|Home | About | Journals | Submit | Contact Us | Français|
Chloride anion is essential for myeloperoxidase to produce hypochlorous acid (HOCl) in neutrophils (PMNs). To define whether chloride availability to PMNs affects their HOCl production and microbicidal capacity, we examined how extracellular chloride concentration affects killing of Pseudomonas aeruginosa (PsA) by normal neutrophils. PMN-mediated bacterial killing was strongly dependent on extracellular chloride concentration. Neutrophils in a chloride-deficient medium killed PsA poorly. However, as the chloride level was raised, the killing efficiency increased in a dose-dependent fashion. By using specific inhibitors to selectively block NADPH-oxidase, MPO and CFTR functions, neutrophil-mediated killing of PsA could be attributed to three distinct mechanisms: 1) CFTR-dependent and oxidant-dependent, 2) chloride-dependent but not CFTR- and oxidant-dependent, and 3) independent of any of the tested factors. Therefore, chloride anion is involved in both oxidant- and non-oxidant-mediated bacterial killing. We previously reported that neutrophils from cystic fibrosis (CF) patients are defective in chlorination of ingested bacteria, suggesting that the chloride channel defect might impair the MPO-H2O2-chloride microbicidal function. Here, we compared the competence of killing PsA by neutrophils from normal donors and CF patients. The data demonstrate that the killing rate by CF neutrophils was significantly lower than that by normal neutrophils. CF neutrophils in a chloride-deficient environment had only 1/3 of the bactericidal capacity of normal neutrophils in a physiological chloride environment. These results suggest that CFTR-dependent chloride anion transport contributes significantly to killing PsA by normal neutrophils and, when defective as in CF, may compromise the ability to clear PsA.
Polymorphonuclear neutrophils (PMNs) play a prominent role in innate immunity against invading microorganisms. For this purpose, neutrophils have evolved a number of mechanisms to generate an arsenal of microbicidal agents including hypochlorous acid (HOCl), hydrogen peroxide (H2O2), proteases and lipases, and defensins among others (see reviews [1-3]). In order to limit collateral damage to surrounding cells or tissues, microorganisms are rapidly ingested into PMNs through a process known as phagocytosis . After ingestion, microorganisms are essentially sealed in membrane-bounded phago(lyso)somes, into which the toxic agents are introduced by either de novo synthesis within the organelle or fusion with neutrophil storage granules .
HOCl is thought to play a major role in the killing of many organisms, although the susceptibilities appear to vary from one strain to another [6-11]. As a potent oxidant, HOCl is produced in large amounts within the phagolysosome lumen and is estimated to account for ~90% of the total oxygen consumed by the neutrophil during phagocytosis . The HOCl-dependent antimicrobial system has been widely studied and depends on the NADPH-dependent oxidase (NOX) as an oxidant source, reducing molecular oxygen to superoxide anion that then dismutates to hydrogen peroxide in the presence of water . Myeloperoxidase (MPO) reacts with hydrogen peroxide to form Compound I, which then reacts with its physiological substrate, chloride anion, to form HOCl [14, 15]. MPO can also oxidize other halides, such as iodine, bromide [16, 17] and the pseudohalide thiocynate . HOCl modifies bacterial proteins, lipids and nucleic acids in a way that inactivates their biological activities and leads to cell death [19-21]. HOCl can also react with amino acids such as taurine to form toxic chloramines . The availability of chloride anion in phagolysosomes is apparently a limiting factor affecting HOCl production and microbicidal capacity of the cells.
Cystic fibrosis (CF) is an autosomal recessive genetic disease. The gene responsible for this disease encodes the CF transmembrane conductance regulator (CFTR), a cAMP-activated chloride channel [22-25]. CF lung disease claims most of the mortality, which is characterized by chronic bacterial infection, small airway obstruction and pronounced airway recruitment of neutrophils [26, 27]. In spite of tremendous efforts made to understand CF pathogenesis, it is unclear why large numbers of neutrophils recruited to the lung fail to eradicate bacterial infections. We previously reported that CFTR is expressed in human neutrophils and their phagolysosomes . Neutrophils from CF patients exhibit a profound loss in the ability to chlorinate ingested bacteria compared to the activity of normal counterparts. The data strongly suggest that the defective CFTR channel significantly compromises the efficacy of the MPO-H2O2-chloride microbicidal system. To define the role of chloride anion and CFTR in neutrophil-mediated bacterial killing, we investigated in the current study the effect of extracellular chloride on killing of PsA by normal neutrophils. Inhibitors for MPO, NOX and CFTR were employed to define the underlying mechanisms. Moreover, the killing of PsA by neutrophils from CF patients and normal subjects was compared to identify the bacterial killing defect of the patients’ cells.
