reduced (Fig. 1 C). Careful quantitation indicated that p67^phox levels in BMNs from p40^phox−/− mice were reduced to 44.8 ± 4.3% of wild-type levels (Fig. 1 E). p40^phox+/− BMNs demonstrated 71.9 ± 3.9% and 15.3 ± 4.0% reductions in p40^phox and p67^phox, respectively (Fig. 1, D and E).

BMNs from p40^phox−/− mice were able to support normal phagocytic uptake of Staphylococcus aureus (Fig. 2 A) and zymosan (unpublished data), and normal activation of PKB, p38 MAPK, and p42/44 MAPK in response to FMLP (Fig. 2 B). These results suggest that the functions of several receptor classes characteristic of differentiated neutrophils are intact in BMNs from p40^phox−/− mice.

We characterized production of ROS in response to FMLP in either unprimed, mTNF-α–, or mGM-CSF–primed BMNs isolated from p40^phox−/−, p40^phox+/−, and p40^phox+/+ mice. ROS production was measured by a horseradish peroxidase (HRP)-dependent chemiluminescence assay that provides a rate measure of O₂⁻ production into the extracellular space. ROS production under each of these circumstances was severely reduced in the p40^phox−/− mice (∼69–84% defective; Fig. 3, A–C). The kinetics of the remaining ROS production were similar in p40^phox−/− neutrophils compared with the wild-type, and the dose-response curves to FMLP showed only a small shift to higher concentrations (Fig. 3, A–C). PMA is often used to bypass cell surface receptors and induce a more direct activation of the NADPH oxidase via, amongst less well-defined pathways, a direct protein kinase C– mediated phosphorylation of p47^phox. PMA-induced production of ROS in p40^phox−/− neutrophils was also substantially reduced compared with the wild-type (Fig. 3 D) and showed negligible shifts in the dose-response curves.

We characterized ROS responses in both primed and unprimed BMNs from p40^phox−/− and p40^pho+/+ mice in response to various particulate stimuli. We initially measured ROS production using luminol-dependent luminescence in the absence of HRP, which previous work has indicated is a rate measure of ROS production in the vicinity of endogenous peroxidases, largely thought to arise from the coincident deposition of ROS and myeloperoxidase into the phagosome (54). This is supported by our own data indicating little effect of adding extracellular HRP in these assays (unpublished data). Under these circumstances, there were significant differences in the relative reductions in ROS production seen with p40^phox−/− neutrophils compared with the wild-type, depending on the nature of the stimulus (Fig. 4).

Priming the cells with mTNF-α and mGM-CSF caused a significant but relatively minor increase in the rate of ROS generation in response to both unopsonized and opsonized zymosan and S. aureus, and a major increase in both the rate and magnitude of ROS generation in response to IgG latex beads. Looking across the full range of the responses to the various stimuli, the deficiencies in ROS production in the p40^phox−/− versus p40^phox+/+ neutrophils were similar between the primed and unprimed cells (Fig. 4 A and unpublished data), suggesting that the p40^phox deficiency was not correlated with a specific deficiency in this type of priming mechanism.

Opsonization of zymosan with IgG or serum increased slightly the rate of ROS generation relative to the unopsonized particles but made relatively little difference to the level of the defect in the p40^phox−/− versus p40^phox+/+ neutrophils (Fig. 4 and not depicted). Serum opsonization of S. aureus increased significantly the rate and magnitude of ROS production in response to these stimuli but again made relatively little difference to the level of defect in the p40^phox−/− neutrophils (Fig. 4). The ROS responses to S. aureus, however, were much more affected by the loss of p40^phox than the responses to zymosan. The amounts of ROS produced by these stimuli over 2 h, across several experiments, are presented in Fig. 4 B (∼84 vs. 30% defective for S. aureus and zymosan, respectively). These differences remained when varying amounts of these two stimuli were used (unpublished data), indicating a fundamental difference in their relative sensitivity to the loss of p40^phox. Zymosan and S. aureus, even in their unopsonized states, represent complex stimuli with respect to the receptors they engage; thus, this difference cannot at this stage be ascribed to particular receptor-associated signaling mechanisms. In an attempt to provide a phagocytic stimulus with a defined mechanism of entry, we made use of latex beads coated with IgG. Under the conditions of our assays, the ROS response to these beads was highly dependent on the IgG coating (Fig. 4); thus, it is reasonable to assume that an Fc receptor of some type plays a central role in the ROS response to this stimulus. p40^phox−/− neutrophils exhibited a very severe defect in ROS production to IgG latex beads (∼85%; Fig. 4), indicating that the signaling pathways from Fc receptors to the oxidase are highly dependent on p40^phox.

