Stochastic Assembly of Bacteria in Microwell Arrays Reveals the Importance of Confinement in Community Development

Abstract

The structure and function of microbial communities is deeply influenced by the physical and chemical architecture of the local microenvironment and the abundance of its community members. The complexity of this natural parameter space has made characterization of the key drivers of community development difficult. In order to facilitate these characterizations, we have developed a microwell platform designed to screen microbial growth and interactions across a wide variety of physical and initial conditions. Assembly of microbial communities into microwells was achieved using a novel biofabrication method that exploits well feature sizes for control of innoculum levels. Wells with incrementally smaller size features created populations with increasingly larger variations in inoculum levels. This allowed for reproducible growth measurement in large (20 μm diameter) wells, and screening for favorable growth conditions in small (5, 10 μm diameter) wells. We demonstrate the utility of this approach for screening and discovery using 5 μm wells to assemble P. aeruginosa colonies across a broad distribution of innoculum levels, and identify those conditions that promote the highest probability of survivial and growth under spatial confinement. Multi-member community assembly was also characterized to demonstrate the broad potential of this platform for studying the role of member abundance on microbial competition, mutualism and community succession.

Fig 1. Microwell array fabrication and design.

(A) Microwell fabrication process: (i,ii) Positive photoresist is patterned over parylene-coated silicon wafers using conventional photolithography. (iii) Dry etching is then used to etch parylene and then silicon to the desired well depth. (iv) The well surface is then modified with a protein layer then (v) a solution of bacterial cells. (vi) Parylene is removed from the substrate and (vii) the substrate is contacted with agar-coated coverslips loaded with the desired chemical media. (B) Dry lift-off procedure involving peel-off of the parylene mask (step vi). (C) Layout of a combinatorial microwell array substrate.

Fig 2. Characterization of the physicochemical features of microwell arrays.

(A) SEM image of an array of 2μm diameter microwells and of an individual well after substrate cleavage (inset). (B) SEM images of a 20μm diameter microwell array. (C) Fluorescent image of microwells after functionalization with WGA-A488 and dry lift-off. The arrow highlights a portion of un-peeled parylene, also containing adsorbed WGA-A488. (D) Fluorescent line plot of WGA-A488 coated wells corresponding to the red line in 2C.

Fig 3. The distribution of bacteria seeded in microwell arrays is guided by well diameter.

(A) Mosaic 10X false-color fluorescent image of a combinatorial microwell array after seeding E. coli-GFP at OD₆₀₀ = 0.3 and dry lift-off to remove background cells. The false color scale denotes fluorescent signal intensities indicative of cell densities. (B) Averaged well fluorescence intensities ± standard deviation measured from individual wells within each array (black line) and CV_array (red dashed line), the standard deviation divided by the average fluorescent signal for each well diameter.

Fig 4. Bacterial well populations follow a Poisson distribution.

(A) 20X false color fluorescent images of 5 μm diameter wells seeded with E.coli-GFP at OD₆₀₀ = 0.01, 0.1, and 1.0. (B) Probability distributions for cell populations at the varied seeding concentrations. Diamonds represent data and solid lines represent a Poisson distribution fit to the data according to eq 2. Seeding at an OD₆₀₀ of 0.01, 0.1, and 1.0 resulted in a λ value of 1.9, 6.2, and 68.6, respectively, and an A value of 1.65, 1.35, and 1.00, respectively.

Fig 5. P. aeruginosa growth trajectories in 5 and 20 μm diameter microwell arrays.

(A) Top: False-color fluorescent images of growth in 20 μm diameter arrays. Bottom: Corresponding growth trajectories. The dashed red trajectory indicates growth in an outlier well. (B) Top: False-color fluorescent images of growth in 5 μm diameter arrays. Solid black trajectories denote wells where growth and colonization occurred, dashed red trajectories denote wells where decay and extinction occurred. Data is representative of 4 independent growth experiments.

Fig 6. Growth of P. aeruginosa in confined volumes depends on inoculum levels.

(A) Scatter plots of initial and final (t = 24 hrs) cell volume fraction in 10 μm diameter wells and (B) 5 μm diameter wells. Growth-decay line deciphers wells that increased or decreased in cell numbers over the incubation period. (C) Probability of well colonization with initial volume fraction of seeded cells for in 10 μm diameter and (D) 5 μm diameter wells. Data was taken from n = 256 wells for 10 μm diameter arrays and n = 840 wells for 5 μm diameter arrays from 4 independent growth experiments.

Fig 7. Multi-member bacterial communities can be assembled at low or high dispersion.

(A) Low dispersion pairing: Seeding a 1:9 mixture of E. coli-mCherry (red) and E. coli-GFP (green) at an overall OD₆₀₀ of 0.4 into 40 μm diameter microwell arrays. (B) High dispersion pairing: Seeding a 1:1 mixture of E. coli-mCherry and E. coli-GFP into 2 μm diameter arrays at an overall OD₆₀₀ of 1.0. (C) Scatter plot of GFP and mCherry signals after low or high dispersion pairing.