Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2024 Jul 5.
Published in final edited form as: Methods Microbiol. 2023 Jun 26;53:309–324. doi: 10.1016/bs.mim.2023.05.005

Use of epifluorescence widefield deconvolution microscopy for imaging and three-dimensional rendering of Pseudomonas aeruginosa biofilms and extracellular matrix materials

Heidi J Smith a,b,∗, Michael J Franklin a,b,∗
PMCID: PMC11225936  NIHMSID: NIHMS2006091  PMID: 38974073

1. Introduction

Biofilms are communities of bacteria and their extracellular matrix material that are attached to surfaces (Costerton, Lewandovski, Caldwell, Korber, & Lappin-Scott, 1995; Flemming & Wingender, 2010; Hall-Stoodley, Costerton, & Stoodley, 2004; Stoodley, Sauer, Davies, & Costerton, 2002). Many species of bacteria have adapted to growth in association with surfaces. During biofilm growth, bacteria are physiologically distinct from planktonic (free-swimming) cells, producing adhesins for attachment to surfaces (Berne, Ducret, Hardy, & Brun, 2015) and extracellular matrix material for adherence and biofilm formation (Flemming & Wingender, 2010; Franklin, Nivens, Weadge, & Howell, 2011). The extracellular matrix material contains proteins, polysaccharides, and extracellular DNA that form the three-dimensional matrix that houses the biofilms (rev. in Flemming, Neu, & Wozniak, 2007). One model for bacterial biofilm formation includes: (1) initial attachment of planktonic bacteria to surfaces through cell-surface associated appendages and adhesins, (2) firm adhesion of cells to the surface through the formation of matrix material, (3) development of complex three-dimensional structures that include the biofilm bacteria, their extracellular matrix material, and any chemicals from the environment that adsorb to the biofilms, (4) returning of cells to the planktonic state through controlled dispersion processes or through mechanical stresses that disrupt the biofilms (Davies & Marques, 2009; Petrova & Sauer, 2012; Sauer, Camper, Ehrlich, Costerton and Davies, 2002; Sauer et al., 2022; Stoodley et al., 2001). Early reports using transmission electron microscopy (Geesey, Richardson, Yeomans, Irvin, & Costerton, 1977) showed that secreted polymers form a conditioning film during the transition from initial adhesion to biofilm matrix formation (steps 1 and 2 above). Traditionally, the extracellular matrix and organic conditioning film have been difficult to image by light microscopy due to the similar refractive index of the matrix material and the surrounding aqueous medium. Several fluorescent stains, such as fluorescently labelled lectins, have now become available for staining certain components of the extracellular matrix (Jennings et al., 2015; Ma et al., 2009). These probes allowed imaging of some matrix materials by epifluorescence microscopy.

Optical microscopy techniques are the backbone of biofilm analyses, and confocal scanning laser microscopy (CLSM) is considered the gold standard for biofilm visualization, due to its ability to provide real-time, three-dimensional information of hydrated biofilms (rev. in Franklin, Chang, Akiyama, & Bothner, 2015; Schlafer & Meyer, 2017). Since biofilms often have extensive three-dimensional architectures, CSLM allows imaging of biofilm structures by reconstruction of a vertical series of two-dimensional images (z-stacks). In CSLM, laser light excites the fluorescent molecules in the biofilms and the emitted light enters the detector through a pinhole, which eliminates all light that is not in the focal plane. Using CSLM, three-dimensional renderings of the biofilms are obtained computationally from analysis of the z-stack series. While CSLM is an excellent tool for imaging three-dimensional biofilms and for computational analysis of biofilm components, it has shortcomings, including high acquisition and maintenance costs. In addition, the relatively slow scan speeds make CSLM not ideal for screening applications, such as mutant phenotypic screening assays.

In the past, other optical imaging approaches have had limited application to biofilm studies, due to high background signal from out-of-focus regions of the three-dimensional biofilm structures. The background signal from out-of-focus light reduces image quality, contrast, and signal to noise ratios. In addition, the background signal reduces the ability to make quantitative estimates of the components of three-dimensional biofilms, such as cell numbers and the volume of matrix materials. One alternative to CSLM for biofilm imaging is epifluorescence wide-field deconvolution microscopy (WF-DCM). In WF-DCM, rather than using laser light to excite the fluorescent molecules, a light-emitting diode (LED) or halogen lamp light source is used to excite the fluorescent molecules. As with CSLM, in WF-DCM a series of z-stacks is obtained for rendering three-dimensional biofilms. However, rather than using a pinhole to filter out-of-focus light, computational deconvolution clearing approaches are used to render the three-dimensional biofilms and eliminate the out of focus haze. Deconvolution clearing approaches are now available in several different software packages including Thunder (Leica), Clear View (Imaris), and BatchDeconvolution (Fiji) (https://github.com/Mechanobiology-Lab/BatchDeconvolution) and ImageJ (https://imagej.net/imaging/deconvolution). The depth of achievable penetration with WF-DCM is dependent on the scattering of the emitted light and has been shown effective at imaging certain samples up to 150 μm deep in the z-direction (Schumacher & Bertrand, 2019). However, compared to CSLM, WF-DCM is still not as effective at imaging thick biological structures, such as mature biofilms.

