Chemicals

All chemicals, of analar grade or better were obtained from Sigma-Aldrich Chemical Co., Poole, UK, unless otherwise stated. All agars and broths were obtained from Oxoid, UK, except where stated.

Bacterial isolates

Ninety-six Pseudomonas aeruginosa isolates were cultured from sputum samples taken from 13 children known to be infected only with P. aeruginosa, who were attending the CF clinic in Belfast City Hospital, N. Ireland at the same time (Andrienne Shaw, pers. Comm. 2003). Isolates were chosen based on their colony morphology on Pseudomonas isolation agar. All isolates were initially confirmed as P. aeruginosa using both the API20 NE identification system (BioMerieux, France) and by subsequent amplification of the P. aeruginosa- specific OprL gene [17]. Isolates were deemed mucoid or non-mucoid by visual inspection on agar and the collection was stored at -70°C on Protect cryobeads (Technical Service Consultants, Heywood, U.K.).

Arbitrarily primed-PCR (AP-PCR)

Motility assays

Microtitre plate assay for assessment of biofilm formation

P. aeruginosa strains were grown to an attenuance (D_{600 nm}) of 0.5 and diluted 100-fold with LB broth following which 100 μl aliquots were dispensed into triplicate microtitre plates which were incubated at 37°C. Three plates were sacrificed for analysis every 4 h and attenuance was measured at 600 nm with a BioTek FL600 microtitre plate reader against uninoculated LB broth; means and standard deviations were calculated. Biofilms were stained with 1% crystal violet, washed with deionised water and quantitated by adding 95% ethanol followed by measurement of the absorbance (OD 595 nm) as per Stepanovic et al. [33]. Strains with no change in O.D over the control were classified non-biofilm producers, weak- (up to a 2 fold change), moderate- (up to 4 fold change) or strong- (greater than 4 fold change) as per Strepanovic et al. [33] All tests were carried out in triplicate and the results were averaged. P. aeruginosa strain PAO1 was included as a positive control.

Biofilms in a capillary flow reactor were grown in glass capillary tubes of square cross sections under continuous flow conditions. The capillaries had a nominal inside dimension of 900 μm and a wall thickness of 170 ± 10 μm (Friedrich & Dimmock, Millville, N.J., USA). The flow cell apparatus consisted of a vented medium feed carboy (four litre capacity), a flow break, a filtered air entry, a peristaltic pump (Watson-Marlow), the capillary and flow cell holder, an inoculation port, and a waste carboy. The components were connected by silicone rubber tubing and were sterilised by autoclaving.

A culture of gfp-P. aeruginosa was grown in LB overnight at 37°C in a shaking incubator at 140 rpm. A 100 μl aliquot of this culture was used to inoculate 10 ml of sterile LB broth in a 250 ml conical flask to achieve good aeration and the culture was grown at 37°C with shaking at 200 rpm for 3 h. The tubing was clamped downstream of the inoculation port and the capillary flow system was inoculated with 300 μl of this fresh culture. The tubing was then clamped upstream of the glass tube and the system was allowed to stand without a flow for 19 h to allow the cells to attach to the glass capillary at 37°C. After initial attachment, the flow of medium (1/10 strength LB, to avoid blockage of the capillary due to excessive biomass production) was adjusted to a flow rate of 20 ml h^-1.

Bacterial staining of mixed biofilms consisting of biofilm⁺ and biofilm^- isolates, were stained with 300 μl of a 5 mg l^-1 rhodamine B (Kodak) solution in water. The stain solution was injected into the capillary reactor through the inoculation port and the cells allowed to stain for 5 min. Biofilms were subsequently observed by confocal scanning laser microscopy with excitation and emission wavelengths of 540 nm at 625 nm respectively for rhodamine B and 475 nm and 510 for GFP.

