Cytosolic pH and the inflammatory microenvironment modulate cell death in human neutrophils after phagocytosis

Neutrophil phagocytosis is an essential and generally effective host response to microbial invasion. However, during necrotizing infections and in abscesses, neutrophils are ineffective at killing bacteria and instead contribute to host tissue destruction. Both intracellular and extracellular factors might contribute to dysregulation of neutrophilic inflammation in such situations, although these are not well characterized. pH may be important in this regard because it has the capacity to modulate neutrophil function. Thus, we investigated the mechanisms of intracellular pH (pH_i) regulation following phagocytosis, under conditions simulating the inflammatory microenvironment, and assessed how pH_i might contribute to the regulation or dysregulation of acute inflammation by altering mechanisms of neutrophil death.

pH_i homeostasis is uniquely challenged in phagocytosing neutrophils because cellular activation generates sufficient acid equivalents to lethally acidify the cell, and extracellular pH (pH_o) at inflammatory sites is often decreased.1,2 The balance between cytosolic proton loading and extrusion is important because many neutrophil functions, including microbicidal behavior, cell migration, intracellular oxidant generation, tumor cell cytotoxicity, and azurophil granule exocytosis, are pH dependent.3-8 Mechanisms potentially contributing to pH_i regulation in neutrophils include Na⁺/H⁺ exchangers (NHEs), proton-translocating adenosine triphosphatases (V-ATPases), and reduced nicotinamide adenine dinucleotide phosphate (NADPH) oxidase–associated proton conductance channels.9-11 Although NHEs are activated following phagocytosis in neutrophils,12 it is not known whether additional mechanisms also contribute to proton extrusion, or indeed whether they are collectively sufficient to preserve pH_iclose to physiologic levels, under conditions of increased proton loading. Further, the combined effect on pH_i regulation of mediators in the inflammatory milieu could also be important in vivo, although it is poorly characterized. Although some soluble inflammatory mediators activate proton extrusion in vitro,13-16others, such as lipopolysaccharide (LPS), may impair it.17Thus, in the first series of experiments reported here, we investigated which pumps and channels are involved in pH_i regulation after phagocytosis and tested how their function is affected by proinflammatory mediators from the lungs of cystic fibrosis (CF) patients and healthy subjects. CF epithelial lining fluid (ELF) is an attractive model of an inflammatory microenvironment for such studies because its pulmonary phenotype is characterized by intense, persistent inflammation and neutrophil-mediated parenchymal destruction.

pH_i and the inflammatory microenvironment may also affect survival of phagocytosing neutrophils by modulating rates of apoptotic and necrotic cell death, both of which may occur in response to bacterial exposure in vitro.18-20 This distinction is important because the former is a protective, orderly involutionary process, whereas the latter likely contributes to disease by exposing host tissues to destructive neutrophil contents.21pH_i and pH_o directly affect apoptosis in many cell types,22-30 whereas severely acidic pH_imight nonspecifically injure cells and thereby promote necrosis.31 Because phagocytosing neutrophils exhibit marked alterations in metabolic acid generation and are subject to variable pH_o, it is surprising that the relationship between pH_i and cell death in neutrophils has not been characterized more rigorously. The literature in this regard is, indeed, conflicting, as different studies have implicated acidic pH_i in both stimulation and inhibition of neutrophil or myeloid cell apoptosis.6,13,32,33 Thus, we addressed the consequences of altered pH_i and pH_o on cell survival in neutrophils exposed to increasing bacterial loads. We focused on pH_i changes over the first 90 minutes to probe early metabolic factors implicated in triggering these processes. Further, because soluble inflammatory mediators affect granulocyte apoptosis, we also evaluated the effect of bronchoalveolar lavage (BAL) fluid from CF patients and healthy subjects on neutrophil survival.

We show that pH_i following phagocytosis is not preserved by active pH_i regulatory mechanisms when neutrophils are exposed to a high-density bacterial challenge, or when neutrophils are cultured at low pH_o, although proton extrusion is elevated by mediators present in the inflammatory milieu. Further, we demonstrate that changes in pH_i/pH_o and the inflammatory microenvironment determine mechanisms of cell death in phagocytosing neutrophils. Our results illustrate that a precarious balance exists in vivo between cytosolic proton loading and extrusion, and that the resultant change in pH_i is a significant determinant of cell survival and mode of death.