Previously, deconvolution steps have been incorporated into diverse biological imaging applications, such as live cell imaging (Swedlow & Platani, 2002), Drosophila spindle imaging (Arigovindan et al., 2013), and plant root tissue studies (Shaw, 2006). There are also studies where deconvolution has been used in biofilm systems. Examples include confocal imaging of catheter associated Staphylococcus aureus biofilms (Hess, Henry-Stanley, Barnes, Dunny, & Wells, 2012), widefield imaging of the spatial arrangement of bacteria in sponge tissue (Manz, Arp, Schumann-Kindel, Szewzyk, & Reitner, 2000), confocal and widefield imaging of bacterial membrane vesicles (Turnbull et al., 2016), and interactions of Lactobacillus challenge on Gardnerella vaginalis biofilms (Saunders, Bocking, Challis, & Reid, 2007). Deconvolution of three-dimensional images is often an additional step requiring a separate software package and downstream analysis of images. Presently, DCM software may be integrated into imaging systems, providing the ability to view in-focus three-dimensional structures instantly, enabling the microscopist to make necessary modifications for increased image quality or for downstream quantitative analyses.

In most biofilm imaging studies, the cells are stained with fluorescent probes or express fluorescent proteins, such as the green fluorescent protein (GFP). The extracellular matrix of biofilms may also be imaged if fluorescent probes that bind the matrix material are available (Jennings et al., 2015; Ma et al., 2009). In the present study, we provide methods for using WF-DCM to evaluate early biofilm formation for obtaining clear in-focus images of bacterial biofilms and for quantitative assessment of biofilm components. The speed of imaging using WF-DCM (compared to CSLM) allows applications such as screening mutant strains that are impaired in biofilm formation in a 96-well microtiter plate format. Biofilms cultured in microtiter plates have been used for many biofilm screening applications, including screening of mutants with reduced ability to form biofilms (O’Toole & Kolter, 1998a, 1998b; O’Toole et al., 1999), response of biofilms to antimicrobial agents (Chlumsky et al., 2021), and response to phages (Knezevic & Petrovic, 2008). Here, using WF-DCM, we show early biofilm formation of GFP-labelled P. aeruginosa PAO1 and show that extracellular matrix material forms in association with the cells on the surface of glass-bottom 96-well microtiter plates. We also show the effect of a ΔplsA mutation on the development of biofilms and the effects of an environmental condition (calcium addition) on early biofilm formation. We provide an example of how deconvoluted images may be used to give quantitative data on biofilm cell numbers and biovolumes of biofilm matrix material. The results here show that WF-DCM may be used as an alternative to CSLM for imaging and quantifying biofilm components during early biofilm formation.

2. Materials and methods

2.1. Strains and biofilm growth conditions

Pseudomonas aeruginosa strain PA01 and its mutant derivative were used in this study. The mutant strain, P. aeruginosa PA01 WFPA60 ΔpslA, which is impaired in the ability to produce the PSL polysaccharide, was graciously provided by Dr. Dan Wozniak. Both strains were engineered to express GFP from plasmid pMF230 (Nivens, Ohman, Williams, & Franklin, 2001). Strains were routinely cultured on LB Broth-Lennox (BD-Difco, Franklin Lakes NJ) containing 150 μg/mL carbenicillin (Anatrace, Maumee, OH).

Biofilms were cultured in biofilm minimal medium (BMM), which contained 9.0 mM sodium glutamate, 50 mM glycerol, 2.0 mM MgSO4, 0.15 mM NaH2PO4, 0.34 mM K2HPO4, and 145 mM NaCl. The pH of the medium was adjusted to 7.0. Autoclaved MgSO4 (0.02 mM final concentration), CaCl2 (0 or 2.0 mM final concentration), and filter-sterilized trace mineral solution (20 μL/L) were added to the medium after autoclaving. Trace metal solution (per litre of 0.83 M HCl) contained 5.0 g CuSO4·5H2O, 5.0 g ZnSO4·7H2O, 5.0 g FeSO4·7H2O, and 2.0 g MnCl2·4H2O.

For biofilm growth, P. aeruginosa PAO1 (pMF230) and WFPA60 ΔpslA (pMF230) were inoculated into LB Broth containing 150 μg/mL carbenicillin and incubated overnight at 37 °C on a test tube roller to an O.D600 of approximately 3.0. An aliquot of the overnight culture (500 μL) was pelleted by centrifugation and washed once by resuspending the pellet in 500 μL of BMM without added CaCl2. The cells were pelleted again and resuspended in 500 μL of BMM. The resuspended cells (2.0 μL) were used to inoculate the wells of a glass-bottom microtiter plate (Greiner CellView 96, Greiner BioOne, Monroe, North Carolina) containing 200 μL of BMM with 0 or 2 mM added CaCl2. The microtiter plates were incubated on a rotary shaker at 200 rpm for 14 h. Following the initial 14 h incubation period, the medium was exchanged with fresh BMM every 1.5 h by drawing off the medium, then adding 200 μL of fresh BMM with 0 or 2 mM CaCl2. The medium was changed three times to mimic a flow-through system, where the biofilm cells remain attached to the glass surface of the microtiter plate and the planktonic cells are removed.

