Microorganisms constitute the most successful form of life on earth, in terms of total number, phylogenetic diversity and extent of habitats colonized. They impact on human existence and well-being, either directly by influencing human development, health and disease, or indirectly by carrying out processes in the natural environment or man-made environments (1–3). Our present state of knowledge on the biology of microbial cells is, due to historical reasons, largely a result of empirical research work on microbial cells living in suspension in a liquid growth medium. However, it is now generally acknowledged that the majority of microbial cells on earth are living in spatially distinct communities, referred to as biofilms. Biofilm cell populations in many cases exhibit distinct features compared to planktonic cell populations (1–5). Therefore, research work aimed at unraveling characteristics of the microbial multicellular lifestyle will provide a more complete understanding and definition of microbial life, and ultimately also human life, in nature.
Microbial Multicellular Communities
Biofilms come in a great variety of sizes and shapes. Some of the most common types contain mushroom-like, pillar-like, hilly, or flat multicellular structures. These structures are formed by cells that are held together by interconnecting compounds, such as self-produced polysaccharides, proteins, extracellular DNA, and cell lysis products, as well as matter from the immediate surrounding environment, which altogether constitute the so-called “matrix.” This allows cells to form long-term relationships, interact with each other and establish metabolic cooperations (6, 7). Biofilms in nature are generally beneficial and frequently established on hydrous solid and semi-solid surfaces, such as soil, rock material, or surfaces of animals and plants. Bacterial communities play key roles in food webs in nature. Many of the underlying processes are interdependent and require cooperation between various bacterial species with different metabolic capacities (3, 4, 6, 8, 9). The fact that in biofilms the participating microbial members are situated in close proximity seems to be advantageous, since metabolites can easily be transferred and metabolized further. In cases of adverse conditions such as desiccation, osmotic shock, or exposure to toxic compounds, UV radiation, or predators, the microbial community as whole can provide protection. Moreover, multicellular communities provide ideal conditions for horizontal gene transfer, which is important for microbial evolution and genetic diversity (4, 6, 10, 11). However, in many man-made environments, such as industrial or medical settings, the formation of sessile microbial communities, for example, in production lines or on indwelling medical devices is unwanted and detrimental. Resultant energy losses, corrosion, fouling, persistent infections, and potential death of humans cause an enormous socioeconomic burden each year worldwide (6, 12).
Microbial Communities as Inhabitants of Humans: Impact on Health and Disease
Microbial communities natively populate human mucous membranes and epithelial surfaces like the gastrointestinal tract, oral cavity, and skin. Each of the body sides is colonized with a mixed microbial community of characteristic composition (2, 13, 14). Intriguingly, for most of our lifespan we do not suffer from harboring these microbial communities. In fact, they are important and beneficial to us as they can degrade nutrients and thereby making them accessible to us, and synthesize some vitamins, which we have not evolved to synthesize on our own. Moreover, these communities play key roles in the development of our immune system and anatomy of the mucosal surfaces and exert protective functions against exogenous pathogens (2, 15, 16).
The relationship between the host and its microbial communities is delicately balanced but under certain conditions, it can break down and result in infectious diseases. These infections can be caused either by members of the indigenous human microbial community or by microorganisms from the environment (1, 2, 6). Under conditions where the host is impaired, for example immunocompromized, injured, or suffering from cancer or cystic fibrosis, harmful biofilms can develop at different body sides and cause persistent infections. Among those infections are various device-related infections, pulmonary infections, periodontitis, wound infections, otitis media, osteomyelitis, infective endocarditis, and chronic prostatitis (5, 17–19). Bacteria, which have been found to be involved in human biofilm-related infections, are for example Pseudomonas aeruginosa, Staphylococcus spp., Escherichia coli, Salmonella spp., Enterococcus spp., Streptococcus spp., Proteus mirabilis, Klebsiella spp., Enterobacter spp., and Haemophilus influenza. Biofilm-related infections are in many cases persistent, that is, they evidently cannot be eradicated by the host immune system and they are difficult to eradicate by antimicrobial chemotherapy (5, 17–20).
Cultivation and Analysis of Multicellular Communities Under Laboratory Conditions
Biofilms are intriguing societies of microbes and it is of general interest to unravel the processes involved in their development, physiology, and adaptation to perturbations. Empirical research work on biofilm biology will ultimately reveal new strategies that contribute to maintenance and restoration of human health through well-informed manipulations of microbial communities. However, due to their complexity, natural microbial communities have been challenging objects of investigation. In addition, biofilms are often located at places that are difficult to access, which makes direct and continuous examinations difficult. Various factors impact on biofilm biology and many of those factors are interrelated. To reduce complexity and facilitate investigations in the laboratory under controlled and reproducible conditions, a number of “simple” biofilm model systems have been established. These include flow-cell-grown biofilms, colony biofilms, microtiter dish-grown biofilms, and pellicle biofilms (21–23). To discover genetic determinants and regulatory pathways impacting on the biofilm mode of life, major focus relies on well-characterized and genetically tractable microorganisms, such as Pseudomonas aeruginosa. The gold standard in biofilm research is an approach, which involves flow-cell technology in combination with confocal laser scanning microscopy (CLSM). Unlike other techniques, this particular methodology allows getting insight into details of developmental processes, spatial organization and function of laboratory-grown biofilms in real-time under continuous and non-invasive conditions down to the single-cell level (21, 24, 25).
CULTIVATION OF BIOFILM CELLS IN A FLUIDIC DEVICE
Biofilm flow-cell setups allow the cultivation of biofilms under continuous hydrodynamic conditions. The biofilm flow-cell system consists of five major components: A medium reservoir, a multichannel peristaltic pump, bubble traps, flow-cells, and an effluent reservoir. All parts are consecutively connected via silicone tubings, splitters and connectors (Fig. 1A) (22).