Scanning Electron Microscopy (SEM)

Prior to SEM, samples were chemically fixed as follows: A 10 μl aliquot of an overnight culture, grown in LB broth at 37°C, with shaking at 140 rpm was placed in a round glass coverslip (10 mm diameter, Chance Proper Ltd., UK) with a 10 μl of fixative (3% glutaraldehyde in 0.1% sodium cacodylate, pH 7.3). The coverslips were previously coated with polylysine (Sigma-Aldrich) to assist adherence of bacterial cells. The cells were allowed to fix for 10 min following which the supernatant was discarded and excess fixative was added to the coverslip for 5 min followed by a final washing step with 0.1 M sodium cacodylate, pH 7.3. In order to dehydrate the bacteria the coverslips were successively placed for 10 min in each one of the following solutions: 30%, 50%, 70%, 90%, and 100% (twice) (v/v) acetone. The coverslips were then dried with a critical point drier and sputter coated with Au: Pt, 60:40 in argon (Polarow E5100). The slides were visualized with a JSM 840 SEM (JEOL Ltd., Herts, UK).

Light and Epifluorescence microscopy examination of P. aeruginosa cells was performed using a Nikon Eclipse E800 microscope equipped with 40 × and 60 × water objectives, differential interference contrast (DIC) polarizing filters and reflectance optics. For epifluorescence microscopy, the microscope was equipped with a 100 W Hg-vapour discharge lamp and fluorescent images were obtained using the following filters: B-2A blue excitation filter with excitation wavelength 470-490 nm, (Nikon) and a Red excitation filter: Cy5 HYQ (Nikon). Images were captured by a Micromax RTE/CCD-732-7 (Princeton Instruments, Trenton, NJ, USA) camera and MetaVue 5.0 software (Universal Imaging Co., Downingtown, PA, USA).

CLSM and image analysis

Glass capillary flow reactors were inoculated with the GFP-P. aeruginosa isolates and biofilms in capillary flow reactors were observed using 40 × magnification lenses with a CLSM (Leica TCS-NT). CSLM image analysis software was Image Pro Plus, Version 3.00.00 (Media Cybernetics, Bethesda, MD, USA). Microscope images were analyzed by use of the line scan fiction of Metamorph image analysis software (Universal Imaging Co., Downingtown, PA, USA). For the depth profile, the interface between the biofilm and the glass wall was set to zero on a spatial axis. Stimulated fluorescence projections and vertical cross sections through the bacterial biofilms were generated with IMARIS (Bitplane AG) software package running on a Silicon Graphics Indigo 2 workstation.

Statistical analysis was performed in order to validate the effect of motility in P. aeruginosa biofilms. The isolates were divided into four groups based on their motility patterns: the first group (C1) consisted of isolates that both swim and twitch, the second (C2) of immotile isolates, the third (C3) of isolates that swim but do not twitch and the forth (C4) of isolates that twitch but do not swim. A one-way ANOVA was performed to test the null hypothesis that there were no differences in the mean motility of the four groups, followed by a Tukey's post-hoc to compare the individual groups' differences. Tukey's post-hoc calculates a 95%-confidence interval for the mean of each group and then substracts the means pair-wise i.e. C1 minus C2, C1 minus C3 etc. If the differences include 0 then the means are not significantly different. The results were verified by performing a two sample Ttest within pairs of groups and the obtained p-values for multiple testing corrected using Bonferroni correction. MiniTab was used for the statistical analysis.

Statement of Ethical Approval

Research carried out in this study was approved by Health and Personal Social Services (HPSS) (Northern Ireland) REC 2, Reference No. 07/NIR02/39.

Diversity in biofilm formation by P. aeruginosa CF isolates

In order to visualise the differences in attachment between strong and weak biofilm forming isolates, bacterial cells were allowed to attach to glass coverslips and subsequently visualized using SEM. Coverslips were immersed vertically in an inoculated culture and two coverslips were removed and fixed for SEM visualization at intervals. Diversity in isolate attachment onto the glass cover slip was observed, with the moderate and strongly adhering isolates from the microplate assay forming clumps of cells (e.g. isolate 17; Fig. 1a). Weakly adherent isolates attached as individual cells (e.g. isolate 80; Fig. 1b) however as both types of biofilm matured, the spaces between the clumps were filled with a cell lawn (Fig. 1c &1d).

Isolates previously characterised as weakly adherent did not form the characteristic biofilm structures, and we observed that relatively few cells were attached to the glass substrate and that biofilm formation was initiated only after the surrounding planktonic culture had reached stationary phase. At this point the cells were elongated, reaching up to 15 μm in length - a potential response to nutrient limitation also observed by other researchers.

P. aeruginosa isolates from CF patients show diversity in motility phenotype

Having observed significant diversity in biofilm formation within the group of clinical isolates we then investigated isolate motility. Swimming motility was initially observed for 48 isolates (50%) with a migration zone of 7 - 40 mm (Table 3, column 7). Twitching motility was distinguished by the presence of an interstitial twitch zone formed by colony expansion. Isolates exhibiting twitching motility (Table 3, column 6) formed flat spreading colonies with a characteristic "rough" appearance and a twitching zone consisting of a very thin layer of cells observed as a halo around the colony. Isolates incapable of twitching formed small, smooth, flat colonies on the agar surface that remained at the inoculation point. Coomassie staining revealed a series of concentric rings in the twitching zone. When P. aeruginosa isolates were inoculated onto the surface of agar to assay swarming motility, 36 (37%) of the isolates (Table 3, column 8) formed characteristic swarming patterns consisting of branches or tentacles radiating from the inoculation point. Movement across the agar surface was rapid, with bacteria having colonised the entire surface of the plate within several hours after inoculation. A lack of twitching motility was not matched by an absence of swarming motility, but did seem to influence the pattern of colony translocation. When twitching motility was present, the swarm edge exhibited a higher bacterial cell concentration than the centre, while in non-twitching isolates, bacterial colonies were denser in the centre and surrounded by a thin film of translocated bacteria. Of the 96 isolates, a number of strains overlapped in terms of motility phenotype: 24 had flagellar and twitching motility, 27 had only twitching motility, 47 had only swarming motility and a total of 45 were non motile.

Given the complex phenotypic diversity of the clinical isolates based on direct observations we recognized the need for a rational approach to selecting the most appropriate isolates for further study. We adopted RAPD as a convenient and quick genotyping method that allowed us to characterise the heterogeneity in the group, using a cut off value of 85% similarity as a threshold to compare strains. Primer 10514 generated a total of 22 different profiles (Table 3), fifteen of which contained more than one isolate. Primer 10514-generated profiles were cross- referenced with those of primer 14306 and showed that similar profiles were generated with both primers.

P. aeruginosa CF isolates exhibit a lack of correlation between motility phenotype and genotype

The observed phenotypic differences in twitching motility led us to consider whether non-twitching isolates were inherently non-motile or whether they possessed the capability to be motile but did not express it. Pilin alleles and associated gene(s) are located in a common chromosomal locus between the conserved pilB and tRNA ^Thr genes [18]. The presence of various tfp accessory genes located upstream of pilA determines amplicon size, thus allowing the delineation of five TFP groups [18, 31]. Seven twitching efficient and 13 twitching deficient isolates were selected (Table 4) and we determined whether or not pilA, the type IV pilus (TFP) gene responsible for the PilA structural protein, was present in the isolates.

Thirteen isolates yielded ~2.8 kbp amplicons with the pilB and tRNA^Thr primers [31], thus the majority of the CF isolates fell into TFP group I (tfpO). Amplicons of ~1.4 kb were produced with the same pilB and tRNA^Thr primer set from isolates 1, 3 and 64 placing them in TFP group II (tfpY). Six other primer combinations were tried with isolates 41, 54, 55 and 72, however a pilA amplicon was generated only from isolate 72 using primers pilA and tRNA^Thr, showing that it belonged to TFP group V (tfpZ). Of the 17 isolates for which pilA presence was confirmed only 7 (41%) actually exhibited twitching motility, demonstrating that the presence of pilA alone does not secure motility. Representative amplicons were cloned and sequenced and subsequent alignments confirmed their categorisation into the groups described by Kus et al. [31]. The fliC structural gene was also detected in all 20 isolates (Table 4), however its presence, like that of pil A, did not guarantee swimming motility as 9 isolates (45%) did not swim. The presence of flagella in isolates was verified with SEM, while full length DNA sequences were obtained for fliC of isolates 1, 40, 41 (motile) and 48 (non motile).

Statistical analysis shows that motility contributes to biofilm thickness but not to biofilm formation in our isolates

It has been reported in a number of studies [16, 25, 35, 36] that motility is required for biofilm formation, whereas in contrast, Klausen et al. [28] reported that mutants deficient in pili and flagella showed no significant differences from wild type. In the current study, biofilm formation was not influenced by the presence of either flagella or type IV pili, since 45 isolates that formed either moderate or strong biofilms were deficient in twitching, swimming, and swarming motility. In contrast however, isolates 5, 6, and 61 (motile) exhibited very poor adhesion in microtitre plates. For the statistical analysis we started with the null hypothesis that motility does not affect biofilm formation and performed a one-way ANOVA that gave an F- value of 9.88, allowing rejection of the null hypothesis. At this point we could not say between which groups the difference was so we performed a Tukey's post-hoc test between all the possible group pairs. Group C1, as it was termed for the analysis, contained the highest percentage of strong biofilm forming isolates - 80% - while in groups C2 and C3 the percentage of strong biofilm forming isolates was only 40% and 33%, respectively (Fig. 2). The results revealed that C1 was different from C2, C3 and C4 but there was no difference among the C2, C3 and C4. The same conclusion was reached using a Ttest with correction for multiple testing. We concluded therefore that the combination of swimming and twitching motility has a positive contribution to biofilm biomass but is not absolutely necessary for the initiation of the process.

Diversity in biofilm architecture among P. aeruginosa isolates

Having shown statistically that the isolates possessing both twitching and swimming motility produced greater biofilm biomass we set out to investigate the architecture of biofilms produced by members of this group. We gfp tagged 5 isolates exhibiting different motility/biofilm biomass combinations: 17 and 40 (twitch⁺, swim⁺, biofilm⁺⁺⁺), 41 (twitch^-, swim⁺, biofilm⁺), 55 and 80 (twitch^-, swim^-, biofilm⁺). The resulting gfp-tagged isolates had growth rates identical to those of the parental strains (data not shown). P. aeruginosa ATCC15442 was used as a control to ensure that reactor did not influence biofilm morphology and following staining with Syto9 and propidium iodide, characteristic mushroom-shaped biofilms of P. aeruginosa ATCC15442 were observed in a number of different reactors. Spatial biofilm distribution in the tagged strains was measured over time in a glass capillary flow reactor and images were acquired with CLSM at regular 12 h intervals at random positions in the flow chambers. Visual inspection revealed that the biofilm architecture of the P. aeruginosa CF isolates was significantly different from that of the ATCC control strain (Fig. 3). Among the isolates tested, 17, 40 and 41 gave biofilm growth while isolates 55 and 80 did not attach to the glass capillary. Isolates were observed as microcolonies on day 1 and formed a biofilm within 48 h of inoculation. They continued to grow until the 7^th day with a maximum thickness of 75 μm for isolates 17 and 40 and 145 μm for isolate 41. Isolate 17 formed a mushroom-shaped biofilm that appeared after 48 h of growth, while isolate 40 formed a flat biofilm with small hilly structures spatially distributed. The biofilm formed by isolate 41, was flat and was the thickest among the isolates. Although stains 55 and 80 showed weak attachment to microtitre dish wells, other than a transient superficial attachment at seven hours no attachment was observed from 12 hours onward in the glass capillary flow reactor. We observed that cell attachment proceeding to biofilm formation was dependent upon the attachment substrate.

Entrapment of non-biofilm forming P. aeruginosa isolates by an already established biofilm

We wished to test the hypothesis that the diversity of isolates derived from clinical samples is due to non-biofilm producing isolates interacting with biofilm producers. Glass capillary flow reactors were inoculated with the GFP-P. aeruginosa 17 isolate and the biofilm formation was followed with CLSM. Following 48 h growth, the capillary reactor was inoculated with isolate 80 and the flow was stopped for 3 h to allow attachment. The bacterial biofilms were stained with rhodamine B (reference colour) and observed with CLSM 24 h after inoculation with isolate 80 (Fig. 4). Isolate gfp-17 was identified by green fluorescence due to the production of GFP, and isolate 80 was identified by rhodamine B. The excitation and emission wavelengths were distant between the fluorophores and did not overlap. Isolate gfp-17 established a green lawn that colonised the reactor surface, while isolate 80 was observed as spatially distributed red cell clumps within the established biofilm. Furthermore, cross sectional analysis of the biofilm (Fig. 5) showed that isolate 80 was not only attached to the surface of the isolate 17 biofilm, but that the cells were incorporated into the three dimensional structure of the established biofilm, suggesting that isolate 80 was able to migrate into the established biofim despite its lack of twitching and swimming motility.