The institution's ethical review board approved the studies. Subjects gave informed consent (14 healthy individuals). Neutrophils were isolated from heparinized peripheral venous blood by density gradient centrifugation, as previously described.34 The neutrophil population was at least 96% pure by morphologic assessment (Cytospin preparations and Giemsa staining). Viability was at least 99% by trypan blue dye exclusion (light microscopy) and propidium iodide (PI) exclusion (flow cytometry). Cells were cultured in RPMI-1640 containing 25 mM NaHCO₃ (Sigma Cell Culture; Poole Dorset) and freshly supplemented withl-glutamine 0.3 g/L (pH 7.35; Sigma Cell Culture). In additional experiments, cells were cultured under HCO$3−$-free conditions in a HEPES (N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid)-buffered medium with pH adjusted to 5.8 with HCl (Sigma Cell Culture). pH_i was measured by flow cytometry (FACScan; Becton Dickinson, Mountain View, CA) using the pH-sensitive fluorescent dye carboxy-SNARF, as previously described.34Briefly, carboxy-SNARF–loaded cells were stimulated with argon laser light (488 nm), and the ratio of emitted fluorescence at 580 nm and 630 nm was measured. pH was extrapolated following calibration based on the nigericin/high K⁺ method, which was carried out on cells from all subjects. Calibration was repeated in stimulated cells to ensure that activation did not affect accuracy of the pH measurements. During sample acquisition, gating according to characteristic side and forward scatter characteristics confined the analysis to neutrophils, excluding clusters and debris.

During experimentation, cells were exposed to heat-killed, human serum–opsonized Escherichia coli (Orpegen Pharma, Heidelberg, Germany) across a range of pathogen-to-phagocyte ratios from 0:1 to 50:1. After addition of E coli, the cells were agitated at moderate speed to promote adherence, and pH_iwas measured. Some pathogen-to-phagocyte ratios induced both alkalinization and acidification in the same cells over time, compared with pH_i in “non–bacteria-exposed neutrophils.” To quantify these complex pH changes, we used commercial software (GraphPad) to calculate the additional area “above” (alkalinization) or “below” (acidification) the pH trace recorded in bacteria-free cells that was induced by exposure to varying densities of E coli, and expressed this area in arbitrary units. This is effectively a measure of the net change in pH_i. Positive values indicate net alkalinization, whereas negative values indicate net acidification.

To assess the roles of individual mechanisms of proton extrusion, we exposed phagocytosing cells to specific inhibitors of pH_iregulatory pumps and channels, including proton-translocating ATPases (bafilomycin A₁, 100 nM), Na⁺/H⁺exchange (amiloride, 0.2 mM), and NADPH oxidase passive proton conductance–associated channels (ZnCl₂, 50 μM). Amiloride, bafilomycin A₁, and ZnCl₂ were obtained from Sigma.

Neutrophil suspensions were incubated with normal saline or human serum–opsonized E coli (pathogen-to-phagocyte ratio 0:1-50:1) at 37°C in the presence of the oxidant-sensitive probe 123-dihydrorhodamine (Orpegen Pharma),35 after which they were immediately placed on ice. Cells were partially fixed for 20 minutes with a solution containing formaldehyde and less than 50% diethylene glycol (FACS lysing solution; Becton Dickinson, Cowley, Oxford, United Kingdom) before washing in phosphate-buffered saline (PBS). Finally, the samples were exposed to a cytoplasmic membrane permeabilizing solution containing PI (Permeabilizing Solution; Becton Dickinson) to stain DNA. At least 10 000 cells in each sample were analyzed by flow cytometry (FACScan; Becton Dickinson). Neutrophils were identified by characteristic size and granularity, as indicated by low-angle forward-scattering and right-angle side-scattering properties of argon laser light (488 nm). Mean emitted fluorescence intensity (mean channel fluorescence; MCF) at 520 nm, expressed on a 4-decade logarithmic scale with constant photomultiplier gain values, was quantified as a measure of intracellular oxidant generation. Analysis of the emitted fluorescence of PI at 520 nm differentiated viable cells from bacterial clumps on the basis of their DNA content. Only cells with the DNA binding characteristics of mammalian cells were included in the analysis. In all cases, background emitted fluorescence at 520 nm in unstained cells was subtracted from fluorescence in stained cells.