The Biofilm Flow-Cell
A central component of the biofilm flow-cell system is the flow-cell, providing chambers for biofilm cultivation. A number of different flow-cell designs exist. Widely used is the flow-cell described here, which is a modified version of a flow-cell originally developed by Wolfaardt et al. (26). The flow-cell is designed so that it can be mounted on nearly any optical microscope. It consists of two parts, a flow-cell base (Fig. 1B) and a conventional microscopy glass coverslip. The flow-cell base is made of a polycarbonate part in which parallel channels with individual dimensions of 40 × 4 × 1 mm have been drilled. In the flow-cell presented here three individual channels have been drilled, meaning that in one flow cell three individual (parallel) biofilm experiments can be performed. To both ends of each channel, ports of 1 mm diameter have been drilled, which serve as medium inlet and effluent outlet, respectively (22). A microscopy glass coverslip (50 × 24 mm) is placed on top of the flow-cell base and thereby covering the open side of the channels to form closed channels. The coverslip is glued with silicone on the flow-cell base by applying silicone glue as thin strings on top of the base along the perimeter of the channels. The microscopy coverslip basically fulfills three functions: it serves as one wall of the channels thereby forming a closed flow channel, it serves as substratum for biofilm formation, and it is optically compatible with microscopic examination techniques (21, 22, 24, 27).
The Biofilm Flow-Cell System
To eventually continuously supply the biofilm cells with nutrients, the flow-cell needs to be connected with the remaining components of the flow-cell system (Fig. 1A). Before the medium reservoir is connected, possible contaminants are removed from the entire system. This is achieved by disinfection of the system with sodium hypochloride or ethanol, or by sterilization with ethylene gas. Most components are also compatible with autoclaving at 121°C. Subsequently the reservoir with sterile nutrient medium is connected to the system. The composition of the medium is chosen dependent on the requirements of the organism(s) of interest. The system is filled with medium using a peristaltic pump. Conventional peristaltic pumps do not deliver an entirely pulseless flow, in contrast to multiroller peristaltic pumps. If the carbon source in the medium has a high hydrophobicity then medium is run through the system for an extended period to equilibriate the silicone tubings, before introducing the organisms of interest (21, 22, 24, 27).
The organisms of interest are introduced at desired initial optical density into the flow chambers via a syringe while medium flow is paused. The syringe is inserted upstream of the flow-cell through the silicon tubing into the inlet of the flow chamber, the cell suspension of microorganisms is injected and the resulting fine hole in the silicone tubing is sealed with a drop of silicone glue. Microbial cells are allowed to attach to the substratum for 1 h, whereupon medium flow is resumed and adjusted to laminar flow. For the biofilm system described here, the flow rate is often set to 0.2 mm/s (approximately 3 ml/h/channel) (21, 22, 24, 27). Universal equations for the calculation of hydrodynamic parameters in a flow-cell can be found elsewhere in the literature (e.g.,28). While fresh medium is continuously transported into the flow cells to allow biofilm development and differentiation, effluent is transported out and collected in the reservoir placed at the most downstream part of the system.
Under certain circumstances air bubbles might arise in the system, which can remove biofilm cells from the substratum while passing through a flow channel. To prevent this, bubble traps have been designed and are mounted between the peristaltic pump and the flow-cells. Bubble traps are composed of three parts: a bubble trap base made of a polycarbonate part (Fig. 1C), a syringe cylinder, and a lid to close the top of the syringe cylinder. Any air bubbles that might pass into the bubble trap will float to the top of the syringe cylinder, and are prevented from passing through the downstream part of the system. Note that the inlet of the bubble trap base is designed so that it is situated higher than the outlet part (Fig. 1C).
The biofilm setup is compatible with various visualization and quantification techniques (see following text). Cell–cell interactions in a biofilm can be studied between (i) cells of a single strain (e.g.,29), (ii) cells of a number of different strains (e.g., mutants) of a particular single species (e.g.,30, 31), (iii) cells belonging to different bacterial species (e.g.,32–35), and (iv) cells belonging to different domains, such as bacterial cells and eukaryotic cells (e.g.,36). Activities in natural multicellular microbial consortia might also be studied in flow-cells after harvesting samples from the natural environment, and transplanting the microbial cells into flow-chambers where they are subjected to further analysis (e.g.,26, 37). The setup might be modified, for example, to study the effect of substratum coatings on biofilm development and physiology, or to study the effect of oxygen up- or down-shifts on biofilm cells (38–40). Biofilm cells might also be harvested from the flow chambers and subjected to further analysis, for example global transcriptional analysis, fluorescence activated cell sorting (FACS), or c.f.u. determinations (31, 41, 42). The entire biofilm flow-cell system can be placed on a rolling table to facilitate secure transport of the system for example between an incubator room with a constant temperature (often 30°C or 37°C) and the room in which microscopic examinations of the biofilms are taking place.
ANALYSIS OF BIOFILM CELLS
Because biofilms are complex 3-dimensional structures the analysis of them is not trivial. While microbial single cells easily can be monitored using a conventional microscope, biofilms require additional resolution in the direction vertical to the substratum (the z-axis). Here we describe the use of confocal laser scanning microscopy (CLSM), fluorescent labeling of biofilm cells, and image analysis in biofilm research.
Confocal Laser Scanning Microscopy (CLSM)
The use of the confocal laser scanning microscope has helped overcoming the apparent shortcomings of the conventional light microscope (e.g., the presence of out-of-focus light) by introducing point illumination and a pinhole, which allows optical sectioning of the specimen. The individual optical sections are subsequently assembled by aid of advanced computer software (some examples are listed in the following text) to a virtual 3D image (e.g., Figs. 2A and 2B and Fig. 3). Typically a biofilm with a thickness of more than ∼150 μm cannot be rendered with a reasonable detail due to physical factors. The implementation of multiphoton excitation is a major step forward. Using a pulsed laser it is possible to guide two (or more) photons to excite a fluorophore simultaneously. This means that the energy of the photons is combined to excite the target molecule. Using this technique the depth resolution (i.e., the minimum distance to resolve two points) is many fold increased, from around 500–700 nm for standard confocal microscopy to 100 nm for two-photon microscopy (43). Neu and co-workers demonstrated two-photon confocal imaging of thick z-sections with very high resolution (43, 44). Recently a new technology, stimulated emission depletion (STED) has been developed into commercial products, increasing the optical resolution even further by using two-photon excitation in combination with quenching of near-by fluorescence which otherwise could deteriorate the image (45, 46). STED, however, is currently of limited use since only few fluorophores are suitable for this particular laser excitation.
