The development of microbial communities depends on a combination of factors, including environmental architecture, member abundance, traits, and interactions. This protocol describes a synthetic, microfabricated environment for the simultaneous tracking of thousands of communities contained in femtoliter wells, where key factors such as niche size and confinement can be approximated.
The development of microbial communities depends on a combination of complex deterministic and stochastic factors that can dramatically alter the spatial distribution and activities of community members. We have developed a microwell array platform that can be used to rapidly assemble and track thousands of bacterial communities in parallel. This protocol highlights the utility of the platform and describes its use for optically monitoring the development of simple, two-member communities within an ensemble of arrays within the platform. This demonstration uses two mutants of Pseudomonas aeruginosa, part of a series of mutants developed to study Type VI secretion pathogenicity. Chromosomal inserts of either mCherry or GFP genes facilitate the constitutive expression of fluorescent proteins with distinct emission wavelengths that can be used to monitor community member abundance and location within each microwell. This protocol describes a detailed method for assembling mixtures of bacteria into the wells of the array and using time-lapse fluorescence imaging and quantitative image analysis to measure the relative growth of each member population over time. The seeding and assembly of the microwell platform, the imaging procedures necessary for the quantitative analysis of microbial communities within the array, and the methods that can be used to reveal interactions between microbial species area all discussed.
Microbial communities are shaped by both deterministic factors, such as the structure of the environment, and stochastic processes, which are associated with cell death, division, protein concentration, number of organelles, and mutation1. Within the natural environment, it can be nearly impossible to parse the individual impact of these influences on community composition and activity. Obscured by natural structures and buried within a chemical and biological milieu, identifying community members and further resolving their spatiotemporal distribution within the natural environment is extremely challenging. Nonetheless, recent efforts have underscored the importance of spatial organization on community function and point towards the need to account for both member abundance and organization in ongoing studies2,3,4.
It is clear that the local chemical environment (i.e. the availability of nutrients and secondary metabolites), the physical structure (e.g., soil architecture, plant roots, ocean particles, or the intestinal microvilli), the presence or absence of oxygen, and the introduction of pathogenic species all affect the composition, architecture, and function of microbial communities5,6,7,8,9,10,11. Nonetheless, traditional techniques for cultures that neglect to capture these factors continue to prevail. Community composition (e.g., the presence of co-dependent species), physical attachment, signaling molecule concentration, and direct cell-cell contact are all important factors for shaping a microbial community and can be lost in conventional culture conditions. These properties are difficult to replicate in a bulk liquid culture or on an agar plate. The availability of microfluidic, micropatterning, and nanofabrication techniques that allow for the replication of key physical and chemical features of natural environments has, however, enabled many researchers to build bacterial communities to study their interactions12,13,14 and to develop synthetic environments that mimic natural conditions4,15,16,17,18,19,20.
This protocol describes a method to fabricate a microwell array device and provides detailed experimental procedures that can be used to functionalize the wells in the array and to grow bacteria, both as single-species colonies and in multi-member communities. This work also demonstrates how bacteria modified to produce fluorescent reporter proteins can be used to monitor bacterial growth within wells over time. A similar array was presented previously and showed that it is possible to track the growth of single-species colonies of Pseudomonas aeruginosa (P. aeruginosa) in microwells. By modulating well size and seeding density, the starting conditions of thousands of growth experiments can be varied in parallel to determine how the initial inoculation conditions affect the ability of the bacteria to grow21. The current work uses a slightly modified version of the microwell array that builds on the previous work by enabling the simultaneous comparison of multiple arrays and by using a more robust experimental protocol. The array used in this work contains multiple subarrays, or array ensembles, containing wells of different sizes, ranging from 15 – 100 µm in diameter, that are arranged at three different pitches (i.e. 2x, 3x, and 4x the well diameter). The arrays are etched into silicon, and the growth of the bacteria seeded in the silicon arrays is enabled by sealing the arrays with a coverslip that has been coated with a medium-infused agarose gel. P. aeruginosa mutants designed to study the Type VI secretion system are used in this demonstration.
The results presented here build toward the ultimate goal of analyzing multimember communities within microwell arrays, enabling researchers to monitor the abundance and organization of bacteria in situ while controlling and probing the chemical environment. This should ultimately provide insights into the "rules" that govern community development and succession.
