|Home | About | Journals | Submit | Contact Us | Français|
Bacterial biofilms are responsible for the majority of all microbial infections and have profound impact on industrial and geochemical processes. While many studies documented phenotypic differentiation and gene regulation of biofilms, the importance of their structural and mechanical properties is poorly understood. Here we investigate how changes in lipopolysaccharide (LPS) core capping in Pseudomonas aeruginosa affect biofilm structure through modification of adhesive, cohesive, and viscoelastic properties at an early stage of biofilm development. Microbead force spectroscopy and atomic force microscopy were used to characterize P. aeruginosa biofilm interactions with either glass substrata or bacterial lawns. Using isogenic migA, wapR, and rmlC mutants with defined LPS characteristics, we observed significant changes in cell mechanical properties among these strains compared to wild-type strain PAO1. Specifically, truncation of core oligosaccharides enhanced both adhesive and cohesive forces by up to 10-fold, whereas changes in instantaneous elasticity were correlated with the presence of O antigen. Using confocal laser scanning microscopy to quantify biofilm structural changes with respect to differences in LPS core capping, we observed that textural parameters varied with adhesion or the inverse of cohesion, while areal and volumetric parameters were linked to adhesion, cohesion, or the balance between them. In conclusion, this report demonstrated for the first time that changes in LPS expression resulted in quantifiable cellular mechanical changes that were correlated with structural changes in bacterial biofilms. Thus, the interplay between architectural and functional properties may be an important contributor to bacterial community survival.
Biofilms are sessile microbial communities growing on a surface or at an interface, often enmeshed in polymeric substances. Being the predominant mode of microbial growth in nature, bacterial biofilms are particularly problematic in the context of human health, accounting for up to 80% of all bacterial infections. In industrial processes, bacterial biofilms cause corrosion and biofouling, resulting in considerable loss of productivity. In the natural environment, biofilms play a role in modulating worldwide geochemical cycles. Given the impact of biofilms in these diverse areas, the need for developing effective strategies to control them is of paramount importance. Since bacterial cell surface structures are convenient targets for control agents, their roles in influencing biofilm function and architecture warrant in-depth investigations. To date, most studies of biofilms have focused on genetic regulation, phenotypic differentiation and their contribution to antibiotic resistance. In contrast, the mechanical and structural properties that link the genotypes to phenotypes of bacterial biofilms are not well understood and rarely studied in a quantitative and correlated manner.
Pseudomonas aeruginosa is a gram-negative opportunistic pathogen implicated in serious infections in patients with cystic fibrosis and immunocompromised patients. This bacterium has a relatively large genome (6.3 Mb) consistent with its propensity to utilize versatile metabolic pathways, thereby developing antibiotic resistance and producing an arsenal of virulence factors, including lipopolysaccharide (LPS) present on the cell surface. LPS is localized in the outer leaflet of the outer membrane of all gram-negative bacteria, forming the first point of contact between the bacterial cell and any surface that it colonizes or therapeutic agents. The LPS of P. aeruginosa consists of three regions: lipid A, core oligosaccharide (core OS), and O antigen. The O antigen is synthesized as two distinct forms with overlapping pathways: the shorter A-band homopolymer is the so-called “common polysaccharide antigen” among this species and consists of repeating d-rhamnose (d-Rha) subunits, while the longer B-band heteropolymer is composed of repeating tri- to pentasaccharide subunits that vary among the 20 serotypes of P. aeruginosa (42). The core OS is conceptually divided into the highly conserved inner core and the more variable outer core. Depending on the linkage of l-rhamnose (l-Rha) with two distinct d-glucose (d-Glc) residues, two main glycoforms of the core OS exist (see Fig. Fig.1A).1A). In the “capped” glycoform, l-Rha is α-1,3 linked to a d-Glc and acts as the acceptor molecule for O antigen, resulting in the production of smooth LPS. In the “uncapped” glycoform, l-Rha is α-1,6 linked to a different d-Glc and is not substituted with O antigen, resulting in the production of rough LPS (39). In addition, the presence or absence of the α-1,6 linked l-Rha substituted with a terminal d-Glc gives rise to the so-called intact or truncated outer core, respectively. The functional significance of this terminal glucose is unclear at present, although a role in host cell binding has been proposed (57).
Mechanical processes that are important in the biofilm life cycle include bacterial adhesion, cohesion, and viscoelasticity. Bacterial adhesion is a prerequisite for surface colonization and the most important functional determinant in the early stages of biofilm development. Accurate measurement of adhesion is therefore essential for monitoring the tendency of bacteria to attach to surfaces and to switch from a planktonic lifestyle to a biofilm lifestyle. Data accumulated in previous studies suggest that LPS is involved in bacterial cell adhesion to both abiotic (2, 8, 20, 32, 35, 54, 56) and biotic (17, 38, 49, 56, 57) surfaces. Moreover, environmental factors, such as growth temperature, pH, ionic strength, nutrient availability, and oxygen levels, may influence cell adhesion via modification of LPS expression and conformation (16, 36, 46, 47, 53). The effect of LPS on bacterial adhesion to various types of surfaces apparently involves distinct and complex mechanisms that remain to be elucidated.