Given the requirement for both ROS and peroxidase activity in the luminol-dependent chemiluminescence assay, and the potential ambiguity in interpretation of a deficiency that this measurement provides, we also sought to measure NADPH oxidase activity in p40^phox−/− and p40^phox+/+ neutrophils by measuring the rate of O₂ consumption. It has been shown previously that NADPH oxidase–catalyzed consumption of O₂ during a stimulated “respiratory burst” dominates O₂ consumption by other routes, e.g., mitochondrial respiration (55), and hence provides a measure of substrate use that should be independent of the location of the oxidase or the pathways of ROS use. PMA-, IgG-Zym–, and S. aureus–stimulated O₂ consumption showed similar relative differences between p40^phox+/+ and p40^phox−/− BMNs as those previously described by the chemiluminescence measurements (Fig. 5, A and B). Thus, the p40^phox−/− neutrophils showed large deficiencies in O₂ consumption when challenged with PMA and S. aureus yet a relatively minor deficiency when challenged with zymosan, despite similar absolute rates of O₂ consumption for each stimulus (Fig. 5 A).

To confirm that ROS accumulation by p40^phox−/− BMNs in response to IgG-Zym was directed into the phagosome, we took advantage of the fact that O₂⁻-catalyzed reduction of soluble formazan salts to form insoluble, colored products can be used as a semiquantitative assay for the location of NADPH oxidase activity. Substantial formation of dark purple nitroblue tetrazolium (NBT) deposits could be seen in zymosan-containing phagosomes in both p40^phox+/+ and p40^phox−/− neutrophils (Fig. 5 C), suggesting that substantial oxidase activity must be present in these structures.

Neutrophils attached to a variety of surfaces via their cell surface integrins can generate substantial quantities of extracellular ROS in response to costimulation by cytokines (e.g., TNF-α), chemoattractants (e.g., FMLP), or Fc receptors (FcγRIII in murine neutrophils) (56–58). We investigated the ability of BMNs from p40^phox−/− mice attached to a fibrinogen surface to produce ROS in response to mTNF-α, a response previously established to be dependent on β2 integrin engagement (59, 60). p40^phox−/− neutrophils exhibited a remarkable deficiency in this response compared with their wild-type controls (96.7 ± 0.7% reduced; Fig. 6, A and C). Similar defects were seen upon costimulation with FMLP and mGM-CSF (unpublished data). p40^phox−/− neutrophils also exhibited a severe deficiency in activation of their oxidase in response to direct cross-linking of their β2 integrins by an anti-β2 antibody (95.1 ± 2.7% reduced; Fig. 6, B and C), a response that recent work has suggested is probably mediated via cooperation between β2 integrin and FcγRIII activation (58). p40^phox−/− neutrophils expressed normal levels of cell surface β2 integrins and demonstrated no defect in TNF-α–induced adhesion and spreading on fibrinogen (Fig. 6, D and E). Thus, p40^phox plays an essential role in the pathways between integrin engagement and oxidase activation in these contexts of neutrophil activation.

Killing of S. aureus by neutrophils in vitro is highly dependent on NADPH oxidase activity, consistent with the high frequency with which S. aureus infections present in cases of CGD (9, 10). In contrast, killing of Escherichia coli is relatively independent of neutrophil NADPH oxidase activity (61). In line with these conclusions, we found that the NADPH oxidase inhibitor diphenyleneiodonium chloride (DPI) had a much greater effect on the killing of S. aureus by BMNs than the killing of E. coli (Fig. 7, A and B). BMNs from p40^phox−/− mice had a clear defect in the killing of S. aureus (∼75% relative to wild-type) but an insignificant defect in the killing of E. coli (Fig. 7, A and B). This defect was mirrored in the killing of S. aureus by whole blood (Fig. S3). Using a range of DPI concentrations to construct a relationship between BMN NADPH oxidase activity and S. aureus killing (61), the NADPH oxidase deficiency of p40^phox−/− neutrophils correlates well with their deficiency in S. aureus killing (Fig. 7 C), suggesting that the two are causally linked.

We sought to establish whether the substantial defect in the in vitro killing of S. aureus by p40^phox−/− neutrophils translated into a defect in the killing of this organism in the whole animal. We injected live S. aureus into the peritoneum of p40^phox−/− and p40^phox+/+ mice and measured the number of surviving bacteria over the subsequent 24 h. We reproducibly recovered significantly more live S. aureus from the p40^phox−/− mice than their wild-type controls (Fig. 7 D).

We observed very substantial defects in NADPH oxidase activity in neutrophils isolated from p40^phox−/− mice, defining an essential role for p40^phox in the physiological regulation of this important enzyme complex. The degree of defect in NADPH oxidase activity varied greatly depending on the nature of the stimulus. Large defects were observed in intracellular ROS production in response to S. aureus or IgG latex beads, and smaller defects were observed in response to the uptake of zymosan. Extracellular ROS generation was absent in response to TNF-α/fibrinogen and greatly reduced in response to FMLP or PMA. ROS stimulation by TNF-α/fibrinogen is thought to occur via costimulation of β2 integrins and TNF-α receptors (59, 62). FMLP is thought to activate the oxidase via formyl-peptide receptors (63) and IgG to act via Fcγ receptors (64). S. aureus and zymosan are complex stimuli that are engulfed and activate the oxidase through a combination of inaccurately defined receptor types, but these are likely to include β2 integrin receptors (64) and β glucan receptors (65). Furthermore, these receptor types are likely to be extended in vivo through opsonization with complement or antibodies. Thus, it is clear that loss of p40^phox has a differential impact in the signaling pathways from different receptor subtypes. The extent of p40^phox involvement in the various physiological mechanisms of NADPH oxidase activation in neutrophils and other cell types will only be resolved with much further work, but our in vitro and in vivo models of oxidant-dependent bacterial killing already indicate that it is a critical component of our innate defense mechanism (Fig. 7). Importantly, the extent of the S. aureus killing defect is as severe as that observed with neutrophils from p47^phox−/− CGD model mice (3, 66).

A key question is clearly the molecular mechanism by which p40^phox acts in the signaling pathways to NADPH oxidase activation. We observed that p67^phox levels are reduced by ∼55% in p40^phox−/− neutrophils. p67^phox and p40^phox are established to exist in a complex in neutrophils, although the stoichiometry and relative inclusion of p47^phox in the resting state are still in debate. In p67^phox-deficient CGD patients, p40^phox expression is reduced (34, 67, 68), suggesting that p40^phox and p67^phox mutually regulate each other's steady-state expression, and this is supported by the reduced p67^phox levels in p40^phox−/− neutrophils. Whether this regulation occurs at the level of protein stability, mRNA stability, or transcriptional/translational control is unknown. It is established that p67^phox is an absolute requirement for reconstituted NADPH oxidase activity in vitro and, through characterized cases of CGD, also in vivo. Thus, the drop in p67^phox levels in the p40^phox−/− neutrophils is a confounding issue in trying to ascribe molecular consequences to the loss of p40^phox. There is the possibility that this reduction in p67^phox has an impact directly on the oxidase, and there is also the possibility that this has an impact indirectly, via altering the dependence on other subunits. Unfortunately, the absence of a p67^phox−/− mouse model prevents a simple assessment of this impact in p67^phox+/− mice; however, we note that human carriers of p67^phox-deficient CGD do not present with a CGD phenotype (69). Further, the very large reductions in oxidase responses to TNF-α/fibrinogen and S. aureus in p40^phox−/− neutrophils suggest that p40^phox has a substantial role in these responses under normal circumstances.

Earlier work has suggested that PtdIns3P binding to the PX domain of p40^phox may play a role linking PI3K signaling pathways to NADPH oxidase activation (22, 37, 46, 47). Furthermore, the discovery that PtdIns3P synthesis can occur on phagosomal membranes provided a context for this potential mechanism (70, 71). Evidence that this pathway does indeed operate in FcγRII receptor stimulation of the NADPH oxidase in the COS phox system is provided in an accompanying article by Suh et al. (on p. 1915 of this issue; reference 72). In this elegantly engineered cell model, an intact PtdIns3P-binding PX domain of p40^phox is shown to be required for FcγRII receptor–initiated phagosomal ROS production. Our data describing the very large reduction in IgG latex bead–stimulated ROS production in p40^phox−/− neutrophils provides strong in vivo corroboration of this model. However, further work must clearly be done measuring elements of the known pathways to NADPH oxidase activation (e.g., p47^phox/p67^phox/p40^phox phosphorylation, p47^phox/p67^phox/rac-2 translocation to cytochrome b₅₅₈) before the molecular consequences of the absence of p40^phox can be properly ascribed, and it is likely that these will need to be done in the context of replenishment of p67^phox and/or knock-in mutations of p40^phox before clear answers emerge. However, the development of the p40^phox−/− mouse does appear to offer a potential route to answering these questions.

The absence of examples of CGD patients characterized by mutations in p40^phox together with enigmatic data arising from attempts to reconstitute the NADPH oxidase in cellular and a-cellular systems lead to the general view that p40^phox does not have a critical role in NADPH oxidase activation. A counter argument arising from data describing widespread tissue expression of p40^phox suggested that this protein may have a wider role outside of phagocytes (73), precluding the presentation of p40^phox dysfunction as CGD. This argument is not supported by our initial characterization of p40^phox expression in the p40^phox−/− mice, which did not provide evidence for high expression of p40^phox outside the haemopoetic system, nor by the fact that p40^phox−/− mice appear healthy when raised under pathogen-free conditions. It seems plausible that either a loss of p40^phox presents a mild phenotype in humans or, indeed, that several uncharacterized cases of CGD-like immunodeficiencies may arise from as yet undescribed mutations in the p40^phox gene.

Murine GM-CSF (mGM-CSF), FMLP, luminol, DPI, and HRP were from Sigma-Aldrich. Murine TNF-α (mTNF-α) was from R&D Systems. All buffer components were from Sigma-Aldrich and were endotoxin free or low endotoxin, as available. The following antibodies were used: anti–phospho- Ser473 PKB (Ab4802; Abcam), anti-PKB (9272; Cell Signaling), anti–phospho-T180/Y182 p38 (9211S; Cell Signaling), anti-p38 (9212; Cell Signaling), anti–phospho-T202/Y204 p42/44 (9106S; Cell Signaling), and anti-p42/22 (606-259-1550; TransLabs).

Several clones encoding the p40^phox genomic sequence were isolated from the RPCI mouse PAC library 21 (Pieter de Jong, UK HGMP Resource Centre) by Southern screens using an NT-cDNA probe. An 11.9-kb SpeI–SpeI fragment encompassing exons 1–4 was isolated from clone RP21-641C7 and inserted into the low copy number plasmid pSC-3Z to form the basis of a p40^phox gene-targeting vector (Fig. S1 A).

A smaller fragment containing exon 3 was subcloned into pBS, and site-directed mutagenesis was conducted to alter codon 73 to a translational stop and to introduce silent mutations creating additional XhoI and ApaI sites (Fig. S1 B), which were used to track the presence of the mutated sequence. This modified segment was sequenced and reintroduced into the targeting vector. This strategy was adopted because the translational stop or p40^phox−/− mouse described here was one of several knock-in mutations planned for exon 3.

A loxP-flanked cassette containing a tACE-Cre expression module and a Neo^R expression module (pACN; A. Plagge, Babraham Institute, Cambridge, UK) (74) was inserted into the SnaBI site in the intron between exons 3 and 4 (Fig. S1 A). The tACE promoter is expected to operate in the testis and drive Cre/Lox-mediated deletion of this cassette on the breeding of targeted chimeras. Deletion is predicted to leave 59 bp of foreign DNA remaining in the intron.

The final targeting vector was digested with DraI and SalI (in the 3′ polylinker) to remove excess pSC-3Z vector sequence and used to transfect E14 129^s/v embryonic stem cells by the Gene-Targeting Facility at the Babraham Institute. 500 clones were initially screened for homologous recombination using a 3′ Southern screen (probe 3′ to targeted sequence, EcoRI digest, 4.6–8.3 kb transition), and positive clones were rescreened by a 5′ Southern screen (probe 5′ to targeted sequence, SpeI digest, 11.9–10.0 kb transition) and for single insertion of the Neo^R cassette (Neo probe, EcoRI/8.3 kb, and SpeI/5.7 kb; Fig. S1, A and C). Three clones were taken forward for blastocyst injection, and male chimeras from these mice were bred with female C57BL/6 animals to generate p40^phox+/− heterozygotes on a mixed 129^s/v/C57BL/6 background. Deletion of the Neo^R cassette was confirmed by appropriate Southern and PCR analyses, and separate p40^phox−/− colonies were created from each of the original targeted embryonic stem cell lines and housed under specific pathogen-free conditions in the SABU facility at the Babraham Institute. Genotyping of the mice was routinely performed by PCR amplification of an approximately 850-bp region flanking exon 3 (to include the inserted translational stop and additional XhoI and ApaI sites) and subsequent diagnosis by susceptibility of the product to cleavage by XhoI or ApaI (to yield ∼525- and 325-bp fragments; Fig. S1 D). All mice used were 2–8-mo old and showed no age-dependent variation. This work was covered by UK Home Office Project License PPL 80/1875.

BMNs were prepared as described previously with minor modifications (75). Bone marrow, from at least three mice per preparation, was dispersed in HBSS (without Ca²⁺ and Mg²⁺), 0.25% fatty acid–free BSA, 15 mM Hepes, pH7.4, at room temperature and purified over discontinuous Percoll (GE Healthcare) gradients. After washes, mature neutrophils were resuspended in Dulbecco's PBS with Ca²⁺ and Mg²⁺, 1 g/liter glucose, 4 mM sodium bicarbonate (DPBS⁺). Purity was typically 70–80% as assessed by cytospin and REASTAIN Quick-Diff (Reagena) staining (non-neutrophils were ∼50% immature white cells, 25% monocytes, and 25% lymphocytes). All assays were performed in DPBS⁺. BMNs were primed at 37°C with 500 U/ml mTNF-α, 100 ng/ml mGM-CSF, and 10% mouse serum for 60 min. In some experiments, as indicated, BMNs were primed individually with 500 U/ml mTNF-α for 30 min, 100 ng/ml mGM-CSF for 60 min, or 500 U/ml mTNF-α and 100 ng/ml mGM-CSF for 60 min.

5 × 10⁶ BMNs were sonicated into 1× SDS loading buffer, and 5 × 10⁵ cell equivalents were subjected to SDS-PAGE, transferred, and blotted for p40^phox (05-539 monoclonal; Upstate Biotechnology; sc-18252 and sc-18253 polyclonals; Santa Cruz Biotechnology, Inc.), p47^phox (07-500 polyclonal; Upstate Biotechnology), and p67^phox (07-502 polyclonal; Upstate Biotechnology). Signal was detected (Image Reader LAS-1000; Fugifilm) and quantified using Aida Image Analyser 2.2. For tissue Westerns, tissues were collected and immediately machinated into ice-cold 20 mM Hepes, pH7.1, at 4°C, 0.1% SDS, 0.4% cholate, 0.1% NP-40, 0.1M NaCl, 0.2 mM PMSF, and 10 μg/ml each of pepstatin A, leupeptin, antipain, and aprotinin and clarified by centrifugation (14,000 g for 30 min at 4°C). Protein concentrations were determined (BCA; Pierce Chemical Co.), and 30 μg of each protein was subjected to SDS-PAGE, transferred, and immunoblotted for p40^phox.

5 × 10⁶ unprimed BMNs from p40^phox−/− and p40^phox+/+ mice were prewarmed for 3 min at 37°C at 5 × 10⁷/ml in DPBS⁺ in duplicate. After 1 min of stimulation with prewarmed FMLP (10 μM final) or salts, reactions were stopped by the addition of excess ice-cold PBS, followed by immediate centrifugation (12,000 g for 10 s). Cell pellets were lysed in 20 mM Tris, pH 7.5 at 4°C, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM β-glycerophosphate, 1 mM sodium orthovanadate, 0.2 mM PMSF, and 10 μg/ml each of pepstatin A, leupeptin, antipain, and aprotinin and incubated on ice for 10 min. Cytoskeletal debris was removed by centrifugation (12,000 g for 20 min at 4°C). Lysates were split between two SDS-PAGE gels and blotted for phospho- and total protein, respectively. Blots were imaged and quantified as described above.

Mouse blood was collected and left to clot at room temperature for 45 min in a glass container before transferal to a 15-ml tube followed by centrifugation (1,500 g for 10 min at room temperature). Serum was removed to a fresh tube, recentrifuged, recovered, placed on ice, aliquoted, and stored at −80°C.

ROS production was measured by luminol-dependent chemiluminescence in polystyrene 96-well plates (no. 23300; Berthold Technologies Ltd.) as described previously (75) in DPBS⁺, except final concentrations of luminol and HRP were 150 μM and 18.75 U/ml, respectively. Prewarmed stimuli were added manually and measurement started immediately. Assays using soluble stimuli (PMA, FMLP, and mTNF-α) were conducted in the presence of exogenous added HRP; the signal was >95% HRP dependent, indicating predominantly extracellular ROS production (unpublished data). Assays using particulate stimuli (zymosan, IgG latex beads, and live S. aureus) were conducted without HRP and thus represent intracellular ROS production. The addition of HRP revealed little extracellular ROS production (unpublished data). Final particle/BMN ratios were as follows: zymosan, 5:1; S. aureus, 20:1; and IgG latex beads, 50:1. In some S. aureus experiments, BMNs were preincubated with varying concentrations of DPI or vehicle (DMSO) alone before stimulation. Data output is in relative light units per second (rlu/s).

Surfaces (96-well plates or 12-well tissue culture plates) were coated overnight at 4°C with sheep fibrinogen (F9754; Sigma-Aldrich) at 150 μg/ml in PBS (100% FCS was used as a nonspecific control). Before use, surfaces were washed three times with PBS. For adhesion-dependent ROS assays, neutrophils at 5 × 10⁶/ml were incubated at 37°C for 1 h before the addition of prewarmed 2× HRP/luminol followed by mTNF-α (final concentration of 20 ng/ml), anti-CD18 (anti–β2 integrin, M18/2; Chemicon), or IgG2a isotype control antibodies (final concentration of 18 μg/ml). Cells were immediately aliquoted into the plate (100 μl/well) and counted. For spreading experiments, preincubated neutrophils were incubated at 37°C in coated 12-well tissue culture plates (6.25 × 10⁵/well) in the presence of 20 ng/ml mTNF-α. After 40 min, wells were aspirated and cells were fixed in 3.8% formaldehyde. Non-adhered cells were washed away and remaining cells were visualized by light microscopy. In both ROS and spreading assays, FCS-coated wells produced minimal responses (unpublished data). For analysis of β2 integrin expression, neutrophils were stimulated with mTNF-α for 30 min at 37°C, placed on ice, and processed for FACS analysis using the anti-CD18 antibody, its isotype control, and an FITC anti–rat secondary.

IgG-opsonized zymosan particles (IgG-Zym) were prepared as per the manufacturer's instructions (unlabeled zymosan A, Z-2849, and rabbit anti-zymosan A, Z-2850; Invitrogen). Zymosan and S. aureus were serum opsonized or mock opsonized by incubation in DPBS⁺ with or without 50% mouse serum at 37°C with end-over-end mixing for 1 h (zymosan) or 15 min (S. aureus) followed by washing. Carboxylate-modified latex beads (0.9-μm diameter; Sigma-Aldrich) were cross-linked to sulfhydryl-modified BSA and coated with an anti-BSA monoclonal antibody (Sigma-Aldrich) or not, as described previously (IgG latex beads) (76). Where appropriate, S. aureus was washed and resuspended in DPBS⁺ (4 × 10⁸/ml), heat killed at 60°C for 30 min, and opsonized in mouse serum as described above.

10⁶ primed BMNs were allowed to adhere to glass coverslips for 20 min at 37°C. They were then aspirated, 10⁷ FITC-labeled, serum-opsonized S. aureus was added, and they were returned to 37°C (FITC labeling of bacteria as described previously [77] and opsonization as detailed above). After 40 min, coverslips were washed, fixed in 4% paraformaldehyde, and mounted. Postfixation probing with a rabbit anti–S. aureus antibody (S-2860; Invitrogen) and goat anti–rabbit Alexa Fluor 568 secondary antibody (Invitrogen) revealed that >95% of bacteria present were internalized. Phagocytosed bacteria were visualized by fluorescence microscopy and enumerated.

Oxygen consumption was measured in a Clark-type oxygen electrode (Rank Brothers Ltd.) at 37°C with rapid stirring. Primed BMNs were added to the rapidly stirred chamber at 5 × 10⁶/ml and equilibrated for 5 min before the addition of prewarmed stimuli. Final concentrations were 1 μM PMA; IgG-Zym, 20:1; and heat-killed S. aureus, 20:1 (particles/BMNs).

Primed BMNs were adhered to a coverslip in 0.5 mg/ml NBT at 37°C. IgG-Zym particles were added, and dark purple formazan deposition was followed during phagocytosis by bright field microscopy.

Bacteria (S. aureus Wood 46 and E. coli E2348169) were subcultured at 37°C to logarithmic growth from an overnight culture. 4 × 10⁷ bacteria were washed in DPBS⁺ and opsonized as described above. Opsonized bacteria (1.5 × 10⁶ S. aureus and 6 × 10⁶ E. coli) were added to 6.2 × 10⁶ primed BMNs (2.5 × 10⁷/ml) at 37°C with rapid orbital mixing. After the indicated times, 50-μl aliquots were removed to 950-μl ice-cold Luria broth (LB) containing 0.05% saponin. Samples were sonicated (output 1.5 for 10 s; Sonicator 3000; Misonix) to liberate intracellular bacteria and returned to ice. Suspensions were serially diluted and plated on LB-agar to enumerate surviving bacteria. A parallel bacterial incubation was also run in the absence of neutrophils. In some experiments, neutrophils were incubated for 5 min with DPI at varying concentrations or vehicle (DMSO) alone before the addition of bacteria (oxidant-dependent killing control).

For S. aureus killing assays in whole blood, 2.5 × 10⁶ bacteria in 1 ml DPBS⁺ were added to 1 ml of fresh whole blood (mixed from at least three animals) and incubated for 20 min at 37°C with end-over-end mixing. 0.8 ml of blood/bacteria mix was added to tubes containing lysostaphin (a final concentration of 2.5 U/ml to kill extracellular bacteria) in duplicate and returned to mixing. Samples were taken after 1 h, added to ice-cold PBS, and pelleted by centrifugation, and the pellets were resuspended in 0.5 ml nutrient broth, 0.05% saponin. Samples were then processed and quantified as described above.

S. aureus (LS-1) was subcultured at 37°C to logarithmic growth from an overnight culture. Bacteria were washed and resuspended in injection-grade PBS at 2.5 × 10⁸/ml. Three animals of each genotype per time point were injected intraperitoneally with 0.2 ml of bacterial suspension (5 × 10⁷ bacteria). After 4 or 24 h, mice were killed and the peritoneal cavity was thoroughly flushed with 10 ml ice-cold PBS, 5 mM EDTA, and 5 U/ml heparin. Aliquots were diluted, sonicated, and plated, and bacteria were enumerated as for the in vitro killing assays.

Fig. S1 illustrates the targeting strategy used to generate p40^phox−/− mice and the subsequent screening strategies used. Fig. S2 denotes normal organ weights of p40^phox−/− animals. Fig. S3 describes the S. aureus killing deficiency of whole blood from p40^phox−/− animals.

We thank Ted Saunders, Martin George, and Debbie Drage of the Gene-Targeting Facility at Babraham for their excellent support; members of the Small Animal Breeding Unit at Babraham for animal husbandry; and Simon Andrews of the Bioinformatics group at Babraham for his help in the design of the gene-targeting strategy. We also thank Stephanie Brown and Michael Murphy for help and advice with oxygen electrode measurements (MRC-Dunn Human Nutrition Unit, Cambridge) and the laboratory of Anthony Green (Dept. of Haematology, Cambridge University) for use of their animal blood cell counter.

This work was supported by grants from the Biotechnology and Biological Sciences Research Council and Arthritis Research Council (no. S0679). C.D. Ellson is a Beit Memorial Fellow.

The authors have no conflicting financial interests.