2.2. Biofilm staining

Following biofilm growth, the BMM medium was removed, and the biofilms were stained with Cell Mask Orange (CMO) (ThermoFisher Scientific, Waltham, Massachusetts) and Bodipy X-SE 630/650 (BOD) (ThermoFisher Scientific). Stains were diluted in 0.85% NaCl to 0.1 μg/mL (CMO) and 1 μg/mL (BOD). Staining was performed by removing the medium then placing 50 μL of stain (ora mix of 25 μL of each stain) into the wells of the microtiter plates. Biofilms were incubated with the stain for 30 min at 23 °C, followed by three 100 μL washes with 0.85% saline. Three independent biological replicates were performed, with the images shown from one of the replicates.

2.3. Biofilm imaging and image processing using wide-field deconvolution microscopy

Images were collected with a Leica DMi8 widefield microscope equipped with the K8 monochrome camera and THUNDER imager system (Leica Microsystems Inc. Wetzlar, Germany). For each sample, multi-stack fluorescent images were acquired for three channels using a multiline tunable LED light source and a quadband filter cube. The filter excitation and emission settings were 475/510 (GFP), 555/590 (CMO), and 635/700 (Bodipy). LED intensities and exposure times were optimized for each strain and growth condition (+/−Ca) for each fluorescent channel, and subsequently kept the same to enable cross image comparisons. Images were acquired with a 100× oil objective (1.4 NA) within LAS X (V3.7.6.25997). Z-stacks were collected for all images using a 0.2 μm step size. Thunder computational clearing that utilizes Leica Microsystems’ proprietary algorithms was used to enhance image contrast and resolution. It should be noted that other software deconvolution algorithms can be used with inverted widefield microscopes that have comparable specifications in terms of light source and filter cubes for excitation and emission objective choice, camera sensitivity, and resolution. Examples of other deconvolution methods include Imaris Clear View (Oxford Instruments, Enkoping, Sweden) BatchDeconvolution (Fiji) (https://github.com/Mechanobiology-Lab/BatchDeconvolution) and ImageJ (https://imagej.net/imaging/deconvolution).

Following image collection and generation of maximum projection images from z-stacks, images were immediately deconvolved using Thunder integrated software with the Instant Computational Clearing method (Schumacher & Bertrand, 2019). For a widefield system, out-of-focus light arises from the out-of-focus point spread function (PSF), which is a characteristic diffraction-based pattern that is specific for every optical configuration. The goal of deconvolution is to separate in-focus signal from the out-of-focus background for each image/stack, which results in the sample signal being directly visible. Leica’s Thunder Computational Clearing (CC) method is based on the principle that recorded background signal has a characteristic behaviour that is independent of its origin. Given that background signal is highly inconsistent and varies greatly across a single field of view (FOV), the out-of-focus PSF is significantly wider than the in-focus PSF. This makes it possible to effectively separate the two contributions by length-scale-discriminating algorithms. Leica has developed a proprietary algorithm based on this principle to separate these two sources of out-of-focus light that results in the removal of the background signal (Schumacher & Bertrand, 2019). Beyond generating an unmasked image, the deconvoluted data may be used for quantification, due to the linear behaviour of deconvoluted data in comparison to raw data.

Biovolumes (μm3) and the number of objects (cells) for a representative widefield image from each strain and growth condition were calculated within Imaris V9.9.1 (Zurich, Switzerland). Given that these biofilms were thin (less than 10 μm), there were no observable issues resulting from the attenuation of light. Therefore, a simple maximum threshold of intensity could be used for biovolume calculations (Costes et al., 2004; Parker, Christen, Lorenz, & Smith, 2020). Within Imaris, surface calculation parameters were optimized to sample and image specific properties for each fluorescent channel. The surface calculation parameters for each channel were as follows: GFP images (used for the quantification of individual cells) were smoothed at a surface area detail level of 0.08 μm, the diameter of the largest sphere was set to 0.50 μm. Thresholding was based on a background subtraction method, and all voxels greater than 10 were collected. CMO and BOD images (used for biovolume generation) were processed as for the GFP images with the diameter of the largest sphere set to 1.0 μm for CMO and 2.0 μm for BOD. Following surface generation with the deconvoluted images, the same calculation parameters were applied to the raw image files (prior to deconvolution) for comparison of the number of identified objects and biovolumes.

3. Results and discussion

3.1. Imaging early biofilm formation by P. aeruginosa PAO1 using epifluorescence wide-field deconvolution microscopy

We used a microtiter plate format to culture early-stage biofilms on the surfaces of glass-bottom microtiter plates. Biofilms were cultured by inoculating cells into biofilm minimal medium (BMM) contained in the wells of glass-bottom microtiter plates. The cells were allowed to attach then incubated in association with to the glass surfaces of the microtiter plate using a rotary shaker at 37 °C for 14 h. To mimic a flow-through system, where attached cells are retained and the planktonic cells are removed, we exchanged the spent medium with fresh medium three times over the course of 4.5 h, following the initial 14 h incubation. One hour following the final medium exchange, the biofilms were stained with the P. aeruginosa PAO1 matrix stains cell mask orange (CMO) and Bodipy X-SE 630/650 (BOD). These stains were identified in a screen of fluorescent molecules that bind P. aeruginosa PAO1 extracellular matrix material (unpublished data). P. aeruginosa PAO1 produces an extracellular polymeric matrix that includes the PSL polysaccharide, which is composed of a repeating pentasaccharide of D-mannose, L-rhamnose, and D-glucose (Byrd et al., 2009; Franklin et al., 2011). Because of the diversity of matrix materials produced by different bacteria, there are no universal biofilm matrix stains. In fact, these dyes are not effective at staining the matrix of other strains of P. aeruginosa, including P. aeruginosa PA14, which produces the PEL polysaccharide, or P. aeruginosa FRD1, which produces alginate (Franklin et al., 2011). However, because these stains effectively bind the matrix of P. aeruginosa PAO1, we used them in the model system here to test the effectiveness of WF-DCM for imaging matrix material formed during early biofilm development of P. aeruginosa PAO1.

