The ΔF508-CFTR mutation results in increased biofilm formation by Pseudomonas aeruginosa by increasing iron availability

Abstract

Enhanced antibiotic resistance of Pseudomonas aeruginosa in the cystic fibrosis (CF) lung is thought to be due to the formation of biofilms. However, there is no information on the antibiotic resistance of P. aeruginosa biofilms grown on human airway epithelial cells or on the effects of airway cells on biofilm formation by P. aeruginosa. Thus we developed a coculture model and report that airway cells increase the resistance of P. aeruginosa to tobramycin (Tb) by >25-fold compared with P. aeruginosa grown on abiotic surfaces. Therefore, the concentration of Tb required to kill P. aeruginosa biofilms on airway cells is 10-fold higher than the concentration achievable in the lungs of CF patients. In addition, CF airway cells expressing ΔF508-CFTR significantly enhanced P. aeruginosa biofilm formation, and ΔF508 rescue with wild-type CFTR reduced biofilm formation. Iron (Fe) content of the airway in CF is elevated, and Fe is known to enhance P. aeruginosa growth. Thus we investigated whether enhanced biofilm formation on ΔF508-CFTR cells was due to increased Fe release by airway cells. We found that airway cells expressing ΔF508-CFTR released more Fe than cells rescued with WT-CFTR. Moreover, Fe chelation reduced biofilm formation on airway cells, whereas Fe supplementation enhanced biofilm formation on airway cells expressing WT-CFTR. These data demonstrate that human airway epithelial cells promote the formation of P. aeruginosa biofilms with a dramatically increased antibiotic resistance. The ΔF508-CFTR mutation enhances biofilm formation, in part, by increasing Fe release into the apical medium.

in cystic fibrosis (CF), abnormalities of the CFTR result in the lung becoming chronically infected with bacteria. The CF lung evolves into a highly inflamed and purulent environment that is the proximate cause of morbidity and mortality in these patients. By late adolescence, 80% of CF patients are chronically infected with Pseudomonas aeruginosa, the dominant pathogen in CF airways (16, 32). Permanent eradication of P. aeruginosa infection in CF airways is impossible because, as current evidence suggests, P. aeruginosa forms antibiotic-resistant biofilms in the CF lung (8, 57, 67). However, there is little direct evidence to support this conclusion. Thus one of the goals of this study is to develop a coculture model to study the ability of P. aeruginosa to form biofilms on human airway epithelial cells expressing ΔF508-CFTR, the most common mutation of CFTR, and on ΔF508-CFTR cells rescued with wild-type (WT) CFTR.

Biofilms are heterogeneous communities of surface-attached bacteria (10). Whereas the structure of these communities varies with environmental conditions, biofilms of P. aeruginosa are often associated with large colonial structures separated by channels (28). Within biofilms, bacteria are encased in an extracellular matrix composed of exopolysaccharides, DNA, and proteins that confers structural stability to the community (10, 65). One hallmark of biofilms grown on abiotic substrates (i.e., glass), including those of P. aeruginosa, is their high level of resistance to antibiotics, which renders most standard antimicrobial therapies ineffective against these communities (34). However, the antibiotic resistance of P. aeruginosa growing in coculture with human airway cells (i.e., biotic substrate) has not been reported. Thus one of the goals of this study is to examine the antibiotic resistance of P. aeruginosa biofilms growing on human airway epithelial cells expressing ΔF508-CFTR and on ΔF508-CFTR cells rescued with WT-CFTR.

The formation of P. aeruginosa biofilms is a complex, well-regulated process, and a number of functions, including flagellar motility, the Psl polysaccharide, and global regulators such as Crc, have been shown to be required for biofilm formation by this microbe on abiotic surfaces (21, 37, 42, 43). The production of acylhomoserine lactone quorum-sensing signals also contributes to the maturation and function of biofilms, at least in some environments (7, 12, 18, 23). In addition, iron is essential for P. aeruginosa biofilm development and maturation on abiotic surfaces (3–5, 53, 56). Increased levels of iron have been described in airway secretions in CF, and it has been suggested that elevated iron levels contribute to the persistent infection by P. aeruginosa. For example, iron in CF lung bronchoalveolar lavage (BAL) fluid and in sputum from CF lungs is ∼8 μM (59, 60), as opposed to 0.018 μM in the BAL isolated from healthy patients (1). Work by Reid and colleagues (50, 51) also reveals that there is increased iron in sputum isolated from the CF lung, and, furthermore, a recent study from this group suggests that increased iron in the CF lung contributes to the persistence of P. aeruginosa infection (49). Reid et al. (49) showed that for CF patients infected with P. aeruginosa, sputum iron concentration was 10–200 μmol/l (median = 46.6), whereas iron concentration in healthy controls was 0–15.8 μmol/l (median = 0). Significantly, even uninfected CF patients had increased iron levels in sputum compared with healthy controls (49), indicating that increased iron was a consequence of the ΔF508-CFTR mutation rather than the bacterial infection promoted by the CFTR defect. Surprisingly, among the relative abundance of data reporting increased iron concentrations in the BAL and sputum isolated from CF patients, nothing is known about the effect of iron on biofilm formation by P. aeruginosa grown in coculture with human airway epithelial cells. Thus one of the goals of this study is to test the hypothesis that iron release by human airway epithelial cells is enhanced by the ΔF508-CFTR mutation and that enhanced iron release contributes to the exuberant formation of P. aeruginosa biofilms.

In this manuscript, we describe a novel coculture model to examine the effects of human airway epithelial cells on the formation of P. aeruginosa biofilms and on the development of antibiotic resistance by P. aeruginosa. We report that coculture with airway epithelial cells increases the resistance of P. aeruginosa to tobramycin (Tb) by >25-fold compared with P. aeruginosa grown on abiotic surfaces (i.e., glass). In addition, we report that enhanced biofilm formation by P. aeruginosa grown on airway (and kidney) cells expressing ΔF508-CFTR is due, in part, to enhanced release of iron into the apical solution and that rescue of CF cells with WT-CFTR reduces both iron release and biofilm formation. Finally, chelation of iron with conalbumin reduces biofilm formation by P. aeruginosa on airway epithelial cells.

MATERIALS AND METHODS

Cell lines and cell culture.

To develop a coculture model system to examine the effects of epithelial cells on biofilm formation and the antibiotic resistance of P. aeruginosa, we used several epithelial cell lines. First, we studied an isogenic set (except for CFTR) of human airway epithelial cells. Human bronchial epithelial cells (CFBE41o⁻) isolated from a CF patient (ΔF508/ΔF508) were originally immortalized and characterized by Dr. D. Gruenert and colleagues (9, 11). Stable lentiviral-based transduction of these cells with either WT-CFTR or ΔF508-CFTR was performed by Tranzyme (Birmingham, AL), and the cells were a generous gift from Dr. J. P. Clancy (17). Cells were maintained in MEM supplemented with 10% FBS, 2 mM l-glutamine, 50 U/ml penicillin, and 50 μg/ml streptomycin in a 5% CO₂-95% air incubator at 37°C. In addition, the WT-CFTR- and ΔF508-CFTR-transduced cells were cultured in 1 μg/ml blasticidin or 2 μg/ml puromycin, respectively. Antibiotics were removed immediately before experiments. CFBE41o⁻ cells were seeded on 40-mm diameter glass coverslips (Bioptechs, Butler, PA) at 1 × 10⁶ cells per coverslip and grown at 37°C for 8–10 days to establish confluent monolayers (62), or they were grown on microscopically clear 12-mm permeable filter inserts (Snapwell; Corning Costar, Kennebunk, ME) at 0.3 × 10⁶ cells per filter and grown at an air-liquid interface at 37°C for 8–10 days.