Fluorescent Labeling of Biofilm Cells
Confocal microscopy and derived methods require the specimen to be fluorescent. The biofilm must therefore either be autofluorescent by means of indigenous fluorescent molecules, or the biofilm cells must express a fluorescent protein (e.g., the green fluorescent protein, Gfp (47)), or individual biofilm cells or other components of the multicellular structure must be stained. Early biofilm studies by the Caldwell group employed a simple, yet efficient way of detecting the biomass in flow cells: The void volume, that is, the liquid phase was supplemented with a solution of fluorescein isothiocyanate (FITC), leaving the biomass unstained. The resulting images were “negatives” and the biofilm could be rendered as the dark portions of the images (48, 49). This gave sufficiently high resolution to determine for example cell sizes and spatial relations. More recently developed stains, such as the Syto stains (Invitrogen, Carlsbad, CA), can efficiently stain cells in virtually any color of the rainbow. In combination with propidium iodide (PI) it is further possible to specifically stain live and dead cells. The dye Syto 9 will stain all cells green regardless if they are dead or alive, while it is generally assumed that only cells with a damaged membrane will be stained by the red PI dye, indicating dead cells. Recent results suggest that propidium iodide might be of limited use as cell viability indicator for some environmental bacterial species (50). Therefore, for each individual bacterial species a fine-tuning of the dye combinations is recommended prior to performing the actual experiments using mixtures of live and dead cells of known ratios (51). Recently the assumption that propidium iodide only targets dead cells was confirmed also for biofilm cells that had been exposed to a membrane damaging agent, using cell sorting of harvested biofilm cells and re-growth test of the separated green (Gfp-tagged) and red (PI-stained) cells on nutrient agar. Only the cells that were not labeled red with PI were able to grow (42, and unpublished results). Stains targeting the extracellular matrix such as lectins (52, 53) or calcofluor white (54, 55) can also be employed to visualize the surrounding of the biofilm cells. In addition, the extracellular DNA component of the matrix can be visualized by the use of different DNA-binding fluorophores (e.g., Ref.56).
If genetic manipulation of the biofilm cells is possible, chromosomal tagging of cells with a gene cassette encoding the green fluorescent protein (Gfp) can be a useful option (e.g.,57). Alternatively, plasmids encoding for the Gfp might be introduced into the cells prior to biofilm examinations. Depending on the construction, this fluorescent tagging can be used as a simple labeling to verify the location of the cells in a biofilm, or, by selecting suitable variants of gfp genes and promoters, it can be used for monitoring gene expression in biofilms. Such tagging of biofilm cells has been done to monitor metabolic/physiological activity in biofilms by introducing constructs encoding for Gfp derivatives with a short half-life, placed under transcriptional control of a ribosomal promoter (58). For example, the gfp[AGA] gene, encoding for a green fluorescent protein (Gfp) with a short half life, was placed under transcriptional control of the ribosomal promoter rrnBP1 and introduced into either E. coli or Pseudomonas spp. wild type strains. Cells which have a high metabolic/physiological activity can be expected to exhibit high Gfp[AGA] expression and emit a high fluorescent signal, whereas cells which have a low metabolic/physiological activity can be expected to exhibit a low or no expression of the fluorescent protein (41, 58). Further, using Gfp variants with different emission spectra, such as the Cfp (cyan fluorescent protein), Yfp (yellow fluorescent protein), and Rfp (red fluorescent protein), the spatial distribution of either cells in a multi species biofilm can be determined, or of a number of (mutant) strains of a single species tagged with different colors (30, 31, 59, 60).
Another way of fluorescently labeling biofilm cells is through the use of fluorescent in situ hybridization (FISH), where specific probes hybridize to the 16S rRNA in the cells. For FISH a DNA probe is designed to match a distinct region of the cell's ribosomal RNA. The probes can be conjugated to a fluorescent dye, such as fluorescein isothiocyanate (FITC) or Rhodamine, or to an enzyme (e.g., horseradish peroxidase), which deposits fluorescent molecules. It might be challenging to introduce larger conjugates such as the horseradish peroxidase into cells of thicker biofilms without destroying biofilm cells due to harsh permeabilization procedures. Therefore FISH involving probes with larger conjugates might preferentially be applied on thin sections of thick biofilms. The number of ribosomes present in a given cell is proportional to the growth potential of the cell, and FISH labeling can consequently also be used to determine the growth status of a biofilm cell (61). However, under certain conditions, for example, stress, cells might have increased numbers of ribosomes, although their actual growth rate is low. The probe design can be adjusted so that the probe only labels a single species by targeting a so called variable rRNA region, or a probe can label all cells belonging to the same domain or phylum by choosing a more conserved region. An example of such a probe is the widely used EUB338, which can hybridize to virtually all bacteria (62). However, the growing knowledge on ribosomal RNA encoding sequences has revealed that probes formerly believed to be universal fail to be able to hybridize to species or entire phyla of microorganisms in the realm they originally were thought to cover completely (63).