Neutrophils were gently stirred while being incubated with fluorescein isothiocyanate (FITC)–labeled human serum–opsonizedE coli (Orpegen Pharma; pathogen-to-phagocyte ratio 0:1-50:1) for 15 minutes at 37°C. The cells were then placed on ice to terminate phagocytosis, and a quenching solution containingN-ethyl maleimide (Becton Dickinson) was added. After sedimentation (500g 10 minutes at 4°C), cells were washed in PBS and sedimented again before fixation, as described above. Following fixation, cells were treated with a permeabilizing solution containing PI to stain DNA, and at least 10 000 neutrophils in each sample were identified as described above and analyzed by flow cytometry. Cells were stimulated with an argon laser at 488 nm. Mean emitted fluorescence intensity (MCF) at 520 nm was expressed on a 4-decade logarithmic scale with constant photomultiplier gain values and was quantified as a measure of bacterial ingestion. Analysis of the emitted fluorescence of PI at 580 nm differentiated cells from bacterial clumps on the basis of their DNA content. Only cells with the DNA binding characteristics of mammalian cells were included in the analysis. In all cases, background emitted fluorescence at 520 nm in unstained cells was subtracted from fluorescence in stained cells.

Cell membrane integrity is preserved until the late phase of apoptosis. Thus, we exploited the intracellular staining with the vital dyes (trypan blue via light microscopy and PI via flow cytometry) to identify necrotic cells. Aliquots of cell suspension were centrifuged and resuspended in a solution containing either trypan blue or PI. In the former case, 200 cells were examined under light microscopy, and the percentage of dye-excluding cells was calculated. In the latter case, the PI-stained solution was assessed by flow cytometry. The cell suspension was excited with argon laser light, and emitted light at 520 nm was quantified in at least 10 000 cells. The percentage of cells with PI staining of DNA (necrotic cells) was quantified. Both techniques were in close agreement (r² = 0.93). Data are thus reported interchangeably.

Neutrophils were cytocentrifuged and stained with Diff-Quick. Slides were mounted and viewed under oil immersion microscopy. At least 500 cells per slide were inspected by an observer who was blinded to the experimental condition. Features that distinguished normal from apoptotic morphology (nuclear pyknosis, chromatin condensation, and cytoplasmic vacuolation) were sought.

An alternative and independent method for quantification of neutrophil apoptosis was also employed: the detection of phosphatidylserine expression on the external surface of the cytoplasmic membrane. Briefly, following washing in PBS, 1 × 10⁶ neutrophils were gently resuspended in a solution containing FITC-labeled recombinant human annexin V in a high-Ca⁺⁺ binding buffer (Boerhinger-Mannheim, Europe). Following incubation in the dark at room temperature for 15 minutes, the binding buffer was diluted and the samples were immediately analyzed by flow cytometry. At least 10 000 cells were excited with argon laser light (488 nm), and emitted fluorescence at 520 nm minus background fluorescence was quantified. Because “annexin positivity” could be due to apoptosis-induced externalization of phosphatidylserine or annexin-V binding to intracellular phosphatidylserine in necrotic cells, additional experiments were always performed to quantify rates of necrosis, and the appropriate correction was made. Reproducibility of this method compared favorably with apoptosis detection by morphologic assessment at 16 hours. To further validate this approach for the detection of apoptosis, we (favorably) compared our results with those of a commercially available enzyme-linked immunosorbent assay of intracellular histone-associated DNA fragmentation. The percentage apoptosis reported is a mean of the results obtained by morphology and annexin-V binding.

Normal and CF epithelial lining fluid (ELF) was derived from bronchoalveolar lavage performed on healthy volunteers or CF patients (exacerbation free for 6 weeks). In brief, during fiberoptic bronchoscopy, 3 × 1 mL/kg body weight aliquots of sterile saline at 37°C were injected into a division of the right middle lobe of the lung and reaspirated immediately.36 BAL fluid was filtered through sterile gauze and centrifuged at 1100 rpm for 10 minutes, and the supernatant was stored at −80°C. The volume of ELF was calculated according to the urea dilution method,37and the lavage was concentrated using Centricon filter devices (Millipore, Bedford, MA). Similar volumes (40 μL/mL) of saline or concentrated normal ELF or CF ELF were added to suspensions of freshly isolated normal neutrophils. We assessed the immediate effects on pH_i, as well as the effects on pH_i in E coli–exposed neutrophils after preincubation (1 hour).

All data are mean ± standard error (SE) of the mean. Statistical analysis was performed using GraphPad Prism software or Statistica software. Linear regression analysis was used to test linearity of the correlation between variables. Differences over time within individual groups were analyzed using repeated-measures analysis of variance (ANOVA) and Bonferroni post hoc correction. Differences over time between groups were compared using either multivariate analysis of variance (MANOVA) or 2-way ANOVA and Bonferroni post hoc correction as appropriate.

Using FITC-labeled E coli, we confirmed that increasing bacterial density led to a linear increase in the mean number of bacteria ingested over the range of pathogen-to-phagocyte ratios we studied (Figure 1A). We then investigated how increasing the number of bacteria ingested affected pH_i. Following exposure to E coli, pH_i exhibited striking dependence on the bacterial load (Figure 1B). The lowest pathogen-to-phagocyte ratios predominantly increased pH_i over 90 minutes. Somewhat higher ratios induced transient alkalinization and later acidification of pH_i. In contrast, marked and rapid declines of pH_i were evident at the highest pathogen-to-phagocyte ratios studied. At all concentrations of E coli, pH_i returned toward baseline by 90 minutes. Pathogen-to-phagocyte ratios correlated strongly with pH_iat 5 minutes and the lowest pH_i value recorded over the first 90 minutes (r² = 0.97 and 0.928, respectively, by linear regression analysis; Figure 1C,D). Viability at 90 minutes (by vital dye exclusion) was preserved at all densities ofE coli exposure at physiologic pH_o (data not shown).

We examined the effect of specific inhibitors of proton extrusive mechanisms on pH_i following phagocytosis. As can be seen in Figure 2, the presence of each inhibitor singly and in combination for 20 minutes did not significantly affect pH_i, which suggests that the proton extrusive mechanisms that they inhibit are inactive under basal conditions. We studied cells exposed to a pathogen-to-phagocyte ratio of 20:1, having established that this caused sustained cytoplasmic acidification and therefore was likely to have stimulated proton extrusion. Inhibition of NADPH-associated passive proton conductance channels (ZnCl₂, 50 μM) enhanced acidification during the first 20 minutes following phagocytosis (Figure 2A). Amiloride (0.2 mM) strongly affected pH_i after phagocytosis, between 10 and 60 minutes (Figure 2B). Amiloride (but not ZnCl₂) treatment reduced pH_i at lower pathogen-to-phagocyte ratios (data not shown). At a pathogen-to-phagocyte ratio of 20:1, inhibition of V-ATPases with bafilomycin had a small (but significant) effect on pH_i(Figure 2C). A combination of ZnCl₂, amiloride, and bafilomycin led to more profound acidification, suggesting that the effects of these 3 proton extrusive mechanisms are additive (Figure2D). We were surprised that bafilomycin did not have a more pronounced effect on pH_i regulation. Because V-ATPases are trafficked to the cytoplasmic membrane on cellular activation,11 we speculated that they may also be rapidly reinternalized at high pathogen-to-phagocyte ratios (due to infolding of the plasma membrane during continuing phagocytosis). Therefore, we investigated the effect of bafilomycin on pH_i regulation at a lower pathogen-to-phagocyte ratio of 3:1. Bafilomycin (100 nM) did induce significant acidification of pH_i under these conditions (Figure 2E). As can be discerned in Figure 2, amiloride, bafilomycin, or ZnCl₂, either alone or in combination, did not affect pH_i before E coli exposure. Further experiments demonstrated that pH_i was not altered in the presence of these inhibitors in the absence of bacteria over 120 minutes (data not shown).

We investigated pH_i regulation in phagocytosing neutrophils at reduced pH_o (5.8). Compared with physiologic pH_o, bacterial ingestion at acidic pH_o led to significantly more acidic pH_i (Figure3), even though lower numbers of bacteria were ingested at acidic pH_o (bacterial ingestion, MCF: pH 7.4, 33 ± 4.3 versus pH 5.8, 22 ± 6.1;P < .05). pH_i failed to return to basal levels over 90 minutes. Under these conditions, a subpopulation of neutrophils with pH similar to that of the extracellular milieu was noted after 30 minutes, potentially representing necrosing cells with impaired membrane integrity (data not shown).

Neutrophils were incubated with saline, normal ELF, or CF ELF. Normal ELF did not affect resting pH_i compared with saline vehicle, but CF ELF induced a sustained alkalinization (Figure4A). CF ELF, but not normal ELF, also increased intracellular oxidant generation in response to bacterial exposure (intracellular oxidant generation, MCF: normal lavage, 400 ± 24.5 versus CF lavage, 550 ± 36.7; P < .05) and efficiency of phagocytosis in neutrophils (bacterial ingestion, MCF: normal lavage, 30 ± 3.6 versus CF lavage, 44 ± 4.2;P < .05). To test the effects of ELF on pH_i following E coli challenge, we incubated neutrophils for 1 hour with equivalent volumes of saline, CF ELF, or normal ELF, and then exposed them to bacteria. Upon E colichallenge, pH_i in normal ELF–treated cells was similar to that in cells treated with vehicle, whereas pH_i remained significantly higher in cells pre-exposed to CF but not normal ELF (Figure 4B), suggesting that inflammatory mediators limit acidification in phagocytosing cells.

At low pathogen-to-phagocyte ratios, E coli ingestion significantly delayed apoptosis, whereas higher ratios accelerated it (Figure 5A). It is puzzling that a quantitatively dissimilar but qualitatively similar signal produced opposite effects on apoptosis. We speculated that this might be the result of differential effects on pH_i, and thus correlated apoptosis with pH_i changes and also with intracellular oxidant generation following phagocytosis, because the latter has previously been implicated in neutrophil apoptosis. Apoptosis correlated significantly with the lowest recorded pH_ifollowing E coli exposure (r² = 0.872) and with pH_i 5 minutes after exposure to E coli(r² = 0.902). Net change in pH_i(calculated by measuring the curve area above and below pH_iin non–bacteria-exposed neutrophils; see “Materials and methods”) over 90 minutes correlated even more closely with rates of apoptosis (r² = 0.95; Figure 5B). Low pathogen-to-phagocyte ratios alkalinized pH_i and inhibited apoptosis, whereas higher ratios acidified pH_i and activated apoptosis. Intracellular oxidant generation correlated less significantly with rates of apoptosis (Figure 5C). Of note, at low pathogen-to-phagocyte ratios, intracellular oxidant generation was modestly increased, whereas apoptosis was suppressed rather than stimulated.

The correlation between pH_i and apoptosis may be an epiphenomenon. To probe a direct link between pH_i and apoptosis, we manipulated pH_i in phagocytosing neutrophils by inhibition of one or more proton extrusive mechanisms and assessed the effect on the mechanism of cell death. Either amiloride or ZnCl₂ accelerated apoptosis in neutrophils exposed toE coli at a pathogen-to-phagocyte ratio of 20:1 (Figure6A). Bafilomycin did not alter rates of apoptosis at a pathogen-to-phagocyte ratio of 20:1, but at lower pathogen-to-phagocyte ratios (3:1), at which we had established that bafilomycin more clearly affected pH_i, apoptosis was promoted (apoptosis at 16 hours, mean ± SE: E coli, 18.9% ± 4.2% versus E coli + bafilomycin, 31% ± 4.8%; P < .05). At lower pathogen-to-phagocyte ratios (3:1), amiloride and ZnCl₂ acidified pH_iand accelerated apoptosis to a modest degree (apoptosis at 16 hours, mean ± SE: E coli (3:1), 18.9% ± 4.2% versusE coli (3:1) + amiloride, 25.7% ± 3.9%;P < .05). The presence of ZnCl₂ at this lower pathogen-to-phagocyte ratio, which had only a minor effect on pH_i, did not affect apoptotic rates significantly (apoptosis at 16 hours, mean ± SE: E coli (3:1), 18.9% ± 4.2% versus E coli (3:1) + ZnCl₂, 21.7% ± 4.8%; P = .17). Exposure of neutrophils to these inhibitors in the absence of E colidid not affect rates of apoptosis, suggesting that they do not exert significant pH_i-independent effects on the cell's apoptotic machinery (apoptosis, mean ± SE: control, 38.5% ± 4.3%, amiloride, 37.7% ± 6.3%, bafilomycin, 40.5% ± 4.0%, ZnCl₂, 43.5% ± 6.1%;P = .3 by ANOVA).