In addition, we also studied an isogenic set (except for CFTR) of kidney epithelial cells to determine whether the effects of the ΔF508-CFTR mutation on biofilm formation and antibiotic resistance were similar in kidney and airway cells. Madin-Darby canine kidney (MDCK) type I cells, stably transfected with either green fluorescent protein (GFP) CFTR (40) or ΔF508-CFTR (31), and parental MDCK cells, which express nominal levels of endogenous CFTR (39), were grown on glass coverslips or Snapwell filters as described above.

Because CFBE41o⁻ cells do not produce mucus, and there is evidence that mucus may be important for P. aeruginosa growth in the CF airways, we also studied immortalized human nasal epithelial cells homozygous for ΔF508-CFTR (JME/CF15; Ref. 22) that produce mucus. These cells were maintained in bronchial epithelial basal media supplemented with bronchial epithelial cell growth medium (BEGM) SingleQuots (Cambrex) and 50 nM retinoic acid in a 5% CO₂-95% air incubator at 37°C. JME/CF15 cells were seeded on 12-mm permeable filter inserts (Snapwell; Corning Costar) at 0.3 × 10⁶ cells per filter and grown at an air-liquid interface at 37°C for 8–10 days to allow polarization and mucus production.

Mucus production.

Muc5ac protein expression was determined by cell surface biotinylation, as described previously (61, 63), using a monoclonal anti-Muc5ac antibody (Abcam, Cambridge, MA) at the concentration recommended by the manufacturer.

Bacterial strains.

P. aeruginosa strain PAO1 and a P. aeruginosa mucoid clinical strain isolated from a CF patient (SMC1587) were grown on rich medium (lysogeny broth, LB) at 37°C. Both strains carry a plasmid (pSMC21) that constitutively expresses GFP, which was used to identify the bacteria by fluorescence microscopy (26). This plasmid is stable in the absence of antibiotic selection (6). The pMH520 plasmid contains a rhlA::gfp transcriptional fusion and was a generous gift from Matt Parsek (University of Washington). Overnight cultures of the rhlA::gfp strain were diluted 1:100 in LB and grown for another 2 h at 37°C to reduce the density-associated fluorescence. GFP-expressing P. aeruginosa strains carrying mutations in genes known to affect biofilm formation on abiotic surfaces were also studied in some of our assays including: the lasI rhlI mutant, a quorum-sensing mutant (44); the crc mutant, a global carbon metabolism regulator that is involved in signal transduction required for biofilm development (42); the flgK mutant that does not produce flagellum and is attachment-deficient on glass and mucin-coated surfaces (27, 43); the psl mutant that is required for adhesion to a variety of surfaces and for maintenance of the biofilm structure (21, 37); and the pilA mutant that lacks functional type IV pili required for twitching motility and biofilm formation on abiotic surfaces (43). For coculture studies, P. aeruginosa strains were grown overnight in LB medium, washed, and resuspended in epithelial cell growth medium without antibiotics or phenol red (see below) and then injected into the flow cell chamber on the apical side of the cells at a multiplicity of infection (MOI) of 25.

Flow chamber imaging experiments.

To image live GFP-labeled P. aeruginosa in coculture with live polarized epithelial cells, we used a FCS2 closed system, live-cell micro-observation chamber from Bioptechs assembled according to the manufacturer's instructions (for details please go to http://www.bioptechs.com/Products/FCS2/fcs2.html) and mounted on the stage of an inverted Olympus microscope. The FCS2 system allows accurate thermal control and laminar flow perfusion of cells grown on a glass coverslip. The chamber containing cells was continuously perfused with a modified cell growth medium (MEM without phenol red, 2 mM l-glutamine) maintained at 37°C and 5% CO₂-95% using a mini-peristaltic pump (Instech Solomon, Plymouth Meeting, PA) set at 20 ml/h, which achieves a flow rate equivalent to that in the lung due to mucociliary clearance. Bacteria were injected into the chamber through a two-way valve, and flow was subsequently halted for 2 h to allow bacteria to attach to epithelial cells. The flow was then reinitiated and maintained at 20 ml/h throughout the rest of the experiment. For biofilm experiments performed on cells grown on permeable filters, bacteria were added to the apical side of cells for 1 h and then removed by carefully suctioning. Thereupon, the filters were transferred to a Delta T Open Dish system (Bioptechs; see http://www.bioptechs.com/Products/Delta_T/delta_t.html for details), and the chamber/cells were continuously perfused with a modified cell growth medium as described above. To facilitate microscopic observation of airway cells grown on filters, nuclei were stained with 10 μg/ml Hoechst 33342 (Molecular Probes) for 30 min and washed twice with the modified growth medium before inoculation with P. aeruginosa.

The integrity of cell monolayers was assessed by Normarski differential interference contrast (DIC) microscopy. Cell viability was also monitored using propidium iodide (1:10,000 dilution; Molecular Probes), which is impermeant to live cells but enters damaged cells, binds to DNA, and is detected using a rhodamine filter set.

In the absence of bacteria, airway cells were viable in both flow chambers for several days without any loss of viability (data not shown). However, 10–12 h after inoculating P. aeruginosa into the chamber, epithelial cells started to accumulate propidium iodide, indicating an alteration of their viability, and 16–18 h postinoculation, the cells were visibly distressed, with patches of cells lifting off the coverslip (data not shown). Based on these observations, we limited our studies to the first 8 h postinoculation.

Microscopy.

An Olympus IX70 inverted microscope equipped with an ORCA-AG deep cooling CCD camera (Hamamatsu Photonics) and a ×60 oil-immersion objective [Olympus, numerical aperture (NA) 1.40] was used for fluorescence microscopy and DIC imaging of epithelial cells and P. aeruginosa in coculture. Digital images were acquired with the OpenLab 4.0.3 software package (Improvision, Lexington, MA) loaded on a G5 Macintosh computer. Volumes were deconvolved by iterative restoration using the Volocity 3.5.1 software (Improvision). Confocal microscopy was performed using a Nikon LiveScan Swept Field Confocal Microscope using a Nikon ×100 Plan Fluor objective (1.3 NA) and a 30-μm pinhole aperture. Images were scanned at a speed of 260 Hz and acquired with the Nikon software suite.

Quantitative analysis of three-dimensional (3-D) biofilm structures was achieved with the COMSTAT image analysis software package (19, 20).

RT-PCR.