The recorded microscopic images can be used immediately or processed further for presentation or quantitative analysis. The images that originate from a confocal microscope are usually grayscale bitmap images, one from each focal plane and one for each detection channel (color). An image of a 30-μm-thick biofilm recorded with a step size of 0.5 μm in three channels will result in 30/0.5 × 3 = 180 individual images. Most microscope software pack the images in containers such as the LIF file format for Leica and the LSM format for Zeiss microscopes. Furthermore the images will be relatively large. In the example above the standard resolution of 256 gray tones and 512 × 512 pixels will result in a file with the size of 45 MB. Special software is required to handle these files and to render the beautiful biofilm representations. While several packages are available, a few seem to dominate the market: Imaris (Bitplane, Bern, Switzerland), Amira (Visage Imaging, Carlsbad, CA), Volocity (Improvision, Coventry, the UK), Voxblast (Vaytek, Fairfield, IA), and Metamorph (Molecular Devices, Sunnyvale, CA) (see43 for a review). Designed for the visualization of larger eukaryotic cells, most 3D presentation packages are not optimized for the small cell sizes involved in microbial biofilms. The average size for a Gram-negative bacterium is 1 μm by 2–3 μm. The optical resolution of a confocal image recorded with one-photon excitation is at best 0.48 μm (64). This means that a sampling of images with a smaller step size than this will not provide more information. Consequently, a single bacterium will not appear in more than one or two independent focal plane images, making a 3D reconstruction of the single cell difficult. However, the biofilm as a whole is much larger and can be rendered in 3D, although it may not be possible to locate the individual cells. Typically a step size of 1 μm is used for bacterial biofilms (e.g., Ref.30). The visualization software packages also include tools for cleaning up the recordings, such as filters for noise and crosstalk (the situation where one fluorophore is recorded in the detection channel of another fluorophore). The main features of these softwares are, however, their capability to visualize the spatial organization of the recorded data. The rendering can be in perspective 3D, or as 2D images in all three axes, x-y, x-z and y-z. Sequential recordings over time can also be rendered, and animated, providing a four-dimensional dataset, x-y-z-t. Such data sets can quickly be very large (sometimes several gigabytes) and it sets new requirements for the software and hardware. Special detector systems mounted on the confocal microscope might also facilitate the simultaneous separation of a number of fluorescent spectra (lambda mode configurations) originating from different fluorescent molecules or proteins with overlapping emission spectra, increasing the possibilities for multi fluorescent labeling of biofilms (Fig. 2) (31).
Quantitative analysis of 3D images can be challenging and several groups have developed special software packages for this purpose. The algorithms start by determining the extent of the biomass, by thresholding each focal plane image. This step is crucial and much effort has been put into optimization of it. It can be done either manually, semi- or fully automated. Some of the first attempts to provide robust quantification software were done by Yang et al. (65) and Heydorn et al. (66) with the programs ISA and COMSTAT, respectively. Both extract a number of parameters, which can be used to characterize the biofilm: biomass (pixels occupied by biomass), biofilm height, height distribution, roughness coefficient, and diffusion distances, to mention a few. Both programs are developed in MATLAB (MathWorks, Natik, MA). COMSTAT utilizes a command-line interface within the MATLAB shell, which is required for operation, whereas ISA and its successor ISA3D are compiled programs that do not require the MATLAB package. Other quantification software is available, for example, the web-based PHILIP (67), a program that has a higher level of automation than the ISA and COMSTAT packages. Further developments of PHILIP have taken the automation and robustness towards the threshold function to a new level (68, 69). COMSTAT is now available in a complete reprogrammed version 2, which is running on the platform independent software foundation Java (unpublished results). This new version of COMSTAT software uses the same thresholding algorithms as the PHILIP software and also incorporates a number of new features such as a user programmable plug-in interface for end user defined image processing functions, and a wider range of image formats.
DEVELOPMENT OF MULTICELLULAR STRUCTURES IN PSEUDOMONAS AERUGINOSA BIOFILMS
Studies involving flow-chamber technology and CLSM have provided knowledge about biofilm formation of numerous bacterial species of both environmental and medical relevance. The studies have, among many other things, given information about the mechanisms, environmental cues and underlying genetic elements involved in attachment of cells to surfaces, formation of multicellular structures, and dispersal of cells from multicellular structures. In the present section we present examples of work done on these topics with biofilms formed by the opportunistic pathogen P. aeruginosa.
Transport of Cells to the Surface
The bacterial cells can reach a surface prior to colonization by means of passive motility mediated by brownian motion or vortex currents, or by active swimming motility mediated by flagella rotation (70). Flagella-driven motility has been reported to enhance the efficiency of surface colonization by P. aeruginosa (71, 72). The dependence on flagella is conditional however, as non-flagellated P. aeruginosa mutants and the isogenic wild type were shown to attach equally well to a surface in flow-chambers under some conditions (29).
Attachment of Cells to the Surface
P. aeruginosa appears to possess a number of different adhesins that can function in attachment to a surface. The conditional dependence on flagella for surface colonization described above may, besides a role in bacterial transport to the surface, be because of adhesion properties of the flagella. P. aeruginosa cells were shown to attach apically to the glass surface in flow-chambers and rotate, indicating that the initial attachment occurred by means of flagella (72). The initial attachment was reversible, but cells became irreversibly attached by progressing from apical to longitudinal attachment. The gene sadB (PA5346) was shown to be required for the progression from apical to longitudinal attachment (73). Non-flagellated mutants of P. aeruginosa attached longitudinally (72), indicating that flagella are not required for attachment per se but may enhance the process. Type IV pili were shown to be important for attachment of P. aeruginosa to various surfaces (74–76), and hyperpiliated variants of P. aeruginosa were shown to rapidly initiate formation of strongly adherent biofilms (75). The dependence on type IV pili for cell attachment appears to be conditional, however, as non-piliated P. aeruginosa mutants and the isogenic wild type were shown to attach equally well to surfaces in flow-chambers under some conditions (29). In addition to flagella and type IV pili, fimbrial appendages termed Cup fimbria were shown to play a role in surface-attachment of P. aeruginosa (77), and evidence has been provided that extracellular DNA plays a role in surface attachment of P. aeruginosa cells under some conditions (78).
