Introduction

Top Abstract Introduction Materials and Methods Results and Discussion References

More than 70 years ago, Fleming reported that human airway secretions can kill bacteria (1). We now know that airways produce many antibacterial proteins and peptides that may contribute to airway defense. Their contribution will depend on their concentration and activity in the environment at the airway surface. The most abundant airway antimicrobial factors are lysozyme, lactoferrin, and secretory leukoproteinase inhibitor (SLPI); these proteins are produced by serous cells of airway glands (2). Lysozyme is found at approximately 10 µg/ml in airway lavage fluid (1, 3, 4) and 1 mg/ml in sputum (5). Lactoferrin, which kills a variety of bacteria (9), is also relatively abundant at 1 to 10 µg/ml in airway lavage fluid (3, 4) and approximately 1 mg/ml in sputum (5, 7, 8). SLPI, a major elastase inhibitor in the airway (10), is also bactericidal (11) and is found at 0.1 to 2 µg/ml in airway lavage fluid (12, 13) and 2.5 µg/ ml in nasal secretions (14). Other bactericidal peptides and proteins in human airway include -defensin peptides (15- 20) and LL-37 (21). The concentrations of these peptides in airway surface liquid (ASL) are not precisely known, but the -defensins HBD-1 and HBD-2 are present at levels about 1,000-fold lower than lysozyme (T. Ganz [University of California Los Angeles], personal communication; 20).

In cystic fibrosis (CF), the airways are colonized and chronically infected with a variety of bacteria, including Staphylococcus aureus and Pseudomonas aeruginosa (22). The fact that the airway infection does not spread systemically suggests that there is a local host defense defect in CF airways. We have proposed that airway defense is disrupted in CF because an increase in the ASL NaCl concentration inhibits the activity of antibacterial factors (23, 24). Consistent with this hypothesis, the antimicrobial activities of lysozyme (1, 25, 26) and -defensins (17, 20, 27) are inhibited by elevated salt concentrations. Although questions remain (28), there is mounting evidence that normal ASL has a low NaCl concentration, and that CF ASL has a higher NaCl concentration (17, 31).

The goal of this study was to evaluate the properties of the antimicrobial factors in human airways as they relate to CF. Therefore, we assessed their activity toward CF pathogens and analyzed the effects of elevated salt concentrations. We focused specifically on the antibacterial factors that are most abundant in ASL because they are likely to play an important role in the defense against infection. To facilitate these studies, we developed a simple, sensitive, and quantitative antimicrobial assay.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results and Discussion References

Materials

Antimicrobial proteins and peptides included: human neutrophil lysozyme (Calbiochem, San Diego, CA), recombinant human SLPI (R&D Systems, Minneapolis, MN), HBD-1 (synthesized at Louisiana State University, Baton Rouge, LA), human neutrophil HNP-2 (Bachem, King of Prussia, PA), and a synthetic protegrin (a generous gift from Intrabiotics, Sunnyvale, CA). Iron-unsaturated human lactoferrin, porcine cecropin P1, human neutrophil elastase, and bovine submaxillary mucin Type I were from Sigma Chemical Co. (St. Louis, MO).

Bacteria and Assessment of Viability

The effects of samples on bacterial viability were measured using a modification of the procedure described previously (23). The bacteria used were clinical strains of Escherichia coli (urinary tract), nonmucoid P. aeruginosa (CF airway), S. aureus (CF airway), and Burkholderia gladioli (CF airway), and laboratory strains of E. coli (DH5; GIBCO BRL, Frederick, MD) and P. aeruginosa (PAO1). Note that B. gladioli closely resembles Burkholderia cepacia. Moreover, B. gladioli is associated with lung disease in patients with CF (34, 35). Bacteria were grown overnight in an appropriate medium, diluted 1:10, and grown to exponential phase. Bacteria were harvested by centrifugation and suspended in 10 mM potassium phosphate, pH 7.4. Generally, 2,000 to 5,000 colony forming units (cfu) were present in a final volume of 33 µl of the standard assay buffer: 1% Luria-Bertani medium and 10 mM potassium phosphate, pH 7.4. After incubation for the indicated time at 37°C in 96-well polypropylene dishes, organisms were diluted, spread on nutrient agar plates, and incubated 1 to 2 d at 37°C, and colony-forming units (cfu) were determined by standard plate-counting procedures.