Of note, all 3 inhibitors together at high bacterial density did not increase apoptosis, but instead promoted necrosis (Figure 6B). In the absence of inhibitors and at physiologic pH_o, little necrosis was induced following high-density bacterial exposure (Figure6B), suggesting efficient function of apoptotic machinery even when bacteria are numerous. However, high-density bacterial exposure when pH_o was low (5.8) induced significant necrosis (Figure 6B), suggesting that excessive declines in pH_i interfere with the inception or perpetuation of apoptosis.

Because the inflammatory milieu may contain many proapoptotic and antiapoptotic substances (eg, proapoptotic inflammatory mediators soluble FAS ligand (CD95) and tumor necrosis factor–α; antiapoptotic inflammatory mediators LPS, N-formyl-methionyl-leucyl-phenylalanine [FMLP], and interleukin-8 [IL-8]), we monitored apoptosis following preincubation with CF ELF and normal ELF. Normal ELF did not affect apoptosis (data not shown). However, CF ELF delayed apoptosis in phagocytosing neutrophils, without inducing necrosis (Figure 6A,B). CF ELF also delayed constitutive apoptosis (apoptosis: control, 39.4% ± 4.2% versus CF ELF treated, 31.0% ± 4.2%; P < .05). However, when neutrophils were exposed to high bacterial densities at low extracellular pH (5.8), pretreatment with CF ELF did not prevent necrosis (necrosis: control, 76.4% ± 9.2% versus CF ELF, 67.8% ± 10.2%;P = .56).

These studies were designed first to elucidate the capacity and mechanisms of pH_i regulation in phagocytosing neutrophils under conditions simulating the inflammatory environment in vivo, and second to establish whether pH_i has a role in regulating neutrophil death. Our results define the relative contributions of NHEs, proton conductance channels, and V-ATPases in neutrophil pH_i regulation following bacterial ingestion and show that cytosolic pH is not preserved when many bacteria are ingested or when extracellular pH is reduced, although proton extrusive capacity may be augmented by components of the inflammatory milieu. Furthermore, they illustrate that early changes in pH_i dictate the fate of the phagocytosing cell. Acidification of pH_i following phagocytosis accelerates apoptosis, whereas more marked decrements in pH_i induce necrosis.

Our results confirm previous observations that NHEs are implicated in pH_i regulation following phagocytosis12 and extend those observations by demonstrating that Zn⁺⁺-sensitive passive proton conductance channels also participate, being relatively more important in combating the early phase of acidification after phagocytosis. V-ATPases are implicated predominantly upon low-level bacterial exposure. These latter pumps are stored internally and are trafficked to the plasma membrane upon cellular activation.11 They have been shown to be reinternalized and incorporated into the phagocytic vacuoles as bacteria are ingested, where they reduce phagosomal pH. Their proportionately lesser role in pH_i regulation when bacterial density is high (and the cell ingests more bacteria) may reflect their rapid reinternalization under these circumstances. It is important to note that all 3 proton extrusive mechanisms acting in combination did not maintain pH_i near resting levels at high-density bacterial exposures. The simplest explanation for this is that increasing bacterial loads trigger metabolic acid production that simply exceeds the cell's maximal capacity to extrude protons. Alkalinizing and acidifying effects of low and high bacterial density thus reflect a scenario where phagocytosis triggers signal transduction events that maximally activate proton extrusive mechanisms. It is, however, also conceivable that early acidification of pH_i, induced by high bacterial density, reflects active inhibition of proton extrusion as a prerequisite for neutrophil apoptosis. Consistent with this postulate, H₂O₂ and caspase-8 (which is activated upstream of mitochondrial changes during apoptosis) induce cytosolic acidification by inhibiting proton extrusion in the course of apoptosis in HL60 cells and MCF-7 cells, respectively.25,27 This is also consistent with our unpublished observations that phorbol-12-myristate-13-acetate (PMA) is more potent than bacterial ingestion as an activator of NADPH oxidase (the major source of metabolically generated protons in activated neutrophils38), but yet does not induce the same rapid acidification of pH_i (May 1998). This lack of a rapid fall in pH_i is also consistent with the observation that, although PMA induces some features of apoptosis in neutrophils, it does not appear to stimulate caspase-mediated DNA fragmentation,39 which may require a decline in pH_i for its promotion.

These studies implicate pH_i in the regulation of neutrophil apoptosis following phagocytosis. There are few previous studies in this regard relating to neutrophils and related cells, and their results are conflicting.6,13,32,33 The role of cytosolic acidification in promoting the apoptotic cascade appears to be more established in other cell types. Critical for the inception of mitochondria-dependent apoptosis, which has been described in neutrophils,40 is activation of the proapoptotic caspase-3 by mitochondrial cytochrome C (cytC). This is maximal at pH values less than 6.8 but minimal at physiologic pH.22 This finding is consistent with the pH_i values that we detected in neutrophils under conditions in which apoptosis was either activated or suppressed. In our study, significant acceleration of apoptosis occurred only when pH_i during the first hour fell below 6.8, and even higher rates of apoptosis accompanied more profound and sustained falls in pH_i (high bacterial loading), whereas increases in pH_i (low bacterial loading) were paralleled by inhibition of apoptosis. Our studies suggest that pH_iactively modulates apoptosis, rather than simply paralleling it, because manipulation of pH_i (by targeted inhibition of NHEs, V-ATPases, or proton conductance channels) after phagocytosis acidified pH_i and accelerated apoptosis. The relationship between pH_i and apoptosis is likely to be complex. Reports that extracellular acidification inhibits constitutive neutrophil apoptosis may reflect differential pH sensitivity of constitutive versus phagocytosis-associated apoptotic pathways. For example, the insertion of the proapoptotic Bax protein into the mitochondrial cell membrane of neutrophils (an event occurring upstream of cytC release) is favored under alkaline conditions and appears to be inhibited at acidic pH_i.41 Phagocytosis may induce signals that overcome the pH dependence of Bax insertion, leading to a cascade of events downstream with an alternative pH sensitivity. This scenario would be teleologically advantageous because acidification of pH_i in an acidic milieu would protect neutrophils from constitutive apoptosis (before they have usefully ingested bacteria), but accelerate apoptosis after successful bacterial ingestion (in cells that have exhausted their useful capacity, but pose a threat to host tissues if they undergo necrosis). Alternatively, it has been suggested that constitutive neutrophil apoptosis is independent of mitochondria, in which case lack of a proapoptotic effect of cytosolic and extracellular acidification would be an anticipated finding.40 Our use of pharmacologic inhibition of individual mechanisms of proton extrusion might be problematic because these inhibitors may have pH-independent effects on apoptotic pathways. However, the weight of evidence from our experiments, particularly the close correlation between pH_i and apoptosis in the absence of these pharmacologic inhibitors, favors a role for pH_i in regulating apoptosis in neutrophils when considered in toto.

pH_i is likely to be an important modulator, rather than an initiator, of agonist-mediated neutrophil apoptosis. Reactive oxygen intermediates have been implicated in the instigation of neutrophil apoptosis.42,43 However, this relationship is apparently not a simple one because oxidants do not appear to be involved in neutrophil apoptosis in some situations,44 and proinflammatory signals, such as FMLP and IL-8, activate the generation of reactive oxygen intermediates but also inhibit apoptosis. Oxidants may be an initial signal for cytC release, with downstream events modulated significantly by pH_i. This is consistent with our observation that low-density bacterial exposure increased the generation of reactive oxygen intermediates but alkalinized pH_i and suppressed apoptosis.