Formation of Initial Multicellular Structures
Shortly after attachment to the surface in flow-chambers the P. aeruginosa cell population consist of a non-motile subpopulation, and a motile subpopulation that moves on the surface via type IV pili or flagella activity (29, 30, 79, 80). Experiments with mixtures of Cfp-tagged and Yfp-tagged cells provided evidence that microcolonies develop by clonal growth in the flow-chambers (30). The initial microcolonies are formed by cells, which do not display motility and therefore proliferate at fixed positions, while the cells that move on the surface do not participate in formation of the initial microcolonies (29, 30). Conditions that promote extensive motility of P. aeruginosa have been shown to prevent microcolony formation in flow-chambers. Interestingly, this could be a mechanism which prevents biofilm formation in humans since lactoferrin, a component of the innate immune system, was shown to induce extensive type IV-driven motility in P. aeruginosa (79).
The cellular adhesiveness mediated by the factors involved in surface attachment probably also plays a role as cell-to-cell adhesin in the earliest phase of microcolony formation. For example, the small microcolonies formed by P. aeruginosa in the initial phase of biofilm formation could be dispersed by exogenous DNase activity indicating a role for extracellular DNA as matrix component in the early phase of P. aeruginosa biofilm formation (78).
Formation of Mature Multicellular Structures
Although the relative importance of the mechanisms involved in initial formation of multicellular structures appears to vary dependent on the conditions, the outcome of initial biofilm formation seems to be similar under various conditions. The early P. aeruginosa biofilm generally consists of a subpopulation of non-motile cells forming small microcolonies, and a subpopulation of motile cells, which move on the surface between the small microcolonies. After formation of the initial microcolonies structural biofilm development by P. aeruginosa depends on the prevailing conditions. For example, flat biofilms are formed in flow-chambers irrigated with citrate minimal medium (29), while heterogeneous biofilms containing mushroom-shaped multicellular structures are formed in flow-chambers irrigated with glucose minimal medium (30).
Formation of the flat P. aeruginosa biofilm in flow-chambers irrigated with citrate medium was shown to occur via expansive surface-migration of cells from the initial microcolonies (29). CLSM time-lapse microscopy indicated that a shift from non-motile to migrating cells occurred when the initial microcolonies reached a certain size, suggesting that the shift may by induced by a limitation arising in the initial microcolonies (For representative movies the reader is referred to Ref.81). Migration of the bacteria appeared to cease with maturation of the citrate-grown biofilm. Since biofilm formation by a P. aeruginosa pilA mutant (deficient in biogenesis of type IV pili) occurred without the expansive phase, and resulted in discrete protruding microcolonies, it was suggested that the expansive migration of the cells on the surface was type IV pili-driven (29).
Formation of the mushroom-shaped structures in glucose-grown P. aeruginosa biofilms evidently occurs in a sequential process where the initial microcolonies formed by the non-motile subpopulation become colonized by cells from the migrating subpopulation that subsequently form mushroom cap-like structures on top of the initial microcolonies, which then become mushroom stalks (for representative movies the reader is referred to Ref.82) (30, 42). Growth of the initial microcolonies in the glucose-grown biofilms continue past the point where spreading by type IV-driven motility prevents further microcolony formation in the citrate-grown biofilms. In glucose-grown biofilms containing a mixture of Cfp-tagged and Yfp-tagged P. aeruginosa wild type cells, mushroom-shaped structures are formed that have single-color stalks and two-color caps, in concordance with the stalks being formed by proliferation of non-motile cells, and the caps being formed via aggregation of motile cells (30). In biofilms containing a mixture of P. aeruginosa wild type and P. aeruginosa pilA mutant, the pilA mutants can only form stalks whereas the wild-type cells form the caps, suggesting that type IV pili are necessary for cap-formation (30). Figure 3A shows a microcolony formed by Cfp-tagged pilA mutants surrounded by Yfp-tagged motile wild type cells, which are beginning to colonize the microcolony. Figure 3B shows mature mushroom-shaped structures with Cfp-tagged pilA mutants in the stalk and Yfp-tagged wild type cells in the cap. In addition to type IV pili, formation of the mushroom cap evidently also depends on flagella, as P. aeruginosa fliM mutants (deficient in biogenesis of flagella) were shown to be deficient in cap-formation (83). Evidence was provided that flagellum-driven surface-associated motility is involved in cap-formation, whereas the dependence of cap-formation on type IV pili may be due to binding of these pili to extracellular DNA which is particularly abundant on the microcolonies that become colonized during mushroom-structure formation (as described in the following text). In accordance with a role of type IV pili-driven motility in the early phase of biofilm formation and a role of flagella-driven motility in the later phase of biofilm formation, flagella-driven surface associated motility (swarming) is known to depend on quorum sensing (e.g.,84), that is, a mechanism by which bacteria can monitor their cell population density through the extracellular accumulation of signaling molecules and express genes when the cell density is high. Production of biosurfactants, which is under quorum sensing control, was shown to facilitate formation of the mushroom cap structures (31). Migration of the bacteria appears to cease with maturation of the glucose-grown biofilm.
Experiments involving the fluorescent reporter rrnBP1-gfp[AGA] (see preceding text) provided evidence that metabolic/physiological activity is highest in cells forming the cap-subpopulation, and lowest in cells forming the stalk-subpopulation (41). As can be seen in Figure 4A, the cells of the cap-forming subpopulation exhibit a high fluorescent signal, indicating high metabolic activity, whereas cells of the stalk-forming subpopulation exhibit a low fluorescent signal, indicating low metabolic activity (41). In the control biofilm formed by a P. aeruginosa wild type strain expressing the stable version of Gfp, all cells exhibit similar levels of fluorescence emission (Fig. 4B). This spatial distribution of high and low metabolically active cells seems plausible as cells in the cap part can obtain oxygen and nutrients from the bulk liquid to drive metabolic processes such as replication, transcription and translation, in contrast to the cells in the stalk part where concentrations of dissolved oxygen and nutrients are likely to be low. It might be speculated that higher levels of oxygen and nutrients in the bulk liquid are sensed by some cells during biofilm development, and that this results in an attraction, driving cells to migrate on top of microcolonies formed by non-motile cells. In support of this, P. aeruginosa strains defective in chemosensory systems show defects in cap-formation [(83) and S.J.P and T.T.N., unpublished].