Fig. 1 shows P. aeruginosa PAO1 biofilms cultured on the glass surface of a glass-bottom microtiter plate in BMM without added calcium. Epifluorescence widefield microscopy was used to image each channel, with the GFP-expressing cells shown as green, the CMO-stained matrix material as false-coloured red, and the BOD-stained material as false-coloured cyan. A series of two-dimensional images was obtained by moving the microscope stage at 0.2 μm increments vertically to obtain z-stack series. Fig. 1 shows maximum projection images of the z-stacks from the same field of view (FOV) for each channel. Fig. 1A shows an example of early biofilm formation, where single cells attach along the glass surface and small colonies (less than 10 μm thick) begin to develop. The individual cells are resolved, but because of the out-of-focus light, the small microcolonies are blurred in the maximum projection images. Following deconvolution, the single cells along the glass surface are in focus, and the cells within the microcolonies are resolved, allowing greater visualization of individual cells within the microcolonies (Fig. 1E). The CMO-stained material also shows out-of-focus light in the maximum projection image (Fig. 1B). Following deconvolution, the CMO-stained material appears more resolved, with most of the stained material in the same regions as the cell clusters. Similarly, deconvolution helps to resolve the BOD-stained material (Fig. 1C, G). Overlays of the three images (GFP cells, CMO- and BOD-stained matrix material) before deconvolution (Fig. 1D) and after deconvolution (Fig. 1H) are shown. The overlay image shows that prior to deconvolution, the matrix and the cells appear as hazy structures, which could indicate that the matrix material forms a gel-like substance with the cells embedded in the matrix. Following deconvolution (Fig. 1H), the cells and the matrix are resolved, with the microcolonies associated with structured matrix material.

FIG. 1.

FIG. 1

Open in a new tab

Maximum projection epifluorescence widefield image of a P. aeruginosa PAO1 (pMF230) biofilm cultured on the glass-bottom surface of a microtiter plate in biofilm minimal medium (BMM). Each image shows the same field of view. (A, E) showing the GFP-labelled cells, (B, F) showing the extracellular material that stains with cell mask orange false coloured red, and (C, G) showing the extracellular material that stains with Bodipy X-SE 630/650 false coloured cyan. Panels (D, H) show an overlay of the three fluorescent images. Panels (A–D) show the early biofilm prior to deconvolution. Panels (E–H) show the biofilms following computational deconvolution using the Leica Instant Computational Clearing method.

3.2. Effect of calcium addition on initial P. aeruginosa PAO1 biofilm formation

Imaging by WF-DCM using a microtiter plate allows simultaneous studies on the effects of different environmental conditions on biofilm development. Since calcium addition affects biofilm formation of an alginate-producing strain, P. aeruginosa FRD1 (Sarkisova, Patrauchan, Berglund, Nivens, & Franklin, 2005), and of other species of bacteria (e.g., Bilecen & Yildiz, 2009; Patrauchan, Sarkisova, Sauer, & Franklin, 2005; Rose, Turner, & Dibdin, 1997; Tischler, Vanek, Peterson, & Visick, 2021), we added calcium to the BMM medium to determine if differences in biofilm structure of the PSL-producing strain P. aeruginosa PAO1 could be resolved by WF-DCM. We cultured P. aeruginosa PAO1 in the wells of microtiter plates in BMM with 2 mM added CaCl2, as in the experiment above. Z-stacks at 0.2 μm increments of biofilms were obtained by wide-field microscopy. Fig. 2 shows maximum projection images of the cellular and extracellular matrix material (CMO-stained and BOD-stained) of these P. aeruginosa PAO1 biofilms from one representative FOV. Prior to deconvolution (Fig. 2A), the GFP-labelled cells appear as a monolayer, with the monolayer interspersed by small clusters of cells. Because the monolayer is thin, there is little out-of-focus light originating from the cells in association with the glass surface. However, cells within the clusters are not well resolved. Fig. 2E shows a maximum projection of the same FOV following deconvolution of the image. The computational clearing did not affect the focus of the monolayer of cells but allowed resolution of individual cells within the cell clusters. The CMO-stained extracellular material shows structure along the surface of the glass substratum. In the non-cleared image, CMO-staining is observed with out-of-focus haze (Fig. 2B). Following deconvolution, the CMO stained material appears as a structured material that forms along the glass substratum (Fig. 2F). Similarly, the BOD-stained material appears hazy prior to computational clearing (Fig. 2C), but as part of the matrix material following deconvolution (Fig. 2G). Overlays of the three maximum projection images show that deconvolution clarifies the image, removing the hazy structures and providing greater resolution of individual cells and matrix material (Fig. 2D, H). The image shows that calcium addition affects early biofilm formation, with the matrix material forming a more structured layer compared to the no-calcium conditions, with the cells appearing to attach to the matrix structure.

