Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Lung disease characterized by ongoing airway infection (typically with Pseudomonas aeruginosa [Psa]) and neutrophilic inflammation is the major cause of mortality in cystic fibrosis (CF) patients (1). The gene defect underlying this disease was identified in 1989 (2), but the link between the product of this gene, the cystic fibrosis transmembrane conductance regulator (CFTR), and CF lung disease has remained obscure. CFTR functions both as a chloride channel itself and as a regulator of other ion channels (5). In CF airway epithelial cells, loss of CFTR function results in abnormal transepithelial electrolyte and fluid transport (8, 9). This plausibly could be expected to result in abnormalities in the ionic composition of the airway surface fluid (ASF) overlying these cells in CF patients. The actual composition of ASF in both CF and normal persons remains uncertain and controversial. Several investigators have found abnormally high ASF chloride concentrations in CF, although it is not yet clear whether this represents a primary abnormality due to loss of CFTR function, or a secondary abnormality occurring as a consequence of chronic airway infection and inflammation. In either case, abnormal ASF composition could contribute to the pathogenesis of CF lung disease, and much attention is currently focused on how an abnormal ASF may impair host defense.

ASF exists only as a very thin layer of fluid, making collection of sufficient volume for conventional analysis difficult (10). Elevated ASF chloride concentrations in CF have been noted by several investigators in both in vivo and in vitro studies. Gilljam and colleagues found a chloride concentration of 170 ± 79 mM in bronchial ASF from CF patients versus 85 ± 54 mM in control subjects (11). Joris and associates found a chloride concentration of 129 ± 79 mM in tracheal and bronchial ASF from CF patients versus 85 ± 54 mM in controls (12). Goldman and coworkers studied ASF in human bronchial xenografts established by seeding denuded rat tracheas with primary human bronchial cells, then implanting the grafts subcutaneously in nude mice (13). These investigators found a chloride concentration of 172 ± 9 mM in ASF from xenografts established with cells from CF patients, versus 83 ± 3 mM in ASF from xenografts derived from non-CF control subjects. Furthermore, transfection of wild-type CFTR into the CF xenografts normalized their ASF chloride concentration to 87 ± 4 mM (13).

Still, the ionic composition of ASF in CF remains controversial. Several investigators have found ASF ion concentrations in CF patients to be similar to those in non-CF controls. Knowles and coworkers found no significant differences in the chloride concentrations in bronchial ASF obtained from uninfected CF patients, uninfected chronic bronchitis patients, and normal volunteers (14). Hull and colleagues compared the chloride and sodium concentrations in tracheal ASF from CF infants, with and without pulmonary infection or inflammation, with the concentrations in ASF from non-CF infants (15, 16). Compared with non-CF infants, these investigators initially reported a significantly elevated sodium concentration in the ASF from CF infants with infection or inflammation, but not in ASF from CF infants without infection or inflammation (15). ASF chloride concentrations demonstrated a similar trend, but did not reach statistical significance. These investigators therefore concluded that increased chloride (and sodium) concentrations found in CF ASF may occur secondary to airway infection or inflammation rather than as a primary abnormality of CF (15). In a subsequent paper, Hull and associates (16) found no differences in sodium concentration in ASF from non-CF infants and CF infants with and without infection. In contrast to their original findings, they reported ASF chloride concentrations lower in CF infants without infection than in non-CF infants, with chloride values in CF infants with infection rising toward that of the non-CF infants (16). It is therefore likely that the definitive characterization of ASF ion composition has not yet been obtained.

In considering the possibility that ASF ionic composition may be aberrant in CF, other recent investigations have focused on the effects these abnormalities might have on antibacterial defense of the respiratory epithelium. Smith and colleagues demonstrated that primary cultures of epithelial cells grown at an air-fluid interface secrete a substance capable of killing a broad spectrum of bacterial pathogens, including Psa (17). Although this substance was secreted by epithelial cells from both CF patients and control subjects, its activity was reversibly abrogated in the CF cultures by the presence of higher surface-fluid NaCl concentrations. Goldman and coworkers subsequently attributed this antimicrobial activity to human -defensin-1 (hBD-1), a peptide previously identified in hemofiltrate of patients with end-stage renal disease (13). The hBD-1 gene was found to be highly expressed throughout the respiratory epithelia of non-CF and CF lungs, and the peptide was demonstrated to be progressively inactivated by increasing NaCl concentrations. In the human bronchial xenograft model used by these investigators, addition of antisense oligonucleotides to hBD-1 abolished the ASF antibacterial activity of non-CF grafts. The impairment of hBD-1 or other epithelial defenses in CF airways would force an increased reliance on polymorphonuclear leukocytes (PMN) to defend against inhaled and aspirated bacteria. Increased reliance on PMN defense, however, might come at the price of the increased airway inflammation characteristic of CF lung disease.