RT-PCR studies of tolA and fliC were conducted to determine when P. aeruginosa changed from a planktonic to a biofilm mode of growth. The tolA gene is involved in the major aminoglycoside resistance mechanism of P. aeruginosa, and its expression is activated in biofilms (35, 52). By contrast, the fliC gene encodes flagellin, the structural subunit of the flagellum, and is expressed only in planktonic P. aeruginosa and not in P. aeruginosa growing as biofilms (54, 55). The rplU gene is constitutively expressed in both planktonic and biofilm modes of growth and was used as a reference gene. Bacteria were grown for 6 h in the flow cell system described above, at a time when, by visual inspection, they formed biofilms. Total bacterial RNA used in RT-PCR assays was purified with the RNeasy kit (Qiagen) and treated with DNase using the DNA-free kit (Ambion). Purified RNA was quantified on an RNA 600 Nano LabChip (Agilent Technologies), and complementary DNA was synthesized using the first-strand synthesis kit for RT-PCR (Ambion). The resulting RNA/cDNA hybrids were used as templates in PCR reactions using the previously reported annealing conditions and primer sets for tolA and rplU genes (35). The primer set for fliC was designed from the P. aeruginosa PAO1 genome sequence: fliC, 5′-GTCAACACGAACATTGCTTCCC-3′ and 5′-TTGCTGCCGACCTGGTAAGAAC-3′. An annealing temperature of 56°C and an extension temperature of 72°C were used for 20 cycles for fliC.

Determination of P. aeruginosa antibiotic resistance.

To determine whether epithelial cells altered the antibiotic resistance of P. aeruginosa, we conducted experiments to determine the concentration of Tb required to eliminate viable P. aeruginosa, known as the minimal bactericidal concentration for Tb (MBC_Tb), by standard methods as described in detail previously (35, 38). Care was taken to assure that the initial colony-forming units (CFU; 1 × 10⁷) was similar for P. aeruginosa grown under planktonic and biofilm conditions.

siRNA knockdown of CFTR.

To examine the role of CFTR in epithelial cells in regulating biofilm formation by P. aeruginosa, CFTR protein expression was reduced in CF-derived CFBE41o⁻ (CFBE) parental and CFBE+WT cells using three different small interfering RNA (siRNA) oligos (Qiagen) tested individually, targeted to unique sequences in CFTR, as described previously, with similar results (62). CFBE cells were transfected with siRNA for CFTR or a scrambled control sequence (siNeg) using HiPerFect Transfection Reagent (Qiagen) on 10⁶ cells per coverslip and grown for 8 days to confluence. After 8 days, CFTR abundance was assessed by Western blot analysis, and plasma membrane abundance was assessed by cell surface biotinylation as described previously (61, 63).

Corr4a and electrophysiological studies.

To examine the effect of increased plasma membrane abundance of ΔF508-CFTR in airway cells on biofilm formation by P. aeruginosa and on iron homeostasis, CFBE41o⁻ cells were grown on Snapwell filters and treated with Corr4a as described previously (24, 62). Corr4a was identified by screening a collection of small molecules and increases the abundance of ΔF508-CFTR in the plasma membrane (46). To assess the effect of Corr4a on ΔF508-CFTR mediated Cl secretion, cells grown on Snapwell filters were mounted in Ussing chambers and the short-circuit current (I_sc) was measured as described previously (17, 30, 41).

Iron analysis.

To determine whether the ΔF508 mutation in CFTR affected the iron content and iron release by CFBE and MDCK cells, intracellular iron and the iron concentration in the apical fluid was measured by inductively coupled plasma mass spectrometry (ICP-MS). To measure intracellular iron, confluent monolayers of CFBE cells and MDCK cells were washed and resuspended in HBSS. Cell pellets were digested with nitric acid, and the concentration of total iron was measured by ICP-MS as previously described (24). To measure the iron concentration in the apical medium, CFBE and MDCK cells were grown as polarized monolayers on filters. The apical solution was aspirated, and then 1 ml of MEM was applied to the apical side of the cells for 2 h, whereupon the medium was collected for ICP-MS analysis of iron. For iron chelation studies, the MEM added to the cells was supplemented with 20 μg/ml conalbumin (Sigma, St. Louis, MO). For iron supplementation studies, a 16 mM stock solution of ferric chloride (Sigma) was prepared in water and filter-sterilized (0.22-μm Millex filter, Millipore). Working concentrations were subsequently diluted into MEM. Human holo-transferrin (Sigma) was prepared in MEM at a 1 mg/ml concentration.

Statistical analysis.

Data were analyzed using GraphPad Prism 4.0 for Macintosh (GraphPad Software, San Diego, CA). One-way ANOVA was performed, followed by a Tukey-Kramer post hoc test using a 95% confidence interval. All samples were successfully run through a normality test (α = 0.05) to check for violation of Gaussian distribution. Data are expressed as means ± SE. A P value <0.05 was considered significant.

RESULTS

Human airway epithelial cells enhance the formation of antibiotic-resistant biofilms.

The first set of studies was conducted to examine the effect of human airway epithelial cells on the formation of antibiotic-resistant biofilms by P. aeruginosa. CF-derived CFBE41o⁻ (CFBE) human airway epithelial cells, homozygous for the ΔF508 mutation, were grown as confluent monolayers on glass, and GFP-labeled P. aeruginosa was applied to the apical face of the cells in a continuous flow cell system. This cell line was chosen because isogenic derivatives expressing ΔF508-CFTR and WT-CFTR are available thus allowing us to assess the contribution of WT-CFTR vs. ΔF508-CFTR to biofilm formation.

P. aeruginosa firmly attached to the apical surface of the parental CFBE cells (i.e., cells expressing endogenous ΔF508-CFTR) and bacterial microcolonies were observed as early as 3 h after inoculation (Fig. 1A). The bacterial colonies on the epithelial cells exhibited the mushroom-like structures previously described for P. aeruginosa biofilms forming on some abiotic surfaces (19, 28). The maximum height of the clusters formed at 3 h was ∼20–22 μm. At 6–8 h, the microcolonies increased to ∼30 μm in height and were dispersed across the airway cells (Fig. 1A). Three-dimensional reconstruction of z-series image stacks allowed the visualization of the microcolonies as well as uncolonized channel regions surrounding the bacterial clusters (Fig. 1B).

Fig. 1.

Representative images of Pseudomonas aeruginosa grown on confluent, human cystic fibrosis (CF) airway epithelial cells and glass. A: time course of biofilm development on parental CF-derived human bronchial epithelial CFBE41o⁻ (CFBE) cells. En face view of green fluorescent protein (GFP)-labeled P. aeruginosa strain PAO1 grown on CFBE cells. Merged and pseudocolored images were viewed by differential interference contrast (DIC), and the corresponding fluorescent images are shown. The observation plane was focused on top of the microcolonies, and the monolayer located underneath appears somewhat blurred. The yellow arrow indicates a microcolony detected at 3 h. Bar, 20 μm. B and C: z-series image stacks acquired with a 1-μm step, 6 h after inoculation, for GFP-PAO1 bacteria grown on parental CFBE cells (B) or on a glass coverslip (C). Images are tilted to facilitate viewing after deconvolution and 3-dimensional (3-D) reconstruction. Experiments were performed a minimum of 3–8 times, and representative fields are shown. Bar, 10 μm.

By contrast, when the substratum was glass instead of CF human airway epithelial cells, microcolonies did not form on this abiotic surface after 6 h (compare Fig. 1, B and C; see Table 1). Individual, GFP-labeled PAO1 attached to the glass substrate but by visual inspection did not form biofilms (Fig. 1C). Indeed, as shown previously, P. aeruginosa must be grown on an abiotic surface (i.e., glass) for 24 h to several days to develop biofilms equivalent to those that form on parental CFBE cells in 6 h (25). The 1,500-fold difference in P. aeruginosa biomass on airway cells vs. glass (Fig. 1, B vs. C) was not due to an increase in the number of bacteria that initially attached to cells vs. glass or to differences in the number of bacteria that detached from cells vs. glass. Indeed, the initial bacterial attachment to glass was ∼5-fold greater than to epithelial cells (see Supplemental Fig. 1A available in the data supplement online at the AJP-Lung Cellular and Molecular Physiology web site). These results reveal that human airway epithelial cells expressing ΔF508-CFTR dramatically enhance the formation of biofilms by 1,500-fold (at 6 h) compared with P. aeruginosa growing on an abiotic substrate.

Table 1.

Quantitative analysis of 6-h biofilms grown on abiotic and biotic substrata

Substratum	Biomass, μm3/μm2‡	Mean Thickness, μm‡	Max Thickness, μm‡
Glass	0.002±0.001	0.003±0.001	3.91±0.29
CFBE (Parental)	3.31±0.26	3.85±0.30	29.61±1.17
CFBE+ΔF508	3.11±0.23	3.60±0.25	30.83±1.11
CFBE+WT	1.35±0.15*	1.52±0.15*	17.94±0.74*
CFBE+ΔF508 + conalbumin	0.86±0.17†	1.02±0.19†	15.17±0.87†
CFBE+WT + conalbumin	0.31±0.06†	0.40±0.08†	10.00±0.46†

^‡Six z-series image stacks were analyzed for each substratum, and experiments were performed in triplicates.

^*P < 0.05 vs. the cystic fibrosis-derived CFBE41o⁻ (CFBE) and CFBE+ΔF508 cell lines;

^†P < 0.05 vs. control without conalbumin. Max, maximum; WT, wild-type.

The imaging studies described above reveal that the structures formed by P. aeruginosa on parental CFBE cells look like biofilms; however, additional studies were conducted to provide support for the conclusion that they are indeed antibiotic-resistant biofilms. Five lines of evidence are consistent with the conclusion that P. aeruginosa forms biofilms on parental CFBE cells. First, as noted above, the images presented in Fig. 1 are consistent with the formation of P. aeruginosa biofilms on abiotic surfaces (Ref. 55; Fig. 1B), including the morphology of biofilms formed on mucin-coated glass slides (27). Second, the quorum-sensing-controlled rhlA gene, which codes for a protein required for the production of rhamnolipid surfactants, was expressed in microcolonies at 8 h but not at 4 h (Fig. 2A). Although the rhlA gene is also expressed in high density planktonic cultures, previous studies indicate that, when grown in flow cell conditions, biofilm colonies must exceed ∼30 μm in diameter before rhlA gene expression is detected (29, 47). Third, P. aeruginosa microcolonies (6 h) expressed a previously reported mRNA marker indicative of biofilm growth, tolA (35, 66), but not the planktonically expressed gene fliC (Ref. 54; Fig. 2B). As an additional control, in our experimental conditions, we demonstrated that planktonic P. aeruginosa, as expected, expressed fliC but not tolA (Fig. 2B). Fourth, we tested strains carrying mutations in genes with known roles in biofilm formation by P. aeruginosa on abiotic surfaces, including the lasI rhlI, crc, flgK, psl, and pilA mutants (12, 21, 33, 42, 43), for their ability to form biofilms on CFBE cells. For all of these mutants, biofilm formation was also defective on CFBE cells (Fig. 2C). The lasI rhlI and crc mutants produced biofilms with altered structure, whereas the flgK, psl, and pilA mutants produced dramatically smaller biofilms than WT-PAO1. Thus mutations that reduced biofilm formation on abiotic surfaces also reduced biofilm formation on biotic surfaces (parental CFBE cells). Fifth, the biofilms that formed on parental CFBE cells developed an exceptionally high resistance to antibiotics, a characteristic of biofilms. The concentration of Tb required to eliminate all viable bacteria (i.e., MBC_Tb) was 8 μg/ml for planktonic P. aeruginosa (Supplemental Table 2 in Ref. 35), 400 μg/ml for biofilms that formed on an abiotic surface (plastic; Table 2 in Ref. 35), and >10,000 μg/ml for biofilms that formed on CFBE cells expressing ΔF508-CFTR (this study). Thus airway cells dramatically (>25-fold) increased the antibiotic resistance of biofilms to Tb compared with the antibiotic resistance of biofilms that form on an abiotic surface.

Fig. 2.

Molecular evidence that P. aeruginosa forms biofilms on CF airway cells. A: en face view of PAO1 carrying the pMH520 plasmid (rhlA::GFP) grown on parental CFBE cells. Any fluorescence observed resulted from the activation of the quorum sensing-regulated gene rhlA. The DIC and the corresponding fluorescent images are shown together with the merged, pseudocolored images. The observation plane was focused on top of the P. aeruginosa microcolonies, and therefore the epithelial monolayer appears somewhat blurred. Bar, 20 μm. B: semiquantitative RT-PCR performed with total RNA isolated from PAO1 grown for 6 h in a flow cell chamber on a monolayer of CFBE parental cells (lane 1) or under planktonic conditions in the same medium (lane 2). The rplU gene is used as a control since it is constitutively expressed under planktonic and biofilm growth conditions. No bands were detected when RNA, instead of cDNA, was used as a template (data not shown). C: mutations in the lasI rhlI, crc, flgK, psl, and pilA genes, genes required for biofilm formation on abiotic surfaces, were also required for biofilm formation on CF airway cells. The merged DIC and corresponding pseudocolored fluorescent images are presented. Bar, 20 μm. D: resistance of biofilms to antibiotic therapy. Biofilms grown on parental CFBE cells for 6 h were treated for 24 h with 1,000 μg/ml tobramycin (Tb), 5 μg/ml ciprofloxacin (Cp), and 5 μg/ml imipenem (Im), applied alone or in combination (Tb+Cp+Im; Combo). Neither individual nor combined treatment eliminated established P. aeruginosa biofilms. The starting bacterial population under these conditions is 6.7 ± 4.0 × 10⁷ colony-forming units (CFU) per well. E: time course of resistance to antibiotics. Bacteria were grown planktonically (P) or as biofilms on parental CFBE cells for the time indicated and then treated with the combination of 3 antibiotics described in D. An antibiotic-resistant subpopulation was apparent at 1.5 h postinoculation, and maximum resistance was developed by 3 h. Data presented in D and E are an average of 6 replicates for each condition. *P < 0.05 vs. t = 0 h postinoculation. WT, wild-type.

Table 2.

Quantitative analysis of 6-h biofilm of a clinical isolate grown on CFBE + WT and CFBE + ΔF508-CFTR cells

Substratum	Biomass, μm3/μm2‡	Mean Thickness, μm‡	Max Thickness, μm‡
CFBE+ΔF508	5.7±0.51	6.9±0.55	27.1±0.89
CFBE+WT	3.4±0.29*	4.8±0.40*	21.2±0.85*

^‡Six z-series image stacks were analyzed for each cell line, and experiments were performed in triplicates.

^*P < 0.05.

Additional studies were conducted to determine whether other antibiotics used to treat CF patients, either alone or in combination with Tb, would be effective in killing biofilms that form on CF airway epithelial cells. Unfortunately, neither Tb, ciprofloxacin (the second most commonly used antibiotic in CF clinic), nor imipenem, alone or in combination at concentrations used in the CF clinic, eliminated P. aeruginosa biofilms on airway cells (Fig. 2D). By contrast, these concentrations of antibiotics eliminate biofilms that form on abiotic surfaces (35). The increased resistance to these antibiotics developed as early as 1.5 h after initiation of biofilm formation (Fig. 2E). Taken as a whole, the five lines of data presented here support the conclusion that airway epithelial cells dramatically enhance the formation of biofilms and also increase, by 25-fold, the antibiotic resistance of P. aeruginosa to Tb.

Mucus-producing airway epithelial cells also enhance biofilm formation.

Recent studies have proposed that P. aeruginosa biofilms form in the anaerobic environment within the thick layers of mucus in the CF airway (8, 36, 67, 69, 70). However, these findings are somewhat controversial. For example, a recent study indicates that P. aeruginosa may grow via microaerobic metabolism in the CF lung rather than by anaerobic respiration (2), and another study suggests that P. aeruginosa may attach to the CF airway directly via a neuraminidase-dependent mechanism (58). Because CFBE cells do not produce mucus, and because of the potential role that mucus may play in biofilm formation, we conducted additional studies on mucus-producing airway cells to determine whether mucus-producing human airway epithelial cells also enhance biofilm formation by P. aeruginosa. Robust biofilms developed on JME/CF15 cells (Fig. 3A), a CF cell line homozygous for the ΔF508-CFTR mutation. JME/CF15 cells produce mucus as determined by cell surface biotinylation of the Muc5ac protein (Fig. 3B). The biofilms that formed on JME/CF15 cells at 3 h were similar in structure and size to biofilms that formed on CFBE cells, a non-mucus-producing cell line at the same time point (Fig. 2A). Biofilm formation on JME/CF15 cells was examined at this earlier time point because these epithelial cells are more susceptible to killing by P. aeruginosa. These results suggest that the presence of mucus is not absolutely required for the formation of bacterial biofilms on CF airway epithelial cells.

Fig. 3.

Biofilm formation on mucus-producing human CF airway cells. Biofilm formation was monitored on polarized JME/CF15 cells at 3 h (A). A representative view of PAO1 grown on Hoechst-stained JME/CF15 cells is shown after fluorescent images were merged and pseudocolored. Bar, 20 μm. B: Western blot detection of Muc5ac protein. JME/CF15 cells were grown for 8 days, and Muc5ac protein was detected in whole cell lysate (Total) and at the apical membrane by cell surface biotinylation (Apical).

The ΔF508-CFTR mutation in human airway epithelial cells facilitates biofilm formation.

Biofilm formation in the CF lung has been proposed to account for the inability of antibiotic therapy to clear these infections. Implicit in this idea is that the formation of biofilms occurs preferentially in the CF lung. Therefore, we conducted experiments to test the hypothesis that rescue of CF airway cells expressing ΔF508-CFTR with WT-CFTR will reduce biofilm formation. To test this hypothesis, we assessed the extent of biofilm formation on the parental CFBE cells, which are homozygous for ΔF508-CFTR, compared with isogenic CFBE cells stably expressing WT-CFTR (CFBE+WT cells) or CFBE cells overexpressing ΔF508-CFTR (CFBE+ΔF508 cells). The parental CFBE cells expressed nominal levels of ΔF508-CFTR in the apical plasma membrane, whereas the CFBE+ΔF508 cell line expresses twofold more ΔF508-CFTR than the parental CFBE line (Supplemental Fig. 2 and Ref. 64). The CFBE+ΔF508 cell line was studied to examine the effect of increased expression of ΔF508-CFTR on biofilm formation.

There was no difference in biofilm formation on the parental CFBE cells compared with the cell line overexpressing ΔF508-CFTR (CFBE+ΔF508 cells; Fig. 4, A–C, and Table 1). Thus increased expression of ΔF508-CFTR does not alter biofilm formation. However, biofilm formation was decreased two- to threefold when CF cells were rescued with WT-CFTR (compare CBFE+WT cells with parental CFBE and CFBE+ΔF508 cells; Fig. 4, A–C). Bacterial biomass was significantly less for the CFBE+WT cell line at every time point tested between 4 and 10 h (Fig. 4C). In addition to the decrease in total biomass, the mean and maximum thickness of the biofilms formed were also reduced when cells were rescued with WT-CFTR (Table 1). Furthermore, biofilm formation by a mucoid clinical isolate of a P. aeruginosa strain isolated from a CF patient was also more robust on CFBE+ΔF508 cells compared with CFBE+WT cells (Table 2).

Fig. 4.

Stable transfection with WT-CFTR reduces P. aeruginosa biofilm formation on CF airway cells. A–C: representative z-stacks from the 6-h time point are shown after deconvolution and 3-D reconstruction of a x-z view (A) and a x-y view (B). In B, note that the z-axis for the parental CFBE and CFBE+ΔF508 cell lines is 3 grid units, whereas the CFBE+WT is 2 grid units (grid unit = 14.5 μm). C: P. aeruginosa biomass on CFBE cells, which provides a measure of the total volume occupied by bacteria on the substratum, was quantified with the COMSTAT program and is plotted for the 4-, 6-, 8-, and 10-h time points. For each cell line, 6 randomly chosen z-series image stacks were analyzed for each time point, and experiments were repeated at least 3 times. (*P < 0.05 vs. CFBE+ΔF508). D: biofilm formation on polarized CFBE cells grown on filters. z-Series image stacks were acquired 6 h postinoculation on a Nikon Swept Field Confocal Microscope. Representative images are shown after 3-D reconstruction (grid unit = 4 μm). Note that the z-scale for the CFBE+WT is 3 grid units and for the CFBE+ΔF508 cell line is 4 grid units.

We also assessed the effect of rescuing ΔF508-CFTR cells with WT-CFTR on biofilm formation by P. aeruginosa grown on fully polarized cells by culturing airway cells on Snapwell filters (Fig. 4D). Similar to observation made on airway cells grown on glass (see above), WT-CFTR reduced the formation of P. aeruginosa biofilms (2-fold) compared with biofilms that form on CFBE+ΔF508 cells. Thus, taken together, these data demonstrate that stable transfection of WT-CFTR in polarized CF human airway epithelial cells reduces the formation of P. aeruginosa biofilms.

In these studies the amount of plasma membrane CFTR in CFBE+WT cells was ∼5-fold higher than the expression of ΔF508-CFTR in CFBE+ΔF508 cells. Thus the reduction in biofilm formation by expressing WT-CFTR could be due to increased abundance of CFTR per se (i.e., WT-CFTR or ΔF508-CFTR). However, when CFTR protein abundance in the plasma membrane in CFBE+WT cells was reduced by siRNA to the same levels observed in CFBE+ΔF508 cells, biofilm formation on CFBE+WT cells was still less in CFBE+WT cells compared with CFBE+ΔF508 cells (Supplemental Fig. 3). Thus the decrease in P. aeruginosa biofilm formation was due to the presence of WT-CFTR in the apical plasma membrane rather than differences in the amount of CFTR protein per se (i.e., WT-CFTR or ΔF508-CFTR).

A difference in initial attachment of P. aeruginosa among the three stable cell lines could not account for the increased biofilm biomass observed on cells expressing ΔF508-CFTR compared with WT-CFTR because the number of adherent bacteria per epithelial cell at 2 h did not differ among the cell lines (Supplemental Fig. 1B). Moreover, the increased biomass of P. aeruginosa on CFBE cells compared with the CFBE+WT cell line was not due to differences in bacterial shedding from the biofilms. During biofilm formation (2–4 h after bacteria attached), 4.9 × 10⁶ bacteria detached from biofilms growing on CFBE cells expressing WT-CFTR, and 4.7 × 10⁶ bacteria detached from biofilms growing on CFBE cells expressing ΔF508-CFTR. These values were not significantly different. Thus, given the measured increase in P. aeruginosa biomass on ΔF508-CFTR cells (2-fold), the data suggest that differences in biomass must be due to enhanced growth and division of P. aeruginosa growing on CFBE cells expressing ΔF508-CFTR.

The ΔF508-CFTR mutation in kidney epithelial cells facilitates biofilm formation.

The results presented above in human airway epithelial cells demonstrate that the ΔF508-CFTR mutation in CF human airway epithelial cells facilitates biofilm formation. To provide additional support for this conclusion, we repeated the experiments described above in an isogenic set of MDCK cells stably expressing either WT-CFTR or ΔF508-CFTR (31, 40) and also included parental MDCK cells that do not express CFTR. P. aeruginosa formed robust biofilms on MDCK+ΔF508-CFTR cells (Supplemental Fig. 4). Biofilm formation was significantly less on parental MDCK cells and on MDCK+WT-CFTR cells compared with MDCK+ΔF508-CFTR cells (Supplemental Fig. 4). These studies on MDCK cells reveal that the effect of ΔF508-CFTR on biofilm formation is not cell line dependent.

Chemical rescue of ΔF508-CFTR reduces biofilm formation.

As demonstrated above, increasing the expression of ΔF508-CFTR in the plasma membrane of human airway epithelial cells by stable transfection with ΔF508-CFTR had no effect on biofilm formation by P. aeruginosa. However, recent studies have identified several potential new drugs to treat CF, including a compound called Corr4a, which increases the abundance of ΔF508-CFTR in the plasma membrane (45). The discovery of Corr4a and other compounds are proof of principle that drugs can be identified to increase ΔF508-CFTR-mediated Cl secretion in CF airway epithelial cells. In addition to enhancing ΔF508-CFTR-mediated Cl secretion, useful drugs for CF would also inhibit the proinflammatory response in CF cells and reduce the ability of P. aeruginosa to form drug-resistant biofilms. However, to our knowledge, nothing is known about the effects of Corr4a on biofilm formation. Thus experiments were conducted to determine whether Corr4a reduced biofilm formation on parental CFBE cells.

We confirmed that Corr4a increased ΔF508-CFTR-mediated Cl⁻ secretion by fourfold in parental CFBE cells (Fig. 5A). As a control, growth of the CFBE cells at 27°C also increased Cl⁻ secretion, as reported previously (Fig. 5A; Ref. 13). Corr4a reduced biofilm formation by P. aeruginosa on parental CFBE cells compared with vehicle-treated parental CFBE cells (Fig. 5, B and C). Thus, although Corr4a both increased the abundance of functional ΔF508-CFTR in the plasma membrane and reduced biofilm formation, increased abundance of ΔF508-CFTR in the plasma membrane by overexpression, driven by a plasmid (see data above), had no effect on biofilm formation on human airway epithelial cells. The possible reasons for this difference will be discussed below.

Fig. 5.

Corr4a increases ΔF508-CFTR-mediated Cl secretion in CFBE parental cells and reduces biofilm formation: A: short-circuit currents (I_sc) attributed to CFTR were measured in Ussing chambers across CFBE cells grown at 37°C, 27°C, or in the presence of Corr4a for 24 h at 37°C. Experiments were performed in triplicate (*P < 0.05 compared with untreated cells maintained at 37°C). B and C: biofilm measurements. Biomass (B) and mean thickness (C) of biofilms formed on CFBE cells grown in the absence and presence of Corr4a. Experiments were performed in triplicate (*P < 0.05 compared with untreated control cells).

The ΔF508-CFTR mutation enhances iron availability, which stimulates biofilm formation.

To begin to elucidate the mechanisms whereby CF airway epithelial cells enhance biofilm formation, we conducted experiments to test the hypothesis that the ΔF508-CFTR mutation enhanced iron release into the cell culture medium and that increased iron facilitates biofilm formation. The rationale for this hypothesis is twofold: 1) iron facilitates biofilm formation on abiotic surfaces (3, 4, 56), and 2) iron concentration is enhanced in the BAL isolated from CF patients.

In the first set of experiments, we measured by ICP-MS the iron content of cells and the iron concentration in the apical fluid secreted by cells. Both intracellular and apical medium iron levels were significantly higher in CFBE+ΔF508 cells compared with CFBE+WT cells (Fig. 6, A and B). In addition, both intracellular and apical medium iron levels were significantly higher in MDCK+ΔF508 cells compared with MDCK+WT-CFTR cells (Supplemental Fig. 4). Thus these data reveal that the ΔF508-CFTR mutation in airway and kidney cells increases the concentration of iron in the apical medium in which P. aeruginosa forms biofilms.

Fig. 6.

Iron facilitates biofilm formation by P. aeruginosa: A: analysis of total cellular iron in airway cells. Intracellular iron concentration was measured by inductively coupled plasma mass spectrometry (ICP-MS) after nitric acid digestion of confluent CFBE+WT (WT) and CFBE+ΔF508 (ΔF508) cell monolayers. Experiments were performed in triplicate (*P < 0.05). B: ICP-MS analysis of iron in the apical medium overlying airway cells. Experiments were performed in triplicate (*P < 0.05). M, medium. C and D: iron chelation with conalbumin inhibits biofilm formation on airway cells. C: z-series image stacks of PAO1 bacteria grown on CFBE+ΔF508 cells with or without 20 μg/ml conalbumin were acquired 6 h after inoculation. Experiments were performed in triplicate, and representative images are shown after deconvolution and 3-D reconstruction. Grid unit, 14.5 μm. D: bacterial biomass was determined with COMSTAT (*P < 0.05 compared with CFBE+WT without conalbumin). No significant difference (ns) was detected between biofilms grown on CFBE+WT or CFBE+ΔF508 cells in the presence of conalbumin. E: iron supplementation increases biofilm formation on CFBE+WT cells. P. aeruginosa biomass on CFBE+WT cells was assessed using COMSTAT in a flow chamber perfused with growth medium supplemented with either 0.5 or 8.0 μM FeCl₃ (−, no supplementation). The iron concentration in unsupplemented medium was ∼0.05 μM, similar to the levels of iron found in the bronchoalveolar lavage of healthy patients. Experiments were performed in triplicate (*P < 0.05 compared with CFBE+WT with no additional iron).

If the increased iron concentration in the apical medium contributes to enhanced biofilm formation on ΔF508-CFTR cells, we predict that treatment with conalbumin, an iron chelator, should decrease biofilm formation. Moreover, we predict that the difference in biofilm formation on CFBE+WT and CFBE+ΔF508 cells should be eliminated by conalbumin. Conalbumin is a protein isolated from chicken eggs that chelates iron and which is known to inhibit P. aeruginosa biofilm formation on abiotic surfaces (56). Indeed, conalbumin reduced biofilm formation on CFBE+ΔF508 cells (Fig. 6C) and eliminated the difference in biofilm formation between CFBE+ΔF508 and CFBE+WT-CFTR cells (Fig. 6D and Table 1). These data are consistent with the conclusion that enhanced iron concentration in the apical medium of CFBE+ΔF508 cells is responsible, in part, for the enhanced biofilm formation on CF cells compared with CF cells rescued with WT-CFTR.

If the elevated iron concentration of the apical medium overlying CFBE cells is responsible, at least in part, for the difference in biofilm formation between CFBE+ΔF508 and CFBE+WT-CFTR cells, we predict that adding iron to the apical medium of CFBE+WT cells should stimulated biofilm formation by P. aeruginosa. Thus studies were conducted to examine the effect of the addition of iron to the apical medium overlying CFBE+WT cells on biofilm formation. In these experiments, the cell culture medium pumped into the chamber was supplemented with different concentrations of iron, and P. aeruginosa biofilm formation was assessed 6 h later. The iron concentration in the apical medium was increased from levels typically found in BAL fluid isolated from healthy individuals (<0.1 μM) to those observed in CF patients (>8 μM; Refs. 1, 59, 60). Although addition of 0.5 μM iron had no stimulatory effect on biofilm formation, addition of 8 μM iron to the medium bathing CFBE+WT cells stimulated biofilm formation to a level indistinguishable from that observed in the CFBE+ΔF508 cell line in which the iron concentration in the fluid added to the apical medium was 0.055 μM (Fig. 6E). By contrast, addition of 8 μM iron to the medium bathing CFBE+ΔF508 cells did not change biofilm formation (Fig. 6E). Thus these data are consistent with the view that enhanced iron release by CFBE+ΔF508 cells is responsible, in part, for increased biofilm formation on CFBE+ΔF508 cells. The reason why addition of iron to the apical medium overlying CFBE+ΔF508 did not enhance biofilm formation will be addressed in discussion.

Iron enhances the formation of biofilms on abiotic substrates.

If iron is important for biofilm formation on airway epithelial cells, it can be predicted that iron should also enhance biofilm formation on an abiotic substratum. Thus iron was added to MEM bathing P. aeruginosa growing in 96-well tissue culture plates, and biofilm formation was assessed by crystal violet staining as described previously (43). Ferric chloride (8–20 μM; Supplemental Fig. 5) increased biofilm formation. A significant increase in biofilm formation was observed with 8 μM, and a maximum effect on biofilm formation was observed at a free iron concentration of 15 μM.

We also used holo-transferrin as a source of iron, instead of ferric chloride, to present the bacteria with a more physiologically relevant source of iron. Holo-transferrin, the iron-saturated form of transferrin, enhanced biofilm formation (Supplemental Fig. 5). A significant increase in biofilm formation was observed with 0.5 μM holo-transferrin, and a maximum effect on biofilm formation was observed at a concentration of 8 μM (Supplemental Fig. 5). Thus, as expected, iron presented as holo-transferrin was more effective in promoting biofilm formation than iron presented as ferric chloride. Moreover, taken together, these data reveal that iron, in the concentrations observed in the medium overlying airway epithelial cells and at levels that are clinically relevant to the CF lung, promote the formation of biofilms.

Corr4a does not reduce the iron concentration in the apical medium.

To determine whether Corr4a decreased biofilm formation by reducing the iron concentration in cells and in the apical medium, parental CFBE cells were treated with Corr4a or vehicle, and the iron concentration in cells and the apical medium was measured by ICP-MS. Interestingly, Corr4a had no effect on the iron levels in parental CFBE cells or on the iron concentration in the apical medium (Supplemental Fig. 6). As presented in discussion, these results suggest that Corr4a reduces biofilm formation by a mechanism that does not involve changes in the iron concentration in the media and is consistent with the view presented below that multiple factors, in addition to iron, are likely to be involved in enhanced biofilm formation by P. aeruginosa on CF epithelial cells.

DISCUSSION

In this study, we developed a novel, in vitro continuous flow cell model of confluent human airway epithelial cells to examine, by high-resolution imaging of living cells, the effects of ΔF508-CFTR and WT-CFTR expression on the ability of P. aeruginosa to form drug-resistant biofilms. By using flow chambers, we were able to monitor, in real-time, the early stages of apical colonization of human airway epithelial cells leading to the formation of bacterial microcolonies and the concomitant interactions developing between P. aeruginosa and epithelial cells. Our observations reveal that P. aeruginosa forms biofilms on airway cells expressing ΔF508-CFTR and that rescue with WT-CFTR reduces biofilm formation, consistent with in vivo observations (8). Our data indicate that the increased biofilm formation on CF-derived cells is not due to increased attachment or decreased detachment but to the promotion of biofilm growth on the CF-derived airway cells. Moreover, we report that airway cells enhance the resistance of P. aeruginosa to Tb by >25-fold, to a level that is ∼10-fold higher than is achieved in the lung of CF patients undergoing treatment with aerosolized Tb. Finally, we report studies to begin to elucidate the mechanism whereby the ΔF508-CFTR mutation enhances biofilm formation. These data reveal that enhanced iron concentration in the apical medium by CF airway cells stimulates the formation of antibiotic-resistant biofilms and that chelation of iron with conalbumin dramatically reduces biofilm formation by P. aeruginosa that grow on cells expressing ΔF508-CFTR.

Most experiments presented here were performed using a set of isogenic cell lines, which were derived from the airway of a patient homozygous for the ΔF508 mutation. The CFBE cell line is useful because isogenic derivatives have been constructed that contain a plasmid expressing WT-CFTR or ΔF508-CFTR. Thus direct comparisons can be made between cell lines expressing WT- and ΔF508-CFTR proteins: there are currently no other airway epithelial cell lines that provide the opportunity for such isogenic comparisons. However, it should be noted that similar results were obtained on an isogenic set of MDCK cells expressing WT-CFTR or ΔF508-CFTR. That is, even in kidney cells, the ΔF508-CFTR mutation enhanced the formation of P. aeruginosa biofilms.

One potential disadvantage of the CFBE cell lines is that they do not produce mucus, and recent studies suggest that biofilms may form in the mucus rather than directly on CF airway cells (57, 67), although these data are far from conclusive. However, using the flow cell system we describe here, we did observe that robust biofilms formed on mucus-producing JME/CF15 airway cells (Fig. 3). In fact, biofilms that formed on mucus-producing JME/CF15 cells were similar to those that formed on CFBE cells. It is also important to note that our in vitro system also lacks other host defense components that are likely to impact bacterial colonization and persistence, including immune cells. However, the flow cell coculture system used here allows us to explore the interactions between P. aeruginosa and airway cells that are not possible with the analysis of biofilms on abiotic surfaces such as plastic and glass. Indeed, our system has led to the novel observations that growth on airway cells enhances the formation of P. aeruginosa biofilms by 1,500-fold compared with the biofilms that form on abiotic surfaces and that coculture with airway cells dramatically enhances the resistance to antibiotics used in the CF clinic including Tb, ciprofloxacin, and imipenem.

Our data also reveal that iron plays an important role in enhanced biofilm formation by airway (CFBE) and kidney (MDCK) cells expressing the ΔF508-CFTR mutation. The concentration of iron in the apical solution by cells expressing ΔF508-CFTR was higher than the iron concentration in the apical medium produced by isogenic cells expressing WT-CFTR. Moreover, chelation of iron with conalbumin eliminated the difference in biofilm biomass between CFBE+ΔF508 and CFBE+WT cells. Consistent with our conclusion that enhanced iron release by CF cells facilitates biofilm formation, we found that addition of iron (8 μM) to the medium overlying CFBE+WT cells increased biofilm biomass to the level observed in CF cells (CFBE and CFBE+ΔF508) that were incubated in an apical medium with a very low iron concentration (<0.1 μM). Addition of iron to P. aeruginosa growing on an abiotic surface also promoted biofilm formation. It is interesting to note that addition of iron (8 μM) to CFBE+ΔF508 reduced biofilm formation, although the decrease did not achieve statistical significance. This observation is consistent with the view that P. aeruginosa growing on CFBE+ΔF508 cells may be optimally obtaining iron from the apical medium secreted by CFBE+ΔF508 cells and that addition of iron to this solution may be cytotoxic to P. aeruginosa. Consistent with this speculation, others (68) and we observed a decreased in biofilm formation when the iron concentration was elevated above 20 μM in the grown medium of P. aeruginosa growing on an abiotic surface.

Thus, taken together, these data suggest that physiologically relevant levels of iron that are observed in the lungs of non-CF and CF patients (1, 3–5, 46–48, 54–56) regulate biofilm formation by P. aeruginosa and that increased iron release caused by the ΔF508-CFTR mutation is responsible, at least in part, for the enhanced biofilm formation in the CF lung.

It is important to note that many other factors, in addition to increased iron release by CF cells and enhanced drug-resistant biofilm formation, are likely to contribute to the chronic P. aeruginosa lung infection. Our model of enhanced biofilm formation is not mutually exclusive with other models put forward to explain the colonization by and persistence of P. aeruginosa in the CF lung. Indeed, it is likely that the persistent infection by P. aeruginosa in the CF lung is also enhanced by reduced mucociliary clearance, increased mucus abundance and viscosity, high DNA concentration, as well as alginate produced by P. aeruginosa (8, 15, 48, 65). The observation that Corr4a reduced biofilm formation without altering iron levels is consistent with the view that many mechanisms, in addition to differences in iron metabolism, contribute to the persistent infection of the CF lung with antibiotic-resistant P. aeruginosa.

Additional studies, beyond the scope of this manuscript, are required to elucidate the mechanism whereby the ΔF508 mutation results in increased cellular iron and increased iron concentration in the medium overlying CFTR+ΔF508 cells. We speculate that there is a CFTR-dependent alteration in iron metabolism that increases iron accumulation in CF airway cells and the airway surface liquid. Consistent with this idea, levels of the iron-binding protein ferritin are increased in sputum isolated from the CF lung (49). Recently, Matsui et al. (36) showed that dehydrated mucus in CF facilitates colony formation by P. aeruginosa, but not bacterial growth, and that the dehydrated mucus rendered the iron-binding protein lactoferrin ineffective. This lack of lactoferrin should increase the availability of iron to P. aeruginosa. Taken together with the present study, these observations suggest that many factors including enhanced iron release by CF cells contribute to the ability of P. aeruginosa to thrive in the CF lung.

Interestingly, we showed that five genes known to be required for biofilm formation on abiotic surfaces were also required for biofilm formation on airway cells. However, the biofilms formed by P. aeruginosa on airway epithelial cells developed much more rapidly and to a greater extent (1,500-fold increase in biomass) than biofilms that form on glass, and furthermore, they exhibited a dramatically increased resistance to a variety of clinically relevant antibiotics, including a MBC_Tb that exceeds 10,000 μg/ml. To our knowledge, this is the first report that biofilms on human airway epithelial cells are dramatically more resistant to antibiotics than biofilms of similar bacterial cell numbers that form on abiotic surfaces. Because current therapy achieves Tb concentrations of only ∼1,000 μg/ml in CF sputum (14) immediately after aerosol therapy, our data provide an explanation why Tb therapy does not eradicate the chronic infection by P. aeruginosa. Thus our findings indicate that although an abiotic surface can serve as an excellent surrogate for airway cells in regard to identifying biofilm formation factors, there are clear differences in at least two properties (biofilm growth and antibiotic resistance) between biofilms formed on biotic and abiotic surfaces, thus emphasizing the need for a model to explore the interactions of P. aeruginosa biofilms with human airway epithelial cells.

A key advantage of the in vitro system we describe here is that it allows manipulation of both the bacterial pathogen and the host; therefore, experiments can be performed to determine whether available treatment strategies that impact CFTR localization and function will also impact the interaction of P. aeruginosa with the host. In our study, genetic overexpression of ΔF508-CFTR did not reduce the ability of P. aeruginosa to form biofilms. However, treatment with Corr4a, which belongs to a class of recently identified CFTR therapeutics, results in a modest reduction in bacterial biomass, providing proof of concept that correction of Cl⁻ ion secretion by a small molecule can indeed reduce the formation of biofilms on CF cells.

In conclusion, these data suggest that the ΔF508 mutation in CFTR increases iron availability in the airway surface liquid of the CF lung, which facilitates the development of highly drug-resistant biofilms by P. aeruginosa. We suggest that the in vitro model of biofilm formation on airway cells we have developed will serve as a valuable tool for studying the interaction of P. aeruginosa with its host, for developing new antibiotic regimens for CF, and for validating therapeutics for the treatment of CF (and possibly other lung diseases) that result in colonization by bacterial pathogens.

GRANTS

This work was supported by the Cystic Fibrosis Foundation Research Development Program to B. A. Stanton (STANTO07R0), and the National Center for Research Resources Centers of Biomedical Research Excellence (COBRE) HL-074175 funds to B. A. Stanton and G. A. O'Toole.