1. Silicon Microwell-array Fabrication
2. Bacterial Culture and Seeding (Figure 1a)
Figure 1: Fabrication and Cell Seeding Procedure. (a) Microwell arrays are etched into silicon wafers coated with a thin layer of parylene (i). To wet the wells and/or functionalize the surface, a protein solution is added in a droplet on top of the arrays (ii). The protein solution is removed, the wafers are dried, and a new solution containing the desired bacteria is added (iii). The bacterial solution is removed after an incubation period, and the wafers are allowed to dry, leaving behind bacteria in the wells and on the surface (iv). The surface-associated bacteria are removed with parylene lift-off, leaving behind bacteria seeded cleanly in the microwells and still viable due to the 2% glycerol medium, which helps to keep the wells hydrated(v). The silicon chips are then placed array-side down on an agarose gel-coated glass coverslip, which feeds bacterial growth in the microwells (vi). (b) Layout of sub-arrays on a single silicon device. Each sub-array contains a set of identical wells. The diameter of the microwells across all sub-arrays range in diameter from 5-100 µm and are organized at 2x, 3x, or 4x the well diameter pitch, which is denoted by the white to dark-gray colors on the bottom panel schematic. When the well depths are shallow (<10 µm), the 5 and 10 µm well diameters are rarely useful, generally because of a lack of cells colonizing these very small wells. In this work, only the data from wells with 15-100 µm diameters were analyzed. Please click here to view a larger version of this figure.
NOTE: As shown in Figure 1b, a complete chip contains sub-arrays of wells, with diameters ranging from 5 to 100 µm, with three different pitches (i.e. 2x, 3x, and 4x the diameter) repeating 4 times.
3. Microscope Set-up
4. Preparation of Agarose-coated Glass Coverslips
5. Sealing the Wells with an Agarose-coated Coverslip and Imaging
6. Analysis
The experimental platform presented here is designed for high-throughput and high-content studies of bacterial communities. The design enables thousands of communities, growing in wells of various sizes, to be analyzed simultaneously. With this microwell array design, the dependence of the final community composition on initial seeding densities, well size, and chemical environment can be determined. This work demonstrates the growth of a two-member community in the microwell array and puts forth methods to analyze community composition and organization.
The model system used here was designed in the Mougous lab to study Type VI secretion in P. aeruginosa. The system is composed of several mutant strains, containing either GFP or mCherry fluorescent reporters. In this work used 2 different strains. The first is a GFP-labeled ΔretS mutant that constitutively expresses the toxic effector proteins associated with Type VI secretion, resulting in higher levels of cell death in susceptible cells. The second is an RFP-labeled ΔretSΔtse/i1-6 strain that is a deletion mutant missing all six known effector proteins and that has been shown to be more susceptible to Type VI pathogenesis22,23,24. The growth trajectories of each species were tracked individually and within two member communities (i.e. co-culture within the array platform).
The ΔretS and ΔretSΔtse/i1-6 mutants were seeded onto three arrays, each separately and also mixed together in a 1:2 ratio, and then imaged every 30 min for 20 h. Bacterial growth in the wells ceased between 6 & 12 h postseeding. Examples of the fluorescent image data are shown in Figures 2 and 3. Figure 2 shows growth over time of the ΔretS and ΔretSΔtse/i1-6 strains in 45 µm diameter wells, and Figure 3 shows the state of these same two-member communities about 6 h postseeding (hps) in wells of various sizes.
Figure 2: Sample Images from a Time-course co-culture Growth Experiment. Two mutants of P. aeruginosa, expressing either mCherry (ΔretSΔtse/i1-6) (red) or GFP (ΔretS) (green), are seeded together in a 2:1 mCherry:GFP ratio in the microwell arrays. Pictured here are composite images taken at 3 different times postseeding in the 45 µm diameter wells. Please click here to view a larger version of this figure.
Figure 3: Range of Well Diameters Included in the Microwell Array Device. Representative images of two-member communities captured at 8 h of growth. Generally, mCherry (ΔretSΔtse/i1-6) (red) or GFP (ΔretS) (green) colonies show minimal colocalization, remaining within distinct areas in the microwells. Please click here to view a larger version of this figure.
Prior to the acquisition of the measurements, a sliding parabaloid background subtraction was performed. An illumination correction step was employed so that the signal from each microwell imaged in a given experiment could be compared quantitatively. These methods have been previously described25. Using the microarray plugin in ImageJ, the average intensity of the red and green bacteria was measured in the corrected images, along with a local background signal. The local background signal was subtracted from the well signal, and the GFP signal was adjusted to correct for the autofluorescence of P. aeruginosa co-cultures (autofluorescence determined from mCherry ΔretSΔtse/i1-6-only arrays). Once all these corrections were performed, the quantitative growth trajectories of mCherry- and GFP-expressing bacteria could be plotted. Example growth trajectories from 20 and 50 µm diameter wells are shown in Figure 4.
Figure 4: Growth Curves of Mono- and Co-cultures of P. aeruginosa Type VI Secretion Mutants. (a and d) GFP-ΔretS monoculture growth curves in 20-µm and 50-µm wells, respectively. (b and e) mCherry-ΔretSΔtse/i1-6 monoculture growth curves in 20 and 50 µm wells, respectively. (c and f) Co-culture of a 1:2 ratio mixture of GFP (green) and mCherry (red) P. aeruginosa strains. Growth curves are plotted from when image acquisition began, approximately 2 h postseeding (hps). Please click here to view a larger version of this figure.
The growth trajectories from the mono- and co-culture of P. aeruginosa strains indicate that there is some variability in growth from well to well and that this variability is increased in wells of smaller dimensions. However, the mean intensity is not significantly affected by well size. There does not appear to be a dramatic or clear negative effect on the growth of the susceptible strain (ΔretSΔtse/i1-6) when the ΔretS strain is present. The overall intensity for each strain decreased, but this was due to a lower starting concentration of each strain in the wells, as the total OD was kept consistent across experiments, not the individual OD of each strain. However, it is difficult to determine what factors are most important to understanding the growth of bacteria in these wells simply by looking at growth trajectories alone. Therefore, each trajectory was fit with a modified logistic function26 (Figure 5a) so that relevant parameters could be extracted and used to analyze bacterial growth data. Each trajectory was transformed by taking the natural log of the signal divided by the initial signal. A modified logistic function, with three parameters corresponding to a maximum signal (A), a maximum rate (µ), and a lag time (τ), was used to fit each trajectory.
Given the known interaction between the GFP-ΔretS and mCherry-ΔretSΔtse/i1-6 strains, one might hypothesize that the growth of the mCherry strain would be inhibited in co-cultures with the GFP-ΔretS strain22,24. Using the extracted parameters, it is possible to look for positive or negative correlations between the parameters extracted from GFP and mCherry trajectories. The initial parameter analysis suggested that the co-culture of these species had a negligible effect on their overall growth. The variability seen in the growth curves is likely due to environmental factors, such as initial seeding density, which becomes more variable at smaller well diameters (Figure 5b). Significant negative effects on the mCherry strain due to the presence of GFP-ΔretS were not observed. For example, when plotting the maximum mCherry signal versus the ratio of the initial GFP-to-RFP signal, no decrease in mCherry expression was found when there was a higher ratio of GFP-ΔretS in the wells.
Figure 5: Fluorescent Growth Trajectories Fit with a Modified Logistic Equation. (a) The example curve is normalized and then fit with a modified logistic equation by adjusting three parameters, a maximum intensity (A), a maximum rate (µ), and a lag time (τ). (b) Initial mCherry signal versus initial GFP signal in individual wells of various diameters. (c) The maximum mCherry reporter signal plotted versus the ratio of the initial intensity of the GFP-expressing bacteria to the RFP-expressing bacteria. Please click here to view a larger version of this figure.
Previous work in the Mougous lab indicated that the effects of Type VI secretion require direct cell-to-cell contact between secreting and susceptible cells and that a longer contact period results in more cell lysis23. Therefore, we believe that, in these arrays, the exponential growth of the two mutants is simply out-pacing the contact-mediated pathogenic effects in this model system. However, in wells of all sizes, the GFP- and mCherry-expressing bacteria grow in distinct patches (Figures 2 and 3). Therefore, rather than focusing on growth rates, bacterial communities containing contact-mediated pathogenic actors may be better studied using spatial analyses. For example, rather than focusing on the integrated intensity, which is representative of the total number of red or green cells in a well, this microwell format allows for the number and location of red and/or green pixels to be counted. It is possible to analyze the growth of individual colonies or patches within these wells, focusing on the borders where the red and green signals overlap (Figure 6a). Quantifying co-localization events and nearest-neighbor analyses can then be executed to look more closely at properties like patch size, patch growth rate, and patch overlap within individual wells. Spatial analyses would be aided by higher-magnification imaging and confocal microscopy (Figure 6b).