The ionic strength of the standard assay buffer was determined by measuring the potassium, sodium, chloride, phosphate, calcium, and magnesium in the potassium phosphate buffer and in the Luria-Bertani medium and calculating the ionic strength. For the standard assay buffer, 10 mM potassium phosphate with 1% Luria-Bertani medium at pH 7.4, the ionic strength was 25 mM.

Luminescence Antimicrobial Assay

For the study of airway antimicrobial factors, we needed an assay that was sensitive, quantitative, and able to handle large numbers of samples. We developed a luminescence assay that measures antimicrobial activity as a decrease in energy-dependent luminescence of bacteria expressing luminescence genes. We used luminescence genes from Photorhabdus luminescens because the luciferase from this organism is more heat-stable than luciferases from other luminescent bacteria (36). E. coli DH5 were transformed with the P. luminescens luminescence genes on the plasmid pCGLS1 (37). This plasmid is a ColE1 replicon containing the luxCDABE operon from P. luminescens and an ampicillin resistance marker. For maintenance of pCGLS1, ampicillin (100 µg/ml) was included in the growth medium. Because the highest levels of luminescence occur in the exponential phase of growth (37), we used cells from cultures at a density of about 10⁹ cells/ml. We also found that although E. coli containing pCGLS1 produce light at 37°C, they are considerably brighter at 30°C; therefore, we chose a temperature of 30°C for our experiments.

Bacteria were grown to exponential phase at 30°C, centrifuged, and resuspended in 10 mM potassium phosphate, pH 7.4, with 1% Luria-Bertani medium. Bacteria (10⁶) were incubated with antimicrobial samples in 96-well plates (Optiplate; Packard Instruments, Meriden, CT) in a total volume of 150 µl. After incubation at 30°C for 4 h, luminescence was measured with a luminometer (Anthos Labtech, Wals/Salzburg, Austria; or Packard Instruments). Luminescence is reported as relative light units. To assess variability in the luminescence assay, we measured the ED₅₀ of lysozyme. We repeated the test on nine different occasions spread out over the course of 9 mo, and found an EC50 of 0.87 µg/ml with a standard error of the mean (SEM) of 0.07.

Lavage Fluids

Procedures were approved by the Human Subjects Review Board of the University of Iowa. Nasal lavage fluid (NLF) was collected from non-CF and CF volunteers. A flexible catheter (18-gauge; Jelco, Tampa, FL) was inserted into each nostril and the area was flushed four times with 4 ml of sterile water. Bronchoalveolar lavage (BAL) fluid (BALF) was obtained from normal volunteers. Aliquots of sterile saline (20 ml) were injected, removed, and filtered through sterile gauze, and cells were removed by centrifugation. Fluids were stored frozen. Fluids were dialyzed with a 1,000 molecular-weight cutoff membrane (Spectra-Por; Spectrum Laboratories, Laguna Hills, CA) for 24 to 48 h against 10 mM potassium phosphate, pH 7.4. Urea, an endogenous marker of lavage fluid dilution (38, 39), was measured to allow estimation of the volume of epithelial lining fluid recovered by the BAL. Urea was measured with a spectrophotometric assay (Sigma Diagnostics Kit 66-UV; Sigma). Protein was measured with the Bradford protein assay (Bio-Rad, Hercules, CA) using bovine serum albumin (BSA) as the standard.