FIG. 2.

FIG. 2

Open in a new tab

Maximum projection epifluorescence widefield image of a P. aeruginosa PAO1 (pMF230) biofilm cultured on the glass-bottom surface of a microtiter plate in BMM with 2 mM CaCl2. Each image shows the same field of view. (A, E) showing the GFP-labelled cells, (B, F) showing the extracellular material that stains with cell mask orange false coloured red, and (C, G) showing the extracellular material that stains with Bodipy X-SE 630/650 false coloured cyan. Panels (D, H) show an overlay of the three fluorescent images. Panels (A–D) show the early biofilm prior to deconvolution. Panels (E–H) show the biofilms following computational deconvolution using the Leica Instant Computational Clearing method.

3.3. Effect of PSL extracellular matrix polysaccharide on early P. aeruginosa PAO1 biofilm formation

To determine if WF-DCM may be used to characterize the effects of specific mutations on biofilms formed in microtiter plates, we cultured P. aeruginosa PAO1 ΔpslA (pMF230) in the wells of glass-bottom plates, using the same protocol as used for the wild-type strain. We cultured the mutant strain in the presence and absence of 2 mM CaCl2. Maximum projection images of the GFP-labelled cells of early biofilms cultured in the absence of added CaCl2 showed that the P. aeruginosa ΔpslA generates dense microcolonies where the cells appear packed in a tight arrangement (Fig. 3A). The dense packing of the microcolonies generated out-of-focus haze so that the individual cells were not resolved. Computational clearing enhanced image quality (Fig. 3E), allowing visualization of single cells within the tightly packed cell clusters. Deconvolution also enhanced the signal to noise ratio of the monolayer of cells along the glass surface, allowing cells to be viewed with improved contrast and resolution. The P. aeruginosa PAO1 ΔpslA biofilms had very little staining of extracellular matrix material within the microcolonies with either CMO (Fig. 3B, F) or BOD (Fig. 3C, G), although the stains bound some material that appears to be cellular. An overlay of the three channels shows that deconvolution provides high resolution of the biofilm cells and little extracellular matrix material associated with the microcolonies (Fig. 3D, H). Although the PSL polysaccharide is required for structural integrity of P. aeruginosa PAO1 biofilms (Colvin et al., 2012), PSL does not appear to be required for initial microcolony formation.

FIG. 3.

FIG. 3

Open in a new tab

Maximum projection epifluorescence widefield image of a P. aeruginosa PAO1 ΔpslA (pMF230) biofilm cultured on the glass-bottom surface of a microtiter plate in BMM. Each image shows the same field of view. (A, E) showing the GFP-labelled cells, (B, F) showing the extracellular material that stains with cell mask orange false coloured red, and (C, G) showing the extracellular material that stains with Bodipy X-SE 630/650 false coloured cyan. Panels (D, H) show an overlay of the three fluorescent images. Panels (A–D) show the early biofilm prior to deconvolution. Panels (E–H) show the biofilms following computational deconvolution using the Leica Instant Computational Clearing method.

P. aeruginosa ΔpslA (pMF230) biofilms cultured in BMM with 2 mM CaCl2 gave similar results to the same strain cultured in the absence of added calcium (Fig. 4). The cells cultured with calcium also had microcolonies of cells that were tightly packed with very little space between cells. Because of the tight packing, the cells were not resolved prior to deconvolution (Fig. 4A). Following deconvolution, the individual cells within clusters were resolved (Fig. 4E), and the monolayers of cells along the surface had increased contrast due to the enhanced signal to noise ratio. As in the absence of added calcium, very little matrix material stained with CMO or BOD, but the stain bound to some of the cells (Fig. 4B, C, F, G). Following deconvolution, an overlay of the three channels provided a clear view of these biofilms (compare Fig. 4D and H).

FIG. 4.

FIG. 4

Open in a new tab

Maximum projection epifluorescence widefield image of a P. aeruginosa PAO1 ΔpslA (pMF230) biofilm cultured on the glass-bottom surface of a microtiter plate in BMM with 2 mM CaCl2. Each image shows the same field of view. (A, E) showing the GFP-labelled cells, (B, F) showing the extracellular material that stains with cell mask orange false coloured red, and (C, G) showing the extracellular material that stains with Bodipy X-SE 630/650 false coloured cyan. Panels (D, H) show an overlay of the three fluorescent images. Panels (A–D) show the early biofilm prior to deconvolution. Panels (E–H) show the biofilms following computational deconvolution using the Leica Instant Computational Clearing method.

3.4. Biovolume analysis of biofilms before and after computational clearing

Given that deconvolution reduces background light that is out of focus, deconvoluted images of biofilm stacks should increase the accuracy of quantification of cellular material and fluorescently stained extracellular matrix components. To test this, we imported both the raw and deconvoluted 3-D images into the image analysis software Imaris V9.9.1 (Zurich, Switzerland) for measurement of individual features (cells) and for biovolume measurements (matrix components). An example of a biovolume analysis is shown in Fig. 5, which is a surface volume map generated from an individual microcolony of a P. aeruginosa PAO1 biofilm cultured without CaCl2. The image shows a microcolony following deconvolution (Fig. 5A), compared to non-deconvoluted image (Fig. 5B). Fig. 5C shows the surface rendering of the cells, and Fig. 5D shows surface rendering of the cells plus matrix material. Fig. 5A and B show that the background haze of a microcolony is removed following deconvolution, and that the surface used to enumerate the number of cells is an accurate match for the GFP signal from the deconvoluted image. Surface rendering shows that the localization of the CMO- and BOD-matrix staining components in relation to GFP cells is resolved by deconvolution, and indicates that the matrix components are associated with the cellular biomass.

