The effects of chemical interactions and culture history on the colonization of structured habitats by competing bacterial populations

Abstract

Background: Bacterial habitats, such as soil and the gut, are structured at the micrometer scale. Important aspects of microbial life in such spatial ecosystems are migration and colonization. Here we explore the colonization of a structured ecosystem by two neutrally labeled strains of Escherichia coli. Using time-lapse microscopy we studied the colonization of one-dimensional arrays of habitat patches linked by connectors, which were invaded by the two E. coli strains from opposite sides.

Results: The two strains colonize a habitat from opposite sides by a series of traveling waves followed by an expansion front. When population waves collide, they branch into a continuing traveling wave, a reflected wave and a stationary population. When the two strains invade the landscape from opposite sides, they remain segregated in space and often one population will displace the other from most of the habitat. However, when the strains are co-cultured before entering the habitats, they colonize the habitat together and do not separate spatially. Using physically separated, but diffusionally coupled, habitats we show that colonization waves and expansion fronts interact trough diffusible molecules, and not by direct competition for space. Furthermore, we found that colonization outcome is influenced by a culture's history, as the culture with the longest doubling time in bulk conditions tends to take over the largest fraction of the habitat. Finally, we observed that population distributions in parallel habitats located on the same device and inoculated with cells from the same overnight culture are significantly more similar to each other than to patterns in identical habitats located on different devices inoculated with cells from different overnight cultures, even tough all cultures were started from the same -80°C frozen stock.

Conclusions: We found that the colonization of spatially structure habitats by two interacting populations can lead to the formation of complex, but reproducible, spatiotemporal patterns. Furthermore, we showed that chemical interactions between two populations cause them to remain spatially segregated while they compete for habitat space. Finally, we observed that growth properties in bulk conditions correlate with the outcome of habitat colonization. Together, our data show the crucial roles of chemical interactions between populations and a culture's history in determining the outcome of habitat colonization.

Figure 1

Colonization of spatially structured synthetic ecosystems. (A) Device of type-1 with 5 parallel habitats (habitats 1 to 5 from top to bottom), each consisting of 85 patches, with separate inlets. Red cells are inoculated on the right (indicated by red inlet holes) and green cells on the left (green inlet holes). (B) Device of type-2 with a single, shared, inlet. Except for the inlet, devices in A and B are identical. (C) Enlarged schematic view of the devices shown in A and B showing an array of patches of 100 × 100 × 5 μm³ linked by connectors of 50 × 5 × 5 μm³. Note that the bacteria are not to scale. (D) Kymograph of fluorescence intensity of the left most 25 patches for strain JEK1036 (green) showing a typical pattern of landscape invasion consisting of three subsequent colonization waves (α at t ≈ 3.5 h, β at t ≈ 5 h and γ at t ≈ 6 h) followed by the expansion front (at t ≈ 6 h); scale bar = 1 mm. The inset at the top shows an enlarged view of the α wave just after entering the habitat from the inlet; scale bar = 100 μm.

Figure 2

The collisions of colonization waves. (A) Occupancy measure (area fraction) calculated per patch for strains JEK1037 (red) and JEK1036 (green) showing the collision between two α waves (at t = 6 h, patch 54). Note how both waves branch: a part of the wave is reflected, a part forms a stationary population, and a part continuous (for a short distance) in the same direction. (B) Kymograph of fluorescence intensity for the collision shown in A. (C) Enlarged view of B, centered at the point of collision. Note how the red and green populations remain largely segregated in space, even though individual cells do mix with the other population. (D) Kymograph of fluorescence intensity of a collision in a different habitat in the same device (with separate inlets; type-2) as the habitat shown in A-C. Note the similarity between B and D. (E) Enlarged view of D, centered at the point of collision.

Figure 3

Decomposition of colliding colonization waves. The top row shows kymographs of fluorescence intensity, the second row shows occupancy levels for strain JEK1037 (red), the third row the occupancy levels for strain JEK1036 (green), and the bottom row the post-collision distributions of bacteria over the reflected, stationary and refracted components (from left to right for green and from right to left for red), as determined from the occupancy distribution 1 hour after the collision. Examples where: (A) Both waves have large reflected parts. (B) Red wave forms a stationary population. (C) Most of the red wave is refracted. Also note how a combined wave (yellow, in top row) is formed when the red β wave collides with a stationary green population (t = 6.5 h, patch 50).

Figure 4

Interactions between expansion fronts. (A) Kymograph of fluorescence intensity for a habitat where a stable boundary is observed. (B) Enlarged view of panel A, for the 6 patches centered at the interface between the green and red populations at t = 19 h. (C) Enlarged view of the 6 patches at the left end of the habitat shown in A at t = 19 h. A few red cells are indicated by the white arrows in the inset. (D) Enlarged view of the 6 patches at the right end of the habitat shown in A at t = 19 h. (E) Kymograph of fluorescence intensity where the green population is expelled from the habitat by the red population, before the two fronts come into physical contact. (F) Kymograph of fluorescence intensity where the green population is expelled from the habitat by the red population, the inset shows that there has not been any physical contact between red cells and the green front before the latter changes direction. Note how the leading edge of the green front and the high density region, roughly 8 patches behind the leading edge, change direction almost simultaneously (G) Kymograph of fluorescence intensity for a habitat inoculated at both sides with a single mixed culture of strains JEK1036 (green) and JEK1037 (red), note that the change in color from red to green with increasing time is mostly due to changes in the fluorescence intensity per cell of RFP compared to GFP and not due to de-mixing of the strains as can be seen by comparing their occupancy patterns (Additional file 7A).

Figure 5

Interactions between chemically coupled, but physically separated populations. (A) Schematic of a microfabricated device of type-3, consisting of two parallel habitats (each of 85 patches) chemically coupled by 200 nm deep nanoslits of 15 × 15 μm, which allow for the diffusion of molecules but are too shallow for bacteria to pass through. (B) Area fraction occupied per patch (occupancy) for the top and bottom habitats, the top habitat is inoculated from the right and the bottom habitat from the left with the same initial culture of strain JEK1036 (green). (C) Kymograph where the fluorescence intensities of the top and bottom habitats are superimposed: cells in the top habitat are shown in red and cells in the bottom habitat in green. Note that both habitats are inoculated from the same (JEK1036) culture and that the bacteria in the upper and lower habitats are spatially confined to their own habitat.

Figure 6

Similarity of spatiotemporal patterns for habitats inoculated with same cultures. Kymographs show the fluorescence intensity of strains JEK1036 (green; inoculated from the left at t = 0 h) and JEK1037 (red; inoculated from the right at t = 0 h). (A) Five parallel habitats in the same device (type 1) with separate inlets, each kymograph shows the spatiotemporal pattern of a single habitat. (B) Habitat on a different device inoculated with a different set of initial cultures (with separate inlets; type-1) than in panel A. (C) Habitat in a device (type-2) with a shared inlet. Note the similarity between the patterns of the five habitats in panel A (all inoculated with the same initial cultures), compared to the patterns of the habitats in panels B and C (inoculated with different cultures than the habitats in A).