Enzyme-Linked Immunosorbent Assays

Enzyme-linked immunosorbent assays (ELISAs) were performed using microtiter plates (Costar E.I.A plates; Corning, Inc., Corning, NY). All reagents were diluted in Tris-buffered saline (TBS; 25 mM Tris [pH 8.0], 137 mM NaCl, and 2.7 mM KCl). Incubations were for 1 h at 37°C, and washes were with 200 µl TBS with 0.1% Tween-20. The human lysozyme concentration was measured with a microtiter plate ELISA. Wells were coated with 100 µl of either lysozyme standards (2 to 300 ng/ml) or other samples. Wells were blocked with 200 µl 5% normal rabbit serum (Sigma). Wells were incubated with 100 µl sheep antihuman lysozyme (1:1,500 dilution; Calbiochem), then with 100 µl goat antisheep immunoglobulin G alkaline phosphatase conjugate (1:5,000 dilution; Boehringer Mannheim, Indianapolis, IN), and finally with p-nitrophenyl phosphate substrate (Calbiochem). Absorbance was read at 420 nm on a microtiter plate reader (Anthos Labtech). Concentrations of lysozyme were calculated from a standard curve (linear range 8 to 200 ng/ml). Human lactoferrin was measured with a sandwich-type microtiter plate ELISA. Wells were coated with 100 µl of 10 µg/ml rabbit antihuman lactoferrin (Jackson ImmunoResearch, West Grove, PA), then blocked with 200 µl 5% normal rabbit serum. Wells were then incubated with either lactoferrin standards (2 to 300 ng/ml) or other samples. Wells were incubated with 100 µl rabbit antihuman lactoferrin alkaline phosphatase conjugate (1:5,000 dilution; Jackson ImmunoResearch), then with p-nitrophenyl phosphate substrate (Calbiochem). Absorbance was read at 420 nm on a microtiter plate reader (Anthos Labtech). Concentrations of lactoferrin were calculated from a standard curve (linear range 8 to 200 ng/ml).

	Results and Discussion

Top Abstract Introduction Materials and Methods Results and Discussion References

Airway Antimicrobial Factors Kill CF Pathogens

Because lysozyme and lactoferrin may be important in airway defense, and because killing will be dependent on the concentration of antimicrobial factor, we estimated the levels of these proteins in ASL. Using ELISA and Western blots, we found that NLF contained 0.8 ± 0.2 µg/ml lysozyme (n = 5) and 3.1 ± 1.0 µg/ml lactoferrin (n = 5). BALF contained 0.7 ± 0.1 µg/ml lysozyme (n = 5) and 0.9 ± 0.2 µg/ml lactoferrin (n = 5). These values are similar to values reported earlier (3, 4, 40). To correct for dilution of ASL during lavage, we measured the urea concentration in lavage fluids and compared them with normal serum urea levels. With this method, we estimate that airway secretions were diluted 25- to 33-fold during lavage, similar to the ~ 100-fold dilution reported previously (38). Therefore, lysozyme and lactoferrin are present in the range of 20 to 100 µg/ml in airway lining fluid.

We measured the antibacterial activity of lysozyme, lactoferrin, and SLPI against several bacteria, including clinical strains of P. aeruginosa, S. aureus, and B. gladioli that were obtained from CF airway. Figure 1 shows that lysozyme killed clinical and laboratory strains of E. coli and P. aeruginosa, as well as a clinical strain of S. aureus. Lactoferrin also killed these bacteria, but was generally less potent than lysozyme. SLPI killed E. coli and a laboratory strain of P. aeruginosa, but it did not kill clinical strains of P. aeruginosa or S. aureus. The amounts of these agents in ASL and their potency suggest that lysozyme and lactoferrin are present in the airway at levels sufficient to kill the CF pathogens P. aeruginosa and S. aureus. Interestingly, lysozyme and lactoferrin were relatively ineffective and SLPI failed to kill B. gladioli. Burkholderia species are also resistant to many pharmaceutical antibiotics and have been associated with severe CF pulmonary infections late in the course of the disease (22, 35, 41). Although the airway environment is very different in early and late CF, it is possible that infection with Burkholderia species may be partly related to their resistance to airway antimicrobial factors.

