Nutrient-responsive regulation determines biodiversity in a colicin-mediated bacterial community

Abstract

Background: Antagonistic interactions mediated by antibiotics are strong drivers of bacterial community dynamics which shape biodiversity. Colicin production by Escherichia coli is such an interaction that governs intraspecific competition and is involved in promoting biodiversity. It is unknown how environmental cues affect regulation of the colicin operon and thus influence antibiotic-mediated community dynamics.

Results: Here, we investigate the community dynamics of colicin-producing, -sensitive, and -resistant/non-producer E. coli strains that colonize a microfabricated spatially-structured habitat. Nutrients are found to strongly influence community dynamics: when growing on amino acids and peptides, colicin-mediated competition is intense and the three strains do not coexist unless spatially separated at large scales (millimeters). Surprisingly, when growing on sugars, colicin-mediated competition is minimal and the three strains coexist at the micrometer scale. Carbon storage regulator A (CsrA) is found to play a key role in translating the type of nutrients into the observed community dynamics by controlling colicin release. We demonstrate that by mitigating lysis, CsrA shapes the community dynamics and determines whether the three strains coexist. Indeed, a mutant producer that is unable to suppress colicin release, causes the collapse of biodiversity in media that would otherwise support co-localized growth of the three strains.

Conclusions: Our results show how the environmental regulation of an antagonistic trait shapes community dynamics. We demonstrate that nutrient-responsive regulation of colicin release by CsrA, determines whether colicin producer, resistant non-producer, and sensitive strains coexist at small spatial scales, or whether the sensitive strain is eradicated. This study highlights how molecular-level regulatory mechanisms that govern interference competition give rise to community-level biodiversity patterns.

Figure 1

Community dynamics in spatially structured microhabitats. (A,B) Schematic (A) and cartoon (B) showing a microhabitat consisting of 85 patches (100 × 100 × 15 μm) connected by corridors (50 × 5 × 15 μm). All 85 patches are connected to growth-medium reservoirs by 180 nm deep slits preventing bacteria from entering the reservoirs but allowing the diffusion of e.g. nutrients and waste. Microhabitats are sealed using a cover slip. (C-H) Kymographs (representing space horizontally and time vertically) of representative sections of microhabitat experiments and the corresponding growth curves in LB (C,F), M9-amino acids (D,G), and M9-glucose (E,H). In all three media, producer (shown in blue) and resistant cells (red) grow readily. Sensitive cells (green) show no growth in LB and M9-amino acids, and sometimes turn filamentous (visible in LB in the connector between the center and rightmost patch). In M9-glucose however, sensitive cells show exponential growth (green line in (H)) in close proximity to producer and resistant cells.

Figure 2

Growth and biodiversity through time in different media. (A-C) Mean habitat-wide per-capita growth rates through time of producer (blue), resistant (red), and sensitive (green) cells calculated from three replicate microhabitat experiments per medium, the dashed line indicates the standard error of the mean (SEM). Insets show the initial phase in detail and demonstrate that the growth rate of the sensitive strain fluctuates around zero in LB (A) and M9-amino acids (B) but is positive in M9-glucose (C). (D-F) Biodiversity through time is calculated at the level of an entire habitat as Shannon’s index ($H=-\sum _{i=1}^3{p}_{i}ln{p}_i$, where p _i denotes the frequency of strain i), full lines indicate the mean of three independent experiments, dashed lines indicate the SEM. The dashed line at ln(2) depicts maximum diversity for 2 types.

Figure 3

Colicin production and release depend on growth phase and medium. (A-D) Mean histograms of E2crimson expression measured using flow cytometry at various stages of growth of duplicate experiments in LB (A), M9-amino acids (B), M9-glucose (C), and M9-glucose + amino acids (D). E2crimson serves as a proxy for colicin production. Colors of histograms correspond to the color-coded time-points on the corresponding growth curves (above). The first bin of all histograms (cells not expressing E2crimson) is not included for clarity. The population fraction that expresses E2crimson (i.e. the area under the curve) increases along the growth curve in all media, the expression distribution has a distinct profile for the four media. (E-H) YFP expression histograms in LB (E), M9-amino acids (F), M9-glucose (G), and M9-glucose + amino acids (H) throughout the growth phase of a strain carrying plasmid pColE2 Δ cel::EYFP. YFP expression is a proxy for lysis protein production.

Figure 4

Colicin release in LB and M9-glucose. Representative microscopy images and induction-lysis curves of cells growing in LB (A,C) and M9-glucose (B,D) expressing E2crimson under control of the colicin E2 promoter (shown in red) and having constitutive yfp expression (green). Panel (C) shows induction-lysis curves of three cells labeled in image t ₁ of (A). E2crimson first rises, 2–3 hours after induction cells show a sharp drop in fluorescence indicating a lysis event. A faint red signal shows the remainder of the cell body after lysis. E2crimson time-traces in M9-glucose (D) of the cells labelled in (B) show that induced cells do not lyse and remain intact for periods longer than 12 hours.

Figure 5

Relief of CsrA repression prevents colicin build-up through increased lysis. Expression histograms of E2crimson through the growth phase of producer cells carrying the mutant pColE2-TT plasmid grown in LB (A), M9-amino acids (B), M9-glucose (C), and M9-glucose + amino acids (D), measured using flow cytometry. Histogram colors correspond to time points along the corresponding growth curve (above), grey lines represent histograms of producers carrying the wild-type plasmid (data from Figure 3A-D). Expression profiles of producers harboring the mutant plasmid have shifted to lower expression values and show a decreased fraction of induced cells when compared to expression distributions of producers carrying the wild-type plasmid.

Figure 6

Relief of CsrA repression increases lysis and results in loss of biodiversity in M9-glucose. (A) Kymograph of a section of a microhabitat showing typical community dynamics of sensitive (green), resistant (red) and pColE2-TT carrying producer cells (blue). Producer and resistant cells grow readily while sensitive cells either show no growth, become filamentous, or die. (B) Corresponding growth curves (time versus occupancy) of the section shown in (A). (C) Microscopy images and induction-lysis curves of producer cells carrying the pColE2-TT mutant plasmid, the E2crimson reporter for colicin induction (shown in red), and expressing YFP constitutively (green) growing in M9-glucose. Induction-lysis curves of the cells labelled in the microscopy images show a steep drop in fluorescence intensity after induction, indicating lysis.