A physical linkage between cystic fibrosis airway surface dehydration and Pseudomonas aeruginosa biofilms

P. aeruginosa Growth and Growth Patterns in 2.5% vs. 8% Mucus.

P. aeruginosa grew and within 48 h reached the stationary phase at ≈10¹⁰ cfu/ml for 2.5% and 8% mucus in both open and closed chambers (Fig. 1C). In closed chambers, pO₂ plunged below 5% ≈9 h after bacterial inoculation and remained low. In open chambers, pO₂ remained higher in 2.5% mucus but was <5% at 24 h in 8% mucus (Fig. 1D). These data suggest that P. aeruginosa grew well in human airway mucus under hypoxic as well as normoxic environments, as previously reported (9, 19).

However, the growth pattern of P. aeruginosa (GFP-PAO1) was distinctly different in 2.5% vs. 8% mucus (Fig. 2). Epifluorescence and confocal microscopy studies of the open-chamber model of CF airway mucus plaques revealed that P. aeruginosa grew as macrocolonies within 8% mucus gels (Fig. 2 B and D). These macrocolonies contained both live and dead cells, and some macrocolonies exhibited band-like patterns of live and dead cells (Fig. 2E). In contrast, no macrocolonies were detected within 2.5% mucus gels, and only small bacterial aggregations were observed on the plastic surface (Fig. 2 A and C). To estimate colony volume, the total fluorescence intensity of P. aeruginosa colonies was measured and was significantly greater in 8% mucus than in 2.5% mucus (Fig. 2Fi). Importantly, the difference in the colony size between 2.5% mucus and 8% mucus was not due to bacterial growth (number of bacteria, Fig. 1C) but was due to growth pattern (planktonic vs. macrocolony).

In the closed-chamber model of mucus plugs, intramucus macrocolonies were again detected in 8% mucus, whereas in 2.5% mucus, small colonies were only detected on plastic surfaces (not shown). The size of the macrocolonies developed in 8% mucus was smaller than in open chambers (Fig. 2Fii), suggesting that oxygen supply can affect macrocolony dimensions, in agreement with previous studies (9).

Because it is possible that CF mucus exhibits features other than increased concentration that promote biofilm formation, we studied 2.5% and 8% mucus from CF cultures. In both the open and closed systems, macrocolonies again were found in 8% but not in 2.5% CF mucus (Fig. 3A), suggesting that the concentration of mucus, rather than the source, was the critical variable. These data are consistent with previous data demonstrating no differences in the biochemical properties between the CF vs. normal mucins, the major component of mucus (20).

Development of P. aeruginosa biofilms is known to require the LasR-LasI quorum sensing system (4). To test whether the macrocolonies observed in airway surface liquid (ASL) were biofilms rather than loose aggregates of bacteria with low motility, we incubated lasI mutants of PAO1 in 8% mucus. These lasI mutants grew similarly to wild-type PAO1 in 8% mucus (≈10¹⁰ cfu/ml) (data not shown). x–y and x–z scanning confocal microscopy demonstrated planktonic single lasI mutant bacteria, but not macrocolonies, within 8% mucus (Fig. 3B iii and iv). The macrocolony phenotype was rescued with the addition of 10 μM N-(3-oxododecanoyl)-l-homoserine lactone (3-oxo C₁₂ HSL) to 8% mucus (Fig. 3Bv). These data demonstrate that the macrocolonies in 8% mucus gels share genetic features with biofilms studied in other systems.

Biophysical Properties of Mucus Gels.

What is unique about an 8% vs. 2.5% mucus gel? The predominant macromolecules in airway mucus are the high-molecular-weight mucin polymers. These polymers form a gel-like mesh in ASL that exhibits an apparent mesh pore width and viscoelastic properties as a function of their concentration. We used a microbead rheological technique that probes an environment in which tracer particles are placed by tracking their thermally driven Brownian motion to characterize the biophysical (mesh, viscoelastic) properties of 8% vs. 2.5% mucus gels. This method provides an excellent insight into the physical environment observed by micrometer-sized particles (21–26) and, thus, is an appropriate method to characterize the environment experienced by P. aeruginosa [0.5 × 1.5 μm (27)].