Figure 6: The Spatial Organization of Different Species may be Analyzed using Epifluorescent and Confocal Microscopy. (a) Series of images from one 100 µm well acquired using a confocal microscope with 20X magnification and NA of 0.8. The white arrows point toward yellow regions where red and green colonies overlap or are mixed. Analyzing how the colonies grow in these regions would be interesting for studying contact-based pathogenesis. (b and c) Confocal images (20X magnification, NA = 0.8) of 15 µm diameter, ~7.5 µm deep wells. Please click here to view a larger version of this figure.
This article presented a microwell array device and experimental protocols designed to enable high-throughput and high-content live-cell imaging-based analysis of bacterial community development. While the focus of the demonstration here was to study the effects of contact-mediated Type VI secretion on community development, the arrays were designed to be flexible and accommodate the study of a broad range of microbial communities and microbe-microbe interactions. The work here focuses solely on the use of bacteria that constitutively express fluorescent markers to allow facile tracking of member abundance and location. However, opportunities to sample biological material from individual wells following growth or perturbation by changing the chemical environment could, in principal, be used to identify community composition and gene expression.
Each silicon device is composed of dozens of arrays containing microwells with various well diameters, organized at either a 2x, 3x, or 4x diameter pitch, and etched to a depth between 3 & 3.5 µm. Because nutrients and secondary metabolites can diffuse freely through and across the agarose lid, different pitch sub-arrays were included to allow for the examination of how signaling between microwells or local nutrient depletion impact community development. In this demonstration, growth and community development did not appear to be impacted by microwell spacing or location within the arrays. The shallow depth was chosen to simplify image analysis, limiting all microbial growth and development to a single image plane. However, deeper wells (20 µm deep) have been used previously and can be easily modulated in depth by varying the duration of the silicon etching process. Increasing the aspect ratio of the wells essentially alters the degree of confinement experienced by the cells in the interior of the wells, effectively changing the ratio of the total well volume to the area through which nutrients diffuse into the top of the well. This variety of array configurations could be used to study the effects of well-to-well or intra-well signaling.
One limitation of using silicon to fabricate the microwells is that it is opaque to visible light. Consequently, routine analyses that rely on transmission of light (e.g., brightfield, phase, or differential interference contrast imaging) cannot be used to monitor growth. Ongoing research and development in our group has focused on adapting this silicon microwell configuration and experimental protocol for work with transparent arrays to allow for higher-resolution fluorescence and brightfield imaging for monitoring and quantification27. By using transparent materials for fabrication, such as SU-8 epoxy on a glass coverslip, it is possible to acquire both brightfield and fluorescence images with objectives operating at shorter working distances. This allows for more optimal resolution to be obtained with water immersion or oil objectives having ≥40X magnification, without the difficulty of imaging through an agar layer, as is the case for silicon arrays.
As described in our prior work, seeding conditions and well geometry impact the seeding of cells within the well arrays. At low seeding densities or in small wells, high levels of variation in the number and types of cells that load into individual wells allows for a broad and rapid exploration of experimental parameter space. Higher seeding densities and/or imaging larger wells yields more consistent ratios of cell types and total inoculum levels, allowing for the analysis of large numbers of experimental replicates. A quick look at images during the earliest stages of growth suggests that cells attach and grow primarily along the edges of the wells (Figure 5a). It is not clear if this is an artifact related to some evaporative drying during the seeding process or suggests a preference for cell attachment to multiple edges. Intentionally altering the topography or surface chemistry of the wells in a deterministic manner could reveal which factors have the greatest impact on cell attachment and biofilm formation.