View larger version (39K): [in this window] [in a new window]   Figure 1.   Antibacterial spectrum of lysozyme, lactoferrin, and SLPI. Bacteria were laboratory strains of E. coli DH5 (open circles) and P. aeruginosa PAO1 (open triangles); and clinical strains of E. coli (filled circles), mucoid P. aeruginosa (filled triangles); S. aureus (filled squares), and B. gladioli (filled diamonds). Bacteria were incubated with the indicated concentrations of antimicrobial factor for 3 h, then surviving organisms were assessed by the plate-count method. Values are percent of control in the absence of antimicrobial factor. Symbols indicate means ± SEM, n = 3. Similar results were obtained in one to three other experiments.

Luminescence Antimicrobial Assay

For a more detailed analysis of airway antimicrobial factors, we developed a luminescence assay that was sensitive, quantitative, and able to handle large numbers of samples. This assay measures antimicrobial activity as a decrease in energy-dependent luminescence of bacteria expressing luminescence genes. Because a sensitive indicator organism would allow us to conserve material, we chose to develop the luminescence assay with E. coli DH5; our studies of several organisms indicated that E. coli DH5 is relatively sensitive to a number of antimicrobial factors (Figure 1).

Because bacterial luminescence requires cellular energy, we expected that the amount of luminescence would be related to the number of living bacterial cells. To determine whether bacterial luminescence was directly related to viability, we compared the amount of luminescence and the number of viable cells remaining after incubation with lysozyme. Lysozyme caused a dose-dependent decrease in bacterial luminescence (Figure 2A). Importantly, we obtained the same relationship when viability was measured with the plate-count assay.

View larger version (31K): [in this window] [in a new window]   Figure 2.   Antimicrobial activity measured with the luminescence assay. (A) Relationship between luminescence (filled circles) and viability (open circles) of E. coli DH5 containing pCGLS1. Bacteria were incubated with the indicated concentrations of lysozyme, then luminescence was measured and surviving organisms were plated and counted. Values are percent of control in absence of lysozyme. Symbols indicate means ± SEM, n = 4. (B) Effect of concentration of several antimicrobial peptides in the luminescence assay. Values are percent of control in the absence of antimicrobial factor. Symbols indicate means ± range, n = 2; similar results were obtained in two to five separate experiments. In some cases, the error bars are covered by the symbols.

We used the luminescence assay to compare the potency of several known antimicrobial peptides and proteins. Figure 2B shows that all samples decreased bacterial luminescence in a dose-dependent manner. The rank order of potency against E. coli was porcine protegrin = human lysozyme > porcine cecropin P1 > human SLPI = human lactoferrin > human HNP-2 > human HBD-1. These results show that the luminescence assay can be used to evaluate a variety of antimicrobials. The data also suggest that both the relative potencies and the abundance of antimicrobial factors in the ASL may be important in considering the contribution to airway defense.

Are Airway Antibacterial Factors Inhibited by Elevated Salt Concentrations?

Because the NaCl concentrations may be elevated in CF ASL, we tested its effect on airway antimicrobial factors. Figure 3 shows that the antibacterial activity of lysozyme, lactoferrin, and SLPI, as well as NLFs and BALFs, was inhibited by elevated NaCl. We also examined the antibacterial activity of CF sputum (Figure 3). CF sputum showed significant antibacterial activity when it was diluted into a low-salt buffer; this activity was inhibited in the high-salt buffer.

View larger version (34K): [in this window] [in a new window]   Figure 3.   Effect of NaCl on antimicrobial activity. Bacteria were incubated with the indicated concentrations of antimicrobial factor in the absence (open circles) or presence (filled circles) of 150 mM NaCl added to the standard assay buffer (ionic strength = 25 mM). Sputum was obtained from an individual with CF who was not taking antibiotics. Sputum was frozen and thawed twice; diluted 10-fold in 10 mM potassium phosphate, pH 7.4; and vortexed for 5 min. Values are percent of control luminescence in the absence of antimicrobial factor; controls were determined for each salt concentration. Symbols indicate means ± range, n = 2; the error bars are covered by the symbols. Similar results were obtained in two or three other experiments.

Is Antimicrobial Activity Inhibited by Ionic Strength or by Osmolarity?