Percoll was obtained from BD Pharmacia (San Diego, CA). Horseradish peroxidase, superoxide dismutase, diphenylene iodonium chloride (DPI), human AB serum, saponin, Aspergillus niger catalase, taurine, diethylenetriaminepentaacetic acid (DEPA) and other common chemicals were obtained from Sigma (St Louis, MO). Aminobenzoic acid hydrazide (ABAH) and GlyH-101 were obtained from Calbiochem (La Jolla, CA). L-[U-14C] amino acid mixtures were obtained from Amersham Biosciences (Piscataway, NJ).
The physiological chloride Ringer’s buffer was composed of 122 mM NaCl, 1.2 mM MgCl2, 1.2 mM CaCl2, 2.4 mM K2HPO4, 0.6 mM KH2PO4, 20 mM HEPES (pH 7.3), and 10 mM dextrose. The gluconate chloride-free Ringer’s buffer was made by substitution of the above chloride salts with equal molar concentration of gluconate salts except 4 mM of calcium gluconate was used to compensate for the mild calcium chelating effect of gluconate. Various concentrations of chloride in Ringer’s buffers were achieved by mixing appropriate proportions of the physiological chloride Ringer’s buffer and the gluconate chloride-free Ringer’s buffer so that the tonicities and ionic concentrations remained unchanged.
Peripheral blood neutrophils were isolated from freshly drawn human venous blood with a human subject protocol approved by the Institutional Review Boards of Louisiana State University Health Sciences Center and Ochsner Clinic Foundation at New Orleans. Neutrophils were isolated from 8-20 ml of blood by fractionation on discontinuous Percoll gradients as previously described . The cells were resuspended in either gluconate chloride-free or physiological chloride Ringer’s buffer at 1 × 107 cells per ml and held on ice for at least 1 hour prior to assays to allow complete intracellular chloride exchange . The genotype of the CF population is as follows: five were ΔF508 homozygotes, one was ΔF508 compound heterozygote ΔF508/1717-1(G→A*), one had a rare mutation (1R347H/1V520F) and, two had unknown mutations with positive sweat tests and pancreatic enzyme supplement.
Pseudomonas aeruginosa (PsA) strain PA01 was cultured overnight in LB broth at 37 °C with vigorous agitation. After centrifugation (5000x g) for 15 min, the cell pellet was suspended in phosphate buffered saline containing 10% fresh human AB serum at a density of 1-2 ×108 (A600nm=0.2) bacteria per ml and incubated for 30 min at 37 °C with agitation. The bacteria were washed twice with chloride-free Ringer’s buffer, resupended at 1-2 × 109 per ml, and held on ice until use.
Bacterial killing assays were performed using a modification of the published methods [31-33]. The aforementioned opsonized PsA bacteria were dissociated by passing through a 25-gauge needle 3 times. Then, the bacteria were mixed with neutrophils (1 × 107 per ml) at a ratio of 1:1 or 20:1, as indicated, in the chloride-free gluconate Ringer’s buffer containing 10% dialyzed chloride-free human AB serum. The mixture was incubated at 37 °C for 20 minutes with shaking and aliquoted for various experimental conditions as indicated. Then, the non-phagocytosed bacteria were removed by low speed centrifugation (81x g for 5 min at 4 °C). The pellets from various aliquots were, respectively, resuspended in various chloride buffers containing 10% dialyzed human AB serum for killing assays. Duplicate determinations were performed. The initial bacterial load was determined by dilution of 10 μl of each suspension in 1 ml of 0.05% saponin in water. After 3 passes through a 25-gauge needle, the sample was further diluted in sterile PBS and 20-25 μl plated on LB agar plates. Then, the suspensions were brought to 37 °C with shaking, 10 μl aliquots withdrawn at various time intervals (10, 20, 30 and 40 minutes or as indicated in the text), and similarly processed for bacterial plating. After overnight culture, viable PsA colonies were counted. Preliminary study proved that lysis of the cells in the saponin water solution and dispersion of bacteria by passing through a fine needle was as effective as the recently published alkaline water method .