Growth of the bacteria in these wells is supported by the use of an agarose gel-coated glass coverslip infused with R2A medium. The methods described to create this agarose-coated coverslip ensure an even thickness through which to image and a fairly constant z-position, allowing for the possibility to simultaneously image multiple silicon chips. The use of ultra-pure agarose as the gelling agent and a minimal medium rather than a complete medium results in a cleaner, less autofluorescent background, increasing the overall signal-to-noise ratio. When relying on fluorescence for the quantitative tracking of microbial communities, it is essential to keep imaging conditions consistent and to be acutely aware of contributions from autofluorescence, photobleaching, and non-uniform illumination. In addition, for experiments requiring long time-lapse image acquisition, a thick layer of agarose, such as used here, is desired. Thinner layers (~100 µm thick) can be made to enable higher magnification imaging. However, it is difficult to keep the assembly humid enough to prevent that thin layer from drying out.
Though the initial analysis of growth parameters and initial inoculum levels has not yet revealed any significant correlations for the simple system described here, the protocols and data shown emphasize the depth of information that can be extracted from images of community development over time. Spatial organization within the wells evolves as single-species colonies expand from initial "nucleation" sites. Thus, factors associated with the initial attachment density (i.e. stemming perhaps from the greater preference of one species for the well surface compared to another) or growth rate of individual patches could dramatically influence community development. Subtler interactions, such as those mediated by contact between species (patches) established very late in the growth process may require more in-depth analysis of the image data.
The microwell array platform and protocol presented here facilitate the rapid assembly and high-content analysis of microbial community development and microbe-microbe interactions. Future work that allows for the control of the local chemical environment, by changing the composition of medium within the agar lid or by directly modulating and sampling via integrated microfluidics, should allow for experimental protocols that move beyond mimicking natural confinement and niche size and begin to explore the impact of dynamically changing environments, similar to those found in nature. Future work will augment the array design with integrated microfluidics that can be used to dynamically modulate the local chemical environment and to sample fluid from individual wells. Additional work on the development of optically clear microwell arrays that can be used for higher-resolution and brightfield tracking of community growth has been described elsewhere27.
The authors have nothing to disclose.
Microwell arrays were fabricated and characterized at the Center for Nanophase Materials Sciences User Facilities Division, Office of Basic Energy Sciences, U.S. Department of Energy. Financial support for this work was provided through the Oak Ridge National Laboratory Director's Research and Development Fund. The authors would also like to thank the J. Mougous Laboratory (University of Washington, Seattle, WA) for the supply of P. aeruginosa strains used in these studies.
Parylene N | Specialty Coating Systems | CAS NO.:1633-22-3 |
Parylene coater | Specialty Coating Systems | Labcoter 2 Parylene Deposition Unit PDS2010 |
Silicon Wafer | WRS Materials | 100mm diameter, 500-550μm thickness, Prime, 10-20 resistivity, N/Phos<100>, |
adhesion promoter | Shin-Etsu Microsci | MicroPrime P20 adhesion promoter |
postive tone photoresist | Rohm and Haas Electronics Materials LLC (Owned by Dow) | Microposit S1818 Positive Photoresist (code 10018357) |
Quintel Contact Aligner | Neutronix Quintel Corp | NXQ 7500 Mask Aligner |
Reactive Ion Etching Tool | Oxford Instruments | Plasmalab System 100 Reactive Ion Etcher |
R2A Broth | TEKnova | R0005 |
Bovine Serum Albumin | Sigma | A9647 |
Multimode Plate Reader | Perkin Elmer | Enspire, 2300-0000 |
Fluorescent Microscope | Nikon | Eclipse Ti-U |
Automated Stage | Prior | ProScan III |
CCD camera | Nikon | DS-QiMc |
Stage-top environmental control chamber | In Vivo Scientific | STEV ECU-HOC |
Phosphate Buffered Saline | ThermoFisher Scientific | 14190144 |
UltraPure Agarose | ThermoFisher Scientific | 16500500 |
25 x 75 mm No. 1.5 coverslip | Nexterion | High performance #1.5H coverslips |
Fluorescence Reference Slides | Ted Pella | 2273 |
Physical Stylus Profilometer | KLA Tencor | P-6 |
lab wipes | Kimberly Clark | Kimipe KIMTECH SCIENCE Brand, 34155 |
commercial software | Nikon | NIS Elements |
Zeiss 710 Confocal Microscope | Zeiss | |
filter cubes | Nikon | Nikon FITC (96311), Nikon Texas Red(96313) |