Bacterial cohesion, herein defined as cell-to-cell adherence, is crucial to the formation of microbial flocs and the growth and detachment of established bacterial biofilms. Quantification of cohesion is important for understanding biofilm biology, and such data are crucial for modeling and forecasting biofilm development so that better control strategies can be developed (55). Previous studies of biofilm cohesiveness have characterized it as highly stratified (3, 4, 18, 43), influenced by ionic strength (14, 34), proportional to shear rate (37), and often variable over 3 orders of magnitude (40, 50). Although an earlier study by Spiers and Rainey (48) provided semiquantitative measurements of the role of LPS on bacterial cohesion within a biofilm, a truly quantitative account of the effect of LPS on biofilm cohesion has not been demonstrated.
Bacterial viscoelasticity refers to the combined liquid-like and solid-like characteristics in the behavior of polymeric systems such that when deformed under stress, their strain can increase over time (i.e., creep) and their original shape may be only partially restored upon stress relief (19). Although earlier reports suggested that LPS modulates bacterial cell compressibility and helps prevent catastrophic structural failure due to mechanical stress (1, 52), no direct physical evidence of its involvement in these processes has yet been presented. Therefore, monitoring biofilm viscoelasticity is crucial for demonstrating how well biofilms resist stresses, due to, for instance, fluid shear and antimicrobial peptides (5, 6). To date, quantitative data on how LPS affects viscoelastic properties of biofilms are lacking, and existing studies have merely focused on elasticity measurements (7, 52). Recently, our group has developed an atomic force microscopy (AFM)-based technique called microbead force spectroscopy (MBFS) to measure the adhesive forces and viscoelastic properties of cells within bacterial biofilms (28). In this study, we expand the application of this MBFS method to measure cohesive forces between cells at an early stage of biofilm development.
Biofilm structure refers to the distribution of biomass or carbonaceous materials associated with cells (including all viable and nonviable cells and their extracellular polymeric substances) within the space occupied by a biofilm. It is known to be very heterogeneous and highly stratified, typically composed of a cohesive basal layer and a relatively fragile top layer (15, 18, 43). Using confocal laser scanning microscopy (CLSM), biofilm structure can be quantitatively described in terms of textural and volumetric parameters (11, 29). Textural parameters characterize the pattern of cell clusters and interstitial voids in a biofilm, whereas volumetric parameters describe the morphological characteristics of bacterial biofilms in three dimensions (3-D) (11). Biomass distribution is affected by the surrounding environment and may reflect fundamental processes occurring within biofilms, such as nutrient transport, accumulation rate, microbial physiology, and mechanical behavior (29). Therefore, quantifying biofilm structure by CLSM will allow us to understand the underlying processes and the relationship between biofilm architecture and behavior (15).
To examine the effects of differential LPS core capping on the mechanical properties of early biofilms (here defined as confluent bacterial lawns that have just begun to develop into full-fledged biofilms) and the structural properties of mature biofilms, we compare P. aeruginosa wild-type strain PAO1 with those of its isogenic migA, wapR, and rmlC mutant strains with defects in the respective genes affecting LPS core biosynthesis (22, 31, 39, 41). The migA gene (PA0705) encodes the putative α-1,6-rhamnosyltransferase necessary for the attachment of the terminal d-Glc to the outer core (39). The wapR gene (PA5000) encodes the putative α-1,3-rhamnosyltransferase crucial to the capping of the core with O antigen (39). The rmlC gene (PA5164) encodes a dTDP-4-dehydrorhamnose 3,5-epimerase essential in the biosynthesis of TDP-l-Rha, which is the precursor for the l-Rha in the LPS core (41). Defects in migA, wapR, and rmlC mutants result in the expression of different LPS phenotypes, including a truncated outer core and/or a lack of capping by O antigen (Table (Table1).1). In this study, we test the hypothesis that LPS contributes to biofilm function and architecture through modulation of cellular mechanics and microcolony structures, thereby contributing to bacterial community survival. By correlating quantitative mechanical changes in early P. aeruginosa biofilms and structural changes in mature biofilms due to differences in LPS chemistry, we aim to elucidate how the properties of these important bacterial cell surface molecules can alter the physical nature of biofilms.
Pseudomonas aeruginosa wild-type strain PAO1 and three isogenic mutant strains with differential LPS core capping were used in this study (Table (Table1).1). Bacteria were grown overnight (16 h) in Trypticase soy broth (TSB) at 37°C on an orbital shaker (125 rpm). The cells were harvested by centrifugation at 2,300 × g for 5 min, and the pellets were washed in sterile water or phosphate-buffered saline and recentrifuged. After the final resuspension, 10-fold dilutions were made, and the optical density at 600 nm (OD600) was measured. The original washed but undiluted cultures were adjusted to the appropriate concentrations (see the individual descriptions of experiments below) for various experiments.
