Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Recent studies suggest that the pulmonary consequences of cystic fibrosis (CF) may be initiated through an early breach in innate immunity (1). A direct link between the genetic defect in CF and host defense was provided by Smith and colleagues, who demonstrated strong antibacterial activity in airway surface fluid (ASF) from normal lung epithelia that was substantially reduced in CF ASF (2). The initial hypothesis was that elevated NaCl in CF ASF inactivates antibacterial molecules (3). Several groups have isolated and characterized host defense proteins from airway secretions. Recent interest has focused on antimicrobial peptides. Members of the -defensins and the cathelicidins are expressed in airway epithelia and secreted into the airways (4). The causative "trigger" in the defensin inactivation hypothesis, i.e., elevated NaCl, has been difficult to confirm due to technical problems in measuring the actual salt concentration of the ASF and modeling the airway in vitro (8). Several hypotheses are discussed at the moment to explain the link between dysfunctional CF transmembrane conductance regulator (CFTR) and the host defense defect in CF. Compromised antimicrobial activity in the CF airway could also be explained by inactivation via other compositional alterations of the ASF or defects in the biogenesis of airway secretions secondary to CFTR expression.

The goal of this study was to characterize ASF biochemically to evaluate abnormalities that could explain decreased bacterial killing function in CF. ASF from normal and CF human bronchial xenografts and bronchoalveolar lavage fluid (BALF) from children with CF or other diseases was characterized for known antibacterial proteins or peptides. Samples were also analyzed biochemically for differences in their profiles of antimicrobial killing in an attempt to characterize relevant components heretofore not identified or implicated.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Preparation of Antibodies Against Human -Defensins 1 and 2 and LL-37

Recombinant human -defensin (hBD)-2 (41 amino acid form, molecular weight [MW] = 4327.8) and hBD-1 (44 amino acid form, MW = 4767.0) were generated or purified as described previously (4). LL-37/hCAP-18 (C-terminal 37 amino acids) was synthesized chemically (Louisiana State University Medical Center, Core Laboratories, Baton Rouge, LA). Antisera against hBD-1, hBD-2, and LL-37 were obtained by injecting 100 µg of peptide coupled to keyhole limpet hemocyanin into two rabbits for each peptide (Research Genetics, Huntsville, AL) and purified using protein A columns.

Human Bronchial Xenograft Model

Human bronchial xenografts were generated using human normal and CF respiratory epithelial cells (4, 9). The CF cells were isolated from four lungs that were removed during lung transplantation (all four donors were F508 homozygous). ASF was expelled with air from the xenograft twice weekly and extracted in acetonitrile and trifluoroacetic acid (TFA). After lyophilization, the substance was resuspended in water and cleared by centrifugation. For further purification, a part of the material was loaded on an Ultrasphere C-18 RP-HPLC column (Beckman, Fullerton, CA) and eluted using a linear gradient of acetonitrile with 0.1% TFA. Fractions were dried and resuspended in distilled water. For reconstitution of the genetic defect of CF cells grown in the xenograft system, E1-deleted adenovirus coding for the full-length complementary DNA (cDNA) of CFTR under the control of the cytomegalovirus promoter (10) was injected into differentiated xenografts at a dose of 5 × 10¹⁰ particles in 100 µl of sterile phosphate-buffered saline (PBS).

BALFs from CF and non-CF Patients

Bronchoscopy and bronchoalveolar lavage were performed as previously described (11, 12). Each CF patient was studied as part of an ongoing protocol to examine inflammation and infection in BALF cross-sectionally in CF. BALF was centrifuged and the supernatant was frozen at 70°C. Control patients underwent bronchoscopy for clinical indications. A total of 19 CF and 10 non-CF patients were included in the study. The CF patients (average age 8.8 yr, range 1 to 24 yr) were characterized regarding their genotype (eight F508 homozygous, one F508/2789 + 5G-A, one F508/2184delA, one F508/W1282X, one F508/ R347P, and one F508/unknown; plus one G542X/unknown, one 621 + 1G > T/W1282X, and four unknown) and their bacteriology (three Staphylococcus aureus positive, three Pseudomonas aeruginosa positive, and four other positive results). The bacteriology results of the control patients (average age 11.3 yr, range 2 to 23 yr) showed four positive results.

Determination of Peptide Concentrations by Dot-Blot and Immunoblot Analysis

From each sample, 2 µl was dotted onto a nitrocellulose membrane and immunolabeling was performed using the polyclonal rabbit antibodies to hBD-1, hBD-2, or LL-37/hCAP-18 (1:500). Bound secondary, peroxidase-labeled antibodies were visualized by chemoluminescent substrate and exposure to X-ray films. The concentration of substances was determined by quantification of the signal intensity and comparison with signals from known amounts of purified peptides. Samples were also used for immunoblotting after separation by sodium dodecyl sulfate polyacrylamide gel electrophoresis using a 15% Tris-tricine gel.