To participate effectively in lung defense, PMN must be recruited from circulation, phagocytose and kill bacteria, and then be cleared in a manner that prevents leakage of their potentially injurious cytoplasmic contents into the airways. PMN can be recruited to the lung by multiple chemoattractants, but interleukin-8 (IL-8) has been shown to account for the majority of PMN chemoattractant activity in the sputum of CF patients (18). Several lines of evidence have demonstrated that a significant amount of IL-8 produced in airways in response to bacterial challenge is secreted by recruited PMN themselves in an autocrine fashion (19, 20). The functional lifespan of PMN in the lung is limited because these cells are preprogrammed to undergo apoptosis relatively quickly (21, 22). Membrane changes associated with apoptosis are believed to mediate PMN clearance from inflammatory sites by signaling for their phagocytosis by macrophages (21). If not cleared, apoptotic PMN undergo secondary necrosis and lysis, with release of proteases and reactive oxygen intermediates that have been implicated in CF lung injury (23).

In light of developments suggesting that altered ionic composition of ASF may impair the primary antibacterial defenses of the respiratory epithelium in CF, we investigated whether this altered ASF would impair secondary, PMN-based defenses. We employed a cell-culture model in which halide concentrations and osmolarity were varied independently. We measured the effects of altered media on PMN synthesis of IL-8, PMN killing of Psa, and PMN apoptosis and lysis. As noted previously, whether impairments in PMN-based defenses would be more relevant to the initiation of CF lung disease, or to its progression once a chronically infected and inflamed airway environment has been established, remains to be resolved.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Media

We prepared five different media based on RPMI-1640 (BioWhittaker, Walkersville, MD) for experiments investigating PMN secretion of IL-8, and PMN apoptosis and lysis. Without modification this medium contains 108 mM chloride, and we refer to it as "normal chloride" medium. The chloride concentration was raised to 134 mM in the "intermediate chloride" medium and to 170 mM in the "high chloride" medium by the addition of 26 and 62 mM choline chloride (Sigma, St. Louis, MO), respectively. To test whether chloride specifically influences PMN functions, we also prepared a high-salt medium whose excess halide was iodide. To test whether osmolarity itself was a significant influence, we also prepared a medium in which the extra halide was replaced with glycerol. The "iodide" and "glycerol" media were formulated by adding either 62 mM choline iodide (Sigma) or 124 mM glycerol (Sigma), respectively, to RPMI to achieve the same osmolarity as the "high chloride" medium. All media used for these experiments also contained penicillin (100 U/ml), streptomycin (100 µg/ml), and amphotericin-B (0.25 µg/ml) (all from BioWhittaker). Media were free of contamination with endotoxin according to the Pyrotell Limulus amebocyte lysate test (BioWhittaker).

For experiments investigating PMN killing of Psa, additional media were prepared. The base medium was a balanced salt solution (BSS) containing 99.6 mM NaCl, 27.6 mM sodium gluconate, 5.8 mM KCl, 0.3 mM CaCl₂, 1 mM MgCl₂, 10 mM dextrose, 0.013 mM phenol red (all from Sigma), and 20 mM N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid (BioWhittaker) at pH 7.4. This medium contains 108 mM chloride, and used unaltered, we refer to it as "normal chloride" medium in these experiments. "Intermediate chloride," "high chloride," "iodide," and "glycerol" media were modified from standard medium as described above. We also prepared media with lower chloride concentration and/or lower osmolarity than standard. The "low chloride, low osmolarity" medium was formulated by decreasing the concentration of NaCl to 76.6 mM, resulting in a final chloride concentration of 85 mM. The "low chloride" medium was formulated from the "low chloride, low osmolarity" medium by adding 23 mM sodium gluconate, thus maintaining chloride at 85 mM but increasing the osmolarity to match that of the "normal chloride" medium. All media used for these experiments were supplemented with 2.5% pooled human serum (type AB; Sigma).

