Dual role of phagocytic NADPH oxidase in bacterial killing

The classical model of bacterial killing by phagocytic cells has been recently challenged by questioning the toxic effect of oxygen products and attributing the fundamental role to K⁺ ions in releasing antimicrobial proteins within the phagosome. In the present study we followed

\(\mathrm{O}_{2}^{{\cdot}-}\)

production, changes of membrane potential, K⁺ efflux, and bacterial killing in the presence of increasing concentrations of the nicotinamide adenine dinucleotide phosphate (NADPH) oxidase inhibitor diphenylene iodonium. Efficiency of bacterial killing was assessed on the basis of bacterial survival measured by a new semiautomated method. Very low rates of

\(\mathrm{O}_{2}^{{\cdot}-}\)

production were accompanied by significant membrane depolarization and K⁺ release and parallel improvement of bacterial killing. When

\(\mathrm{O}_{2}^{{\cdot}-}\)

production exceeded 20% of its maximal capacity, no further change was detected in the membrane potential and only minimal further K⁺ efflux occurred, yet bacterial survival decreased parallel to the increase of

\(\mathrm{O}_{2}^{{\cdot}-}\)

production. The presented results indicate that both electrophysiological changes (depolarization and consequent ion movements) and the chemical effect of reactive oxygen species play a significant role in the killing of certain pathogens. The observation that an increase of membrane depolarization can compensate for decreased

\(\mathrm{O}_{2}^{{\cdot}-}\)

production may be important for potential therapeutic applications.

Phagocytic cells play a fundamental role in the antimicrobial defense by engulfing and killing various potentially pathogenic microorganisms. Patients with chronic granulomatous disease (CGD) suffer repeated, often fatal infections due to severe impairment of the antimicrobial defense mechanisms.¹  The discovery that CGD is the consequence of a genetic defect in one of the essential subunits of the

-producing nicotinamide adenine dinucleotide phosphate (NADPH) oxidase provided strong support for the role of

\(\mathrm{O}_{2}^{{\cdot}-}\)

and other reactive oxygen species (ROS) in the mechanism of bacterial killing.² 

When activated, NADPH oxidase transfers electrons from intracellular NADPH to extracellular (or intraphagosomal) O₂, forming

anions. The enzyme carries out an electrogenic function (ie, with each turn one negative charge is removed from the cell).³  This charge movement is reflected in depolarization of the plasma membrane that was shown to accompany

\(\mathrm{O}_{2}^{{\cdot}-}\)

production. In fact, intensive

\(\mathrm{O}_{2}^{{\cdot}-}\)

generation is able to reverse the polarity of the plasma membrane: in resting polymorphonuclear neutrophils (PMNs), values around -60 mV have been measured,⁴  whereas upon phorbol myristate acetate (PMA) stimulation the membrane potential approached +60 mV.⁵  The electromotive force of the enzyme is similar to that of mitochondrial cytochromes: electron current via the phagocytic NADPH oxidase ceases only around +200 mV.⁶ 

Decrease or reversal of the plasma membrane potential alters the driving force for all other charged particles, favoring outward movement of cations and inward movement of anions. Indeed, compensating H⁺ efflux has been demonstrated by many laboratories^7-12  although the underlying transport pathway(s) is still the subject of intensive investigations.⁴  More recently K⁺ efflux has also been reported from activated PMNs.^13,14  On the other hand, plasma membrane depolarization impedes Ca²⁺ influx induced either by store depletion or by stimulation of chemotactic receptors.^15,16 

In view of all these ion movements initiated by the charge separation via the NADPH oxidase, we had proposed that alteration of the ionic composition of the intraphagosomal space may contribute to the impairment of bacterial killing in CGD.¹⁷  Recently Reeves and coworkers questioned the toxic effect of various ROS under the conditions of the phagosome.¹⁸  Instead, they suggested a specific role for K⁺ ions enriched in the intraphagosomal space in the release of granule enzymes from their polyanionic support and thereby in their activation. In fact, this hypothesis reduces the role of

generation to providing a driving force for K⁺ movement into the phagosomes.^13,14 

The aim of the present study was to gain insight into the relative importance in bacterial killing of (1)

as a reactive molecule and (2) membrane potential changes initiated by electron transfer via the

\(\mathrm{O}_{2}^{{\cdot}-}\)

-generating NADPH oxidase. Our approach consisted of a stepwise decrease of the rate of

\(\mathrm{O}_{2}^{{\cdot}-}\)

production by applying increasing concentrations of the inhibitor diphenylene iodonium (DPI) and relating

\(\mathrm{O}_{2}^{{\cdot}-}\)

generation to changes of the plasma membrane potential, K⁺ release, and bacterial killing. Our quantitative analysis provides experimental support that both

\(\mathrm{O}_{2}^{{\cdot}-}\)

and ion movements contribute to the killing of some microorganisms (eg, Staphylococcus aureus), whereas in other cases (eg, Escherichia coli) none of them seem to be decisive.