The thermally driven motion of microbeads greatly depends on the percentage of solids of the mucus sample in which they diffused. At the long time scales (≈5 sec), the means squared displacement curve, Δr², of 1-μm beads is 50 times greater in 2.5% mucus than in 8% mucus (Fig. 4A). The slope of the means squared displacement curve, α (see Methods), indicates that the fluid is liquid-like (purely viscous) when α = 1 and is solid-like (purely elastic) when α = 0. In 8% mucus, α was ≈0.2, whereas in 2.5% mucus, α was ≈0.8 (Fig. 4A), suggesting that the 8% mucus is highly elastic in nature.

Tracking motion of beads of varying sizes provides information about the mesh properties of mucus (28, 29). In 2.5% mucus, 200-nm beads diffused over much greater distances than in 1- and 3-μm beads, even when corrected for bead radius (Fig. 4B). These results indicate that the effective “mesh size” within the 2.5% mucus is between 0.2 and 1 μm. In 8% mucus, all three bead sizes were effectively trapped (Fig. 4B), indicating that all were trapped in a tight mesh smaller than 200 nm. This tighter mesh is also reflected in the decreased slope of the means squared displacement curve for 8% mucus in Fig. 4A.

Interaction of P. aeruginosa with Mucus Gel.

We tested predictions from the biophysical characterization of mucus in a series of microbiological experiments. First, consistent with analyses of mesh size and bead motility, swimming (flagellar) motility of PAO1 was found to be greatly reduced in 8% vs. 2.5% mucus (Fig. 4C). Second, consistent with a measured mesh size of 0.2–1 μm, an immobile PAO1 mutant (without pili and flagella) placed in 2.5% mucus sedimented out of mucus onto the plastic surface (data not shown). Third, it is predicted that small molecules (much smaller than mesh size) will have equivalent diffusion in both mucus concentrations. However, 3-oxo C₁₂ HSL (estimated to be smaller than 1 nm) diffused 10 times more slowly in 8% mucus vs. 2.5% mucus (calculated diffusion coefficient was 4.8 × 10⁻¹⁰ m²/s in 2.5% mucus and 4.9 × 10⁻¹¹ m²/s in 8% mucus) (Fig. 4D). We speculate that the slower diffusion reflects either increased interactions between the hydrophobic 3-oxo C₁₂ HSL molecules and the hydrophobic sites of the mucin mesh or HSL binding to mobile proteins larger than mesh size.

Effect of Mucus Gel on Soluble Innate Defense Molecules.

Finally, we asked whether there were unique aspects of thickened mucus that allow P. aeruginosa macrocolonies to evade soluble secondary-defense mechanisms that protect airway surfaces. Because of reports that lactoferrin was effective in inhibiting P. aeruginosa biofilm formation in well oxygenated, flow-through abiotic chamber surfaces (7), we studied lactoferrin effectiveness in 8% mucus gels.

Although the iron content (42.9 μM) of 8% mucus was in the range reported for CF mucus (17–134 μM) (30), we prepared low-iron mucus (8.4–12.3 μM) so that the iron-chelating capacity of physiological concentrations of lactoferrin (1 mg/ml, ≈22 μM) (31) was not exceeded. Pseudomonas formed macrocolonies in low-iron mucus (Fig. 5A), and this process was not inhibited by lactoferrin addition in concentrations predicted to be sufficient to chelate all iron (1 and 10 mg/ml) (Fig. 5B). A high concentration of conalbumin (10 mg/ml), reported to inhibit Pseudomonas macrocolony development (7), also had no effect on macrocolony formation in concentrated airway mucus (Fig. 5C). These results appear to mimic the condition in CF airways in vivo when biofilms form, despite apparently high concentrations of lactoferrin (2, 32). We also measured diffusion of fluorescently labeled lactoferrin from a point source in shallow chambers filled with either 2.5% or 8% mucus, mimicking submucosal gland secretion of lactoferrin into a mucus layer in the normal and CF airways. Like HSL, lactoferrin diffusion was markedly slowed in 8% mucus gels (Fig. 5D).

Preparation of Concentrated Mucus.

Human tracheo–bronchial epithelial cells (CF and normal) were isolated and cultured under air–liquid conditions as previously described (38). Mucus harvested from ≈200 concurrent cultures was pooled and stored at 4°C. The mucus samples were placed in dialysis tubes (molecular weight of CO = 3,500; Spectrum Laboratories, Rancho Dominguez, CA) and concentrated with a polymer absorbent (Spectra/Gel; Spectrum Laboratories) for 1–5 days at 4°C. The concentrated mucus was dialyzed against PBS (3 liters, four times) containing 500 μM MgCl₂ and 800 μM CaCl₂ at 4°C. The percent dry weight of mucus was calculated by measuring the wet and dry weights with a microbalance before and after drying the sample at 80°C overnight (39). Because of the difficulty in depleting harvested mucus samples of exact volumes, a 10% difference from the target concentration was allowed.

Determination of Mucin Concentration.

The concentration of mucins in mucus was assessed by differential refractometry (40). A 500-μl sample was chromatographed on a Sepharose CL-2B column (Amersham Pharmacia, Piscataway, NJ), which was eluted with 200 mM sodium chloride/10 mM EDTA at a flow rate 0.3 ml/min. The column effluent was passed through an in-line Dawn EOS laser photometer coupled to a Wyatt/Optilab DSP inferometric refractometer to measure light scattering and sample concentration, respectively. The concentration of the mucins was calculated by integrating the refractive index peak associated with the material eluted in the void volume of the column and employing a value for the refractive-index increment (dn/dc) of 0.165 ml/g, which has been measured previously at 650 nm. The reproducibility of this procedure was within 5%.

Preparation of Bacteria and Autoinducers 3-oxo C₁₂ HSL.

P. aeruginosa PAO1 (41), an isogenic lasI mutant (42), and a nonpiliated, nonflagellated mutant (PAO-NPNF, kind gift from A. Prince) were maintained on trypticase soy agar (Becton-Dickinson, Sparks, MD) with appropriate antibiotics as needed (gentamycin, 50 μg/ml; Sigma). In addition, the plasmid pTdK-gfp (43) was introduced by using electroporation (44) into these strains to allow visualization of the bacteria. Overnight cultures of bacteria were prepared for each assay with appropriate antibiotics when needed (200 μg/ml carbenicillin; Sigma). Autoinducer 3-oxo C₁₂ HSL was synthesized and purified as previously reported (45) and used as received.

Bacterial Culture in Mucus.

To measure bacterial growth, ≈5 × 10³ PAO1 were deposited on 270 μl of mucus of varying concentrations in a plastic flat chamber (1 mm in depth, 20 mm in diameter) with or without a plastic lid. At 0, 6, 24, 48, and 72 h later, bacteria in mucus were collected, diluted with PBS containing 10 mM DTT, and plated on tryptic soy agar plates after serial dilution. DTT treatment disaggregated the bacteria and did not affect the viability of the bacteria.

To measure macrocolony development, ≈5 × 10³ GFP-PAO1 were deposited onto mucus in the same plastic chambers and observed by epifluorescent (Nikon TE300) or confocal (Zeiss 510) microscopy at 24, 48, and 72 h. For quantitation, five pictures for each culture were digitally acquired and analyzed by using the Metamorph image analysis system (Universal Image). A bacterial colony was defined as a structure with a discrete edge and size larger than 50 μm². Dead bacteria were stained at day 3 with 5 μg/ml propidium iodide (molecular probes), which is impermeable into intact bacteria. The resolution of these images was too low to determine the internal structures of the biofilms.

Measurement of Oxygen.

Oxygen partial pressure was measured with an oxygen-sensitive fluorescent probe, the ruthenium complex Tris(4,7-diphenyl-1,10-phenanthroline)-ruthenium hexafluorophosphate (Sigma–Aldrich, St. Louis, MO). The ruthenium complex, adsorbed onto silica particles (1–5 μm; Sigma), was embedded in silicone rubber (silicone II; GE) and pasted onto glass slides (46). The coated glass slides were then used as the bottom of the bacterial culture chamber (Fig. 1B). Fluorescence intensity was measured by confocal microscopy at 3, 6, 9, 24, and 48 h to calculate oxygen partial pressure. Each glass-slide probe was calibrated with uninfected mucus for 21% pO₂ and with saturated sodium sulfite solution for 0% pO₂ at time 0. At each time point, artificial fluctuation of fluorescence intensity, caused partly by laser power instability, was corrected by measuring fluorescence intensity of distilled water (pO₂ = 21%) and sodium sulfite solution (pO₂ = 0%).

Bacterial Motility in Mucus.