Data from multiple time points were obtained according to experimental designs. The rate of killing was linear with time when plotted on a semi-log plot (first-order; see Fig. 1A), which was consistent with previous publications [33, 35]. The first-order rate constant was determined by plotting the percent of initial remaining viable bacteria on a log scale as a function of incubation time to determine the rate of loss of viable bacteria per min relative to the initial bacteria added. The loss of viability was best approximated by a first-order fit according to the equation V= e−kt, where V equals the relative fractional rate of loss of viability at any given time t in minutes, and k equals the first-order rate constant.
In the case of using MPO, NOX and CFTR inhibitors, the drugs were introduced 10 min prior to phagocytosis and were included throughout the killing assays. Preliminary studies established that none of the drugs used had appreciable effects on PsA viability in the absence of neutrophils. Sample collection, lysis, dilution and plating were the same as above.
PsA bacteria were metabolically labeled with a 14C-L-amino acid mixture by growing for 18-19 hrs in 1 ml LB medium containing 2 μCi of the 14C-labeled amino acid mixture. The labeled bacteria were washed with PBS three times by centrifugation to remove free isotopes, opsonized with human serum, and then washed twice with chloride-free Ringer’s buffer. To disperse and remove bacterial aggregates, the bacterial suspension was passed through a 25-gauge needle three times and centrifuged at 250x g for 5 min. The supernatant was retained. The bacterial density and radioactive isotope incorporation were determined by spectrophotometry and liquid scintillation counting, respectively. The 14C-labeled PsA (2 × 108 per ml) were mixed with neutrophils (1 × 107 per ml) in the indicated Cl−-Ringer’s buffers containing 10% (v/v) of the dialyzed human AB serum. Inhibitors, when present, were pre-incubated with PMNs for 10 min prior to the addition of bacteria. After incubation for 20 min at 37° C with shaking, 50 μl aliquots were layered on Nyosil M-25 silicon oil and centrifuged at 9500x g. After being frozen at −80 °C, the tips of the tubes containing neutrophil pellets were excised with a razor blade and radioactivity associated with the cell pellets was determined by liquid scintillation spectrometry. In the absence of neutrophils no bacterial-associated radioactivity over background levels was recovered in the tube tips.
Wells containing 100 μl of a test Ringer’s buffer with an indicated concentration of chloride were set up in triplicates. MPO-dependence of the reaction was determined by adding 150 μM ABAH, a highly specific inhibitor for MPO. Neutrophils (1×107) and serum-opsonized FITC-zymosan particles (1×108) were added to each well. After 30 min at 37 °C, the wells containing adherent cells with ingested particles were gently rinsed with ice-cold gluconate Ringer’s buffer. Then, 100 μl of a 127 mM KCl Ringer’s buffer (pH 8.2) containing 0.004% trypan blue, 7 μM nigericin and 5 μM monensin were added. This medium allowed a quick equilibration of the phagolysosomal pH with the extracellular medium pH, abolishing any fluorescence differences due to acidification of the phagolysosomal lumen during phagocytosis . Trypan blue was included also to quench the fluorescence of non-internalized FITC-zymosan particles. The FITC fluorescence associated with intracellular zymosan was then measured with a Biotek FL600 fluorescence plate reader (BioTek Instrument, Inc., Winooski, VT). Excitation was at 585 nm and emission at 520 nm using standard filters provided with the instrument. After correction for cellular background fluorescence (< 5-10% of the total signal), the fluorescence relative to that observed in gluconate Ringer’s buffer was calculated. Neutrophil-mediated loss of fluorescence that was both chloride- and MPO-dependent reflected the intraphagolysosomal bleaching or chlorination. Microscopic examination at low magnification revealed a uniform monolayer of cells with ingested particles (see Fig. 2A) under all the conditions used, indicating that the loss of fluorescence was not due to differential cell adherence to the plates.
Real-time superoxide anion release by the reduction of cytochrome c was measured spectrophotometrically at 550 nm under the ambient conditions as described elsewhere . Briefly, 5 × 105 PMNs were resuspended in 1 ml of Ringer’s medium containing various chloride concentrations with 50 μM cytochrome c. PMA at a final concentration of 500 ng/ml in 2 μl of DMSO was added. The increase in absorbance at 550 nm was monitored over a 5-minute period at 15-second intervals. The initial maximal rate of O2− synthesis was calculated from the data by converting the rate change in absorbance to nmol of O2− produced per minute using the extinction coefficient of reduced cytochrome c of 21, 500 M−1cm−1. Addition of superoxide dismutase (50 μg/ml) reduced O2− production by more than 90%. In addition, inhibition of the NADPH oxidase by addition of 20 μM DPI inhibited O2− production by more than 95%. In the experiments indicated, superoxide anion production was measured by the enhanced chemiluminesence of lucigenin. Briefly, neutrophils were incubated with opsonized PsA (PsA/PMN=20:1) in Na gluconate Ringer’s buffer containing 10 % dialyzed serum at 37 °C. Aliquots (25 μl) containing 250,000 neutrophils were diluted into 0.8 ml of NaCl Ringer’s containing 125 μM lucigenin (Invitrogen) and the chemiluminesence measured over 30 seconds in a Turner Model TD 20/20 luminometer at selected time points. Twenty minutes after PsA addition, the cells were centrifuged at 81 x g for 5 minutes at 4 °C and resuspended in either Na gluconate or NaCl Ringer’s buffer supplemented with 10% dialyzed serum. Incubation at 37 °C was continued and lucigenin-mediated chemiluminescence (LucCL) monitored as before at the indicated time points.
Student’s t-tests were performed by standard procedures. P-values smaller than or equal to 0.05 were considered to be significantly different.
Figure 1A shows the effect of extracellular chloride on the intracellular killing of PsA by normal human neutrophils. In these experiments, modified Ringer’s buffers with varied chloride concentrations were used, as detailed in the Materials and Methods. All the buffers had an identical isotonicity and free ionic concentration. Varied chloride levels were achieved by substituting gluconate salts for chloride salts, proportionally. Serum-opsonized PsA bacteria and Percoll-gradient purified neutrophils were resuspended in the chloride-free gluconate Ringer’s buffer at a ratio of 20:1 (PsA:PMN). After 20 minutes of phagocytosis at 37 °C with shaking, the samples were aliquoted for various experimental conditions and centrifuged at low speed at 4 °C to remove uningested PsA. The cell pellets were then resuspended in Ringer’s solutions containing varied concentrations of chloride with 10% (v/v) dialyzed human serum. Performing phagocytosis in a master tube and aliquoting for individual experimental treatments eliminated any possible variations arising from phagocytosis. The samples collected at various time points (0, 10, 20, 30 and 40 minutes) under various chloride conditions were lysed in 0.05 % saponin water solution with homogenization, serially diluted in PBS and plated onto LB agar plates to determine the viable bacteria. After overnight growth, colonies were counted to determine percent of bacterial killing as compared to viability in the control samples without any added neutrophils. Data from a representative experiment (Fig. 1A) was analyzed and found to be best fitted by the exponential first-order kinetic model (V=e−kt), where V equals the rate of loss of viability relative to 0 time at any given time, t indicates time in minutes and k equals the first-order rate constant. Statistical analyses of multiple data sets demonstrated that when the killing rate constants expressed as % per minute were plotted as a function of extracellular chloride concentration, neutrophil-mediated intracellular killing of PsA was poor (2.85 % per min ± 0.16, N=8) in the medium containing no chloride. The killing efficiency increased in a dose dependent manner as extracellular chloride levels were raised toward the physiological chloride level (127 mM). Neutrophils killed PsA efficiently in the 127 mM chloride-rich medium with a rate constant of 6.08 % per min ± 0.6 (N=8), twice as fast as in the chloride-free medium. The killing rate constants plateaued at 60 mM or higher concentrations of chloride. Notably, phagocytosis in this study was all done in chloride-free gluconate Ringer’s buffer and the extracellular chloride levels were manipulated only after phagocytosis had occurred. Therefore, all the differences expressed in the PsA killing rates were totally attributed to post-phagocytic chloride availability to the cells.
In order to assess the effect of extracellular chloride on the NADPH oxidase system, we next examined the phorbal myristyl acetate (PMA)-induced superoxide anion production by monitoring the reduction of ferricytochrome c spectrophotometically. The activity of the NADPH oxidase was slightly increased by 15- 20 % at higher chloride levels relative to that seen in the chloride-free medium (Fig. 1C). However, the two-fold difference of PsA killing between the chloride-rich and chloride-poor conditions cannot be explained by the minor effect of the oxidant response by the cells. Therefore, the observed chloride-dependent killing of PsA implicated an important role for chloride anion in post-phagocytic, neutrophil-mediated innate defense against this particular organism.
The method used for the above killing assays inherently had two phases: the phagocytosis phase and the post-phagocytosis phase. We examined if neutrophils in the post-phagocytosis remained active with regard to oxidant production. Lucigenin-enhanced chemiluminesence (LucCL) at the selected time points in Figure 1D was measured. Neutrophils responded maximally to the PsA stimulus within 10 min after addition. After centrifugation to remove the non-phagocytosed free bacteria and transfer to the gluconate medium, the cells had a LucCL response that very slowly declined over the 40-min period after the buffer change. When the cells were suspended in the NaCl Ringer’s buffer, the superoxide production continued at similar levels for the first 5-10 min after buffer change as compared to that in the gluconate buffer and then declined to lower but still productive levels over the next 20-30 min. Addition of PMA at the end of the incubation elicited an addition elevated response in both cases although the response was more robust in the gluconate buffer. The differences of superoxide levels in the two buffers may reflect consumption of superoxide by MPO after dismutation to H2O2 or inactivation of NADPH oxidase activity by HOCl. Importantly, addition of PMA at the end of the experiment shows that the cells remained responsive and had additional stores of oxidase available for activation if needed.
To determine if the chloride-dependent killing of PsA correlated with HOCl production, we investigated the effects of extracellular chloride level on HOCl production by neutrophils. Stimulated neutrophils generate HOCl intraphagosomally and, when MPO is released into the medium during cell activation, extracellularly. To examine the intraphagolysosomal HOCl production, we generated a fluorescent probe by chemically conjugating fluorescein (FITC) to zymosan-A particles. The FITC-zymosan were opsonized and fed to neutrophils. After 30 min at 37 °C, the cells adhered to the wells, which was followed by gently rinses with ice-cold gluconate Ringer’s buffer. Then, 100 μl of a 127 mM KCl Ringer’s buffer (pH 8.2) containing 0.004% trypan blue, 7 μM nigericin and 5 μM monensin were added. Bleaching of fluorescein was observed by fluorescence microscopy and quantitated using a fluorescence plate reader. Figure 2A displays the images of the FITC-zymosan probe in neutrophils 30 minutes after incubation under 0, 63.5 or 127 mM chloride condition in the presence or absence of MPO-specific inhibitor ABAH. Without the inhibitor, the fluorescence intensity of the zymosan probe progressively decreased as the chloride concentration increased from 0 mM to 63.5 mM and to 127 mM. In contrast, the fluorescence intensity of the probe did not respond to chloride level changes significantly when ABAH was present. Figure 2B shows that the relative FITC-Zymosan fluorescence intensity was inversely correlated with extracellular chloride levels, and that the fluorescence was MPO-dependent. Given that MPO-mediated bleaching of fluorescein by neutrophils is known to be due to its chlorination by HOCl , we interpret these findings to indicate that extracellular chloride levels affect intraphagolysosomal HOCl production.
To define the potential mechanisms involved in chloride-dependent bacterial killing by neutrophils, we evaluated the impact of inhibitors of NOX, MPO, or CFTR on the rate of killing PsA by normal neutrophils in the presence or absence of chloride. To ensure that intracellular bacterial killing was exclusively measured, we again allowed phagocytosis (PsA:PMN=20:1) first for 20 minutes in chloride-free Ringer’s buffer. The inhibitors or no drug vehicles were included 10 minutes prior to and throughout the phagocytosis process. After removal of extracellular bacteria by means of low speed centrifugation at 4 °C, bacteria-laden neutrophils were resuspended in either a chloride-rich or chloride-free buffer, with or without the inhibitors. The percent of viable PsA remaining at 0, 20 and 40 minutes after phagocytosis was determined by LB agar plating and bacterial colony counting. Initial phagocytosis in a no chloride medium ensured that no extracellular chloride was taken up from the medium either by phagocytosis or by pinocytosis. Thus, chloride potentially contributing to any killing would have to be transported directly by membrane channels from the extracellular medium to the cytosol and from the cytosol to the lumen of phagolysosomes. Because CFTR is expressed in neutrophil phagolysosomes , this chloride channel may be involved in regulating phagolysosomal chloride levels. We predicted that CFTR, if blocked, should interfere with the function of neutrophil MPO-H2O2-chloride microbicidal system. To test the hypothesis, we selected the highly specific CFTR inhibitor, GlyH-101, which at micromolar concentrations blocks CFTR-mediated chloride transport . We also used the peroxidase inhibitor ABAH and the NOX inhibitor DPI, to elucidate the relative contributions of MPO and oxidant species to bacterial killing in our system. Figure 3A shows the effects of all three inhibitors on the rate constant of PsA killing, which were determined by plotting the exponential rate of loss of viable bacteria per min relative to the initial bacteria loaded. In a chloride-deficient medium normal neutrophils killed PsA at a rate constant of 2.45% per min ± 0.16 (N=4), whereas in a physiological chloride medium neutrophils had a rate constant of 5.27% per min ± 0.32 (N=4). In 127 mM chloride, DPI, ABAH, or the CFTR inhibitor GlyH-101, lowered the constants significantly from ~5.27% per min to ~3.76%, ~3.66% or ~3.84% per min, respectively. Moreover, the three inhibitors when used in combination gave rise to a killing rate constant of 3.15 % per min ± 0.50 and did not inhibit killing any further over that seen with any of the drugs alone. However, in the absence of extracellular chloride, the three drugs did not alter the initial killing rate significantly from ~2.45% of the no drug control to ~2.22%, 2.29% or 1.95% per min, for DPI, ABAH or GlyH101, respectively. None of these agents affected the rate of phagocytosis (Figure 3B) or directly compromised PsA viability (data not shown). These data suggest that three components contributed to PsA killing by the normal phagocytes. First, about 25-30% of the killing was inhibited by GlyH-101, ABAH or DPI, suggesting the contribution by CFTR and oxidants to the overall PsA killing under our experimental conditions. Second, an additional 25-30 % of the killing is dependent on chloride, but not on CFTR or oxidant-related mechanisms, and the remaining 40-50% of killing is independent of any of above mentioned factors.
It was asked that to what extent the bacteria were killed in the first (phagocytosis) phase. During all of our killing assays we measured the initial number of viable PsA added to the original cell suspension as well as viable PsA recovered after the 20 min incubation. The estimated killing rate constant (% per min) for control cells or cells treated with GlyH-101, ABAH or DPI over the 20 min period in the chloride-free gluconate medium was 1.94±0.57, 2.49±0.66, 2.71±0.69 or 2.32±0.82, respectively. This corresponds to 32.2±7.3%, 39.2±7.5%, 41.8±7.5% or 37.1±9.5% of the initial total PsA inoculums, which were killed during the first phase of the assay. As can be seen the rate constants are not significantly different in gluconate medium without or with any added drug. Further, the estimated rates are similar, but slightly less, than those seen if Na Gluconate is used in the second phase of the killing assay (Figure 3). This slightly slower rate in the first 20 minutes likely reflects the delay between bacterial engagement of cell receptors, phagocytosis and subsequent killing, because all events engaged were in an asynchronous fashion. In the second phase these steps are completed and the rate constant is closer to the true killing constant.