Study Subjects

Each set of experiments used PMN isolated from blood obtained from at least six healthy adult volunteers. All samples were collected in accordance with a protocol approved by the Massachusetts General Hospital's Committee on the Use of Human Subjects.

PMN Isolation

Human PMN were isolated from venous blood by dextran sedimentation (30 min at 21°C, in 2.5% dextran [Sigma] in normal saline); then density centrifugation on commercial lymphocyte separation media (Organon Teknika, Durham, NC; 35 min at 21°C at 450 × g), followed by hypotonic lysis of residual red blood cells (exposure to 0.2% NaCl at 4°C for 20 s) according to our laboratory's standard method (27). Cells prepared in this manner were > 95% PMN, as determined by Wright-Giemsa staining, and demonstrated > 95% viability, assessed by exclusion of trypan blue (GIBCO, Gaithersburg, MD).

PMN Synthesis of IL-8

Following purification, PMN were resuspended in 1-ml aliquots of each of the five RPMI-based media described previously, in replicate 14-ml round-bottom polypropylene tubes (Falcon, Franklin Lakes, NJ) at 2 × 10⁶ PMN/ ml. To determine baseline values of IL-8 present in culture supernatants at the start of each experiment, the supernatant from one aliquot of PMN in "normal chloride" medium was harvested immediately following resuspension of PMN in the various media. Culture supernatants were harvested after PMN were pelleted by centrifugation at 16,000 × g at 4°C for 5 min. To determine baseline values of IL-8 already synthesized and present within PMN at the start of each experiment, PMN in another aliquot of "normal chloride" medium were lysed by sonication (30 s; Vibra Cell ultrasonic processor; Sonics & Materials, Danbury, CT) and the supernatant from this sample was collected following centrifugation. PMN cultures in each of the five RPMI-based media were then incubated for 24 h at 37°C in 5% CO₂. After this incubation, these cultures were centrifuged and the supernatants collected. Human IL-8 concentrations in these samples were determined by a commercially available enzyme-linked immunosorbent assay (ELISA) (Genzyme, Cambridge, MA). As recommended, samples were stored in polypropylene at 70°C between collection and assay. Values of IL-8 present in the various culture supernatants were standardized for the number of PMN present in those cultures. The measured value of IL-8 in each medium was divided by the baseline number of PMN present in the cultures, which had been counted in two replicate aliquots of cells in that medium. The values we report for IL-8 therefore represent IL-8 per million PMN in culture.

PMN Killing of Psa

The assay used to measure PMN killing of Psa was based on that used by Mizgerd and colleagues (28). Psa were grown from a clinical isolate of nonmucoid Psa. Prior to the experiments, bacteria were incubated with 10% pooled human serum (type AB; Sigma) in RPMI for 20 min (rocking at 36 rpm, 37°C). Psa (1 × 10⁶ colony-forming units [cfu]) then were incubated in 1-ml aliquots of each of the BSS-based media described previously, both with and without 2.5 × 10⁶ PMN. Incubations were performed at 37°C in 14-ml polypropylene tubes, with the same gentle rocking. After 90 min, PMN were lysed with saponin (final concentration 0.05%) and the cultures were serially diluted and plated on bacteriologic agar (Sigma) containing tryptic soy broth (DIFCO, Detroit, MI), then cultured overnight at 37°C prior to counting. Adsorption of the tenaciously adherent Psa to plastics can be minimized by the inclusion of protein in the medium (29). To achieve equivalent protein concentrations in all Psa cultures just prior to the serial dilutions and plating, cell lysates from 2.5 × 10⁶ PMN were added to the cultures without PMN. This step also controlled for any decrement in Psa growth that might have been caused by bactericidal material contained in the PMN lysates present in the cultures originally containing PMN, released by the addition of saponin. PMN killing in a particular medium was determined as the decrement in Psa recovered from cultures with PMN present during the incubation, compared with PMN-free cultures in that medium. This step controlled for potential differential growth rates of free Psa in the various media:

(1)

To compare the effects of the various media on the ability of PMN to kill Psa, percent killing values in each test medium were normalized to the percentage of bactericidal activity observed in "normal chloride" medium controls run in parallel. All statistical analyses, however, were performed on the raw data (before this normalization).