To learn whether specific ions inhibit the antimicrobial factors or whether inhibition results from high ionic or high osmotic strength, we examined the effects of various salts and osmolytes on antimicrobial activity. Figure 4 shows that the activity of lysozyme, lactoferrin, NLF, and BALF was inhibited by several salts. There was near-complete inhibition of activity at 150 mM NaCl, and in some cases there was stimulation of bacterial growth. Inhibition by KCl and Na gluconate were indistinguishable from inhibition by NaCl, suggesting that there was no specific requirement for either Na⁺ or Cl. When Na⁺ was replaced by the divalent cation Mg²⁺ or Ca²⁺, there was significant inhibition of antimicrobial activity at lower concentrations (1 to 5 mM). The total Ca²⁺ in ASL is about 4 mM; much of this, however, may be bound to anionic macromolecules (31). Nonionic osmolytes had little effect on antimicrobial activity. Glycerol was not inhibitory, even at concentrations of 300 mM, comparable in osmotic strength with 150 mM NaCl. Mannitol was only slightly inhibitory at 300 mM. These results indicate that the antimicrobial activity is active at low ionic strength, but does not require a hypotonic environment. These experiments also suggest that inhibition does not require specific ions and that some divalent cations are more effective inhibitors than monovalent cations. Cations also inhibit the antimicrobial activity of defensin antimicrobial peptides (42). Cations may compete for binding sites on the lipopolysaccharide surface of the outer membrane (43). It is also possible that cations bind bacteria and shield them from antimicrobial factors.

View larger version (24K): [in this window] [in a new window]   Figure 4.   Effect of ionic and nonionic osmolytes on antimicrobial activity. Bacteria were incubated with lysozyme (15 µg/ml), lactoferrin (230 µg/ml), NLF (9 µg/ml), or BALF (34 µg/ml) in the presence of the indicated concentrations of CaCl2 (closed triangles), MgCl2 ( filled circles), NaCl (open triangles), KCl (open circles), Na gluconate (open squares), glycerol ( filled squares), or mannitol ( filled diamonds). Values are percent of control luminescence in the absence of antimicrobial factor. Symbols indicate means ± SEM, n = 3. The error bars are covered by the symbols. Similar results were obtained in two or three separate experiments.

What Ionic Strength Inhibits Antimicrobial Activity?

We examined the relationship between ionic strength, the concentration of several antimicrobial factors, and antimicrobial activity (Figure 5). We found that the inhibitory effect of increased ionic strength could be partially overcome when we added more of each antimicrobial protein or lavage fluid. For example, relatively low ionic strength (35 mM) strongly inhibited the activity of 2 µg/ml lysozyme, but had little effect on 70 µg/ml lysozyme. However, 70 µg/ml lysozyme was almost completely inhibited by higher ionic strength (125 mM). We also tested killing of P. aeruginosa. Figure 6 shows that killing of P. aeruginosa by lysozyme was inhibited by increased ionic strength, and inhibition depended on the concentration of lysozyme. These results indicate that there is no one ionic strength that inhibits antimicrobial activity; inhibition depends on the concentration of antimicrobial factor. These data also suggest that a CF defect in bacterial killing due to increased ASL NaCl concentration might be overcome by increasing the concentration of antibacterial factors in the airway.

View larger version (35K): [in this window] [in a new window]   Figure 5.   Relationship between the ionic strength, antimicrobial factor concentration, and antimicrobial activity. Bacteria were incubated with various amounts of antimicrobial factor in the standard assay buffer (ionic strength = 25 mM) with NaCl added to obtain the indicated ionic strength; luminescence was measured. The effects of salt were tested on three amounts of each factor: low ( filled circles), intermediate (open triangles), and high (closed triangles). Factors were lysozyme: 2, 7, and 70 µg/ml, respectively; lactoferrin: 70, 200, and 700 µg/ml; SLPI: 5, 20, and 80 µg/ml; NLF: 2, 20, and 200 µg protein/ml; and BALF: 3, 30, and 300 µg protein/ml. Values are percent of control in the absence of antimicrobial factor (open circles); controls were determined in each experiment for each ionic strength. Symbols indicate means ± range, n = 2; in some cases the error bars are covered by the symbols. Similar results were obtained in two separate experiments.

