Biofilms are communities of cells attached to a surface. Significant effort is being placed into understanding how they form and the basis for their unique properties, perhaps most notably their resistance to antimicrobials. Imaging technology has and continues to play a very important part in these analyses. Yet, many microbiologists may not be familiar enough with the new technologies to fully maximize their potential. On this site we outline the choices we made to image biofilms, explain what the choices are based on, and list any pitfalls we encountered during imaging and solutions we found. By sharing this information, we hope we can expedite the progress other groups make.
The choice of how to grow a biofilm and how to image should in first instance be driven by the questions that are being asked and realistically, on the technology available and costs. Our overall interest relates to heterogeneity of gene expression in populations, and in this context our goal was to image bacterial biofilms that are growing in the lab in real time with a magnification and resolution that allows individual cells to be identified. Thus, the choices described here are based on this.
This work and the related publication in the Journal of Microscopy (2009, Volume 235, Pt 2, 128-137) is based on a collaborative effort between University of York microbiologist Marjan van der Woude and imaging experts Peter O’Toole and Jo Marrison from the Technology Facility and has been carried out by Matt Lakins (CII).
Modifications to flow cells were carried out by Biosurface Technologies, and other bespoke adaptations by the Department of Biology workshops at University of York, UK .
A detailed hands on course that discusses all aspects of imaging technology is provided by the Technology Facility at the University of York, who also can be approached for contract work. Please go to the Imaging and Cytometry homepage for more information.
Many devices have been developed to grow biofilms in the lab for different purposes. For our purpose, it was essential to use material that permits fluorescent light penetration and does not have autofluorescence, i.e. glass with high optical properties. The next point to consider is whether you require a static biofilm or a biofilm grown in a flow cell. For mutagenesis screens for example the static biofilm will be more suitable. In contrast, to analyze biofilm formation the flow cell may provide valuable additional information. For our long term goals, we wanted to be able to analyze biofilms in a flow cell in real time.
Based on these requirements, we chose to use the FC-281 flow cell from Biosurface Technologies shown below.
The microscope stage-mounting tab was removed and modified with a widened viewing window (see figure below) to facilitate access of the flatter, wider Plan-Apochromat 63x /1.4NA oil DIC objective and other high NA lenses and retain space for movement in the X,Y plane. Biosurface Technologies carried out this modification so any similar adaptation of the product is accessible to all research groups.
|Original window||Modified window|
Our set up involves placing the flow cell in an incubator, either on the microscope stage or as shown here in the lab during biofilm establishment (left hand image). The media was supplied at a controlled temperature using a water bath. A bubble trap (centre image) was used to limit air bubbles entering the flow cell. A highly accurate low speed Watson-Marlow 250U pump was used with 3.2mm diameter tubing (1.6mm bore, 1.6mm wall) for the whole system.
The bacterial strains used here were chosen to facilitate the technology development. Clearly, the species/strain of interest can be used as long as it will attach to a glass surface. With the strain described here, we were able to use untreated glass.
All biofilms consisted of derivatives of the E. coli strain TG1 18, which constitutively produces the F pilus (Ghigo, 2001). This strain was transformed with plasmids pZEtetR21-GFP (Da Re et al., 2007) or pRSETB-tdTomato (Shaner et al., 2004) resulting in strains MV1257 and MV1264, respectively.
The strains were grown in M9 minimal medium with 0.4% glucose at 37˚C (Ghigo, 2001; Reisner et al., 2003). 30mg/ml kanamycin and 100mg/ml ampicillin were added to maintain the plasmids pZE21-gfp or pRSETB-tdTomato respectively. Antibiotics were omitted from the medium during biofilm growth with no evidence of plasmid loss. Gravity facilitated adhesion to the coverslip occurred over 30-60 minutes prior to initiating flow. A biofilm up to 180µm thick was obtained reproducibly within 48 hours after seeding with approximately 1x108 log-phase cells.
Da Re, S., Le Quere, B., Ghigo, J.M. and Beloin, C. (2007) Tight modulation of Escherichia coli bacterial biofilm formation through controlled expression of adhesion factors. Appl Environ Microbiol, 73, 3391-3403.
Ghigo, J.M. (2001) Natural conjugative plasmids induce bacterial biofilm development. Nature, 412, 442-445.
Shaner, N.C., Campbell, R.E., Steinbach, P.A., Giepmans, B.N., Palmer, A.E. and Tsien, R.Y. (2004) Improved monomeric red, orange and yellow fluorescent proteins derived from Discosoma sp. red fluorescent protein. Nat Biotechnol, 22, 1567-1572.
The fluorescence from GFP or tdTomato expressing strains was analyzed directly or alternatively, the BACLight LIVE/DEAD stain (Invitrogen) (1ml of 1:1000 in media) was added just prior to imaging. This stain consists of SYTO9 (green) that is taken up by all cells, and Propidium Iodide (red) that is taken up by cells with a compromised membrane. With the strain and growth conditions used here the fluorochromes completely penetrated the biofilm biomass.