Ficoll was from Pharmacia (Peapack, NJ); 3-3′-dipentyloxacarbocyanine

from Molecular Probes (Eugene, OR); saponin from Calbiochem (Darmstadt, Germany); and valinomycin, cytochrome c, lucigenin, cytochalasin B, phorbol myristate acetate (PMA), formyl-methionyl-leucylphenylalanine (fMLP), and DPI, from Sigma (St Louis, MO). ⁸⁶RbCl was purchased from the Izotóp Intézet (Budapest, Hungary). All other reagents were of research grade. The extracellular medium (called H-medium) contained 145 mM NaCl, 5 mM KCl, 1 mM MgCl₂, 0.8 mM CaCl₂, 10 mM HEPES (N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid), 5 mM glucose, pH 7.4. The Luria-Bertani (LB) medium contained 10 g/L tryptone, 5 g/L yeast extract, 80 mM NaCl, and 1 mM NaOH.

Human neutrophils were obtained from blood of healthy volunteers by dextran sedimentation followed by Ficoll-Paque gradient centrifugation using endotoxin-free solutions and sterile conditions at 4°C to prevent premature activation of the cells. Contaminating red blood cells were removed by hypotonic lysis. Cells were finally resuspended in sterile endotoxin-free H-medium and kept on ice until use. Preparations contained more than 95% neutrophils; their viability, as determined by erythrosin B dye exclusion, exceeded 95%.

Superoxide production of the cells was tested with the superoxide dismutase-inhibitable cytochrome c reduction or the lucigenin-based chemiluminescence. To measure superoxide production in the extracellular medium with cytochrome c, cells (10⁶/mL) were suspended in H-medium containing 100 μM cytochrome c. Control samples contained 12.5 μg/mL superoxide dismutase. Aliquots (200 μL) of the suspension were added into wells of a 96-well plate and prewarmed at 37°C for 5 minutes in a shaking enzyme-linked immunosorbent assay (ELISA) reader (Labsystems iEMS Reader MF, Helsinki, Finland). The cells were stimulated with the appropriate stimuli by the addition of 5 μL of the stimulus solution. The changes in the absorption were recorded at 550 nm for 10 minutes with 2 measurements per minute at 37°C with gentle shaking. After subtracting the background values, superoxide production was calculated with the use of an absorption coefficient of 21 mM^-1cm^-1 for cytochrome c.

To measure reactive oxygen species production with lucigenin-based chemiluminescence, the cells (10⁷/mL) were suspended in H-medium containing 5.1 mg/mL lucigenin (dissolved in dimethyl sulfoxide [DMSO]). Aliquots (200 μL) of the suspension were added into wells of a 96-well plate and prewarmed at 37°C for 5 minutes in a shaking Fluoroskan Ascent FI luminometer (Labsystems). Cells were stimulated with the appropriate stimuli by the addition of 5 μL of the stimulus solution. Changes in the luminescence were recorded for 20 minutes at 37°C with gentle shaking. In 6 independent experiments linear correlation was obtained between values of cytochrome c reduction and chemiluminescence; thus, relative fluorescence units (RLUs) were converted to nanomoles of superoxide on the basis of parallel measurement of PMA-stimulated

production by cytochrome c reduction.