GFP-PAO1 was deposited onto mucus (2.5% and 8% solids) in the plastic chamber and photographed by epifluorescent microscope with a ×40 objective at an interval of 0.25–1 sec. Only planktonic bacteria observed to undergo unidirectional motion were used to calculate bacterial velocity.

Diffusion of 3-oxo C₁₂ HSL in Mucus.

A 10 mM 3-oxo C₁₂ HSL solution in DMSO was mixed with a 10× volume of mucus (2.5% or 8% solids) to prepare the source of HSL. Capillary tubes (40 mm in length, 25-μl capacity; Statspin Technologies, Norwood, MA) containing 2.5% or 8% mucus were placed in the tubes containing HSL in an equivalent mucus concentration and had the free end sealed with silicone. After 24 h at 37°C, the capillary tubes with mucus were frozen and cut into eight pieces to measure the 3-oxo C₁₂ HSL concentration in each section.

Autoinducer was extracted from each mucus cross-section by using ethyl acetate (J. T. Baker) acidified with 0.001% glacial acetic acid (J. T. Baker) as previously described (47) with the following modifications. Twice the sample volume of acidified ethyl acetate was added to the mucus sample suspended in 100 μl of PBS in a 1.5- ml Eppendorf tube. Samples were subsequently vortexed for 3 min, and the two phases were separated by centrifugation at 10,000 × g for 10 min. The ethyl acetate phase was then transferred to a clean 1.5-ml Eppendorf tube. The extraction process was repeated once more, and the ethyl acetate phases were pooled. The concentration of 3-oxo C₁₂ HSL was assayed in each sample by using the previously described bioassay (47). The number of moles of 3-oxo C₁₂ HSL in each sample was then normalized to the total number of moles of 3-oxo C₁₂ HSL extracted from all samples in each assay and reported as percent occurrence. Each experiment was repeated in quadruplicate.

The diffusion coefficient was calculated by fitting a Gaussian to data collected for the distribution of radioactive molecules in 1D dimension. The distribution was defined by:

(48), where D is the diffusion coefficient and t is the time over which diffusion has occurred.

Microbead Rheology.

MyOne 1-μm-diameter carboxylated beads (Dynal), 2.9-μm-diameter yellow-green fluorescent carboxylated beads (Polysciences), and 200-nm-diameter red fluorescent carboxylated beads (Polysciences) were added to mucus samples, and their diffusion was recorded at 120 frames per second with a Pulnix PTM-6710 8-bit grayscale video camera (Jai-Pulnix) on a Nikon Eclipse TE2000-E inverted microscope with a ×60 water immersion objective (NA = 1.2). Beads were tracked in postprocessing by using Video Spot Tracker (software available for free download at http://www.cs.unc.edu/Research/nano/cismm/download/spottracker/video_spot_tracker.html). The average noise signal of a fixed bead was 5 nm.

Eq. 1 is used to calculate the mean squared displacement of particles undergoing Brownian motion within the fluid, Δr̄²(τ), over three decades of data (where τ is the time interval between measurements) (24).

The slope of the mean squared displacement curve in log–log space,

ranges between 0 for purely elastic fluids and 1 for purely viscous fluids and is used to characterize the degree of elasticity measured in a fluid (24).

Effects of Lactoferrin on Pseudomonas Macrocolony Formation.

GFP-PAO1 was deposited onto concentrated mucus of 8% solids containing lactoferrin (iron-unsaturated, Sigma) or conalbumin (Sigma) (final concentrations, 1 and 10 mg/ml). Macrocolony formation was monitored as mentioned above at days 1, 2, and 3.

Measurement of Iron in Mucus Samples.

Total iron content of mucus samples was measured after trichloroacetic acid precipitation (20%, 1:1) with a flame atomic absorption spectrometer (49) (AAnalyst 100; Perkin–Elmer, Wellesley, MA).

Diffusion of Lactoferrin.

One microliter of human lactoferrin solution labeled with Oregon Green (1 mg/ml; Molecular Probes) was deposited at four points onto an open chamber containing 270 μl of mucus (2.5% or 8% solids). After 0, 3, 18, 24, and 48 h of incubation at 37°C, fluorescent intensity was measured at the brightest and dimmest points, by using Nikon TE300 with a ×4 objective. Fluorescence intensities were normalized against control images of mucus with no lactoferrin.