The multicellular structures in P. aeruginosa biofilms are stabilized by a matrix consisting of exopolysaccharides, extracellular DNA, and proteins. The alg operon (PA3540-PA3551) encoding alginate polysaccharide appears not to be expressed in P. aeruginosa flow-chamber biofilms (85), but evidently plays a role in biofilm formation by P. aeruginosa in particular in the lungs of cystic fibrosis patients (86). The psl genes (PA2231-PA2245) encode production of a mannose rich exopolysaccharide, which was shown to play a role in P. aeruginosa biofilm formation (87–89). The pel (PA3058–PA3064) genes encode production of a glucose rich matrix component, which was shown to facilitate biofilm formation of some P. aeruginosa strains (90). Extracellular DNA was shown to be present in high concentrations particularly on the microcolonies in young P. aeruginosa biofilms and between the stalk-forming and cap-forming subpopulations in mature biofilms (56). Type IV pili bind to DNA (91, 92), and evidence has been presented that the high concentration of extracellular DNA on the mushroom stalks may cause accumulation of the migrating piliated cells and thereby facilitate formation of the mushroom caps (83). Production of large amounts of extracellular DNA in P. aeruginosa biofilms has been shown to be dependent on the PQS quorum-sensing system (56). Evidence is accruing that Cup fimbria in addition to their role in initial biofilm formation also play a role as matrix components in mature biofilms. The sadARS genes (PA3946-3948, also termed rocARS) were shown to regulate biosynthesis of Cup fimbria (93), and mutations in any of these genes resulted in biofilms with an altered mature structure (94).
Dissemination of Cells from Multicellular Structures
In addition to the mechanisms involved in biofilm formation, bacteria also possess mechanisms to reduce their adhesiveness and to break down or modulate the biofilm matrix. Emigration of cells from biofilm communities is necessary to spawn novel communities at new locations, and it may be induced if the biofilm cells face unfavorable conditions (e.g.,95, 96). Migration of cells may also allow sessile communities to change spatial organization in response to changing environments (e.g.,34). After prolonged biofilm development of P. aeruginosa in flow-chambers local dispersion was observed as a hollowing out of the mature microcolonies (97). Through careful microscopic inspection it was observed that two subpopulations existed in the mature multicellular structures. The outer parts of these structures contained a wall-forming subpopulation of non-motile cells, whereas a motile rapidly moving cell subpopulation was present inside the multicellular structures. The motile subpopulation coordinately evacuated the multicellular structures from local break out points resulting in structures with a central void. This phenomenon has been termed “seeding dispersal,” and was shown to be dependent on the mature multicellular structures reaching a critical size, suggesting that it may be induced by substrate limitation or accumulation of signal molecules or waste products. Biosurfactants appears to have multiple roles in P. aeruginosa biofilm development, as production of large amounts of rhamnolipid biosurfactant has been associated with dispersal of cells from P. aeruginosa biofilms (98).
ANTIMICROBIAL ACTION ON BIOFILM CELLS
Biofilms in the environment as well as those associated with animals and plants or present in man-made environments are frequently exposed to antimicrobial compounds, both of natural and synthetic origin. Flow-chamber-grown biofilms appear to be a useful model system to study antimicrobial action in biofilms, for example, as the spatial appearance and distribution of dead and surviving cells in a biofilm upon antimicrobial attack can be followed in real time. Here we will give an overview of the spatial antimicrobial susceptibility and tolerance phenotypes of P. aeruginosa biofilms living in flow-chambers exposed to a number of different antimicrobial compounds.
As described in the previous section, P. aeruginosa mushroom-shaped biofilms are commonly found to be composed of two major subpopulations, a subpopulation situated close to the substratum and a subpopulation on top. Intriguingly, it appears that antimicrobial compounds seem to exert their antimicrobial effects on only one of the two subpopulations, whereas the other subpopulation survives the treatment. In most cases the surviving subpopulation of cells exhibits phenotypic tolerance and not resistance, as surviving biofilm cells harvested from antimicrobial-treated biofilms exhibit the same antimicrobial susceptibility phenotype as the cells, which were used to initiate the biofilm (e.g., Refs.41, 42).
Effect of Conventional Antibiotics on Biofilm Cells
Most conventional antimicrobial agents used to treat bacterial infections in humans and animals, interfere with fundamental physiological processes of bacterial cells, such as replication, transcription or translation processes. Here we describe the effect of three such conventional antibiotics on mushroom-like shaped P. aeruginosa biofilms, namely ciprofloxacin, tetracycline, and tobramycin.
The fluoroquinolone ciprofloxacin induces bacterial cell death by interfering with the replication process due to inhibition of the DNA gyrase. Ciprofloxacin is administered for treatment of various infections caused by Gram-negative and Gram-positive bacteria. Exposure of P. aeruginosa flow-cell-grown biofilms was found to preferentially induce cell death in the cap-forming cell subpopulation (Fig. 5B) (41). Using in situ gene expression analysis, involving a growth activity-dependent fluorescent reporter (see details about the fluorescent reporter above), it was found that ciprofloxacin specifically targets the P. aeruginosa biofilm cells exhibiting high metabolic activity in the upper part of the multicellular structures. By contrast, the biofilm cells in the deeper layers exhibiting low metabolic activity survive ciprofloxacin treatment (Fig. 5B) (41). A similar phenotype with respect to the distribution of growth activity and ciprofloxacin-induced cell death was observed for a P. aeruginosa colony biofilm (99).