Changes in the membrane potential were followed by the potential-sensitive fluorescent dye 3-3′-dipentyloxacarbocyanine,

⁠, as described by Seligmann and Gallin.¹⁹  Cells (10⁶/mL) were suspended in H-medium in a stirred cuvette at 37°C, and the changes in fluorescence were followed in a Deltascan dual-wavelength spectrofluorimeter (Photon Technology International, South Brunswick, NJ) with 484 nm as excitation and 510 nm as emission wavelengths. The cells were preincubated for 5 minutes before 100 nM

\(\mathrm{di-O-C}_{5}^{3}\)

was added. Fluorescence reached a constant value in 6 to 7 minutes. Thereafter, the cells were stimulated with different stimuli (fMLP, PMA), and the changes in the fluorescence were monitored. The obtained fluorescence data were calibrated to membrane potential values (millivolts [mV]) using 2 μg/mL valinomycin in a suspension of cells with different external K⁺ concentrations as described by Geiszt et al.¹⁵  The values attained at 1 or 2 minutes after stimulation with fMLP or PMA, respectively, were used for further calculations.

Neutrophils (10⁷/mL) were incubated in H-medium with ⁸⁶RbCl (0.25 μCi [0.0093 MBq]) for 30 minutes at 37°C, washed twice in cold phosphate-buffered saline (PBS), and resuspended in H-medium at the density of the killing assay (10⁷/mL). After incubation with or without stimuli at 37°C for 5 minutes, the cells were immediately centrifuged (4 minutes, 500g). Radioactivity was determined in an aliquot (100 μL) of the cell-free supernatant using an Automatic Gamma Counter (1470 Wizrad; Wallac, Turku, Finland). The ⁸⁶Rb efflux detected in the unstimulated samples was subtracted from the values of the stimulated ones. The ⁸⁶Rb release of the PMA-stimulated, DPI-free sample was taken as 100%. DPI by itself did not induce any ⁸⁶Rb efflux from the cells.

Bacteria (9 × 10⁸ cells in 900 μL H medium) were opsonized with 100 μL mixed human blood serum of at least 5 different donors for 5 minutes at 37°C. PMNs (9 × 10⁶ in 900 μL H-medium) were then incubated with 100 μL (10⁸/mL) opsonized bacteria for 30 minutes at 37°C. Samples (100 μL) were taken at indicated time points and lysed in 900 μL ice-cold H-medium containing 1 mg/mL saponin. In the case of S aureus, additional treatment at -80°C for 20 minutes was applied. Freezing did not impair subsequent growth of S aureus.

Bacterial growth was followed in a plate reader (Labsystems iEMS Reader MF) on the basis of changes in optical density, taking into account that in the exponential phase of bacterial growth the division time is constant. For calculations we applied the principles of real-time polymerase chain reaction (PCR); thus we characterize the samples containing unknown amount of living bacteria with an incubation time (t_inc) the length of which is only dependent on the initial bacterial concentration. Specifically, lysed PMN samples were diluted in LB to decrease the initial optical density (OD) (typically a 5-fold dilution to a final volume of 200 μL per well was applied). A parallel series of 1:2 dilutions from the stock bacterial suspension (10⁸/mL) was also prepared in LB. Then bacteria were grown in a shaking plate reader at 37°C for 5 to 8 hours, and the OD was followed continuously at 650 nm. Samples were run in 4 to 8 parallels. The data obtained in a typical experiment are shown in Figure 1A. The time required for achieving a certain OD value (in most experiments 0.2 was chosen) was determined for each dilution and plotted as initial bacterial count versus incubation time (t_inc). A straight line was fitted on the experimental points (Figure 1B), and the number of surviving bacteria was calculated with the following formula: E coli or S aureus =

⁠, where t_inc indicates the incubation time required to reach the threshold OD value, f is a numerical factor, and g is the slope of the line, indicating the rate of bacterial reproduction. This calibration was then used to determine the number of viable bacteria in the lysed samples taken in the killing assay. Each experiment was calculated on the basis of its own calibration curve. The slope of the curve shown in Figure 1B allows the determination of the average division time of the investigated microorganism. Summarizing all experiments, 17.1 ± 1.91 minutes (SEM, n = 29) and 28.6 ± 1.1 minutes (SEM, n = 63) were obtained as the division times for E coli and S aureus, respectively. Thus, experiments carried out on different days showed high reproducibility.

Killing activity of neutrophils was assessed on the basis of the changes in the number of viable bacteria. Because the 30-minute incubation time with PMNs exceeded the division time of both bacteria under some experimental conditions, the final bacterial count was higher than the initial one. In Figures 1, 5, and 6, “bacterial survival” is represented as the final bacterial count expressed as percentage of the initial count.