View larger version (19K): [in this window] [in a new window]   Figure 6.   Relationship between ionic strength and lysozyme concentration on killing of P. aeruginosa PAO1. Bacteria were incubated for 24 h at 37°C with 3 (open circles), 15 (open triangles), or 30 ( filled squares) µg/ml lysozyme in the standard assay buffer (ionic strength = 25 mM) with NaCl added to obtain the indicated ionic strength. Surviving organisms were assessed by the plate-count method. Values are percent of control in the absence of lysozyme. Symbols indicate means ± SEM, n = 3. The error bars are covered by the symbols.

Are the Airway Antimicrobial Factors Inhibited by Mucin or Elastase?

Colonized CF airways contain high levels of mucin and elastase. Mucins are epithelial proteins containing highly anionic sugars, and they are known to bind lysozyme (7). Elastase is a protease produced by bacteria and neutrophils. Bacterial elastase cleaves human airway lysozyme in vitro, destroying its bacteriolytic activity (44), but the effects of neutrophil elastase on lysozyme bactericidal activity have not been tested. In CF airways, lactoferrin and SLPI were found to be cleaved, but the effects on bactericidal activity were not examined (12, 45). Figure 7 shows that addition of mucin to lysozyme, lactoferrin, and NLF had no effect on their antimicrobial effectiveness, and mucin did not alter the inhibition by elevated ionic strength. Mucin alone did not have antimicrobial activity. Figure 8 shows that neutrophil elastase did not alter bacterial killing by lysozyme, lactoferrin, NLF, or BALF. The elastase concentration used was sufficient to proteolyze BSA, but it did not cleave lysozyme or lactoferrin (not shown). Elastase was not directly bactericidal in this assay, although higher concentrations of neutrophil elastase are bactericidal (46). We also found that porcine pancreatic elastase, which did cleave lysozyme and lactoferrin, failed to inhibit activity of the antibacterial factors (not shown). These results suggest that mucin and neutrophil elastase do not inhibit antimicrobial factors.

View larger version (26K): [in this window] [in a new window]   Figure 7.   Effects of mucin on antimicrobial activity. Bacteria were incubated with lysozyme (7 µg/ml), lactoferrin (67 µg/ml), or NLF (80 µg/ml) in the absence (open circles) or presence ( filled circles) of bovine submaxillary mucin (a 2- to 20-fold molar excess). Assays contained the standard assay medium (ionic strength = 25 mM) with NaCl added to obtain the indicated ionic strength. Mucin had been dialyzed with a 10,000 molecular weight cutoff membrane for 24 to 48 h against H2O. Control panel shows the effect of mucin in the absence of antimicrobial factors. Values are percent of control luminescence in the absence of antimicrobial factor and mucin. Symbols indicate means ± range, n = 2. In some cases, the error bars are covered by the symbols.

View larger version (28K): [in this window] [in a new window]   Figure 8.   Effects of human neutrophil elastase on antimicrobial activity. Antimicrobial factors were preincubated with 30 µg/ml human neutrophil elastase (or H2O) for 3.6 h at 30°C. Bacteria were incubated with elastase alone (open triangles) or the indicated concentration of antimicrobial factor which had been pretreated with elastase (open circles) or with H2O ( filled circles), and luminescence was measured. Control experiments showed that elastase was able to proteolyze several control proteins (not shown). Values are percent of control in the absence of antimicrobial factor. Symbols indicate means ± range, n = 2. In some cases, the error bars are covered by the symbols.

Do CF Airway Secretions Have Reduced Antimicrobial Activity When Studied at a Low NaCl Concentration?