How might our findings be integrated into a model of neutrophil behavior in inflammatory microenvironments? Optimally, neutrophil survival at inflammatory sites should be lengthy enough to facilitate ingestion of the maximum number of bacteria before an orderly process of cell death is instigated, thus restricting harm to host tissues. Although neutrophils were reported to undergo apoptosis after E coli ingestion,18 other studies suggested inhibition of apoptosis at low pathogen-to-phagocyte ratios, but induction of necrosis at higher ratios.19,20 This report refines previous observations on the effects of bacterial load on neutrophil apoptosis, suggesting that neutrophil survival is promoted by ingestion of relatively few bacteria. Where bacterial density in vivo is low (in mild or early infection, for instance), neutrophil lifespan would be prolonged, increasing the probability of useful subsequent phagocytosis if bacterial density subsequently increases. Because the capacity of neutrophils to kill bacteria is limited by their inability to regenerate intracellular microbicidal substances (such as lysosomal granule contents), it is desirable that apoptosis is triggered when the ingestion of a larger number of bacteria exhausts these stores. Because many diseases involve apparently dysregulated neutrophil inflammation, a limited ability of homeostatic mechanisms to support a favorable mode of cell death under all pathologic conditions is hardly surprising. In this respect, although we found that these processes were efficient at physiologic extracellular pH, they were compromised in acidic environments (pH 5.8), where significant necrosis (instead of apoptosis) occurred on exposure to bacteria at high density. Although the exceptional buffering capacity of the plasma compartment means that the tissue interstitium will generally not acidify to this degree (even allowing for anticipated local proton production due to immune-cellular and bacterial respiration), situations may arise in which microenvironmental pH does fall significantly. It has been established that serious and poorly resolving bacterial infections (eg, lung abscesses and pulmonary empyemas) are associated with very low pH, and that low pH under these circumstances is a marker of tissue destruction.1 The inability of neutrophils to regulate pH_i while ingesting bacteria in such a milieu may contribute to the poorly resolving and necrotizing nature of these infections by promoting neutrophil necrosis. How the host manipulates micro-milieu pH in response to infection and inflammation is largely unknown and merits scrutiny, particularly in relation to sites that are not in continuity with the well-buffered plasma and interstitial compartments, such as the luminal epithelial surfaces of the pulmonary, gastrointestinal, and urogenital tracts. Regulation of the surface liquid pH on these epithelia may have significant effects on neutrophil function and survival. This may be particularly pertinent in the case of CF. Lung disease in this condition is characterized by severe and unremitting bacterial infection and parenchymal lung destruction, which are mediated in large part by neutrophil-derived proteolytic and oxidant species. Lack of a CFTR (cystic fibrosis transmembrane conductance regulator)–dependent apical epithelial bicarbonate conductance has been hypothesized to cause hyperacidified airway surface liquid.45 Our results predict that in this abnormally acidic milieu, bacterial ingestion may induce neutrophil necrosis, rather than apoptosis, and thus promote lung parenchymal degradation. Although we demonstrated that the net effect of CF ELF was to increase proton extrusion and inhibit apoptosis in phagocytosing neutrophils, this effect was insufficient to prevent significant necrosis of neutrophils when pH_o was low and bacterial density was high, a circumstance that likely replicates the situation in vivo. This would contrast with the situation in the non-CF lung, where the antiapoptotic effect of the inflammatory microenvironment coupled with the ability of normal epithelia to alkalinize surface liquid may protect neutrophils therein from the deleterious effects of low pH. In keeping with this notion, apoptosis is suppressed in lung neutrophils from patients with pneumonia.21 Manipulation of the inflammatory microenvironmental pH may therefore provide a means of attenuating tissue destruction in necrotizing infections.

In conclusion, we have demonstrated that following phagocytosis in vivo, a precarious balance between cytosolic proton loading and extrusion likely exists. The resultant changes in pH_imodulate cell death in neutrophils in a manner likely to be relevant in disease pathology and potentially amenable to therapeutic manipulation.

We thank Dr Barbara Grubb for her useful suggestions with regard to this manuscript.