Bacterial growth in the central or peripheral wells of the plate differed by less than 2%, and it was not influenced by the presence of nonviable bacteria or saponin-treated PMNs. The utilized concentration of saponin and dilutions of DMSO or the opsonizing serum had no effect on the growth of bacteria. DPI (0 to 5000 nM) was without effect on the viability and rate of reproduction of S aureus but inhibited E coli. Therefore, the lysed samples of the E coli killing assays using DPI were diluted 1:500 instead of 1:5. In this dilution DPI had no inhibitory effect on E coli cells even in the case of 5000 nM in the killing suspension. Varying the ratio of bacteria to PMNs between 1 and 10 had no significant influence on the killing efficiency in the presence of various DPI concentrations. Figure 1 summarizes the results of a typical experiment where bacterial survival was compared by the semiautomated plate reader assay (Figure 1C) and conventional counting of colony-forming units (CFUs) on LB agar plates (Figure 1D). In the latter experiments, the lysed samples have been diluted further, 1:2500, in cold H-medium, and 100 μL was smeared on Petri dishes containing LB agar and cultured overnight at 37°C. The CFUs have been counted and presented as the percentage of the initial CFU number. Disruption of the microfilament structure by cytochalasin B is known to block phagocytosis and, hence, killing. In a concentration of 10 μM, cytochalasin B inhibited bacterial killing to similar extent, independent of the applied method.

Data are presented either as representative traces of the indicated number of experiments or as the mean ± SEM of the number of determinations indicated (n).

Neutrophilic granulocytes were stimulated either with phorbol myristate acetate (PMA), an activator of protein kinase C, (Figure 2A-C) or with the chemotactic peptide formyl-methionyl-leucylphenylalanine, fMLP (Figure 2D-F). Changes of membrane potential and

production were measured in parallel samples.

\(\mathrm{O}_{2}^{{\cdot}-}\)

production induced by PMA (Figure 2A) was detectably inhibited by 50 nM DPI, and the inhibition was almost complete in the presence of 5 μM DPI. In contrast to

\(\mathrm{O}_{2}^{{\cdot}-}\)

production, membrane depolarization following PMA stimulation was not influenced by 50 nM DPI, and even in the presence of 5 μM inhibitor there was only a 50% decrease in PMA-induced depolarization (Figure 2B). Data of 11 experiments are summarized in Figure 2C, where both measured parameters are expressed in percentage of the values obtained in the absence of any inhibitor. The discrepancy in the sensitivity of

\(\mathrm{O}_{2}^{{\cdot}-}\)

production and membrane depolarization toward DPI is remarkable. In the case of fMLP stimulation, both the onset, the extent, and duration of

\(\mathrm{O}_{2}^{{\cdot}-}\)

production and also the kinetics of membrane depolarization are different from those evoked by PMA (Figure 2D-E). Nevertheless, fMLP-initiated

\(\mathrm{O}_{2}^{{\cdot}-}\)

generation was also inhibited at a significantly lower DPI concentration than the accompanying membrane potential change (Figure 2F). DPI did not interfere with accumulation of the potential-sensitive fluorescent dye in the cells or alter the linearity of the calibration curve obtained in solutions of different K⁺ concentrations. The H⁺ ionophore carbonylcyanide M-chlorophenylhydrazone (CCCP) completely prevented depolarization when it was added before the stimulus and reversed the fluorescence signal when it was applied after the stimulus-induced depolarization had taken place (data not shown). Notably, the same fluorescent dye was applied in our previous experiments, where we have shown that neither PMA nor fMLP induced any detectable change in the membrane potential of neutrophil granulocytes from CGD patients¹⁵  or in gp91^phox-/- PLB-985 cell lines.¹⁶  Thus, we interpret the fluorescence signal detected in the presence of 5 μM DPI as depolarization induced by the remaining low rate of

\(\mathrm{O}_{2}^{{\cdot}-}\)

production. In a few experiments PMA- or fMLP-induced increase of O₂ consumption has also been measured. The inhibitory effect of 5 μM DPI was similar to that determined by the cytochrome c assay (data not shown).

Further analysis of the data summarized in Figure 2 emphasizes the nonlinear relation between the rate of

production and the change in the membrane potential induced thereby. As shown in Figure 3, less than 5% of the maximal

\(\mathrm{O}_{2}^{{\cdot}-}\)

-generating capacity is sufficient to induce 50% of the maximal depolarization. On the other hand, near-maximal depolarization is attained at 20% of the maximal

\(\mathrm{O}_{2}^{{\cdot}-}\)

-producing capacity, and further increase of the rate of

\(\mathrm{O}_{2}^{{\cdot}-}\)

production does not initiate significant changes in the membrane potential. The rather stable membrane potential prevailing under conditions when the rate of

\(\mathrm{O}_{2}^{{\cdot}-}\)

production increases from 20% to 100% maximal capacity indicates that in this phase electron transport via the NADPH oxidase is fully compensated by movement of another ion(s).