PMN Apoptosis and Lysis

PMN were resuspended (2 × 10⁶ PMN/ml) in 1-ml aliquots of the RPMI-based media described previously, in replicate 14-ml round-bottom polypropylene tubes (Falcon) and incubated at 37°C in 5% CO₂. Previously, we reported that PMN in culture progress sequentially through the following stages: fresh, nonapoptotic cells; apoptotic cells; necrotic cells; and lysed cells (27, 30). PMN in the first two stages, nonapoptotic and apoptotic, retain their membrane integrity and thus their ability to exclude vital dyes such as trypan blue. Necrotic PMN lose their membrane integrity, and thus stain with vital dyes. The number of necrotic cells present at all times is quite small, suggesting that only a short time elapses between PMN necrosis and subsequent disappearance due to cell lysis.

PMN apoptosis was determined according to our laboratory's standard method (27, 30). Briefly, the percentages of apoptotic PMN were determined from Wright-Giemsa stains of the cells cytocentrifuged onto glass slides (500 rpm, 5 min, Shandon cytocentrifuge). Apoptosis was indicated by a change in nuclear morphology: the multisegmented nucleus characteristic of fresh PMN condenses into one to two chromatin-dense apoptotic bodies (Figures 1a and 1b). The appearance of apoptotic bodies has been shown by others (21) and confirmed by our laboratory (30) through DNA gel electrophoresis to be the morphologic representation of apoptosis in PMN. At least 500 PMN were examined per slide in a blinded fashion. Apoptosis assessments were obtained at baseline and after 16, 24, and 48 h of culture.

View larger version (35K): [in this window] [in a new window]   Figure 1.   Photomicrographs of Wright-Giemsa-stained cytocentrifuged (a) nonapoptotic PMN and (b) apoptotic PMN. (a) PMN immediately after isolation, demonstrating the multisegmented nuclear morphology characteristic of nonapoptotic PMN. (b) PMN incubated for 24 h after isolation, demonstrating the chromatin-dense nuclear morphology characteristic of apoptotic PMN. Bars represent 25 µm.

We also determined the number of PMN that were viable at baseline and after 16, 24, and 48 h in culture. These were the number of cells excluding 0.04% trypan blue dye. Because both nonapoptotic and apoptotic PMN exclude trypan blue, both of these categories were counted as viable. PMN survival in culture was determined as the number of viable PMN at a given time divided by the number of viable PMN present in the culture at baseline. PMN percent lysis was 1  PMN survival at a given time:

(2)

This equation actually determines the percentage of cells that have either lysed or become necrotic by a given time. However, because the number of necrotic cells noted at all times was quite small, the equation represents a reasonable reflection of the percentage of PMN that had lysed.

Data Analysis

Data are shown as means ± SEM. Results were analyzed by Student's t test, two-tailed, using Statview statistical software (Abacus Concepts, Inc., Berkeley, CA).

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Effect of Chloride Concentration on PMN Synthesis of IL-8

IL-8 already synthesized and present in PMN at the time of their isolation was measured in the PMN cultures sonicated at the start of the experiments (referred to as "lysate" values, Figure 2). The quantities of IL-8 measured in culture supernatants at the start of the experiments ("supernatant" values, Figure 2) were low, indicating that the processes of isolating and resuspending PMN in the culture medium did not stimulate much PMN secretion of preformed IL-8. Values of IL-8 measured in the culture supernatants after incubation for 24 h are referred to in Figure 2 by the particular medium used, e.g., "normal chloride." In some cultures IL-8 levels exceeded "lysate" values, indicating new PMN IL-8 synthesis during the incubation. PMN IL-8 synthesis was significantly induced by 24-h incubation in the "high chloride" medium (P = 0.0007 versus lysates), but not in the "normal chloride" or "intermediate chloride" media (Figure 2). PMN secretion of IL-8 was also significantly induced in the "iodide" medium (P = 0.044 versus lysates), but not in the "glycerol" medium. IL-8 secreted by PMN and hence detected in the culture supernatants after these 24-h incubations represented approximately 50% of the total IL-8 synthesis induced by exposure to the hyperosmolar media: IL-8 levels measured in cultures with hyperosmolar media in which the PMN were lysed by sonication prior to supernatant collection were approximately twice as high as those measured in cultures in which the PMN remained intact (data not shown).

View larger version (10K): [in this window] [in a new window]   Figure 2.   Neutrophil production of IL-8. "Lysate" refers to IL-8 measured in PMN cultures sonicated at the start of the experiments, immediately after isolated PMN were resuspended in medium. "Lysate" IL-8 therefore represents IL-8 already synthesized and present in PMN at the time of their isolation. "Supernatant" refers to IL-8 measured in the culture supernatants at the start of the experiments. "Supernatant" IL-8 levels were low, indicating that the process of resuspending PMN in the culture medium did not stimulate much secretion of the preformed cytokine. All other IL-8 measurements were obtained from supernatants of PMN incubated for 24 h in the medium noted. IL-8 exceeding "lysate" values indicated that new IL-8 was synthesized during the incubation. PMN IL-8 synthesis was significantly induced by 24-h incubation in the "high chloride" medium (*P = 0.0007 versus lysate), but not in the "normal chloride" or "intermediate chloride" media. PMN secretion of IL-8 was also significantly induced in the "iodide" medium (#P = 0.044 versus lysate), but not in the "glycerol" medium.

Effect of Chloride Concentration on PMN Killing of Psa

In the "normal chloride" media, PMN killed an average of 66.5 ± 2.2% of Psa, comparable with the 64.3 ± 17.0% killing reported by Mizgerd and associates (28) in similar BSS. For purposes of comparison, results of experiments investigating PMN killing in the altered media were standardized to the percentage of bactericidal activity observed in normal medium controls run in parallel (Figure 3); all statistical analyses were performed on raw (not standardized) data. PMN bacterial killing was significantly reduced in the "medium chloride" medium (P = 0.029), and further impaired in the "high chloride" medium (P = 0.0051). PMN killing was also reduced in the "iodide" medium, although the difference was not statistically significant. PMN killing was not significantly impaired in either the "low chloride" or the "low chloride, low osmolarity" medium, and was not altered at all by the "glycerol" medium.

View larger version (13K): [in this window] [in a new window]   Figure 3.   Neutrophil killing of Psa. In this assay, PMN incubated in the "normal chloride" medium killed 66.5 ± 2.2% of Psa. For purposes of comparison, PMN bactericidal activities obtained in the various other media have been normalized relative to the killing activity observed in "normal chloride" medium controls run in parallel with cells from the same donor. Killing was significantly reduced in the "intermediate chloride" medium (*P = 0.029) and further inhibited in the "high chloride" medium (#P = 0.0051). Bactericidal activity was also reduced in "iodide" medium, although the effect did not achieve statistical significance. Other media had no significant effects on PMN ability to kill Psa.

Effect of Chloride Concentration on PMN Apoptosis and Lysis

PMN apoptosis was measured in the various media at baseline and after 16, 24, and 48 h. At 16 h, the percentage of PMN in "high chloride" medium that were apoptotic was significantly greater than the percentage in "normal chloride" medium (P = 0.0097, Figure 4), indicating that PMN apoptosis was accelerated in media with high chloride concentrations. The percentage of PMN in "intermediate chloride" that were apoptotic at 16 h was not significantly different than the percentage in "normal chloride." The comparisons at baseline and after 24 and 48 h were not as informative: at baseline, few PMN were apoptotic in any medium; at 24 and 48 h, almost all viable PMN were apoptotic (data not shown).

View larger version (9K): [in this window] [in a new window]   Figure 4.   Neutrophil apoptosis at 16 h. After 16 h in culture, the percentage of PMN in "high chloride" medium that were apoptotic was significantly greater than the percentage in "normal chloride" medium (*P = 0.0097), indicating that PMN apoptosis was accelerated in the presence of high chloride concentration. The percentage of PMN in "intermediate chloride" that were apoptotic at 16 h was not significantly different than the percentage in "normal chloride."

PMN lysis was also determined in the various media at baseline and after 16, 24, and 48 h. At both 24 and 48 h, the percentages of PMN in "high chloride" medium that had lysed were significantly greater than the percentages lysed in "normal chloride" medium (P = 0.0003 at 24 h, Figure 5a; P = 0.0011 at 48 h, Figure 5b), indicating accelerated PMN lysis in the "high chloride" medium. The percentages of PMN in "intermediate chloride" that had lysed at 24 and 48 h were not significantly different from the percentages in "normal chloride." The comparisons at 16 h were not as informative because at this time, relatively few PMN had lysed in any of the media (data not shown).

View larger version (13K): [in this window] [in a new window]   Figure 5.   Neutrophil lysis at (a) 24 and (b) 48 h. After either 24 or 48 h in culture, the percentage of PMN in "high chloride" medium that had lysed was significantly greater than the percentage in "normal chloride" medium (*P = 0.0003 at 24 h, and #P = 0.0011 at 48 h), indicating that PMN lysis was also accelerated in high chloride concentration. There was a trend toward increased lysis in "intermediate chloride" medium, but the difference was not statistically significant.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

Evidence that host defense is impaired in the lungs of CF patients comes from their early vulnerability to bacterial infections of the airways and their inability to eradicate these infections once initiated. In a recent study of CF infants with a mean age of only 6 mo, nine of 16 infants already had evidence of airway infection by quantitative culture of bronchoalveolar lavage fluid (BALF). In a subsequent investigation, even some infants with negative BALF cultures were found to have evidence of Psa airway infections, demonstrated by polymerase chain reaction or ELISA of BALF (31). Furthermore, it appears that once initiated, bacterial infection in CF airways is rarely eradicated, even with intensive antibiotic treatment. In a study of 18 adolescent and adult CF patients who were > 2 mo out from their last clinically manifest pulmonary exacerbation, all patients were still found by BALF culture to have evidence of ongoing airway infection (32).

One line of host defense that may be impaired in the CF lung is the recently characterized peptide hBD-1. The gene for hBD-1 is widely expressed in epithelia throughout the lung, including the conducting airways, submucosal glands, and alveoli (13). The hBD-1 has broad-spectrum antimicrobial activity, which is abolished by increasing NaCl concentration. If ASF chloride concentration is abnormally high in CF, hBD-1 function would be impaired, thereby providing a causal link between loss of CFTR function and compromised lung defense.

Inactivation of hBD-1 in the CF lung would force a greater dependence on PMN to defend against bacterial infection. This increased reliance on PMN of itself could contribute to CF lung injury, as persistent PMN influx into the airways would be expected to increase the amount of PMN proteases and reactive oxygen intermediates released there. In addition, we have now found evidence that if ASF chloride concentration is abnormally high in CF, PMN defense functions would likely be impaired as well. In this study, we observed that high chloride concentration can alter PMN recruitment, bacterial killing, and PMN clearance, all in a manner that would promote inflammation and could contribute to injury in the lung.

We have demonstrated that increased chloride concentration, in the range that several investigators have observed in CF ASF, can induce PMN synthesis of IL-8, one of the most important PMN chemoattractants present in CF airways. High concentrations of IL-8 have been noted both in sputum (18) and BALF (33) of CF patients, and the majority of PMN chemotactic activity in CF sputum can be inhibited by a blocking antibody to IL-8 (18). Although IL-8 is produced by a wide variety of other cells, including bronchial epithelial cells, monocytes, alveolar macrophages, endothelial cells, and fibroblasts (34), several lines of evidence have demonstrated that recruited PMN themselves are a significant source of IL-8 present in airways, where IL-8 is secreted in an autocrine fashion. Using in situ hybridization and immunohistochemistry, one recent study demonstrated intense IL-8 expression in CF airway luminal PMN, but not in those PMN in airway vessels or submucosal tissue. Somewhat surprisingly, expression of IL-8 in CF respiratory epithelium was not striking, and often was absent altogether (20). In a dog model of PMN recruitment, introduction of cell-free Psa supernatant into the trachea caused accumulation of IL-8 and PMN in the tracheal lumen (19). Inhibition of PMN chemotaxis into the trachea with the leumedin NPC 15669 markedly reduced luminal IL-8 levels in response to introduction of the Psa-derived inducers, suggesting that the recruited PMN contribute significantly to airway IL-8 accumulation in response to Psa (19).