Killing of PsA by normal neutrophils was significantly depressed when CFTR function was blocked by inhibitor GlyH-101 (Figure 3A), implicating CFTR as an important component of normal neutrophil function. Therefore, we reasoned that neutrophils from CF patients, which lack a functional CFTR, would behave similarly to normal neutrophils in which CFTR function was inhibited by GlyH-101. To test this prediction, neutrophils were isolated from normal subjects and CF patients, and resuspended in the chloride-free gluconate Ringer’s buffer. Phagocytosis of opsonized PsA was performed at a ratio of 1:1 or 20:1 (PsA:PMN) for 10 minutes. After removal of the non-ingested bacteria, the PsA-loaded neutrophils were resuspended in 0 mM or 127 mM chloride Ringer’s buffer and incubated for killing for 0, 15 or 30 minutes. The killing rate constant was determined similarly as above to indicate the first order loss of viable bacteria per min relative to the initial bacteria ingested. Figure 4A shows the rate constant of killing PsA by normal and CF neutrophils in 0 mM and 127 mM chloride media at an initial ratio of 1:1 (PsA:PMN). In low chloride, the rate constant of PsA killing by normal neutrophils was 4.91%±0.63 (N=5) and was significantly higher than that of the CF cells (2.98%±0.22, N=5). Moreover, when physiological chloride (127 mM) was present, the normal cells killed PsA at a rate of 8.96%±0.22 (N=5). However, under the identical condition the CF cells had a rate of 6.55%± 0.77 (N=5), which is significantly lower than that for normal neutrophils. It is worth pointing out that normal neutrophils in a chloride-rich environment killed PsA three-times as efficiently as the CF neutrophils in the chloride-deficient environment. At an MOI of 20:1 (PsA:PMN), neutrophils killed bacteria at a much slower rate (Fig. 4B) which was reduced by ~30-50% over that seen at the 1:1 ratio. However, as was the case of MOI of 1:1, normal neutrophils killed PsA significantly faster than did CF neutrophils at either chloride concentration. The reduction in the rate of killing at MOI of 20:1 could be explained by an overwhelmed microbicidal capacity of the cells, quenching of oxidant by bacterial factors such as catalases, and/or the effects of bacterial toxins on PMN function. This was supported by experiments performed at an MOI of 50:1 (PsA:PMN), which showed rates of killing that were only ~10% of that seen at the MOI of 1:1 (data not shown). Thus, the following two conclusions can be drawn from the above data: 1) CF neutrophils were less effective in killing PsA than were normal neutrophils, regardless of the chloride environment; 2) the killing deficiency of CF neutrophils displayed in a chloride-poor environment was more severe than that seen in a chloride-rich environment. The bacterial killing rate of CF neutrophils in a chloride-deficient environment was only 1/3 of that of normal neutrophils in a chloride-rich environment.
Neutrophils represent the predominant cell type in the early innate host defense to bacterial infection. Neutrophils contain in their azurophilic granules an abundant amount of MPO (~1-5% of dry weight of the cells), a peroxidase that uniquely catalyzes the two electron oxidation of chloride anion in the presence of H2O2 to generate HOCl, which plays an important role in killing microorganisms, including those that are resistant to non-oxidant killing mechanisms . Given the significance of HOCl for optimal neutrophil antimicrobial action, we speculated that an ample supply of chloride in neutrophils and, specifically in their phagolysosomes, would be extremely important for maintaining the maximal destructive capacity of the phagocytes. In this report, we have demonstrated that the extracellular chloride level significantly affects neutrophil-mediated bacterial killing. Our data indicated that any extracellular environment with a sufficient supply of chloride would be expected to enhance neutrophil bactericidal function. When neutrophils migrate out of blood vessels, chloride availability to neutrophils relies exclusively on two sources: 1) that pre-stored in the cells and 2) that obtained by transport from the extracellular microenvironment. The pre-stored chloride in neutrophils, unless bound with macromolecules inside the cells, very likely equilibrates with extracellular fluid within minutes , so maintenance of extracellular chloride is vital for maintaining maximal killing capacity of neutrophils.