A reduction in the net amount of antimicrobial activity in CF airways might explain the susceptibility to infection. Previous studies have shown that lysozyme and lactoferrin levels are not decreased in CF airway secretions, and may even be increased (5, 47, 48). Those studies, however, did not examine antimicrobial activity. To assess the activity in airway secretions, we diluted airway lavage fluids into a low-salt buffer and used the luminescence assay to measure activity. We found that antimicrobial activity was similar in NLF from non-CF and CF individuals (Figure 9). The similarity in antimicrobial activity suggests that something about the environment of the CF airway may prevent these factors from killing bacteria. Our current data, combined with earlier measurements of the NaCl concentration in CF airways, suggest that an abnormally high salt concentration may impair airway antimicrobial activity.

View larger version (17K): [in this window] [in a new window]   Figure 9.   Antimicrobial activity of NLF from CF ( filled circles) and non-CF (open circles) airways. Bacteria were incubated with the indicated amount of NLF and luminescence was measured. Values are percent of control in the absence of NLF. Symbols indicate means ± SEM for CF (n = 5 specimens) and non-CF (n = 12 specimens) samples.

Implications

The data show that the abundant endogenous antimicrobial factors in ASL are inhibited by a high ionic strength. The level of inhibition depends on the concentration of salt, the concentration of the antimicrobial factor, and perhaps other components of ASL. Thus, there is no one unique salt concentration that inhibits antimicrobial activity; rather, the greater the amount of antimicrobial factor or the lower the ionic strength, the greater the bacterial killing. These results suggest that at least partial disruption of this pulmonary defense system may contribute to the pathogenesis of CF. The data also show that endogenous antimicrobials kill different organisms in a salt-sensitive way. For example, killing of E. coli was salt-sensitive. Yet E. coli pulmonary infections are rare in both normal and CF individuals. Why is it that CF airways are predisposed to infections with P. aeruginosa, S. aureus, and Burkholderia species? It is possible that other factors contribute to the prominence of specific bacterial species as the disease progresses. For example, bacterial exposures, differential adherence, specific virulence factors, resistance to pharmaceutical antibiotics, and defects in phagocytosis may favor emergence of common CF pathogens with disease progression.

Our data, combined with previous studies, suggest that antibacterial defense at the airway surface is a very complex process. Several considerations must be kept in mind. First, ASL contains a variety of antimicrobial factors; these include lysozyme, lactoferrin, SLPI, -defensins (HBD-1 and HBD-2), and the cathelicidin LL37, as well as neutrophil- and macrophage-derived factors such as HNP1-3 (49). Second, the potency of each factor toward specific bacteria will determine, in part, the activity toward inhaled and aspirated bacteria. Third, the amount of the individual factors in ASL is a very important issue, especially because our data suggest that the potencies of the different factors are not dramatically different, whereas the amounts of the factors may differ substantially. Fourth, the various factors may show additive, synergistic, or antagonistic interactions; this needs to be tested. Fifth, airway antibacterial factors may interact with other proteins, lipids, or carbohydrates in ASL; although our data suggest that mucin and elastase do not affect activity of the abundant factors, other components could be important. Finally, the electrolyte composition of the ASL environment may be critical to antimicrobial function. For future studies designed to evaluate this process in CF and non-CF airways, these issues may be key.

Our data suggest several possible approaches to prevent or treat bacterial infections in the CF lung. First, the data show that inhibition of the antimicrobial factors by NaCl can be partially compensated by increased amounts of antimicrobial factor. If this finding holds in vivo for clinically relevant CF pathogens such as P. aeruginosa, then addition of exogenous antimicrobial factors might have therapeutic value. It is possible that stimulation of synthesis of antibacterial factors could also be beneficial. A second potential therapeutic approach would be to decrease the salt concentration in CF ASL, allowing endogenous factors to kill bacteria. Our finding that isosmotic concentrations of nonionic osmolytes do not inhibit killing suggests that they might be used to draw water to the apical surface of epithelia, thereby reducing the NaCl concentration and removing inhibition of the antimicrobial factors. Third, it may be possible to treat CF with salt-insensitive antimicrobial factors such as the porcine protegrins (50). Protegrins, or other salt-insensitive antimicrobials, are candidates for replacing the inactive antimicrobial factors in CF airways. Additional understanding of this defense system should yield new approaches to the therapy of CF as well as other infectious lung diseases.