Biofilms of P. aeruginosa strain PAO1 and its three isogenic mutants were grown in continuous-culture flow cells constructed of polycarbonate (Biosurface Technologies Inc., MT) with chamber dimensions of 40 mm by 10 mm by 0.1 mm. Briefly, 75% (vol/vol) ethanol was first pumped from a reservoir through silicone tubing at 1 ml/min using a peristaltic pump (Minipuls 2; Gilson, Inc., Middleton, WI) to sterilize the flow cells for at least 12 h. Subsequently, 1/10 strength TSB (BD) was pumped from a 6-liter flask (medium reservoir) to as many as eight flow cells set up in parallel at a rate of 1 ml/min for 2 h, before inoculation of washed and diluted overnight cultures (OD600 of 0.1) via upstream injection ports (while the line from the medium reservoir was clamped).The flow of culture medium was suspended during inoculation to facilitate adhesion and resumed 1 h postinoculation. The spent medium exiting the flow cells was collected in a 20-liter carboy (waste reservoir). Flow cells were operated for 3 days in experiments cultivating biofilms that were used to isolate and characterize LPS and for 7 days in experiments cultivating biofilms that were evaluated by confocal microscopy (see below).
Lipopolysaccharide was prepared by the method of Hitchcock and Brown (23) and resolved by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis. Briefly, overnight bacterial cultures or bacterial cells collected from disrupted 3-day-old biofilms were washed and diluted to an OD600 of 0.5 in a volume of 1 ml per sample. Bacterial suspensions were centrifuged at 2,300 × g for 5 min and resuspended in 250 μl of Hitchcock and Brown lysis buffer (2% SDS, 4% β-mercaptoethanol, 10% glycerol, 1 M Tris HCl [pH 6.8], 0.002% bromophenol blue). The samples were heated at 100°C for 30 min, cooled to room temperature, and incubated with 1.5 μl of proteinase K (20-mg/ml stock; Roche Diagnostics, Mannheim, Germany) at 56°C for 2 h. For each sample, 5 μl of the preparation were resolved by electrophoresis on a 12.5% polyacrylamide gel at 150 V. After electrophoresis, polyacrylamide gels were silver stained according to the ultrafast method of Fomsgaard et al. (21).
Topographic atomic force microscopy images were obtained using an MFP-3D atomic force microscope (Asylum Research, Santa Barbara, CA). For imaging bacterial lawns, rectangular silicon cantilevers CSC38 (no Al, type B) (Mikromasch USA, Wilsonville, OR) with manufacturer's quoted resonance frequencies of ~10 kHz (range, 7 to 14 kHz) and force constants of ~0.03 N/m (range, 0.01 to 0.08 N/m) were used. Twenty microliters of washed and diluted overnight bacterial cultures (OD600 of 1.0) was deposited onto freshly cleaved mica and air dried for 20 min. Atomic force micrographs were collected by raster scanning the sample under contact mode in air with the set point adjusted to 0.2 V and the integral gain at 30 or under contact mode in water with the set point adjusted to 0.05 V and the integral gain at 50. Height, deflection, z-sensor and lateral signals were simultaneously collected for real-time image construction. The atomic force microscope was controlled using the MFP-3D software version 070111+217 (Asylum Research, Santa Barbara, CA) operating within the Igor Pro 6.02A software environment (Wavemetrics, Inc., Lake Oswego, OR). The root mean square (RMS) roughness of cell surfaces was calculated from height images after the background areas had been masked. Three-dimensional renditions of height images were constructed offline using the Argyle function in the MFP-3D program.
The MFP-3D atomic force microscope was also used in force mode to quantify the force of bacterial adhesion based on a method called microbead force spectroscopy that was recently developed by our group (28). For quantitative measurement of forces, rectangular tipless silicon cantilevers (CSC12 [tipless, no Al, type E]; Mikromasch USA, Wilsonville, OR) with the manufacturer's quoted resonance frequencies of ~10 kHz (range, 7 to 14 kHz) and force constants of ~0.03 N/m (range, 0.01 to 0.08 N/m) were calibrated in air by the thermal method (24) to derive an accurate spring constant for each individual cantilever. This was a two-step process involving measuring the inverse of the optical lever sensitivity by pressing on a hard surface and collecting the thermal nose power spectrum. Only cantilevers with a calibrated spring constant within the range of 0.015 to 0.045 N/m were accepted for use in force spectroscopy studies. Microsized glass beads with diameters of 50 μm (Polysciences, Inc., Warrington, PA) were attached to the distal ends of the cantilevers with two-component epoxy glue using a micromanipulator and dried overnight. Beaded cantilevers were then coated with 0.01% poly-l-lysine by exposure for 1 min and air dried for 10 min. Overnight bacterial cultures that were washed and adjusted to an OD600 of 2.0 were applied to the poly-l-lysine-coated beads three times to ensure confluence. After the cantilever was mounted onto the head of the atomic force microscope and centering the infrared laser spot behind the bead, the cantilever assembly was immersed in 100 μl of sterile deionized water on a precleaned glass slide or a mica surface precoated with a bacterial lawn as was done for imaging studies. The microbead was lowered gradually to approach the glass surface or the cell layer, upon which force curves (force-separation plots) and creep curves (indentation-time plots) were gathered simultaneously. In each individual experiment, 10 force curves and 10 creep curves were collected for each sample under standard conditions, defined as a loading force of 10 nN, contact time of 1 s, ramp velocity of 2 μm/s, and ramp distance of 3 μm (28).
After 7 days of growth, biofilms in flow cells were stained in situ using reagents from the Live/Dead BacLight bacterial viability kit (BD Biosciences). Equal proportions of the green SYTO 9 stain for live cells and the red propidium iodide stain for dead cells were mixed and diluted in sterile deionized water (3:1,000), and 1 ml of this mixture was injected into the upstream port of each flow cell. Biofilms were incubated in complete darkness for 30 min (which was found to be sufficient time for stain penetration), and unbound stains were removed by injection of 3 ml of 1/10 strength TSB into each upstream port. Imaging by CLSM was performed immediately and completed within 24 h using a confocal microscope (model TCS SP2; Leica Microsystems Canada Inc., Richmond Hill, Ontario, Canada). For each of the four bacterial strains investigated, optical sectioning was performed for three representative microcolonies to obtain image stacks for structural quantitation. Excitation wavelengths were set at 488 nm and 543 nm, while emission bandwidths were set at 500 to 535 nm and 555 to 700 nm, respectively, for detecting the green and red channels. Image stacks consisted of image slices spaced 1 μm apart, starting from the substratum and ending at the top of microcolonies. Image slices were gathered at a scan speed of 400 Hz and a resolution of 1,024 by 1,024 pixels using a Leica PL Fluotar 40.0 × 1.00 oil immersion objective lens (with the 10× eyepiece and 2× digital zoom, a total magnification of ×800). The microscope was controlled using the proprietary Leica confocal software, a platform in which data for cross sections and average projections of microcolonies could be obtained.
Biofilm structural quantitation was performed on the CLSM image stacks by the methods of Lewandowski and Beyenal (29). In such 3-D analyses, image stacks were exported as individual images in TIFF format from the Leica application suite advanced fluorescence lite program (Leica) into XnView (freeware by Pierre Gougelet, available at http://www.xnview.com), batch converted to gray-scale images, and resized to 188 by 188 pixels such that “voxel” dimension was 1 μm by 1 μm by 1 μm. Using the ISA-3D package for biofilm image analysis (Center for Biofilm Engineering, Montana State University, Bozeman, MT) within the MATLAB and Image Analysis Toolbox environments (Mathworks, Inc., Natick, MA), textural parameters (i.e., textural entropy, energy, and homogeneity) were extracted from the gray-scale images, whereas volumetric parameters (i.e., average run lengths, aspect ratios, diffusion distances, fractal dimension, porosity, biovolume, biomass and biofilm thicknesses, biofilm and biomass roughnesses, biomass surface area, and biovolume-to-biomass surface area ratio) were calculated from binary images after an automatic threshold protocol was applied.
For the measurement of contact angles, a sessile drop technique based on the method of Korenevsky and Beveridge was performed (27). Briefly, overnight cultures of P. aeruginosa cells were harvested and washed twice with a sterile buffer (0.1 M NaCl-0.05 M HEPES [pH 7.4]) and adjusted to an OD600 of 1.0 in a 10-ml volume of the buffer. The bacterial cells were then deposited on cellulose acetate membrane filters (AcetatePlus, supported, plain, 0.22 micron, 47 mm; GE Water & Process Technologies) via vacuum aspiration to produce an even, confluent bacterial lawn. Bacterial layers were dried for 30 min before the so-called “plateau contact angles” were measured using sterile deionized water droplets. Water droplets of approximately 2 μl were deposited onto the bacterial lawns using a micropipette (Gilson, Inc., Middleton, WI). To obtain static contact angles, water droplets were allowed to settle for 2 s before digital images were captured from the side using a 12.1-megapixel charge-coupled-device camera (Sony of Canada Ltd., Toronto, Ontario, Canada) operating in macro mode, against a bright illuminating background. Saved images were processed by a program (designed by Joop der Vries, University of Groningen, The Netherlands) based on the approximation that sessile droplets form spherical caps. Mean contact angles for each sample were calculated from five different droplets deposited on different areas of the membrane, with five replicate measurements per droplet.
The relative hydrophobicity of cells was examined using the bacterial adhesion to hydrocarbon (BATH) assay (45). Briefly, cells cultured in LB broth were standardized to an OD600 of 0.3, harvested by centrifugation at 5,000 × g, and washed once in 0.1 M potassium phosphate buffer (pH 7.0) before resuspension in 1.5 ml of the same buffer. The cell suspension (1.2 ml) was transferred to a glass test tube (10-mm diameter) and incubated with 200 μl of hexadecane (Fisher Scientific) for 10 min at 30°C. The cell and hexadecane mixture was vigorously mixed by using a Vortex (VWR Scientific) for a total time of 2 min, followed by further incubation for 25 min at room temperature. The OD595 of the aqueous layer (ODaq) was measured on a BMG FluoStar Optima plate reader spectrophotometer and compared to the OD595 of the cell suspension before the addition of hexadecane (ODcs) in order to determine the percent adherence using the following equation: percent adherence = [1 − (ODaq/ODcs)] × 100.
The clumping and resultant precipitation of bacterial cells were monitored and quantified over a 24-h period as follows. Each strain of P. aeruginosa was grown overnight (16 h) in 10 ml of TSB in glass test tubes at 37°C with shaking. The cultures were then placed upright in a test tube rack to allow flocculation to proceed. Finally, at 0, 1, 2, 3, 4, and 24 h after the beginning of the assay, 100-μl aliquots were taken from near the surface (5-mm depth) of each culture and diluted 10-fold for OD600 measurements.
The three isogenic mutants of P. aeruginosa PAO1 investigated in this study, namely, the migA, wapR, and rmlC mutants, have differential core capping in their LPS moieties (Fig. (Fig.1A1A and Table Table1).1). Silver-stained SDS-polyacrylamide gels of LPS from planktonic cells revealed that the rough strains lacking LPS O antigen (i.e., wapR and rmlC mutants) are devoid of long-chain O polysaccharides as well as core OS capped with one O-antigen repeat unit, i.e., the “core-plus-one” entity (Fig. (Fig.1B).1B). Interestingly, the biofilm cells of these strains appeared to have LPS profiles similar to those of their planktonic counterparts. In contrast, the PAO1 strain and migA mutant, which both possess O polysaccharides, showed different profiles between planktonic and biofilm LPS. Specifically, the LPS bands at intermediate-molecular-weight O antigen present in the LPS sample from planktonically grown PAO1 cells were absent in LPS prepared from the same strain grown in biofilm mode. These intermediate-molecular-weight bands were also absent in LPS samples prepared from the migA mutant grown planktonically as well as in migA cells grown as biofilm cultures in the flow chamber. Furthermore, production of both the full-length O antigen and the core-plus-one entity present in the former sample were significantly reduced in the latter sample, likely due to downregulation.
Atomic force micrographs of P. aeruginosa cells scanned in air revealed changes in cell surface morphology in the LPS truncation mutants compared to the wild type (Fig. (Fig.2).2). While deflection images were useful for the emphasis of textural details, height images were obtained for the calculation of surface roughness after masking to eliminate background areas (Fig. 2A to D). The values for RMS roughness for strain PAO1 (wild type) and migA, wapR, and rmlC mutants were found to be 49.43 nm, 68.41 nm, 44.40 nm, and 34.14 nm, respectively. Images collected in aqueous medium did not show any discernible differences in appearance among the cells of the different strains (data not shown). Three-dimensional rendering of the height images in air were produced to highlight changes in surface topography (data not shown). The textures of cells shown in these images corresponded with RMS roughness data and further demonstrated that bacterial cells from strains possessing O antigen have rougher cell surfaces, while strains devoid of O antigen have a smoother appearance.
Using microbead force spectroscopy with two different bead and surface configurations (see Fig. S1 in the supplemental material), we were able to precisely quantify the forces of early biofilm adhesion to glass and cell-cell cohesion (see Fig. S2A in the supplemental material) for the wild-type strain PAO1 and its three isogenic mutants. Overall, cell adhesion and cohesion were enhanced in the LPS core truncation mutants than in the wild type (Fig. (Fig.3).3). The average adhesive force of bacteria to glass measured for wild-type strain PAO1 and migA, wapR, and rmlC mutants under standard conditions were 0.66 ± 0.27 nN, 5.52 ± 1.83 nN, 6.90 ± 1.22 nN, and 5.12 ± 1.12 nN, respectively (Fig. (Fig.3A).3A). The corresponding average cell-cell cohesive forces for strain PAO1 and migA, wapR, and rmlC mutants were 1.00 ± 0.31 nN, 2.99 ± 0.13 nN, 5.15 ± 1.09 nN, and 12.18 ± 0.15 nN, respectively (Fig. (Fig.3B).3B). The inverses of adhesion and cohesion as well as their ratios were plotted to reveal other possible trends when comparing the four bacterial strains to each other (see Fig. S3 in the supplemental material). These trends allowed the assessment of correlation between quantitative changes in mature biofilm structure with variations in early biofilm mechanics.
For the determination of early biofilm viscoelastic properties, fitting of data from the MBFS experiments to a Voigt standard linear solid model (28) yielded values for instantaneous and delayed elastic moduli, viscosity, and characteristic response time (i.e., retardation time) (see Fig. S2B in the supplemental material). The viscoelastic properties of all the LPS core truncation mutants were different from those of the wild type (Fig. (Fig.4).4). Instantaneous elastic moduli were found to be larger for wild-type strain PAO1 and migA mutant (1.69 × 105 Pa ± 0.80 × 105 Pa and 1.88 × 105 ± 1.08 × 105 Pa, respectively) than for wapR and rmlC mutant strains (6.25 × 104 ± 1.90 × 104 Pa and 7.96 × 104 ± 2.38 × 104 Pa) (Fig. (Fig.4A).4A). Delayed elastic moduli were larger for the PAO1 strain (1.09 × 106 ± 0.10 × 106 Pa) than for the migA, wapR, and rmlC mutants (4.78 × 105 ± 0.61 × 105 Pa, 6.12 × 105 ± 4.60 × 105 Pa, and 7.48 × 105 ± 6.15 × 105 Pa, respectively) (Fig. (Fig.4B).4B). The viscosity values for strain PAO1 and migA, wapR, and rmlC mutants were found to be 2.31 × 105 ± 1.06 × 105 Pa·s, 1.28 × 105 ± 0.05 × 105 Pa·s, 2.12 × 105 ± 1.66 × 105 Pa·s, and 3.72 × 105 ± 3.25 × 105 Pa·s, respectively (Fig. (Fig.4C),4C), whereas the characteristic response times were measured and found to be 0.204 ± 0.078 s, 0.274 ± 0.046 s, 0.327 ± 0.026 s and 0.432 ± 0.079 s, respectively, for these strains (Fig. (Fig.4D4D).
Bacterial cultures of LPS core mutants exhibited a higher degree of clumping than those of the wild-type strain (see Fig. S4 in the supplemental material). Previous work carried out by our group also indicated that the larger the truncation of the core OS and/or the O antigen, the greater the resultant level of cell clumping in bacterial biofilms grown on glass slides (30). We therefore carried out semiquantitative bacterial clumping assays to verify these observations. Cells in broth cultures of strain PAO1 and migA, wapR, and rmlC mutants exhibited increasing tendencies, in that order, to flocculate (form clumps) and settle at the bottom of stationary test tubes (Fig. (Fig.5).5). These observations showed good correlation with quantitative measurements of cell-cell cohesion forces for these respective bacterial strains (Fig. (Fig.3B3B).
The bacterial adhesion to hydrocarbon assay, though indirect, has been routinely used for examining the hydrophobicity of cells and is based on the fact that cells with more hydrophobic surfaces will adhere more readily to a droplet of hydrocarbon (44). Under the standardized experimental conditions described in Materials and Methods, cells of the wild-type strain PAO1 and the migA, wapR, and rmlC mutants with LPS variants adhered to n-hexadecane with increasing affinity, with 33.5%, 38.2%, 40.2%, and 45.2% of cells adhered on average, respectively (Table (Table2).2). These data should theoretically correspond with contact angle measurements between water and bacterial lawns using a sessile drop technique, since contact angles are characteristic of interfacial energies and thus also dependent on cell surface hydrophobicity (51). Accordingly, bacterial lawns of the wild-type strain PAO1 and its isogenic LPS migA, wapR, and rmlC mutants were shown to contact water droplets in an increasing trend with angles of 33.3°, 34.0°, 34.1°, and 37.5°, respectively (Table (Table22).
Confocal micrographs of the bacterial strains examined showed that while cells of wild-type strain PAO1 form microcolonies with round perimeters, the microcolonies formed by the LPS mutants have more irregular edges (Fig. (Fig.6).6). Interestingly, biomass thickness was found to be consistently higher for strain PAO1 and rmlC mutant. These qualitative observations were confirmed by quantitative comparison of average run lengths and biomass thicknesses, respectively (Table (Table3).3). Quantitative biofilm structure analysis of microcolony image stacks in 3-D also indicated that there were significant changes of textural and volumetric parameters between wild-type strain PAO1 and its isogenic mutants that could be correlated with differences in force and viscoelasticity measurements (Table (Table4)4) (see Fig. S5 in the supplemental material).
Until this study, the contribution of LPS to the mechanical properties of bacterial biofilms had not been defined. The few existing quantitative studies of how LPS affects bacterial adhesion have given inconclusive results. For example, Burks et al. (13) performed nanoscale adhesion measurements by conventional force spectroscopy on three E. coli strains of differing LPS phenotypes but were unable to correlate the results with data from macroscale assays they gathered based on binding to glass bead columns. Likewise, Abu-Lail and coworkers (2) observed that bacterial adhesion to silica is significantly higher for P. aeruginosa strain AK1401, a mutant that lacks B-band O antigen, than for P. aeruginosa PAO1, while adhesion to organics is stronger in the wild-type strain. In separate studies, however, this research group observed that adhesion to silicon was stronger for strain PAO1 than for strain AK1401 (8), while adhesion to serum albumin was three times higher in the mutant (9). These conflicting results in comparing the adhesive strengths of wild-type and LPS variant strains might have been caused by the use of sample preparation techniques and testing environments that have not been standardized in the different studies (28). To address these issues, we employed the MBFS method to measure the mechanical properties of P. aeruginosa cell layers derived from strains with different LPS phenotypes and further correlated the differences in mechanical behavior at this early stage of biofilm formation to matrix architectural changes in fully developed biofilms, as quantified by image analysis of confocal laser scanning micrographs.