Our results show that high ASF chloride, if present in the CF airway, would provide a stimulus for PMN IL-8 production, in addition to the stimulus provided by the presence of bacteria and bacterial products (19, 35). The additional stimulus provided by high ASF chloride concentration could augment the amount of IL-8 secreted by recruited PMN, and at least partially account for the recent finding that BALF IL-8 levels in CF children with lower respiratory-tract infections were markedly elevated compared with those in non-CF children who also had lower respiratory-tract infections, even when the patients were matched for the pathogen present and for bacterial colony counts (36).

We have demonstrated that high chloride concentration impairs PMN killing of Psa. This disadvantage could contribute to the vulnerability of CF patients to infections with Psa, and to their difficulty in eradicating these infections once initiated. Impairment of PMN phagocytosis of Staphylococcus epidermidis and Escherichia coli by high salt concentration was previously noted in a study of how PMN perform when exposed to the wide variation in solute concentration and osmotic pressure found in the kidney and the urine (37). The observed decrement in PMN killing of Psa in high chloride noted in our study could be attributable to impairment of PMN phagocytosis of Psa, to impairment of PMN killing of Psa once phagocytosed, or both. In a separate set of experiments, we have observed that high chloride concentration does impair PMN phagocytosis of Psa (data not shown). We have not yet examined whether high chloride also affects PMN killing of Psa once phagocytosed.

We have demonstrated that exposure to elevated chloride levels accelerates PMN apoptosis and lysis. PMN are short-lived cells, preprogrammed to undergo apoptosis. Apoptosis is known to be associated with a loss of PMN function, including chemotaxis, phagocytosis, and degranulation (22). Acceleration of PMN apoptosis by aberrant ASF would decrease the functional lifespan of PMN in CF airways, further impairing defense against bacterial infection.

Additionally, accelerated PMN lysis would be expected to increase the quantity of toxic PMN cytoplasmic contents released into CF airways, thereby promoting lung injury. Clearance of apoptotic PMN from sites of inflammation is normally accomplished by macrophages in a manner that minimizes the indiscriminate release of potentially injurious PMN cytoplasmic contents (21). Macrophage phagocytosis of apoptotic PMN normally is completed while the PMN membrane remains intact. If not cleared, apoptotic PMN lose their membrane integrity as they subsequently undergo secondary necrosis and lysis. Therefore, macrophage phagocytosis of apoptotic PMN prior to PMN lysis is believed to allow for the safe clearance of PMN from inflamed sites. Acceleration of PMN apoptosis and lysis by high chloride concentration could overwhelm the clearance capacity of pulmonary macrophages. This would increase the number of PMN undergoing lysis prior to being cleared, and consequently increase the quantity of PMN toxic products released into the airways. Proteolytic enzymes and reactive oxygen intermediates released from PMN have both been implicated in the pathogenesis of CF lung disease (23).

With respect to the mechanism by which high chloride concentration affects PMN function, high salt concentration has previously been shown to induce IL-8 synthesis in human peripheral blood mononuclear cells and human THP-1 monocyte-like cells (38). High salt concentrations induced IL-8 at the transcriptional level, and this induction of IL-8 was inhibited by the p38 MAP kinase inhibitor SB 203508 (39), indicating the involvement of this signal transduction pathway. These investigators found that IL-8 could be induced by three different salts (NaCl, NaI, and KI), and concluded that the signal for IL-8 synthesis was more likely to be osmotic stress, rather than the presence of specific ions. They also found that IL-8 was not increased by the addition of glycerol, which achieved extracellular osmolarity equivalent to those of the media containing the additional dissociable salts. Because glycerol is a cell membrane-permeable solute, and therefore increases tonicity but does not establish an osmotic gradient across cell membranes, these investigators further concluded that hyperosmolarity itself is not a sufficient signal to generate IL-8. Rather, the signal that induces IL-8 in these experiments is likely the establishment of an osmotic gradient across cell membranes. Similarly, we found that PMN synthesis of IL-8 was induced by high extracellular chloride or iodide concentration, but not by glycerol added to achieve the same degree of hyperosmolarity. That the establishment of an osmotic gradient across PMN cell membranes is the signal inducing PMN IL-8 synthesis in our experiments is also suggested by experiments performed using two additional media: "sodium chloride" medium, consisting of RPMI with the chloride concentration increased to the same level as the "high chloride" medium (170 mM) by adding 62 mM sodium chloride rather than choline chloride; and "sodium gluconate" medium, consisting of RPMI with the osmolarity increased to match "high chloride" and "sodium chloride" media by adding 62 mM sodium gluconate. In these experiments, similar amounts of IL-8 were produced by PMN cultured in "high chloride," "sodium chloride," and "sodium gluconate" media (data not shown).

As noted previously, hyperosmolarity has also been shown previously to impair PMN phagocytosis. In a study of PMN function following exposure to the wide variation in solute concentration and osmotic pressure found in the kidney and the urine, Chernew and Braude found a direct correlation between the hyperosmolarity of a solution and the degree to which it depressed PMN phagocytosis (37). Depression of PMN killing by the establishment of an osmotic gradient across PMN cell membranes is also supported by our experiments performed using the "sodium chloride" and "sodium gluconate" media described previously. We observed similar impairments in PMN killing of Psa in "high chloride," "sodium chloride," and "sodium gluconate" media (data not shown). In the PMN killing experiments, as in the IL-8 experiments, raising medium osmolarity with glycerol did not impair PMN function, indicating again that the significant feature is the establishment of osmotic gradients across cell membranes rather than extracellular hyperosmolarity itself. However, our experiments showed that salt concentrations may also influence PMN killing of Psa through mechanisms other than the establishment of osmotic gradients across PMN cell membranes. This is suggested by our observation that PMN killing was affected differently by "high chloride" and "iodide" media, even though these media were formulated to establish equivalent osmotic gradients across cell membranes. In "high chloride" media, PMN killing was 36.8 ± 12.8% of that seen in "normal chloride" media (P = 0.0051), whereas in "iodide" media, PMN killing was less impaired: 65.9 ± 20.8% of that seen in "normal chloride" media (P = 0.18). In his examination of the effectiveness of the PMN myeloperoxidase-halide-H₂O₂ antibacterial system utilizing different halides, Klebanoff demonstrated that iodide ions were more effective on a molar basis than were chloride ions (40). This increased effectiveness of oxidative killing with iodide may contribute to the better-preserved bacterial killing that we observed in the "iodide" media compared with the "high chloride" media.

The establishment of osmotic gradients across PMN cell membranes, rather than the presence of elevated extracellular chloride concentration per se, also appears to be the stimulus responsible for the observed acceleration of PMN apoptosis and lysis in the "high chloride" medium. This was suggested by experiments using the hyperosmolar "sodium chloride" and "sodium gluconate" media, in which we observed similar accelerations of apoptosis and lysis (data not shown) as in the "high chloride" medium.

In our experiments examining PMN synthesis of IL-8 and PMN apoptosis and lysis, we compared PMN function in media with elevated chloride concentration and media with a chloride concentration in the range of normal for human serum, 108 mM. As noted, the chloride concentrations reported in non-CF ASF in both in vivo and in vitro studies has been lower than that of serum, ranging from 83 to 85 mM (11). We did examine the effect of chloride concentration in this lower range in our experiments examining PMN killing of Psa. We found that PMN killing in medium containing 85 mM chloride was not significantly reduced compared with PMN killing in medium containing 108 mM chloride, indicating that PMN function would be well preserved in the lower chloride concentration reported in normal (non-CF) ASF (11).

We have shown that the same abnormally high ASF salt concentration that may compromise defensin function in the CF airway may also impair PMN-based defense, while promoting inflammation and lung injury. Abnormally high ASF chloride and/or sodium or other ions may increase IL-8 synthesis by recruited PMN, thereby recruiting additional PMN in a positive feedback loop and promoting inflammation. Abnormal ASF chloride or other ions may also reduce killing of Psa by PMN recruited into the airways, which promotes infection. Finally, abnormal ASF may accelerate PMN apoptosis and lysis, thereby increasing the quantity of PMN toxic products released into the airways and promoting lung injury. These effects of abnormal ASF chloride or other ions on PMN function could therefore provide another causal link between loss of CFTR function and CF lung disease. If abnormal ASF ionic composition is present in CF as a primary abnormality occurring because of loss of CFTR function, then the effects on PMN function that we have observed could contribute to the early pathogenesis of this disease. Alternatively, if abnormal ASF is a secondary issue in CF due to chronic airway infection and inflammation, then these effects could contribute to progression of CF lung disease later in its course, once an infected and inflamed airway environment has been established.