FIG. 5.

FIG. 5

Open in a new tab

Expanded image of a microcolony of P. aeruginosa PAO1 (pMF230) from Fig. 1, showing the workflow used to enumerating biofilm cells and extracellular matrix components. All panels show the same field of view. (A) Microcolony showing the GFP-labelled cells following deconvolution using the Leica Instant Computational Clearing method, and (B) microcolony prior to deconvolution. (C) Surface map of biofilm cells generated from the deconvoluted image, using Imaris software to enumerate individual cells within the microcolony. (D) Surface map of deconvoluted microcolony image, generated using Imaris software. The surface map was used to calculate biovolumes for the extracellular matrix material within a microcolony with the cells shown in green, the cell mask orange-stained material shown in red, and the Bodipy X-SE 630/650-stained material shown in blue.

Using Imaris, surface volume maps were generated for each field of view (FOV) for the images shown in Figs. 1–4. The GFP channel was used to enumerate the total number of cells and the fluorescence intensity of the cells. The CMO- and BOD-staining channels were used to quantify the volume of fluorescently stained extracellular matrix present (Table 1). Surprisingly, the image analysis software was able to quantify number of cells even in the absence of computational clearing. The number of cells per FOV was similar for the raw images and for the computationally cleared samples, even for the P. aeruginosa ΔpslA strain, where individual cells from within microcolonies were difficult to resolve (Table 1). However, computational clearing reduced the mean intensity of the GFP fluorescence by 33–55% for wild-type cells and 53–55% for the P. aeruginosa Δpsl cells. Similarly, for P. aeruginosa PAO1, the biovolumes for the CMO-stained components were reduced by 22% following deconvolution and the BOD-stained components were lower by 26–37% following deconvolution. For P. aeruginosa ΔpslA, the biovolumes for both CMO- and BOD-stained material was lower than for the Psl-producing wild-type strain. Computational clearing reduced the biovolumes measured for P. aeruginosa ΔpslA CMO- and BOD-stained material by 14–31%. The results indicate that by removing out-of-focus light from epifluorescence widefield microscopy images, the biovolume measurements may be more accurately determined than from epifluorescence microscopy images without deconvolution.

Table 1.

Quantification of P. aeruginosa biofilm components prior to and following computational deconvolution.

Strain Condition Deconvolution Number of cells Mean intensity (GFP—μm3) Δ% CMO volume (μm3) Δ% BOD volume (μm3) Δ%
P. aeruginosa PAO1 − CaCl2 N 5008 1128 10,774 6524
Y 5039 751 33 8402 22 4808 26
P. aeruginosa PAO1 + CaCl2 N 5124 1464 4943 4985
Y 5191 649 55 3835 22 3127 37
P. aeruginosa ΔpslA − CaCl2 N 4746 751 496 1356
Y 4850 347 53 422 14 937 30
P. aeruginosa ΔpslA + CaCl2 N 7980 1386 1692 1875
Y 8210 622 55 1299 23 1283 31
Open in a new tab

4. Conclusions

Epifluorescence widefield microscopy is a useful imaging strategy for studying microbial biofilm development. With appropriate fluorescent stains, epifluorescence microscopy may be used to characterize both the cellular and extracellular components of developing biofilms. However, when compared to confocal scanning laser microscopy, epifluorescence widefield microscopy suffers from low resolution of biofilm structures, due to background light originating from the out-of-focus regions of three-dimensional biofilm structures. In this study, we demonstrate the utility of combining epifluorescence widefield optical imaging with computational deconvolution for characterizing early biofilm formation of P. aeruginosa PAO1. The speed of imaging and immediate computational deconvolution provides an effective strategy for obtaining clear in-focus images of biofilms, including biofilms formed on the surface of glass-bottom microtiter plates. Using microtiter plates to culture biofilms provides the ability to test the effects of many different environmental conditions on biofilm development simultaneously, and here we show that calcium addition affects biofilm and matrix structure of P. aeruginosa PAO1 biofilms. The microtiter plates may also be used to screen mutant strains for impaired biofilm formation using optical imaging as a readout for biofilm structure, and here, we provide an example of the effect of a mutation in a gene required for PSL polysaccharide production on P. aeruginosa PAO1 biofilm structure. The results indicate that WF-DCM is a useful tool for obtaining clear in-focus images of early biofilm formation, and that the deconvoluted images may be used to obtain quantitative information on biofilm components and structures.