In previous investigations H⁺ and K⁺ were suggested as cations potentially compensating for the electron efflux via the oxidase. In the last decade ample information has been gained on the H⁺ conductance of phagocytic cells under resting and activated conditions^7,10-12,20  whereas the importance of K⁺ movements has been emphasized recently.^13,14  To gain further insight into the relation of

generation and K⁺ release, we measured the effect of DPI upon PMA-induced efflux of ⁸⁶Rb, the tracer generally used for assessing K⁺ movements.^13,14  As shown in Figure 4A, at a concentration of 50 nM, DPI had no measurable effect on ⁸⁶Rb release, and even in the presence of 5 μM DPI it decreased only to 38%. Relating ⁸⁶Rb efflux to

\(\mathrm{O}_{2}^{{\cdot}-}\)

production measured in parallel (Figure 4B) revealed that less than 5% of

\(\mathrm{O}_{2}^{{\cdot}-}\)

production is accompanied by the release of 38% of the mobilized K⁺ and that only 30% of K⁺ movement takes place in the phase where the rate of

\(\mathrm{O}_{2}^{{\cdot}-}\)

generation increases from 20% to 100% of maximal capacity. The relation is very similar to that obtained between

\(\mathrm{O}_{2}^{{\cdot}-}\)

production and membrane potential changes (compare Figure 4B with Figure 3). In fact, plotting ⁸⁶Rb efflux against membrane potential changes revealed a nearly linear relationship (Figure 4C), confirming the data obtained by the fluorescent technique.

In the next experiments bacterial survival was measured by a new, highly effective semiautomated method using a microtiter plate reader (see “Materials and methods” and Figure 1). Intracellular ROS production occurring during bacterial phagocytosis and killing was followed in parallel by the lucigenin-based chemiluminescence method in a luminescent plate reader. DPI inhibited the chemiluminescence signal induced by either E coli (Figure 5A) or S aureus (Figure 5B) with similar efficiency as observed in the cases of

production induced by soluble stimulants. Killing of E coli was only weakly dependent on

\(\mathrm{O}_{2}^{{\cdot}-}\)

production. In the absence of DPI, 14.5% ± 3.8% of the bacteria survived the 30-minute incubation with PMNs, and this value increased to 26.1% ± 7.2% in the presence of 5 μM DPI. In contrast, survival of S aureus depended significantly on the rate of

\(\mathrm{O}_{2}^{{\cdot}-}\)

generation. In the presence of 5 nM DPI, when

\(\mathrm{O}_{2}^{{\cdot}-}\)

production was still at 80% of the maximal capacity, bacterial survival was already increased 3-fold. Most significantly, when 5 μM DPI was present during the 30-minute incubation with PMNs, instead of decreasing, the bacterial count was doubled (Figure 5B). Other studies indicated that 5 μM DPI does not decrease the number of engulfed bacteria (W. Nauseef, personal communication, February 2004). Thus, the increase in survival must be interpreted as impairment of the killing process.

Analyzing the survival of S aureus in the function of the rate of

generation (Figure 6) reveals a multiphase relation. Decreasing the rate of

\(\mathrm{O}_{2}^{{\cdot}-}\)

production to 20% of the maximal capacity results in 90% bacterial survival. (Because bacteria multiply during the assay, this value still indicates elimination of bacteria by neutrophils, but evidently the killing function is seriously impaired.) It should be recalled that between 20% and 100% of maximal

\(\mathrm{O}_{2}^{{\cdot}-}\)

-generating capacity the membrane potential is stable and only minor K⁺ efflux takes place. Thus, impairment of bacterial killing occurs under conditions where electron flow via the assembled NADPH oxidase is completely compensated (mostly by H⁺ ions); consequently, the driving force for other ion movements remains unchanged. We interpret the parallel increase in killing efficiency and rate of

\(\mathrm{O}_{2}^{{\cdot}-}\)

production in the presence of a stable membrane potential as an indication for the chemical effect of

\(\mathrm{O}_{2}^{{\cdot}-}\)

and the products formed thereof.

Figure 3 revealed a very sharp relation between membrane potential change and

production when the latter fell below 20% of the maximal capacity. This relation is reflected in the survival of S aureus also (Figure 6). Decreasing the

\(\mathrm{O}_{2}^{{\cdot}-}\)

generation from 20% to approximately 5% of the maximal capacity results in an increase of bacterial survival from 90% to 140%. Finally, a decrease of

\(\mathrm{O}_{2}^{{\cdot}-}\)

production by only 2% (from 5% to 3%) leads to further increase in bacterial survival (from 140% to 200%).

To investigate further the importance of depolarization versus

in the killing of S aureus by neutrophils, we tried to uncouple

\(\mathrm{O}_{2}^{{\cdot}-}\)

generation and membrane potential change. Unfortunately, all the agents generally used for depolarizing cells, such as Na⁺ ionophores or lipophilic cations, are toxic for bacteria. Zn²⁺ ions were shown to inhibit the electrogenic H⁺ pathway already at low concentration,²¹  and in our pilot experiments they did not impede bacterial growth. Similar to previous findings,^6,20  Zn²⁺ increased the membrane depolarization by 31.7% ± 7.1% and decreased the rate of

\(\mathrm{O}_{2}^{{\cdot}-}\)

generation detected upon stimulation of neutrophilic granulocytes by 30.8%± 8.7%. The rate of

\(\mathrm{O}_{2}^{{\cdot}-}\)

production was slightly lower in the presence of 10 μM ZnSO₄ than in the presence of 5 nM DPI (69.2% ± 8.7% versus 80.4% ± 2.4% of the control value, n = 5). However, unlike DPI, Zn²⁺ did not impair bacterial killing. In fact, bacterial survival was slightly but consistently lower than the control value: in the absence of any inhibitor 6.7% ± 6.2%, in the presence of 5 nM DPI 21.3% ± 8.3%, whereas in the presence of 10 μM ZnSO₄ 3.5% ± 0.53% (n = 10) S aureus survived. Apparently, differences in the extent of membrane depolarization (ie, in the driving force for K⁺ efflux) are able to modify the efficiency of bacterial killing independent of the rate of

\(\mathrm{O}_{2}^{{\cdot}-}\)

generation.

Our experiments revealed a nonlinear relation between

production on one hand and membrane potential change or K⁺ release or bacterial killing on the other hand. At very low intensity,

\(\mathrm{O}_{2}^{{\cdot}-}\)

generation is accompanied by significant depolarization and K⁺ efflux: 38% of total K⁺ release takes place when

\(\mathrm{O}_{2}^{{\cdot}-}\)

production proceeds at less than 5% of its maximal capacity. Because K⁺ movement accounts only for 5% to 6% of total charge compensation,¹³  it can be calculated that in this initial phase roughly half of the electron efflux is compensated by parallel K⁺ efflux. Our experiments indicated -60 mV as the membrane potential in resting neutrophils (Figure 2), and this value corresponds well to earlier findings.⁴  The closeness of the resting membrane potential to the calculable equilibrium potential for K⁺ (-80 mV) indicates that under resting conditions K⁺ conductance dominates. This conclusion is further supported by our observation that in the presence of 5 μM DPI, the H⁺ transport inhibitor Zn²⁺ had no detectable influence on the membrane potential (data not shown). However, the relatively high K⁺ conductance of the plasma membrane of neutrophilic granulocytes prevailing at rest is still limiting the rate of K⁺ movements and results in the abrupt change of the membrane potential upon initiation of

\(\mathrm{O}_{2}^{{\cdot}-}\)

production. In our experiments (Figure 3) approximately 50% of the maximal depolarization occurs in this phase; thus, a change by 30 to 50 mV can be estimated. Elimination of bacteria is severely impeded at this low rate of

\(\mathrm{O}_{2}^{{\cdot}-}\)

production, as survival of S aureus attained 200%.

At moderate rate of

production (between 5% and 20% of the maximal capacity) both the membrane potential change and K⁺ release are attenuated. Electrogenic H⁺ efflux pathways were shown to be potential sensitive and exhibit a threshold for opening that is, depending on the exact conditions, 20 to 40 mV more positive than the resting membrane potential.^10-12  With channel-opening processes being statistical events, probably in this range of

\(\mathrm{O}_{2}^{{\cdot}-}\)

generation H⁺-conducting pathways open gradually, and charge compensation for electron transfer is taken over more and more by H⁺ efflux. Bacterial killing capacity is significantly increased as indicated by the drastic fall of survival of S aureus from 200% to 90%. Assuming that the proportion of oxidase molecules to ion transporters does not basically differ between the plasma and the issuing phagosomal membrane, we related bacterial survival to the membrane potential change and to K⁺ efflux. Both relations resulted in a straight line (data not shown), supporting the concept that depolarization and ion movements initiated by the electrogenic oxidase may have a decisive role in the killing process. Of note, membrane potential and K⁺ efflux exhibited a linear relationship (Figure 5C), indicating constant K⁺ conductance in the measured range of membrane potential (approximately between -20 and +60 mV). This observation can be accorded with the recent report on the appearance of an outward current at -30 mV in PMA-activated neutrophils,¹⁴  although many details of these latter experiments have been criticized.²² 

Finally, when

production exceeds 20% of the maximal capacity, alteration of the membrane potential becomes minimal and the remaining K⁺ efflux accounts only for approximately 1% to 2% of the electron efflux. In activated eosinophils, under current-clamped conditions it was shown that the value of the membrane potential followed the equilibrium potential of H⁺ ions, both in the absence and in the presence of K⁺ in the pipette solution.²⁰  These findings indicate that in the case of fully activated oxidase, the conductance for protons largely exceeds the conductance of any other ions. Thus, at high rate of

\(\mathrm{O}_{2}^{{\cdot}-}\)

generation protons serve as dominant charge-compensating ions. At the stable membrane potential no dramatic change occurs in the driving force for any other mobile ion; thus, no serious change in the ion composition of the phagosome can be expected except for an increase of

\(\mathrm{O}_{2}^{{\cdot}-}\)

itself and of the compensating H⁺ ions. On the other hand, survival of S aureus diminishes from 90% to 5% as

\(\mathrm{O}_{2}^{{\cdot}-}\)

generation increases from 20% to 100% capacity, indicating a significant increase in the bacterial killing activity of neutrophils. Under the apparently stable electrophysiological conditions of the phagosome, an enhancement of the killing capacity developing parallel to the increase in [

\(\mathrm{O}_{2}^{{\cdot}-}\)

] and [H⁺] must be ascribed to the chemical effect of these products (plus their derivatives).

In our experiments the variable parameter was the rate of

production gradually decreased by increasing doses of the inhibitor DPI. During phagocytosis, the observations detailed in the previous paragraphs may correspond to different phases following the onset of

\(\mathrm{O}_{2}^{{\cdot}-}\)

production initiated by particular stimuli. We suggest that at the beginning of the process K⁺ efflux dominates, causing a rapid increase in [K⁺] inside the phagosome and allowing for the temporary alkalinization of the intraphagosomal space.^14,23  As the process progresses, H⁺ efflux takes over and contributes to the acidification of the phagosomal milieu observed in earlier experiments.^14,23  Furthermore, the experimental condition in the presence of high DPI concentration is relevant to the pathological situation when neutrophils become activated in an environment of very low oxygen tension, such as inside abscesses or in poorly perfused regions. When the availability of oxygen limits the rate of electron transfer via the NADPH oxidase, membrane depolarization and consequent ion movements may become critical in the battle with the invading pathogen.

Taken together, the presented data provide experimental support for our earlier hypothesis¹⁷  that both depolarization and secondary ion movements initiated by electron transfer via the NADPH oxidase, and the produced

and its derivatives as chemical compounds, participate in the killing of certain bacteria. The significance and contribution of the 2 mechanisms to the entire killing process may depend on the exact conditions, such as availability of oxygen or the species of the invading microorganism. The experiments carried out in the presence of the H⁺ channel inhibitor Zn²⁺ suggest that an increase of the membrane potential change may compensate for a decrease of

\(\mathrm{O}_{2}^{{\cdot}-}\)

production. Improvement of bacterial killing by modification of ion channel activity would be a basically new concept in the therapy of infectious diseases.

The authors are indebted to Profs D. Roos and P. Enyedi and Dr A. Mócsai for critical reading of the manuscript and to Ms Erzsébet Seres-Horváth, Edit Fedina, and Krisztina Sütő for expert and devoted technical assistance.