Although bacterial cells growing in a biofilm have been shown to lose B-band O antigen from their LPS (10), the role of core OS structure in this process is unclear. The LPS-banding profiles of samples prepared from strain PAO1 and its core-deficient mutants in SDS-polyacrylamide gels clearly revealed that strain PAO1 and migA mutant had substantially different patterns when grown planktonically and in a biofilm, albeit in different ways. In strain PAO1, the medium-length O antigen present in LPS prepared from planktonically grown cells was absent when the wild-type strain was grown as biofilms, whereas in the migA mutant, no medium-length O antigen could be discerned in the LPS prepared from planktonically grown cells. In addition, O antigen and “core-plus-one” bands were also absent from LPS samples of migA mutant grown as biofilms. These observations suggested that the outer core defect caused by mutation in the migA mutant might lead to instability or downregulation of O-antigen capping. A more intriguing possibility is that core OS integrity may have been compromised, as suggested by the appearance of a new band running slightly faster than core OS in the migA mutant biofilm sample (Fig. (Fig.1B1B).
Bacterial cells with and without LPS O antigen generally exhibit “smooth” and “rough” colony morphology of bacterial growth, respectively. Interestingly, high-resolution images of strain PAO1 and its LPS core variants captured by AFM showed that the roughness of the surface of individual bacterial cells is inversely related to colony roughness. The “smooth” strains possessing O antigen exhibited rough cell surfaces, likely due to the presence of higher proportion of exopolysaccharide, while “rough” strains without the O polysaccharide showed smoother cell surfaces. Also, the observation that the migA mutant strain has rougher cell surfaces than strain PAO1 (as measured by the RMS roughness) is logical because the loss of O antigen from some migA mutant LPS can lead to a mix of smooth and rough LPS, resulting in greater variations in surface topography. Despite these differences, the contribution of cell surface roughness to bacterial interaction with external surfaces remains unclear, since surface roughness simultaneously increases friction and reduces contact area, two factors that have opposing effects on adhesion. Finally, although drying time had been standardized for AFM imaging, differential contribution of drying artifacts in the various strains to RMS roughness measurements could not be ruled out.
Bacterial adhesion is a fundamental requirement for cells to attach to substrata for biofilms to form. Measurements of bacterial adhesion at an early stage of biofilm formation using MBFS showed that strains with rough LPS (wapR and rmlC mutants) have quantitatively stronger adhesion to glass than strains with smooth LPS do (strain PAO1 and migA mutant). These results are consistent with data from assays that measured bacterial adhesion to glass that were previously performed by our group, suggesting that increasing defects in the LPS core and O antigen lead to an increasing degree of adhesion to glass (30). Intriguingly, the trend in adhesive force measured by MBFS (strain PAO1 < migA mutant < wapR mutant > rmlC mutant) was not monotonically increasing, with the wapR mutant having a stronger adhesion to glass than the rmlC mutant strain did. This trend contrasted with that found in our earlier study performed using planktonic cells and suggested a role for the terminal d-glucose in the intact core OS of the wapR mutant in strengthening adhesion to glass within the context of a biofilm.
Bacterial cohesion is a critical link between adhesion and viscoelasticity. Force measurements by MBFS showed that cells with increasing core and O-antigen defects exhibit increasing cell-cell cohesion. As expected, this trend in cohesive force (strain PAO1 < migA mutant < wapR mutant < rmlC mutant) correlated well with visual observations and semiquantitative assays of bacterial clumping. In addition, cohesion may also be linked to cell surface hydrophobicity as measured by bacterial adhesion to hydrocarbons and by contact angles. However, the minor variations in hydrophobicity compared to the much larger differences in cohesion among the four strains tested suggest that hydrophobicity played only a minor role in cell-cell cohesion.
The balance between bacterial adhesion and cohesion is crucial to understanding the structural development of a biofilm. For instance, biofilm adhesive and cohesive strengths may contribute to the differential effects of solid surface tension, hydrodynamic shear, and disinfecting agents on different biofilms (12, 33). Remarkably, certain trends in adhesion and cohesion and their derived quantities (i.e., inverses and ratios) among the four tested P. aeruginosa strains proved to be highly correlated with some viscoelastic properties and quantitative structural data (see below).
Bacterial viscoelasticity is an important determinant of how bacteria growing as biofilms respond to physical stress. Since biofilms are known to be viscoelastic materials that exhibit both viscous and elastic behaviors, we modeled them using a simple mechanical analog (the Voigt standard linear solid model) consisting of a spring in series with a spring-dashpot element in parallel (28). Measurements of biofilm viscoelasticity of the bacterial strains used in this study showed that four main trends exist: (i) link to the presence of O antigen (strain PAO1 ≈ migA mutant > wapR mutant ≈ rmlC mutant), as seen for instantaneous elastic modulus; (ii) correlation with inverse adhesion (PAO1 > migA mutant > wapR mutant < rmlC mutant), as seen for delayed elastic modulus (correlation coefficient [R] = 0.9119); (iii) correlation with the cohesion-to-adhesion ratio (PAO1 > migA mutant < wapR mutant < rmlC mutant), as seen for viscosity (R = 0.9496); and (iv) correlation with cohesion (PAO1 < migA mutant < wapR mutant < rmlC mutant), as seen for characteristic response time (R = 0.9799) (see Fig. S6 in the supplemental material). Thus, the presence of O antigen appeared to be important in the immediate response of a biofilm to mechanical stress (instantaneous elasticity). In contrast, inverse adhesion influenced the continued response to sustained indentation (delayed elasticity). Moreover, the ratio of cohesion over adhesion was important for determining the rate of irreversible deformation (viscosity). Finally, cohesion affected the initial interval in which rapid creep (~63% of total creep) occurs after a constant stress is applied, a characteristic response time derived by dividing viscosity over delayed elasticity (28).
Biofilm structure can be defined as the spatial distribution of biomass within a biofilm. The most practical approach for quantifying biomass distribution is by analyzing confocal microscopy images and taking into consideration parameters that characterize biofilm structures. Since these parameters are only mathematical functions (with arbitrary units) characterizing pixel distribution in the biofilm images, they must be chosen carefully for their relevance to the underlying biofilm processes. Following the method of Lewandowski and Beyenal (11, 29), we calculated selective parameters that are deemed useful for describing biofilm structure, and we further correlated these parameters to mechanical properties of cell layers at an early stage of biofilm formation (see Fig. S7 in the supplemental material). When the confocal image stacks of representative microcolonies were subjected to structural quantification, 3-D textural parameters and volumetric parameters were derived. Trends for textural parameters correlated either with cell adhesion, as seen for energy (R = 0.9477) and homogeneity (R = 0.9701), or with inverse cohesion, as seen for textural entropy (R = 0.9327). The volumetric parameters followed five different trends: (i) correlation with adhesion, such as for average x and y run lengths (R = 0.7358 and 0.7227, respectively), average and maximum diffusion distances (R = 0.8965 and 0.9715, respectively), average biofilm roughness (R = 0.8982), and biovolume-to-biomass surface area ratio (R = 0.7605); (ii) correlation with inverse adhesion, as for fractal dimension (R = 0.9945); (iii) correlation with cohesion, such as for average z run length (R = 0.8122) and aspect ratio xy (R = 0.9932); (iv) correlation with the adhesion-to-cohesion ratio, such as for aspect ratios xz and yz (R = 0.9691 and 0.9899, respectively); and (v) correlation with the cohesion-to-adhesion ratio, such as for biovolume (R = 0.9475), biomass and biofilm thicknesses (R = 0.9514 and 0.9937, respectively), average biomass roughness (R = 0.9908), and biomass surface area (R = 0.8020). On the basis of the above analyses, we were able to conclude that in general, textural parameters were most related to early biofilm adhesion, whereas volumetric parameters were variously linked to adhesive and cohesive properties and the balance between them. In a conceptual model linking mechanical and structural properties, it is reasonable to assume that a biofilm in a flowing environment grows thicker if cohesion is stronger than adhesion but becomes more spread out over a surface if adhesion predominates. This conjecture has indeed been proven correct by our observations that the wild-type strain PAO1 and rmlC mutant have both higher cohesion-to-adhesion ratios and thicker biofilms than the migA and wapR mutants do. Interestingly, the only structural parameter not correlated with force measurements was porosity. Since this parameter, which had previously been related to biofilm accumulation rate (29), was instead linked to the presence of an intact outer core in this study, our observation implicates the integrity of the LPS outer core in optimal metabolic activity. For a more detailed description of the biofilm structural quantification parameters, interested readers are referred to Table S1 in the supplemental material.
While unicellular organisms can simply modulate their structure to bring about a functional response to an external stimulus, the coordinated responses of multicellular systems such as bacterial biofilms to environmental stresses are much more complex. Specifically, structural changes in individual cells need to be translated into mechanical and architectural modifications in the entire cell assembly for responses to be effective. Previous studies linking biofilm mechanics and structures (25, 26) merely provided qualitative observations and conceptual assessments of the properties of biofilms but did not take into account the roles of specific cell surface features such as LPS expression. In this study, we measured the mechanical properties in the early biofilms of P. aeruginosa wild-type strain and mutants with LPS core variants and related them to quantitative structural data in mature biofilms. The evidence presented here suggests that the structure of bacterial biofilms is most strongly correlated to the strength of adhesion, the balance between adhesion and cohesion, and the structural integrity of the LPS outer core OS. For the first time, changes in bacterial mechanical behavior during an early stage of biofilm formation caused by differential LPS core capping have been correlated with matrix architecture modifications in fully developed biofilms. Also, the presence of a terminal glucose in the LPS core of P. aeruginosa appeared to be linked to core OS stability, increased adhesiveness in the absence of O antigen, and optimal structural development within a biofilm. Evidently, LPS expression and the resultant interplay between biofilm mechanics and structure may contribute importantly to bacterial community survival. Therefore, further investigation correlating these properties in other model microbial systems under various environmental stimuli will be of tremendous value to microbiological research.
This work is supported by AFMNet funding and Discovery Grants to T.J.B. and J.R.D. from the Natural Sciences and Engineering Council of Canada and an operating grant to J.S.L. from the Canadian Cystic Fibrosis Foundation. T.J.B. was a Canada Research Chair in the Structure of Microorganisms, J.R.D. is a Canada Research Chair in Soft Matter Physics, and J.S.L. holds a Canada Research Chair in Cystic Fibrosis and Microbial Glycobiology.
Published ahead of print on 28 August 2009.
†Supplemental material for this article may be found at http://jb.asm.org/.