Acknowledgements

We thank Kerry Williamson and Jill Story for their helpful input during the writing of this manuscript. Imaging was made possible by The Center for Biofilm Engineering Imaging Facility at Montana State University, which is supported by funding from the National Science Foundation MRI Program (2018562), the M.J. Murdock Charitable Trust (202016116), the US Department of Defence (77369LSRIP), and by the Montana Nanotechnology Facility (an NNCI member supported by NSF Grant ECCS-2025391). Additionally, research reported in this publication was supported by the National Institute of General Medical Sciences of the National Institutes of Health under Award Number P20GM103474 and R21AI154171 (M.J.F.). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

References

  1. Arigovindan M, Fung JC, Elnatan D, Mennella V, Chan YH, Pollard M, et al. (2013). High-resolution restoration of 3D structures from widefield images with extreme low signal-to-noise-ratio. Proceedings of the National Academy of Sciences of the United States of America, 110(43), 17344–17349. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Berne C, Ducret A, Hardy GG, & Brun YV (2015). Adhesins involved in attachment to abiotic surfaces by Gram-negative bacteria. Microbiology Spectrum, 3(4). 10.1128/microbiolspec.MB-0018-2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Bilecen K, & Yildiz FH (2009). Identification of a calcium-controlled negative regulatory system affecting Vibrio cholerae biofilm formation. Environmental Microbiology, 11(8), 2015–2029. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Byrd MS, Sadovskaya I, Vinogradov E, Lu H, Sprinkle AB, Richardson SH, et al. (2009). Genetic and biochemical analyses of the Pseudomonas aeruginosa Psl exopolysaccharide reveal overlapping roles for polysaccharide synthesis enzymes in Psl and LPS production. Molecular Microbiology, 73(4), 622–638. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Chlumsky O, Smith HJ, Parker AE, Brileya K, Wilking JN, Purkrtova S, et al. (2021). Evaluation of the antimicrobial efficacy of N-acetyl-l-cysteine, rhamnolipids, and usnic acid—Novel approaches to fight food-borne pathogens. International Journal of Molecular Sciences, 22(21). 10.3390/ijms222111307. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Colvin KM, Irie Y, Tart CS, Urbano R, Whitney JC, Ryder C, et al. (2012). The Pel and Psl polysaccharides provide Pseudomonas aeruginosa structural redundancy within the biofilm matrix. Environmental Microbiology, 14(8), 1913–1928. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Costerton JW, Lewandovski Z, Caldwell DE, Korber DR, & Lappin-Scott HM (1995). Microbial biofilms. Annual Review in Microbiology, 49, 711–745. [DOI] [PubMed] [Google Scholar]
  8. Costes SV, Daelemans D, Cho EH, Dobbin Z, Pavlakis G, & Lockett S (2004). Automatic and quantitative measurement of protein-protein colocalization in live cells. Biophysical Journal, 86(6), 3993–4003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Davies DG, & Marques CN (2009). A fatty acid messenger is responsible for inducing dispersion in microbial biofilms. Journal of Bacteriology, 191(5), 1393–1403. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Flemming HC, Neu TR, & Wozniak DJ (2007). The EPS matrix: The “house of biofilm cells”. Journal of Bacteriology, 189(22), 7945–7947. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Flemming HC, & Wingender J (2010). The biofilm matrix. Nature Reviews. Microbiology, 8(9), 623–633. [DOI] [PubMed] [Google Scholar]
  12. Franklin MJ, Chang C, Akiyama T, & Bothner B (2015). New technologies for studying biofilms. Microbiology Spectrum, 3(4). 10.1128/microbiolspec.MB-0016-2014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Franklin MJ, Nivens DE, Weadge JT, & Howell PL (2011). Biosynthesis of the Pseudomonas aeruginosa extracellular polysaccharides, alginate, Pel, and Psl. Frontiers in Microbiology, 2. 10.3389/fmicb.2011.00167. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Geesey GG, Richardson WT, Yeomans HG, Irvin RT, & Costerton JW (1977). Microscopic examination of natural sessile bacterial populations from an alpine stream. Canadian Journal of Microbiology, 23(12), 1733–1736. [DOI] [PubMed] [Google Scholar]
  15. Hall-Stoodley L, Costerton JW, & Stoodley P (2004). Bacterial biofilms: From the natural environment to infectious diseases. Nature Reviews. Microbiology, 2(2), 95–108. [DOI] [PubMed] [Google Scholar]
  16. Hess DJ, Henry-Stanley MJ, Barnes AM, Dunny GM, & Wells CL (2012). Ultrastructure of a novel bacterial form located in Staphylococcus aureus in vitro and in vivo catheter-associated biofilms. The Journal of Histochemistry and Cytochemistry, 60(10), 770–776. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Jennings LK, Storek KM, Ledvina HE, Coulon C, Marmont LS, Sadovskaya I, et al. (2015). Pel is a cationic exopolysaccharide that cross-links extracellular DNA in the Pseudomonas aeruginosa biofilm matrix. Proceedings of the National Academy of Sciences of the United States of America, 112(36), 11353–11358. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Knezevic P, & Petrovic O (2008). A colorimetric microtiter plate method for assessment of phage effect on Pseudomonas aeruginosa biofilm. Journal of Microbiological Methods, 74(2–3), 114–118. [DOI] [PubMed] [Google Scholar]