For intracellular bacterial killing, HOCl must be produced in phagolysosomes where the bacteria are physically located after phagocytosis. There are two general sources of chloride in the phagosome: passive acquisition of chloride from the extracellular space as the nascent phagosome forms, and transport from the cytoplasm across the membrane of the established phagosome. In the latter case, extracellular chloride anion must be first transported from the extracellular medium to the cytosol, and from the cytosol to the phagolysosomes. Transport of chloride by the plasma membrane is well-studied and chloride influx and efflux from the cytosol has been shown to be rapid and fully exchangeable . Moreover, upon stimulation with neutrophil agonists, such as opsonized zymosan or formylated peptides, there is a rapid efflux of chloride from the cytosol resulting in lowering of the cytosolic chloride level from about 90-110 mM to about 50-60 mM . Recent data from our laboratory show that the phagolysosomal chloride level in zymosan-activated cells is ~72 mM, similar to that in the cytosol . In contrast, it is unknown how chloride anion is transported from the cytosol to the lumen of phagolysosomes. At least two chloride channels, CFTR and ClC3, are expressed in phagolysosomes [28, 43] and we have previously shown that the CFTR chloride channel defect in CF leads to a profound deficiency in chlorination of ingested bacteria . In this report, we document that the neutrophils from CF patients had defective bactericidal activity, which is consistent with the impaired chlorination of bacteria. Importantly, the bacterial killing defect in CF neutrophils is further compromised in a chloride-deficient microenvironment. In low Cl−, CF neutrophils presumably acquire less Cl− to phagolysosomes directly through phagocytosis or pinocytosis. The CFTR channel dysfunction in CF would be expected to further limit the Cl− supply to phagolysosomes. Moreover, efflux of Cl− from the cells to the low Cl− environment would lead to a net loss of cytoplasmic Cl−, which also would decrease the gradient-dependent Cl− secretion intracellularly from the cytosol to the phagolysosomes through other Cl− channels such as ClC3. Taken together, these factors constitute the molecular mechanism underlying the CF neutrophil defect with respect to bacterial clearance.
The chloride level in the airway surface liquid (ASL) secreted by normal and CF airway epithelia has been measured by many groups with discrepant results . Two major concerns with these studies are that 1) none of the measurements were performed without perturbing the investigated system and, 2) none of the measurements were in the actual disease setting, such as ongoing bacterial infections and the associated inflammation. Impermeability of chloride across the airway epithelium due to the mutant CFTR channel could potentially limit the chloride availability to neutrophils mobilized to CF lungs. Changes in the ASF composition due to defective chloride transport of epithelial cells may affect its viscosity, salt balance and ionic distribution, which could potentially impede the chloride exchange with neutrophils in the airways. In addition, certain microorganisms, such as PsA, express high levels of catalase which could potentially divert H2O2 from producing HOCl and further compromise oxidant-dependent defenses. Moreover, when mucoid forms of PsA are present, the bacteria are highly resistant to killing by non-oxidant killing mechanisms. Therefore, we postulate that phagocyte function, the airway environment and the nature of the microorganisms present in the lung, collectively, contribute to the pathophysiology in the CF lung. Thus, we propose that CF lung disease is virtually an innate immunity deficiency due to insufficient chloride availability to phagocytes and their phagolysosomes, which in turn leads to evasion of phagocytic host defenses by microorganisms such as PsA.
In this report, we found that PsA killing by normal neutrophils was only partly dependent on oxidant production, since MPO and NOX inhibitors blocked 25-30% of the killing efficacy. These inhibitors had no significant effect in chloride-poor media, suggesting that oxidant-mediated killing was chloride-dependent. Interestingly, an additional component amounting to 25-30% of the total killing capacity was dependent on extracellular chloride but not compromised by oxidant or CFTR inhibitors. This result suggested the existence of an additional non-oxidant-mediated killing mechanism that was chloride-dependent. Future studies will be required to identify the mechanisms underlying these processes.
In summary, our current data together with our previous studies indicate that CF neutrophils have an impaired ability to chlorinate and kill PsA. These findings suggest that the defective CFTR present in the CF neutrophil phagolysosomal membrane limits the concentration of chloride anion in the phagosomal compartment, which in turn compromises their ability to produce hypochlorous acid within the organelle. This defect is further exacerbated in a chloride-deficient environment, which ultimately undermines the overall antimicrobial capacity of CF neutrophils.
The authors would like to acknowledge critical reading of the manuscript by Ms. Marla Gomez and Dr. Martha Aiken. We are grateful to the normal and patient volunteers that generously provided their blood samples for this study. This work is supported by grants to G. Wang from NIH (1R01-AI72327-01A1), Cystic Fibrosis Foundation (CFF WANG07I0) and Louisiana Gene Therapy Consortium, and by a Merit Review award to W.M. Nauseef from the Veterans Administration.