Tetracycline is an antimicrobial agent, which originates from secondary metabolites produced by Streptomyces spp. It can inhibit bacterial protein synthesis by preventing attachment of aminoacyl-tRNA to the ribosomal acceptor site (A-site) and thereby induce cell death. When mature P. aeruginosa wild type biofilms were exposed to tetracycline, the cells of the cap-forming subpopulation were killed, whereas the cells situated in the deeper areas survived the treatment (41). An experiment involving treatment of a biofilm formed by a strain, which harbors the growth activity-dependent fluorescent reporter fusion, indicated that tetracycline preferentially kills the cells in the upper area, which exhibit high metabolic activity. By contrast, cells in the deeper areas of the biofilm, which exhibit a lower metabolic activity, were not killed by tetracycline (S.J.P. and T.T.N., unpublished observation).
The aminoglycoside tobramycin is a secondary metabolite derived from Streptomyces spp. and it can inhibit protein synthesis in Gram-negative bacteria by preventing translocation of peptidyl-tRNA from the A-site to the P-site of the ribosome, and thereby induce cell death. When mature P. aeruginosa biofilms were exposed to tobramycin, the cells situated in the upper area of the multicellular structures were killed, whereas the cells situated in the deeper areas survived the treatment (36, 100). The efficiency of tobramycin-induced killing in P. aeruginosa biofilms was increased by co-administration of furanone C-30, a compound, which was identified to inhibit quorum sensing regulated gene expression (100). A biofilm formed by a mutant strain, which was defective in las- and rhl-mediated cell-to-cell-communication showed increased sensitivity to tobramycin, indicating a possible role of quorum sensing in tolerance towards tobramycin (36). Mah et al. presented results, which indicate that in biofilm cells of strain P. aeruginosa PA14 periplasmatic glucans might sequester tobramycin and hence prevent the interaction of tobramycin with its target (101). An ndvB-mutant, which is deficient in the synthesis of periplasmatic glucans, exhibited increased sensitivity to tobramycin in biofilms (101).
Altogether this indicates, that conventional antimicrobial compounds, which interfere with fundamental physiological processes of bacterial cells preferentially induce cell death in P. aeruginosa biofilm cells that have a high metabolic/physiological activity in the top layer of the multicellular structures. In contrast, cells that have a low metabolic/physiological activity are able to survive exposure to the antimicrobial compound.
Effect of Membrane-Targeting Compounds on Biofilm Cells
The rise in appearance of multiresistant bacteria and the persistence of biofilms in medical settings has increased the interest in alternative antimicrobial compounds. Among the new potential antimicrobial therapeutics are antimicrobial peptides, as resistance to antimicrobial peptides has rarely been observed so far (102–104). In addition, antimicrobial peptides are promising new drugs due to their ability to modulate immune responses (103, 105). Antimicrobial peptides are found to exert their primary antibacterial activities by interfering with the bacterial membrane, resulting in leakage and eventual death of the bacterial cell (e.g.,106, 107). Here we describe the effect of the antimicrobial peptide colistin (polymyxin E) on P. aeruginosa biofilms. In addition we describe the effects of two other membrane-targeting compounds on P. aeruginosa biofilms, namely the detergent sodium dodecyl sulfate (SDS) and the chelator ethylenediaminetetraacetic acid (EDTA).
Colistin belongs to the polymyxin group of antimicrobial peptides, and it is synthesized by strains of Paenibacillus spp., such as P. polymyxa (107–111). Colistin is administered as treatment against infections caused by Gram-negative bacteria, for example, pulmonary infections caused by P. aeruginosa in CF patients, or sepsis, wound infections, and urinary tract infections caused by a variety of Gram-negative bacteria (112–115). When mature mushroom-like shaped P. aeruginosa biofilms were exposed to a clinically relevant concentration of colistin, only the stalk-forming cell subpopulation was killed, whereas the cap-forming subpopulation survived the treatment (Fig. 5C) (41, 42). Detailed investigations on the spatiotemporal-dependent effects of colistin exposure on P. aeruginosa biofilms using fluorescent reporters and knock-out mutants provided evidence, that the metabolic/physiological active cells of the cap-subpopulation were able to adapt to colistin by inducing two antimicrobial tolerance mechanisms: the polymyxin resistance (pmr) LPS-modification system, and the antimicrobial efflux pump MexAB-OprM (41). In contrast, cells with low metabolic/physiological activity in the deeper layer of the multicellular structures were not able to adapt to colistin exposure, and were therefore killed by the action of the antimicrobial peptide (41). Although tolerance development in initial multicellular structures (2-day-grown biofilms) in addition appeared to be dependent on type IV pili driven motility (42), cellular migration evidently does not play a role in colistin-tolerance development in mature multicellular structures (4-day-grown biofilms) (41). Interestingly, the observed spatial distribution of live and dead cells upon colistin exposure appears to be independent of the actual three-dimensional structure of the biofilm and the carbon source used for biofilm-growth. Mushroom-shaped as well as irregular-shaped glucose-grown mutant biofilms, and flat-structured citrate-grown biofilms exhibited a subpopulation of dead cells close to the substratum and a subpopulation of live cells in the upper cell layer of the multicellular structures upon colistin treatment (41, 42, and J. Haagensen, personal communication).
SDS is a synthetic anionic surfactant, and this compound and its derivatives can be found in many household products (e.g., soaps). Due to its amphiphilic characteristics SDS interferes with biological membranes and is also known for its ability to denature proteins. When initial and mature multicellular structures formed by P. aeruginosa were exposed to SDS, cell death was induced in the cell subpopulation close to the substratum, whereas the cells in the top layer of the biofilm survived the treatment (42). Another study reported disruption of P. aeruginosa biofilms via hollowing after exposure to SDS (98). However, as the reported experiment was performed in the absence of a fluorescent indicator for dead cells (e.g., propidium iodide) it is unclear whether cells from the interior area of the biofilm detached or were killed. The genetic determinants and mechanisms, which facilitate tolerance development of the surviving fraction of cells are unknown at present, and are subject to ongoing research in our laboratory.