  19. Ma L, Conover M, Lu H, Parsek MR, Bayles K, & Wozniak DJ (2009). Assembly and development of the Pseudomonas aeruginosa biofilm matrix. PLoS Pathogens, 5(3), e1000354. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Manz W, Arp G, Schumann-Kindel G, Szewzyk U, & Reitner J (2000). Widefield deconvolution epifluorescence microscopy combined with fluorescence in situ hybridization reveals the spatial arrangement of bacteria in sponge tissue. Journal of Microbiological Methods, 40(2), 125–134. [DOI] [PubMed] [Google Scholar]
  21. Nivens DE, Ohman DE, Williams J, & Franklin MJ (2001). Role of alginate and its O acetylation in formation of Pseudomonas aeruginosa microcolonies and biofilms. Journal of Bacteriology, 183(3), 1047–1057. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. O’Toole GA, & Kolter R (1998a). Flagellar and twitching motility are necessary for Pseudomonas aeruginosa biofilm development. Molecular Microbiology, 30, 295–304. [DOI] [PubMed] [Google Scholar]
  23. O’Toole GA, & Kolter R (1998b). Initiation of biofilm formation in Pseudomonas fluorescens WCS365 proceeds via multiple, convergent signalling pathways: A genetic analysis. Molecular Microbiology, 28, 449–461. [DOI] [PubMed] [Google Scholar]
  24. O’Toole GA, Pratt LA, Watnick PI, Newmann DK, Weaver VB, & Kolter R (1999). Genetic approaches to study of biofilms. Methods in Enzymology, 310, 91–109. [DOI] [PubMed] [Google Scholar]
  25. Parker AE, Christen JA, Lorenz L, & Smith H (2020). Optimal surface estimation and thresholding of confocal microscope images of biofilms using Beer’s law. Journal of Microbiological Methods, 174, 105943. [DOI] [PubMed] [Google Scholar]
  26. Patrauchan MA, Sarkisova S, Sauer K, & Franklin MJ (2005). Calcium influences cellular and extracellular product formation during biofilm-associated growth of a marine Pseudoalteromonas sp. Microbiology, 151(Pt. 9), 2885–2897. [DOI] [PubMed] [Google Scholar]
  27. Petrova OE, & Sauer K (2012). Dispersion by Pseudomonas aeruginosa requires an unusual posttranslational modification of BdlA. Proceedings of the National Academy of Sciences of the United States of America, 109(41), 16690–16695. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Rose RK, Turner SJ, & Dibdin GH (1997). Effect of pH and calcium concentration on calcium diffusion in streptococcal model-plaque biofilms. Archives of Oral Biology, 42(12), 795–800. [DOI] [PubMed] [Google Scholar]
  29. Sarkisova S, Patrauchan MA, Berglund D, Nivens DE, & Franklin MJ (2005). Calcium-induced virulence factors associated with the extracellular matrix of mucoid Pseudomonas aeruginosa biofilms. Journal of Bacteriology, 187, 4327–4337. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Sauer K, Camper AK, Ehrlich GD, Costerton JW, & Davies DG (2002). Pseudomonas aeruginosa displays multiple phenotypes during development as a biofilm. Journal of Bacteriology, 184(4), 1140–1154. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Sauer K, Stoodley P, Goeres DM, Hall-Stoodley L, Burmolle M, Stewart PS, et al. (2022). The biofilm life cycle: Expanding the conceptual model of biofilm formation. Nature Reviews. Microbiology, 20(10), 608–620. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Saunders S, Bocking A, Challis J, & Reid G (2007). Effect of Lactobacillus challenge on Gardnerella vaginalis biofilms. Colloids and Surfaces. B, Biointerfaces, 55(2), 138–142. [DOI] [PubMed] [Google Scholar]
  33. Schlafer S, & Meyer RL (2017). Confocal microscopy imaging of the biofilm matrix. Journal of Microbiological Methods, 138, 50–59. [DOI] [PubMed] [Google Scholar]
  34. Schumacher J, & Bertrand L (2019). THUNDER imagers: How do they really work. THUNDER Imager Technical; Note. [Google Scholar]
  35. Shaw PJ (2006). Comparison of widefield/deconvolution and confocal microscopy for three-dimensional imaging. Handbook of Biological Confocal Microscopy, 3, 453–467. [Google Scholar]
  36. Stoodley P, Sauer K, Davies DG, & Costerton JW (2002). Biofilms as complex differentiated communities. Annual Review of Microbiology, 56, 187–209. [DOI] [PubMed] [Google Scholar]
  37. Stoodley P, Wilson S, Hall-Stoodley L, Boyle JD, Lappin-Scott HM, & Costerton JW (2001). Growth and detachment of cell clusters from mature mixed-species biofilms. Applied and Environmental Microbiology, 67(12), 5608–5613. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Swedlow JR, & Platani M (2002). Live cell imaging using wide-field microscopy and deconvolution. Cell Structure and Function, 27(5), 335–341. [DOI] [PubMed] [Google Scholar]
  39. Tischler AH, Vanek ME, Peterson N, & Visick KL (2021). Calcium-responsive diguanylate cyclase casa drives cellulose-dependent biofilm formation and inhibits motility in Vibrio fischeri. MBio, 12(6), e0257321. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Turnbull L, Toyofuku M, Hynen AL, Kurosawa M, Pessi G, Petty NK, et al. (2016). Explosive cell lysis as a mechanism for the biogenesis of bacterial membrane vesicles and biofilms. Nature Communications, 7, 11220. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES