Hv1 proton channels are required for high-level NADPH oxidase-dependent superoxide production during the phagocyte respiratory burst

Abstract

Granulocytes generate a “respiratory burst” of NADPH oxidase-dependent superoxide anion (O₂^−∙) production that is required for efficient clearance of bacterial pathogens. Hv1 mediates a voltage-gated H⁺ channel activity that is proposed to serve a charge-balancing role in granulocytic phagocytes such as neutrophils and eosinophils. Using mice in which the gene encoding Hv1 is replaced by β-Geo reporter protein sequence, we show that Hv1 expression is required for measurable voltage-gated H⁺ current in unstimulated phagocytes. O₂^−∙ production is substantially reduced in the absence of Hv1, suggesting that Hv1 contributes a majority of the charge compensation required for optimal NADPH oxidase activity. Despite significant reduction in superoxide production, Hv1^−/− mice are able to clear several types of bacterial infections.

The voltage-gated H⁺ channel Hv1 (also termed VSOP or BTS) was identified by its homology to the voltage sensor domain of voltage-gated cation channels and a voltage-regulated phosphatase Ci-VSP (1–4). Hv1 is encoded by the Hvcn1 gene, and both protein function and genetic sequences are well conserved from primitive chordates to mammals (1, 2). Expression of Hv1 in either mammalian cells or Xenopus oocytes is sufficient to reconstitute the hallmark biophysical properties of native voltage-dependent H⁺ conductances (G_vH+): activation by transmembrane depolarization or net intracellular acidification, exquisite H⁺-selective conductance, and sensitivity to inhibition by micromolar Zn²⁺ (1, 2, 5–8). Under resting physiological conditions in most cells (pH_i ≈ pH_o), the channel's voltage-dependent gating dictates that protons flow only outward through Hv1, suggesting that a prominent physiological role for the channel is to secrete H⁺ (5).

Voltage-gated H⁺ currents and Hv1 protein are prominently expressed in granulocytic phagocytes such as neutrophils and eosinophils (1, 3, 5, 9); granulocytes are characterized by their ability to mount a “respiratory burst” of high-level NADPH oxidase-dependent superoxide anion (O₂^−∙) production that is required for clearance of bacterial pathogens (10). NADPH oxidase would cause a rapid intracellular acidification and profound depolarization of the cell membrane in the absence of a balancing charge movement, and G_vH+ mediated by Hv1 is ideally suited to meet this physiological requirement (11–14). To investigate whether Hv1 is required for innate immune function, we generated an Hv1 knockout mouse (Hv1^−/−) line and assessed voltage-gated H⁺ current, O₂^−∙ production, and bacterial clearance in Hv1^−/− mice and their wt/wt littermates. Our results demonstrate that Hv1 is required for G_vH+, a robust respiratory burst, and efficient bacterial clearance in vitro.

Results

The Bay Genomics genetrap of the Hvcn1 locus leads to transcription of a transgenic mRNA that contains a β-Geo reporter gene sequence in place of the Hv1 protein coding sequence, such that only the 6 N-terminal residues of Hv1 are translated (Fig. S1). RRN293 ES cells bearing the Hvcn1 genetrap allele were used to generate a transgenic mouse line that that is expected to lack Hv1 protein expression and voltage-gated proton channel activity. Triplex genomic PCR demonstrated the expected Hvcn1 size difference for the wt and transgenic alleles (Fig. 1A), and was used in conjunction with quantitative genomic PCR for genotypic screening during breeding. Homozygous RRN293 mice lack detectable levels of Hv1 protein in spleen, bone marrow cells (BMC), and peripheral blood leukocytes (Fig. 1B), indicating that the genetrap strategy results in an effective Hv1 knockout (Hv1^−/−). Consistent with the reported expression of G_vH+ in alveolar epithelial cells (15, 16), we detected Hv1 protein in lung homogenates from wt/wt but not Hv1^−/− mice (Fig. S1).

Fig. 1.

Generation of Hv1^−/− mice. RRN293 ES cells contain a Bay Genomics genetrap vector insertion in the Hvcn1 gene. The transgene serves as a surrogate splice acceptor for wt exon 2 that produces an in-frame chimeric mRNA encoding Hvcn1 exons 1 and 2 followed by β-Geo sequence. (A) Triplex genomic PCR using reverse primers specific for both Hvcn1 and β-Geo amplifies a ≈1.5-kb band from the wt allele (lane 1) and a ≈1.0-kb band from the transgenic allele (lane 3); both alleles are detected in heterozygous wt/RRN293 mice (lane 2). (B) Expression of Hv1 protein (32 kDa) is readily detected in spleen (lanes 1 and 2), BMC (lanes 3 and 4), and PBL (lanes 5 and 6) isolated from wt/wt mice (+/+, lanes 1, 3, and 5) but is absent in RRN293/RRN293 mouse tissue (−/−, lanes 2, 4, and 6).