Antimicrobial Assays and Desalting Procedures

For direct measurement of the antimicrobial activity of ASF from xenografts, 10⁴ colony-forming units (cfu) of P. aeruginosa PAO1 in 5 µl of PBS were added to 50 µl of the ASF. After 2 h of incubation at 37°C the material was plated onto LB agar plates, and bacterial growth was determined after 24 h of incubation. To detect high-performance liquid chromatography (HPLC) fractions that have antimicrobial activity, an agar diffusion assay was used as described earlier (4). To quantify the antimicrobial activity of ASF samples or individual HPLC fractions, 2-fold dilutions of the material were prepared in 10 mM sodium phosphate buffer (pH 7.0), each tube containing a final volume of 100 µl. A total of 10⁴ cfu of P. aeruginosa PAO1 were added to each tube. After 2 h of incubation at 37°C the material was plated, and bacterial growth was determined after 24 h of incubation. These assay conditions were developed empirically and allow a sensitive and reproducible measurement of antimicrobial activity in a fluid. For desalting of crude ASF, pooled flushings from five to ten xenografts were loaded onto a SepPak C-18 column (Waters, Milford, MA), washed with water, and eluted with acetonotrile (all solutions contain 0.1% TFA). After lyophilization, the material was resuspended in 10 mM sodium phosphate buffer. Samples from normal xenografts were treated in different ways to further characterize the fraction of interest obtained from reverse-phase (RP)-HPLC. In one set of experiments, the NaCl concentration of the samples was adjusted to 200 mM in each sample. In other sets, ethylenediaminetetraacetic acid (EDTA) was added to each sample (final concentration, 20 mM). In two sets the samples were heated to 100°C for 30 min or treated with trypsin (0.5 mg/ml) and proteinase K (0.1 mg/ml) for 60 min before diluting. All samples were normalized according to their total protein concentrations.

Statistical Analysis

Data are displayed as means ± standard error of the mean (SEM) and were compared using analysis of variance; two-tailed, unpaired Student's t test; or Rank sum test, when indicated.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

A Salt-Independent Abnormality in Antimicrobial Activity of CF ASF

Previous reports have described diminished antimicrobial activity in ASF from cultured CF airway cells. We systematically evaluated this finding in a human bronchial xenograft model. As shown in Figure 1, ASF harvested soon after seeding of normal and CF cells in the xenograft showed antimicrobial activity, likely due to the presence of residual antibiotics in the cell culture medium. Xenografts made from normal cells acquired substantial bacterial killing activity after 3 wk, whereas CF grafts became and remained relatively incompetent in bacterial killing. This killing defect was completely reversed after adenovirus-mediated gene transfer of CFTR (Figure 1).

View larger version (20K): [in this window] [in a new window]   Figure 1.   Antimicrobial activity of ASF from normal and CF xenografts, shown as a function of time after seeding the cells into the grafts. Activity against P. aeruginosa PAO1 was measured as described in MATERIALS AND METHODS. ASF harvested soon after seeding of the cells in the xenograft showed antimicrobial activity, likely due to the presence of antibiotics in the cell culture medium. Xenografts made from normal cells acquired substantial bacterial killing activity after 3 wk whereas CF grafts remained relatively unable to kill bacteria. Adenoviral gene transfer of CFTR cDNA corrected the killing defect of CF xenografts. Each data point represents mean ± SEM for n = 15.

The previously reported finding that diminished bacterial killing of CF ASF is reversed by lowering the concentration of salt suggests that the defect is reversible and a product of its milieu (2). We expanded this work by performing a controlled and detailed comparison of normal and CF ASF from human bronchial xenografts. Proteins and peptides from xenograft ASF were desalted by fractionation through a C-18 SepPak column. The material was lyophilized, resuspended in 10 mM sodium phosphate buffer, and assayed for antimicrobial activity in 10 mM sodium phosphate buffer using a quantitative dilution assay. The antimicrobial activity was calculated from serial 2-fold dilutions of the ASF and determined to be the dilution at which material no longer inhibited bacterial growth. The desalting step resulted in an increase in antimicrobial activity of the samples from both CF and normal xenografts as compared with the crude material, although the normal, desalted sample titrated to much higher levels of antimicrobial activity than did the corresponding CF sample (Figure 2). Following desalting, normal ASF maintained antimicrobial activity after being diluted 512-fold; however, CF ASF could only be diluted 32-fold before losing killing activity (Figure 2). Mixing experiments indicated that the abnormality in CF ASF was due to the absence of activity rather than the presence of an inhibitor (data not shown). This suggests that a deficiency of CFTR leads to a defect in the production of an important antimicrobial molecule whose dysfunction is apparent even in the absence of NaCl.

View larger version (27K): [in this window] [in a new window]   Figure 2.   Antimicrobial activity of ASF from human bronchial xenografts. ASF from CF and normal xenografts was desalted using a SepPak C-18 column, and antimicrobial activity was determined by diluting the material in 10 mM sodium phosphate buffer in 2-fold steps and determining the ability to kill P. aeruginosa PAO1. Whereas crude material showed relatively low antimicrobial activity, after desalting, both CF and normal material revealed increased antimicrobial activity. The normal material could be diluted many more times than the CF material. Absorbance at 220 nm was recorded.

Concentrations of Known Antimicrobial Peptides Are Equivalent in CF and Normal ASF

Several experiments were performed to characterize known antimicrobial protein/peptides in the context of the CF defect. ASF from CF and normal xenografts was fractionated by RP-HPLC and individual fractions with antimicrobial activity were immunoblotted using antibodies against hBD-1, hBD-2, the cathelicidin LL-37hCAP-18, lysozyme, and lactoferrin. There was no apparent difference in the electrophoresis migration patterns or concentrations of the molecules between normal and CF (Figure 3). A sensitive and specific quantitative assay was developed to more accurately quantititate hBD-1, hBD-2, and LL-37/hCAP-18 in ASF from xenografts and BALFs from CF and normal infants. Samples from xenografts were treated by organic extraction, resulting in better recovery of antimicrobial peptides. Initial studies with xenograft material isolated at different time points after seeding of the cells showed that levels of hBD-1 and hBD-2 increased as a function of time (data not shown). Patient samples were available only after centrifugation, as per the protocol of the clinical investigation. As shown in Figure 4, concentrations of antimicrobial peptides are not significantly different in ASF from xenografts (Figure 4A) or BALFs of CF patients (Figure 4B) as compared with control samples. Repeated measurements of samples from the same patient were reproducible for all peptides except hBD-2 from patient samples (data not shown). Differences in the sampling procedures (organic extraction only for the xenograft material) likely contribute to this finding. Therefore, these data were not included in this analysis.

View larger version (49K): [in this window] [in a new window]   Figure 3.   Presence of several known host defense proteins in the ASF from CF and normal xenografts. ASF was isolated from xenografts and separated by RP-HPLC. Individual fractions were assayed for antimicrobial activity using an agar diffusion assay. Fractions that revealed antibacterial activity were used for immunoblots using antibodies against lactoferrin, lysozyme, hBD-1, hBD-2, and LL-37/hCAP-18. All these substances could be detected in the ASF of CF and normal xenografts after elution at characteristic time points during the chromatography. MW standards (in kD) are shown along the left border.

View larger version (43K): [in this window] [in a new window]   View larger version (39K): [in this window] [in a new window]   Figure 4.   Concentrations of antimicrobial peptides in ASF collected from normal and CF xenografts (each group n = 15) (A) or obtained by bronchoscopy of CF (n = 19) and control patients (n = 10) (B). Samples were extracted in acetonitrile/TFA and concentrations were determined by a quantitative dot-blot assay and normalized to the total protein concentration. Concentrations of the antimicrobial peptides were not decreased in samples from patients with CF compared with samples from normal patients. Demographic data of the patients are described in MATERIALS AND METHODS.

The Salt-Independent Defect in Antimicrobial Killing of CF ASF Is Due to an Abnormality in Its Macromolecular Composition

To further evaluate differences in the biochemical composition of normal and CF ASF, samples were separated by RP-HPLC and each fraction was assayed for antimicrobial function (Figure 5). The profile of the chromatogram as measured by optical density at 220 nm was identical between CF and normal ASF. Quantitative analysis of fractions for bacterial killing showed equivalence in the majority of the peaks; however, diminished activity in CF samples was found in peaks eluting at 38% acetonitrile. Dot-blot analysis indicated that these fractions contained small amounts of hBD-1, hBD-2, and lysozyme, with equivalent concentrations in CF and normal samples, although many other proteins were present (data not shown). Comparisons with assays of the purified -defensins and lysozyme indicated that the total bactericidal activity in this HPLC fraction could not be explained by the presence of these known molecules.

View larger version (24K): [in this window] [in a new window]   Figure 5.   Antimicrobial activity of ASF from CF and normal xenografts separated using RP-HPLC. The fractions were individually assayed for their antimicrobial activity. Certain fractions showed decreased antimicrobial activity in CF samples as compared with normal samples (arrows), whereas other fractions had equivalent activity. Experiments were repeated five times with identical results.

The causal relationship between decreased bacterial killing and CFTR deficiency was confirmed in CF xenografts infected with a CFTR-expressing adenoviral vector, which resulted in restoration of the antimicrobial activity in the total ASF and 38% acetonitrile fraction (Figures 1 and 5). Antimicrobial assays of HPLC samples of the active fraction from normal xenografts (corresponding to the 38% acetonitrile peak) were performed after treatment with heat or proteases or in the presence of NaCl or EDTA. Protease treatment and addition of salt resulted in the reduction of antimicrobial activity of normal samples, whereas heat treatment did not change the activity. EDTA was not informative because it killed bacteria directly (Table 1).

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

Defects in the innate immune system of the airways have been proposed to play a role in the pathogenesis of CF lung disease. Investigations of the relationship between CFTR dysfunction and the development of CF lung disease have provided two contrasting explanations (8). The "isotonic/low volume" hypothesis of Matsui and colleagues states that underhydrated CF ASF causes impaired clearance of secretions and bacteria (13). The "high salt/defensin" hypothesis of Zabner and colleagues implies that CFTR dysfunction results in an elevated salt concentration in CF ASF, which inactivates antimicrobial substances (14). The actual salt concentration of human ASF (CF or normal) is not known and is exceedingly difficult to measure. It was the aim of the present study to investigate whether antimicrobial function of CF airway secretions is impaired for reasons other than salt content or surface fluid volume.

We quantified the antimicrobial activity of ASF obtained from a human bronchial xenograft model and compared samples from CF and normal xenografts. The antimicrobial activity of ASF samples was determined under native and low-salt conditions. Reconstitution of the samples in low-salt medium resulted in significantly increased activity of both normal and CF samples, confirming that cationic antimicrobial substances have decreased activity in high salt concentrations (3, 15). However, desalted ASF samples from normal xenografts had a much greater antimicrobial activity when compared with CF samples, indicating that the differential measured under native conditions cannot be eliminated by lowering the NaCl concentration. Data presented here indicate that the CF-specific defect in antimicrobial activity is not caused by decreased total concentrations of known host defense factors; our studies imply that an unidentified antimicrobial factor is absent from CF ASF. Using RP-HPLC, the abnormality in antimicrobial activity was localized to a specific fraction that eluted at 38% acetonitrile. Reconstitution of CFTR function with adenoviral gene transfer restored the antimicrobial activity of this fraction. Further experiments using different treatments to inhibit the antimicrobial function of these fractions suggested that the putative substance is sensitive to high salt concentration and proteases but not heat, consistent with it being a protein or requiring a protein as a critical cofactor.

The data presented in this study show that the absence of a factor from CF ASF contributes to the defect of antimicrobial activity in ASF. It has been suggested that CFTR functions in the regulation of cellular processes, such as exocytosis, endocytosis, and regulation of the pH of cellular compartments (16). Dysfunctional CFTR may result in alteration of these processes and therefore in abnormal biosynthesis or secretion of antimicrobial substances. CFTR and most of the previously described host defense molecules are expressed in essentially the same cells of the human airways (17). Our hypothesis regarding abnormal production of an important antimicrobial protein is independent of the controversy about the ASF salt concentration (8) and does not necessarily exclude other hypotheses about the development of a host defense defect in CF. Relevant to this issue is our previous study, in which hBD-1 was inhibited in a normal xenograft by antisense oligonucleotides that partially inhibited the ability of ASF to kill P. aeruginosa (3). Direct comparison with the current study is difficult because the previous experiments used non-desalted ASF that was not quantitatively titrated for bacterial killing. However, measurement of remaining bacteria in the presence of undiluted and non-desalted ASF suggests that bacterial killing in oligonucleotide-treated normal xenografts was less than that observed in untreated normal xenografts but greater than in CF xenografts, suggesting partial inhibition of hBD-1 or the contribution of other factors such as the salt-dependent activity described in this study.

The fact that the salt-independent defect could be corrected by adenovirus-mediated gene transfer, where no greater than 10% of the cells are transduced, is encouraging. This also bears on the underlying pathogenetic mechanism where full functional correction is achieved with partial genetic correction, suggesting a threshold effect (i.e., expression of this molecule is rate-limiting) or a paracrine or bystander effect (i.e., corrected cells influence neighboring noncorrected cells).

In summary, this study suggests a mechanism by which a deficiency of CFTR compromises innate immunity that is independent of NaCl and H₂O metabolism. Further studies are needed to identify the molecule that mediates this effect.