As well as affecting growth rates we noticed that within the closed system of the flow cell rich media caused the production and accumulation of gas in the chamber. This resulted in a flow cell void of biofilm and/or media but entirely filled with froth. LB had this effect, whereas growth on minimal media supplemented with glucose eliminated this problem altogether.
Minimal growth medium also has the advantage of minimal autofluorescence where as cells grown in LB often show autofluoresence.
More technical details can be found in our publication in the Journal of Microscopy (2009, Volume 235, Pt 2, 128-137)
Microbes that express a fluorescent protein or are stained with fluorescent dyes can be used to image living biofilms with fluorescence microscopy. This can be either standard microscopy, single photon laser scanning confocal microscopy (SP-LSCM) or multiphoton laser scanning confocal microscopy (MP-LSCM).
There are several advantages of MP-LSCM over SP-LSCM some of which are:
The biofilm is a highly scattering sample so for the above reasons we chose to use multiphoton laser microscopy.
Imaging was carried out on a Zeiss LSM 510 NLO META LSCM using a bespoke configuration with three non-descanned detectors on the reflective port (NDD, left hand image) coupled to a Coherent Chameleon Ultra laser (centre image). This was fitted onto a Zeiss Axiovert 200 microscope in a fully enclosed, blacked out, bespoke temperature controlled chamber (Solent Scientific, right hand image). Images were acquired using various objectives including a Plan Apochromat 63x/1.4 oil DIC objective, C-Apochromat 40x/1.1W Corr UV-VIS-IR objective , C-Achroplan NIR 40x/0.8 objective and a W Plan Apochromat 20x/1.0 DIC VIS-IR objective
When fully assembled the injection ports and outlet ports on the sides of the flow cell prevented the flow cell from resting flat on the microscope stage. One solution to this problem was to raise the flow cell using a bespoke stage insert. However, by raising the flow cell, the objective could no longer reach the coverslip and a threaded riser was also required. Together these modifications (shown below) allowed the flow cell to be placed on the microscope stage and full X, Y plane mobility retained and the plane of focus could be reached throughout the flow cell depth.
|Flow cell with side ports||Bespoke stage insert|
|Objective lenses and threaded risers||Flow cell on microscope stage|
These are discussed in context of multiphoton laser scanning confocal microscopy but some will also be applicable to single photon-LSCM imaging.
This is a compromise between the desire to use high laser power to visualize even faintly fluorescent material and low laser power to avoid inducing unintentional damage/alterations to the sample. Increased laser power causes fluorescence bleaching, death of the biofilm, ablation and even localized water expulsion or “explosions” in extreme cases.
This is a compromise between increased signal and increased noise. In the cases where high gain is necessary, multiple scans can be performed and by using frame averaging so the noise is filtered out.
The wavelength of the signal should be considered, since both the excitation and emission of longer wavelengths (red) can be obtained to greater depth than the shorter (green) wavelengths. An example is shown in our publication. This would be of particular concern in very thick biofilms (also see below).
All the normal variables associated with the systems should be carefully considered, such as intensity, half life, maturation time and whether the coding sequence is suitable for the organism.
The concentration of the dyes used is important. In some instances we noticed increased fluorescence after attempting to ‘bleach’ an area which can be attributed to excitation of excess dye.
Keep in mind the biology
What is the parameter being reported and how might biofilm growth (physiology, heterogeneity) affect the signal or the interpretation of the signal. For example, there is an ongoing discussion about the cells that stain “dead” with BACLight Live/Dead and whether they are indeed always non-viable.
Biomass limits the signal output. Biofilm age and the growth medium may affect the quality of the image. Growth rates will differ with medium composition, and thus age is not an "absolute" variable.
Imaging pitfalls - Things you may see that are artifacts of imaging.
A lack of a signal in an image is readily interpreted as absence of biofilm material but care has to be taken that it is not a case of lack of fluorescence intensity. This is particularly important when interpreting results using two colours of considerably different wavelength. Using the BACLight Live/Dead stain one can interpret the image incorrectly: all cells will take up SYTO 9 (green) but only the cells with altered membrane integrity will take up PI ("dead"). Thus the interpretation is based on the fact that the red signal overpowers the green signal. As mentioned above, red signal can be detected to greater depths, and thus it is possible that all the visible cells are "dead" (red) but that the live cells that are present (green) in the same plane are just not visible. Similar problems could theoretically arise with a mixed biofilm of GFP and tdTomato cells, etc.
In some instances individual bacteria proved to be considerably brighter than others. If imaging is optimised on the bright bacteria the dim bacteria become invisible and this leads to misinterpreted data so to is important to ensure all cells can be visualised. Autofluorescence can be discounted or controlled by using an unstained specimen.
There is an online hypertextbook on all aspects of biofilms targeted for different levels of expertise. Note that it is only partially complete however.
Some of the many other resources on general aspects of biofilms and methods to grow and analyze biofilms include chapters in several issues of Current Protocols in Microbiology.
The website of the Montana State University Center of Biofilm Engineering has background information and many examples of their imaging work.
G.A. O’Toole’s website has information on their biofilm research and pictures of their set up.