Hv1 protein was not detectable in hippocampal microsomes, suggesting that the lack of detectable voltage-gated H⁺ current in mammalian neurons results from low Hv1 expression (17). Consistent with previous reports (18, 19) we measured G_vH+ in acutely isolated non-neuronal hippocampal cells (presumably microglia) but not in Hv1^−/− cells. The relative paucity of microglia could account for undetectably low levels of Hv1 protein in brain, and future studies should investigate Hv1 expression and function in microglia in situ. We next measured voltage-gated whole-cell H⁺ currents in leukocytes isolated from wt/wt or Hv1^−/− mice as described (1). Purified resting (round and nonadherent) granulocytes exhibited G_vH+ at varying levels, but Hv1^−/− mice lacked detectable H⁺ current (Fig. 2 B and D). Consistently larger currents were measured in cells displaying the morphological characteristics (presence of pseudopodia and a uropod, spontaneous adherence and cytokinesis on glass substrate) of circulating neutrophils (Fig. 2A). The apparent threshold for voltage-dependent activation (V_thr) of H⁺ currents elicited by depolarizing voltage steps was approximately +40 mV under symmetrical recording conditions (Fig. 2C; pH_i = pH_o = 6.5). As expected for authentic G_vH+, V_thr shifted negatively ≈40 mV when a 1-log unit outward [H⁺] gradient (pH_i = 6.5, pH_o = 7.5) was imposed (Fig. 2C) (5, 7). In no instance were we able to measure detectable voltage-activated H⁺ currents in cells isolated from Hv1^−/− mice (Fig. 2 B and D), indicating that Hv1 expression is required for G_vH+. In combination with previous results demonstrating that Hv1 is sufficient to reconstitute G_vH+ (1, 2), these results suggest that Hv1 intrinsically forms the voltage-dependent proton channel that underlies G_vH+ in various cell types. The existence of additional genes encoding voltage-gated proton channels therefore seems unlikely. This conclusion is further supported by the observation that native G_vH+ and expressed Hv1 share unique biophysical properties (especially the invariant coupling of V_thr to the pH gradient) that are conserved in a wide variety of cell types (5, 7). Cell-specific factors are likely to modulate Hv1 channel properties, particularly in cells such as neutrophils, where G_vH+ appears to be posttranslationally regulated by signaling cascades (20–22).

Fig. 2.

RRN293/RRN293 leukocytes lack detectable voltage-gated H⁺ currents. (A and B) Representative whole-cell currents (Lower) resulting from the indicated voltage steps (Upper) in neutrophils (Lympholyte M) purified from wt/wt (+/+, Left) or RRN293/RRN293 (−/−, Right) B5 mouse blood. (Inset) The pH of solutions (TMA6.5 pipette, TMA7.5 bath) is shown. (C) Representative I_STEP vs. voltage relations for the cells shown in A under symmetrical [H⁺] conditions and after bath superfusion to impose a 1 pH unit outward [H⁺] gradient (○, +/+ TMA6.5 bath; ●, +/+ TMA7.5 bath; □, −/− TMA6.5 bath; ■, −/−TMA7.5 bath). (D) Whole-cell current (I_STEP, +60 mV) in granulocytes (Robbins PMN) purified from wt/wt (+/+) or RRN293/RRN293 (−/−) mouse blood (Na6.5 pipette and Na7.5 bath solutions). Solid line indicates the mean current level, and box indicates SEM (+/+, 58.6 ± 10.5 pA; −/−, 2.7 ± 0.4 pA; n = 14 cells per genotype).

Two lines of evidence suggested that G_vH+ might be necessary for sustained O₂^−∙ production during the respiratory burst: first, outward H⁺ current through Hv1 should balance the outward e⁻ current produced by the active NADPH oxidase, thereby limiting membrane depolarization that would otherwise inhibit NADPH oxidase activity (11–14); second, Hv1 allows for extrusion of intracellular protons that are produced by NADPH oxidase, thus limiting intracellular acidosis and maintaining optimal O₂^−∙ production (23–25). To test the hypothesis that Hv1 is required for high-level superoxide production, we isolated BMC from wt and Hv1^−/− mice and measured phorbol 12-myristate 13-acetate (PMA)-stimulated O₂^−∙ and hydrogen peroxide (H₂O₂) production spectrophotometrically (Fig. 3 and Fig. S2). PMA (200 nM) elicited a robust time-dependent increase in O₂^−∙ production that was absent in vehicle-treated control cells and largely blocked by dephenylene iodonium (DPI; 10 μM), an inhibitor of NADPH oxidases (Fig. 3A). After an initial delay, the rate of O₂^−∙ production (measured 1–3 min after PMA addition at 37 °C) was attenuated ≈65% in cells from Hv1^−/− mice compared with wt/wt littermates (Fig. 3B), and a similar decrement was observed 15 min after PMA addition (Fig. 3A). DPI (10 μM) inhibited 79% and 73% of the PMA-stimulated O₂^−∙ production (at 15 min) in wt/wt and Hv1^−/− BMC, respectively (Fig. 3A). Similar results were observed in peripheral blood leukocytes (PBL) purified from either Lympholyte M or discontinuous Percoll gradients, and BMC isolated from Hv1^−/− mice with mixed C57BL/6J host and 129 ES cell genetic background were not significantly different from back-crossed Hv1^−/− B5 mice. Finally, PMA-stimulated H₂O₂ production was also substantially decreased in Hv1^−/− BMC compared with wt/wt controls (Fig. S2).

Fig. 3.

Superoxide anion (O₂^−∙) production is decreased in Hv1^−/− BMC. O₂^−∙ secretion was measured by using the cytochrome c reduction assay in intact BMC isolated from wt/wt (+/+, filled symbols) or RRN293/RRN293 (−/−, open symbols) B5 mice in the absence (black circles) or presence of 200 nM PMA (blue triangles) or 200 nM PMA + 10 μM DPI (red squares). (A) Time course of O₂^−∙ production after vehicle or PMA addition at t = 0 min in intact BMC. O₂^−∙ production after 10-min incubation (37 °C) in the presence of 200 nM PMA was 9.93 ± 1.24 nmol/10⁶ cells (+/+) and 3.60 ± 0.18 nmol/10⁶ cells (−/−); mean ± SEM, n = 4 mice per genotype). (B) The rates of PMA-stimulated O₂^−∙ production in intact BMC between 1 and 3 min after PMA addition were: +/+, 1.39 ± 0.23 nmol/min per 10⁶ cells;−/−, 0.47 ± 0.04 nmol/min per 10⁶ cells; mean ± SEM, n = 4 (solid lines). Data and symbols are as shown in A. (C) Time course of O₂^−∙ production after vehicle or AA (100 μM) addition at t = 0 min in a representative pair of wt/wt and RRN293/RRN293 littermate BMC sonicates assayed in parallel. (Inset) Time course of O₂^−∙ production after vehicle or PMA addition at t = 0 min in intact BMC before sonication. Data are expressed as nmol O₂^−∙/min per 10⁶ cells. Open and filled symbols are as in A. (D) Western blot showing p22^phox expression in lysates prepared from nontransfected COS-7, COS-phox, or HL-60 cells (lanes 1–3) or BMC isolated from 2 pairs of RRN293/RRN293 (−/−, lanes 5 and 6) or wt/wt (+/+, lanes 7 and 8) mouse littermates.

In contrast to intact cells, O₂^−∙ production was not different in a broken-cell preparation made from either wt/wt or Hv1^−/− BMC (Fig. 3C). Consistent with the idea that Hv1 is required for charge compensation and H⁺ efflux in intact cells but is irrelevant in membrane fragments that do not experience changes in voltage or pH gradients after NADPH oxidase activation. In addition, we observed similar levels of p22^phox expression in wt/wt and Hv1^−/− BMC microsomes (Fig. 3D). Together, these results suggest that attenuated O₂^−∙ production in Hv1^−/− cells is unlikely to result from a decrease in the activity or expression of NADPH oxidase complex components per se. Although our data do not rule out the possibility that other gene products or ion transport pathways also contribute to charge compensation during the phagocyte respiratory burst, they suggest that Hv1 is likely to play a central role in electrical and H⁺ gradient homeostasis that is required for optimal NADPH oxidase activity under physiological conditions in mouse granulocytes (12, 25–32).

The role of the respiratory burst in innate immune function is illustrated by the observation that deletion or mutation in any of several protein components comprising the NADPH oxidase complex impairs bacterial clearance in mice and causes chronic granulomatous disease in humans (10). To determine whether the attenuated O₂^−∙ production that we observed in Hv1^−/− cells would adversely affect bacterial clearance, we incubated wt or Hv1^−/− BMC with live, serum-opsonized Staphylococcus aureus and assayed bacterial survival. BMC from both wt/wt and Hv1^−/− mice cleared >90% of the added bacteria (Fig. 4A), suggesting that the viability and bacterial killing function of BMC isolated from both wt/wt and Hv1^−/− mice is similar. NADPH oxidase-dependent bacterial killing was significantly attenuated in Hv1^−/− compared with wt/wt mice (Fig. 4 A and B). Bacterial survival was similar in the presence of DPI, suggesting that NADPH oxidase-independent mechanisms of bacterial clearance were unchanged in Hv1^−/− cells and that Hv1 deletion selectively affects NADPH oxidase-dependent killing. Furthermore, phagocytosis of heat-inactivated, fluorescently-labeled, serum-opsonized S. aureus or E. coli was not different in Hv1^−/− and wt/wt cells, suggesting that the observed defect in bacterial clearance is caused by reduced O₂^−∙ production and not by a change in the abilty of Hv1^−/− cells to phagocytose bacteria.

Fig. 4.

Hv1 deletion decreases bacterial clearance in vitro. (A) Mouse BMC (5 × 10⁵) from wt/wt (solid columns, n = 8) or RRN293/RRN293 (hatched columns, n = 7) mice were incubated (30 min, 37 °C) with S. aureus in the absence (white columns) or presence of DPI (10 μM; gray columns) before addition of gentamicin and washing. Cell pellets were lysed in 1% saponin, and surviving bacteria were serially diluted and elaborated overnight on selective agar plates. (B) Data shown as the fraction of surviving CFU in the absence of DPI (control) divided by the CFU surviving the presence of DPI to isolate the DPI-sensitive component of bacterial clearance (wt/wt, white column; RRN293/RRN293, gray column). (C) Bacterial survival 24 h after i.p. inoculation of 1 × 10⁸ CFU S. aureus in wt/wt (white column, n = 6) or RRN293/RRN293 (gray column, n = 8) mice. (D) Bacterial survival 6 h after i.p. inoculation of 5 × 10⁹ CFU S. aureus in wt/wt (white column, n = 7) or RRN293/RRN293 (gray column, n = 7) B5 mice. Data represent mean values ± SEM; indicated P values are from Student's unpaired t test.

We next tested whether Hv1^−/− mice are able to clear bacterial infections in vivo. Hv1^−/− or wt/wt mice were inoculated i.p. with S. aureus, and surviving bacteria were elaborated on selective agar from peritoneal lavage fluid h 6–24 later. In mice inoculated with 1 × 10⁸ S. aureus we observed a >2-log unit decrease in bacterial CFU after a 24-h incubation in vivo (Fig. 4C). Whereas mice that received 1 × 10⁸ CFU of S. aureus did not visibly manifest sepsis, mice of either genotype that received 5 × 10⁹ CFU were typically moribund within 6 h of inoculation. Although bacterial load in the peritoneal lavage from Hv1^−/− mice suggested a trend toward decreased bacterial clearance, the data did not reach statistical significance. We also observed robust clearance of experimental pneumonia 24 h after intranasal inoculation of 1 × 10⁸ CFU Pseudomonas aeurginosa or Burkholderia cepacia in both wt/wt and Hv1^−/− mice. Although Hv1 deletion reduces Hv1 protein and G_vH+ to undetectably low levels, decreases O₂^−∙ production by ≈75%, and significantly attenuates bacterial killing in vitro, we were unable to see a significant effect on bacterial clearance in vivo under our experimental conditions.

Discussion

In summary, we demonstrate here that Hv1 is necessary for measurable G_vH+ in a variety of leukocytes, and previous studies showed that expression of Hv1 is sufficient to reconstitute the hallmark biophysical properties of G_vH+ in heterologous systems (1, 2, 5–8). Superoxide production by intact (but not broken) cells is substantially reduced in the absence of Hv1. The simplest interpretation of these results is that Hv1 serves to provide a majority of the charge compensation needed to support NADPH oxidase-mediated e⁻ export. Despite a significant reduction in superoxide production, wt/wt and Hv1^−/− mice were similarly able to clear S. aureus infections in vivo, but differences in innate immune function may become apparent under different experimental conditions.

Similar to human patients with chronic granulomatous disease, mice bearing targeted deletions or mutations of genes encoding NADPH oxidase components manifest symptoms (e.g., dramatically decreased or undetectable O₂^−∙ production and decreased bacterial clearance in vitro and in vivo) (33–36). Interestingly, mice lacking either gp91^phox (the product of the Nox2 gene) or inducible nitric oxide synthase (the Nos2 gene product) exhibited a similarly modest impairment in the innate immune response to S. aureus challenge. In vivo, Nox2^−/−/Nos2^−/− (double-knockout) mice required administration of oral antibiotics for survival even in the absence of bacterial challenge, suggesting that normal innate immune function relies on at least partially redundant systems for bacterial clearance (37). Thus, superoxide- and nitric oxide-dependent systems appear to function synergistically for bacterial clearance in vivo. Reconstitution of a small fraction of the total superoxide-generating capacity of the innate immune system in gp91^phox (Nox2^−/−) knockout mice was sufficient to significantly restore bacterial clearance, suggesting the presence of enormous excess capacity for NADPH oxidase-dependent superoxide production in normal granulocytes (38–40). Because O₂^−∙ generation was decreased ≈75% in Hv1^−/− mice, the lack of a significant in vivo impairment in bacterial clearance in these mice is perhaps not surprising. Hv1^−/− mice should be useful for investigating the physiological role of Hv1 in cells and tissues outside of the innate immune system in the future.

Materials and Methods

Hv1 Knockout Mice.

Mice lacking functional Hv1 protein were derived from the Bay Genomics genetrap of the Hvcn1 locus (dbGSS7585445; gi37500117). The exogenous genetrap vector sequence, containing an in-frame splice acceptor site followed by coding sequence for a β-galactosidase/neomycin (β-Geo) fusion protein, was inserted into a large (13.6 kb) intron following exon 2, which contains the Hv1 translation initiation codon. Chimeric male mice were derived from sequence-verified RRN293 genetrap ES cells and used to establish heterozygous wt/RRN293 founders; homozygous offspring were generated from heterozygous mating pairs (Mutant Mouse Regional Resource Center, Davis, CA). Hv1^−/− mice backcrossed against C57/BL6J for 5 generations are designated B5. All animal husbandry and experimental procedures were approved by and performed in accordance with guidelines of the Children's Hospital Boston Institutional Animal Care and Use Committee.

Mouse Genotyping.

The genomic locus of the Bay Genomics genetrap vector (pGTOLxf) insertion was identified by PCR using a series of forward primers spaced at 1- to 2-kb intervals in the predicted intronic Hvcn1 sequence, along with a β-Geo-specific reverse primer (β-GeoR: 5′-GACAGTATCGGCCTCAGGAAGATCG-3′) to amplify a product containing the pGTOLxf insertion site. DNA sequencing of the PCR product (Mental Retardation/Developmental Disabilities Research Center Molecular Genetics Core Facility at Children's Hospital) indicated that the pGTOLxf vector was inserted after nucleotide 13899 (of 17054) in predicted intron 3 (variant 1, Fig. S1A). Genotypes were confirmed by triplex PCR performed on genomic DNA isolated from mouse tails using 2 gene-specific primers (Hvcn1F: 5′-GAGATCCATCTGCCTCCGTTATGAGTG-3′; Hvcn1R: 5′-CATGGTCTCAGTGATGTTAGGACGTC-3′) and β-GeoR. The genotypes of mice used for breeding were determined by quantitative genomic PCR (Transnetyx) using primers specific to Hvcn1 intron 3 and pGT0Lxf; results from the Transnetyx genotyping strategy were verified by triplex genomic PCR. Although quantitative RT-PCR confirmed that the RRN293 transgene was expressed at the expense of the wt Hvcn1 allele (Fig. S1C), we were unable to detect either β-galactosidase protein [using several different anti-β-galactosidase antibodies in Western blot analysis, immunocytochemistry (ICC) or immunohistochemistry (IHC)] or activity (using X-gal staining) in spleen or BMC from Hv1^−/− mice.

Quantitative RT-PCR.

Total RNA (500 ng) extracted from mouse BMC (Rneasy; Qiagen) was reverse-transcribed by using an oligo(dT) primer (SuperScript 3; Invitrogen). Quantitative PCR (Real time PCR systems 7300; Applied Biosystems) was performed in a final volume of 25 μL containing 40 ng of cDNA, 300 μM primers (Hvcn1 Ex1 Fwd: 5′-TGCAAAGGAGTGCTGCAAACTA-3′; Hvcn1 Ex1 Rev: 5′-TCGAGTAGACGCTCCGCAAT-3′; Hvcn1 Ex2–3 Fwd: 5′-CTCTCACTGTTGGATCTTGAGACAA-3′, Hvcn1 Ex2–3 Rev: 5′-GGGAGCCACCTTGGTTCTG-3′; β-Geo Fwd: 5′-GGACGCGCGAATTGAATTA-3′, β-Geo Rev: 5′-CTGTTGACTGTAGCGGCTGATG-3′), and 12.5 μL of SybrGr PCR master mix (Applied Biosystems). Triplicate quantitative PCRs were performed by using BMC cDNA synthesized from wt/wt or RRN293/RRN293 mRNA isolated from 3 mice of each genotype. Normalized values for cycle threshold (ΔCt_wt/wt and ΔCt_RRN/RRN) were calculated by subtracting the observed cycle threshold (C_t) for each cDNA and primer pair from the C_t observed for L8 ribosomal RNA primers (2). Relative expression of β-Geo (Fig. S1, primers H and I) or Hvcn1. Relative expression of β-Geo mRNA (Fig. S1, primers H and I) in wt/wt BMC and Hvcn1 mRNA in RRN293/RRN293 BMC [Fig. S1, primers D and E (Hvcn1 Ex 1′, 1″) or primers F and G (Hvcn1 Ex 2–3)] was calculated by using Eqs. 1 and 2.

Eq. 1.

Fractional β-Geo expression in wt/wt = 2^{−ΔΔCtβ-Geo}, where ΔΔCt_β-Geo = ΔCt_wt/wt − ΔCt_RRN/RRN (primers H and I) and ΔCt_wt/wt = Ct_Hvcn1Ex1 − Ct_L8RNA and ΔCt_wt/wt = Ct_β-GeoRNA − Ct_L8RNA.

Eq. 2.

Fractional Hvcn1 expression in RRN293/RRN293 = 2^{−ΔΔCtHvcn1}, where ΔΔCt = ΔCt_RRN/RRN − ΔCt_wt/wt (primers D, E and F, G).

Isolation of Mouse Leukocytes.

For measurement of O₂^−∙ and H₂O₂, phagocytosis, and in vitro bacterial killing assays, BMC were dispersed by trituration through a 22-G needle in ice-cold Dulbecco's Ca²⁺- and Mg²⁺-free PBS (D-PBS) or Ca²⁺-free modified Tyrode's Solution (TS+EGTA: 138 NaCl, 5 KCl, 1 MgCl₂, 1 EGTA, 10 Hepes, pH 7.4 at 23 °C) and passed through a 40-μm mesh filter (Becton-Dickinson). Cells were centrifuged (250 × g, 10 min, 4 °C), resuspended to 5 × 10⁶/mL in TS, and kept on ice until use. In some experiments, whole blood (collected in 2–10 mM EGTA and diluted in D-PBS) or BMC were layered onto 1.5 vol PMN isolation medium (Robbins Scientific) or 0.75 vol Lympholyte M (Cedarlane), or before centrifugal sedimentation. For Western blot analysis, the mixture of PBL contained in the fractions above the erythrocyte pellet were removed, diluted into 20 vol ice-cold D-PBS containing 1 mM EGTA, pH 7.4 (PBS/EGTA), and centrifuged for 10 min at 250 × g. Differentiated HL-60 and COS-phox cells (gift of M. Dinauer, Indiana University School of Medicine) were cultured as described (1, 41).

Immunoblotting.