EDTA has the ability to form complexes with metal ions, such as Mg2+, Ca2+ and Fe3+. Exposure of proteobacterial cells to EDTA has been reported to result in removal of divalent cations (Mg2+, Ca2+) from LPS of the outer membrane and consequently in disruption of the outer membrane (116, 117). When mature multicellular structures formed by P. aeruginosa were exposed to EDTA, cell death was induced in the cell subpopulation close to the substratum, whereas the cells in the top layer of the biofilm survived the treatment (118). In addition, EDTA exposure was also reported to induce dispersal of some cells from the multicellular structures (118). Induction of cell death and dispersal by EDTA was found to be inhibited by the addition of Mg2+, Ca2+, or Fe3+ ions, supporting the notion, that EDTA exerts its effects on biofilm cells via complex-formation with metal ions, present in the outer membrane of the cells and possibly also part of the stabilizing extracellular matrix (118). The reason for why the upper subpopulation of cells survived the EDTA treatment is unclear at present.
Altogether this indicates, that compounds, which interfere with membrane function of bacterial cells preferentially induce cell death in P. aeruginosa biofilm cells in the deeper layer of the multicellular structures having a low metabolic/physiological activity. In contrast, cells in the top layer of the multicellular structures are able to survive exposure to the antimicrobial compound.
Effect of Combined Antimicrobial Treatment on Biofilm Cells
As described above, exposure of P. aeruginosa biofilms to a single antimicrobial agent was able to kill only a subpopulation of the biofilm cells, whereas the remaining biofilm cells were able to withstand the action by the particular antimicrobial agent. It was observed that conventional antimicrobial agents preferentially targeted the subpopulation of cells forming the upper layer of the biofilm, whereas membrane targeting compounds preferentially targeted the subpopulation of cells situated in the deeper layer of the multicellular structure (41). It was therefore of interest to investigate, if a combined antimicrobial treatment strategy, involving a compound targeting the biofilm cells in the upper layer and a compound targeting the biofilm cells in the deeper layer, could be potentially successfully used to kill all biofilm cells. Indeed, combined antimicrobial treatments involving either ciprofloxacin + colistin, or tetracycline + colistin, were found to kill nearly all biofilm cells. By assessing the number of surviving cells it was found that in particular the exposure of the P. aeruginosa biofilm cells to ciprofloxacin + colistin was very efficient leaving only a few single surviving cells (41). Interestingly, it was observed that combined antimicrobial treatment using ciprofloxacin + colistin in cystic fibrosis (CF) patients is effective in reducing the onset of chronic P. aeruginosa infection (119), and hence this particular treatment strategy was recently implemented in the recommended early intervention and prevention therapy in CF (120, 121).
CONCLUSION AND FUTURE PROSPECTS
The biofilm mode of living is assumed to be the predominant bacterial life style in nature. These complex microbial communities carry out processes in the natural environment as well as man-made environments, and thereby contribute to human development, health and disease. Investigations of biofilms established under controlled conditions in laboratory setups have provided fascinating insights into the fundamental capacities of bacteria to form multicellular structures. Studies of model systems, such as flow-chamber-grown biofilms, have revealed a set of inherent elements of the participating microbes that can facilitate their organization into multicellular communities. Among these factors are for example the production of matrix compounds, cell-surface bound proteins, the secretion of biosurfactants, cellular migration, and regulatory elements such as signal transduction systems, and intra- and extra-cellular signal messenger molecules. CLSM monitoring of fluorescently color-coded bacteria, grown in mixed-strain biofilms under continuous hydrodynamic conditions, has provided intriguing insights into the spatiotemporal developmental processes, in some cases down to the single cell level (e.g.,30, 31). Detailed microscopic examinations have revealed that, even in monospecies biofilms, a number of physiologically distinct cell subpopulations exist and differentiate during biofilm development (e.g.,41). Intriguingly, recent investigations have revealed that these distinct cell subpopulations exhibit differential sensitivity towards antimicrobial agents (e.g.,41, 42). One cell subpopulation, which exhibited high metabolic/physiological activity, was sensitive towards conventional antimicrobial agents, whereas the second subpopulation was refractory (41). However, the second cell subpopulation exhibited increased sensitivity towards a number of membrane-targeting compounds, and hence a combined antimicrobial treatment using a conventional antibiotic and a membrane-targeting compound was able to kill both cell subpopulations of the biofilm (41). These, and other studies highlight the importance of studying the characteristics of biofilm cells at the subpopulation, and single cell level.
Future studies, facilitated by sophisticated approaches and new technologies, will increase our understanding of microbial life in multicellular communities. Investigations of spatiotemporal gene expression in biofilms by the use of fluorescent reporter genes and CLSM will continue, and expand through the use of combinations of gene expression markers. Cell subpopulations will be isolated from complex communities, for example, by the use of microdissection, microfluidic devices or cell sorting, and subjected to further analysis, such as RT-PCR measurements of specific gene expression or DNA array analysis of global gene expression (e.g., Ref.122). Moreover, new technologies are emerging that might enable analysis of global transcription profiles of single biofilm cells (123, 124). Studies of the metabolic functions of subpopulations of biofilm cells or single cells in a biofilm community, might be done by the use of NanoSIMS (nanometer-scale secondary-ion mass spectrometry) or related techniques, which have recently been successfully employed on microbial cells from the environment and microbial cells associated with animals (e.g.,125, 126). Insight into the features of distinct cell subpopulations and single cells in microbial communities will provide a more complete understanding of the microbial multicellular lifestyle, and open up new strategies to manipulate harmful biofilms to restore and maintain human well-being.
The authors acknowledge the scientific contributions of their colleagues at the Technical University of Denmark, and elsewhere.