Membranes were prepared from cell pellets or tissues by homogenization (1–2 min; Polytron) in ice-cold homogenization solution, 150 mM NaCl, 10 mM Tris·HCl (pH 7.2), 1 mM EDTA containing 0.3 M sucrose and protease inhibitor mixture (Complete mini; Roche; mammalian cell protease inhibitor mixture; Sigma) and sonicated (3 × 5 s, 40% maximum power, Heat Systems–Ultrasonics W225) before centrifugation (10 min, 4,000 × g, 4 °C) to clear nuclei and particulate material. The supernatant was then centrifuged (30 min, 25,000 × g, 4 °C; Sorvall Ultrafuge) to pellet cell membranes. Membranes were solubilized by sonication (3 × 5 s, 25% maximum power) in 0.2 mL of ice-cold lysis buffer (TS+EGTA containing 1% Triton X-100 and protease inhibitors). Protein concentration was determined (Bradford, BioRad protein assay) and 30 μg per lane was loaded onto 4–12% Bis-Tris Novex Nu-PAGE gels (Invitrogen) for electrophoresis in Mops running buffer before transfer to a PVDF membrane. Western blots were incubated with rocking in PBST (D-PBS containing 0.5% Tween-20) blocking solution containing 5% nonfat dry milk (60 min, 24 °C). Blots were first incubated with an anti-actin (Chemicon) primary antibody (1:1,000, 2 h, 24 °C) followed by secondary antibody (goat anti-mouse HRP-conjugated IgG; Zymed; 1:100,000, 45 min, 24 °C). Immunoreactivity was detected by using ECL Advance (Amersham). Blots were stripped (20 min, 50 °C) in 67.5 mM Tris·HCl (pH 6.8), 100 mM β-mercaptoethanol, and 2% SDS and washed with PBST before sequential incubation with anti-Hv1 primary (4234, 2 h) (1) and secondary (goat anti-rabbit HRP-conjugated IgG; Zymed; 1:100,000, 45 min) antibodies.

Electrophysiology.

Mouse PBL or PMN were resuspended in either TMA6.5 [100 mM Bis·Tris (pH 6.5), 80 mM TMA-MeSO₃, 2 mM MgCl₂, 4 mM HCl, 1 mM EGTA] or Na6.5 [100 mM Bis·Tris (pH 6.5), 80 mM NaCl, 10 mM MgCl₂, 1 mM EGTA]. Cells were visually selected and subjected to whole-cell voltage clamp under symmetrical conditions (pipette solution same as bath solution). We also recorded currents from either crude BMC or Percoll gradient-purified BMC neutrophils; in no case did we measure voltage-dependent H⁺ current in cells isolated from RRN293/RRN293 mice.

O₂^−∙ and H₂O₂ Production Assay (Cuvette Format).

BMC were resuspended in modified Tyrode's solution [TS: 138 NaCl, 5 KCl, 1.8 CaCl₂, 1 MgCl₂, 10 Hepes (pH 7.4) at 23 °C] on ice. Phase-dark cells were counted (after 5 min at 24 °C) with a glass hemacytometer and diluted to 5 × 10⁶/mL in ice-cold TS. Cells (0.1× final volume) were added to individual cuvettes containing TS plus 100 μM cytochrome c (or 5 μM Amplex Ultra Red and 0.5 unit/mL horseradish peroxidase for H₂O₂ production assay) with or without DPI (10 μM) or ZnCl₂ (1 mM). O₂^−∙ and H₂O₂ production were measured at 550 nm, 37 °C, in 1.0 mL of cuvettes by using a temperature-controlled spectrophotometer fitted with an automated 6-cuvette sample changer (DU800; Beckman–Coulter). Medium containing assay reagent and indicated drugs was preincubated (37 °C) for 10 min; cells were added 5 min before addition of PMA (200 nM) or arachidonic acid (AA, 100 μM) and absorbance was measured in a final volume of 1.0 mL (cuvette assay). O₂^−∙ and H₂O₂ production in wt/wt and Hv1^−/− BMC were measured in parallel assays conducted simultaneously using the automated sample changer.

H₂O₂ Production Assay (96-Well Format).

BMC or HL-60 cells were centrifuged (250 × g, 10 min) and resuspended in TS on ice. Cells were counted and diluted to 5 × 10⁶/mL in ice-cold TS. Cells (20 μL) were added to individual wells containing ice-cold TS plus indicated assay reagent with or without PMA (200 nM), DPI (10 μM), or ZnCl₂ (1 mM). Assay plates containing added assay reagents and indicated drugs (minus PMA) were preincubated in a warm-air incubator (37 °C) for 10 min Cells were added and incubated (37 °C) for an additional 5 min. Plates were removed and t = 0 assigned as immediately after PMA addition. H₂O₂ secretion was measured at room temperature by using a plate-reading spectrophotometer/fluorimeter (550 ± 8-nm filter, Victor3; Wallac) in a final volume of 0.2 mL per well. All assay wells contained 5 μM Amplex Ultra Red (Invitrogen) and 0.5 unit/mL horseradish peroxidase (Zymed). In control experiments where PMA-stimulated O₂⁻ (cytochrome c reduction) and H₂O₂ (Amplex Ultra red) were measured side by side in a heated cuvette-reading spectrophotometer (1.0-mL assay volume), we observed only a small delay in H₂O₂ generation compared with O₂⁻ production, suggesting that the slow time course observed in some experiments (Fig. S2) was attributable to slow heating and reheating of the reagents and plates in the 37 °C warm air incubator between time points that were measured using the nonheated spectrophotometric plate reader.

In Vitro Bacterial Killing.

Aliquots of bacterial stocks (stored at −80 °C in 20% glycerol) on dry ice were scraped with a sterile needle and bacteria were spread on selective agar plates (S. aureus: mannitol, Molecular Toxicology; P. aeurginosa: cetrimide, PML Microbiologicals; B. cepacia: BCSA, PML Microbiologicals) and grown overnight at 37 °C. LB medium was subsequently inoculated and liquid cultures were grown with shaking at 37 °C to an OD₆₀₀ ≈ 0.5. The concentration of bacterial CFU was calculated (OD₆₀₀ = 0.5, CFU = 5 × 10⁵/μL) and verified for each bacterial preparation by serial dilution and plating onto selective agar plates. Bacteria were centrifuged (10 min, 1,000 × g, 4 °C) and resuspended and opsonized in mouse serum containing 0.05% Tween-20 (30 min with rotation, 24 °C). Opsonized bacteria were then diluted into TS containing 0.05% Tween-20 to 2 × 10⁴ CFU/μL. BMC were resuspended in ice-cold TS (10⁶ cells per 100 μL) and added to round-bottom tubes containing TS solution with or without DPI (10 μM). The assay was initiated by addition of bacteria (50 μL) and allowed to proceed for 30 min (37 °C, orbital shaking) before addition of gentamicin (50 μg/mL; Sigma) and centrifugation (10 min, 2,800 × g, 24 °C). Cell pellets were washed in TS + EGTA and lysed by resuspension in saponin (1% in water) and 1 freeze-thaw cycle. Lysates were serially diluted and plated onto selective LB agar for overnight culture (37 °C) for manual determination of bacterial CFU. Although DPI reproducibly inhibited killing of bacteria under these experimental conditions, a large fraction of the inoculum survived even in the presence of DPI (Fig. 4A), suggesting that mouse BMC may use NADPH oxidase-independent mechanisms for bacterial killing that are not prominent in purified, circulating human neutrophils and/or that incomplete cell lysis may have led to an underestimate of the DPI-dependent fraction of bacterial killing (42). Our results should therefore be interpreted as a lower limit on the effect of Hv1 deletion on bacterial killing by the NADPH oxidase system.

In Vivo Bacterial Clearance.

Bacteria were grown to OD_600nm ≈0.5, centrifuged (10 min, 1,000 × g, 4 °C), and resuspended in sterile 0.9% NaCl or D-PBS to a final concentration of 2 × 10⁸ CFU/mL (i.p.) or 2 × 10¹⁰ CFU/ml (intranasal) for inoculation. Mice were allowed to recover for 18–24 h after i.p. or intranasal inoculation before peritoneal (3 × 5 mL of D-PBS, 4 °C) or lung lavage (3 × 0.8 mL of D-PBS, 4 °C), respectively. Lavage fluid was centrifuged (10, 1,000 × g, 4 °C) and the pellet was resuspended in D-PBS containing 0.05% Tween-20 (PBST), then serially diluted in PBST before plating on selective agar plates for culture at 37 °C for 24–48 h.

Phagocytosis Assay.

BMC (1 × 10⁵ cells per well) were added to a 96-well plate and incubated in 0.2 mL of TS (60 min, 37 °C) to promote adherence. The supernatant was aspirated and replaced with prewarmed TS with or without DPI (10 μM) and the assay was initiated by addition of heat-inactivated Alexa Fluor 488-labeled E. coli K-12 Bioparticles (Invitrogen) resuspended in TS per the manufacturer's directions. After phagocytosis (120 min, 37° C), trypan blue solution (100 μL) was added to each well to quench fluorescence of nonphagocytosed extracellular bacterial particles and quickly (1 min, 24 °C incubation) aspirated. Fluorescence of phagocytosed bacterial particles was measured on a plate-treading fluorimeter (Wallac Victor3; excitation 488 nm, lowpass filter 520 nm).