The weak shall inherit: bacteriocin-mediated interactions in bacterial populations

Abstract

Background: Evolutionary arms race plays a major role in shaping biological diversity. In microbial systems, competition often involves chemical warfare and the production of bacteriocins, narrow-spectrum toxins aimed at killing closely related strains by forming pores in their target's membrane or by degrading the target's RNA or DNA. Although many empirical and theoretical studies describe competitive exclusion of bacteriocin-sensitive strains by producers of bacteriocins, the dynamics among producers are largely unknown.

Methodology/principal findings: We used a reporter-gene assay to show that the bacterial response to bacteriocins' treatment mirrors the inflicted damage Potent bacteriocins are lethal to competing strains, but at sublethal doses can serve as strong inducing agents, enhancing their antagonists' bacteriocin production. In contrast, weaker bacteriocins are less toxic to their competitors and trigger mild bacteriocin expression. We used empirical and numerical models to explore the role of cross-induction in the arms race between bacteriocin-producing strains. We found that in well-mixed, unstructured environments where interactions are global, producers of weak bacteriocins are selectively advantageous and outcompete producers of potent bacteriocins. However, in spatially structured environments, where interactions are local, each producer occupies its own territory, and competition takes place only in "no man's lands" between territories, resulting in much slower dynamics.

Conclusion/significance: The models we present imply that producers of potent bacteriocins that trigger a strong response in neighboring bacteriocinogenic strains are doomed, while producers of weak bacteriocins that trigger a mild response in bacteriocinogenic strains flourish. This counter-intuitive outcome might explain the preponderance of weak bacteriocin producers in nature. However, the described scenario is prolonged in spatially structured environments thus promoting coexistence, allowing migration and evolution, and maintaining bacterial diversity.

Figure 1. Colicin toxicity and cross-induction.

The relationship between colicins titer and the expression these colicin extracts triggered in the reporter strains (Table 1) were tested. The colicins’ titer correlated with the measured light emitted by the reporter strains. Results are presented as average with standard deviation of the light emission of all reporters tested. Each experiment was preformed in duplicates and repeated at least three times.

Figure 2. Community dynamics of bacteriocin producers in an unstractured environment.

Competitions between bacteriocin producers in an unstructured environment were tested (A)Competition in unstrauctured habitat cwas tested empirically and (B) numerically. Both simulations demonstrate that competition between equal bacteriocinogenic strains (in this case mutually triggering mild bacteriocin expression) resolves in the slightly more potent bacteriocinogenic strain prevailing when initial concentrations are equal. (A) We followed the fluorescently labeled ColA over time to illustrate the competition between two pore formers (ColA and ColK) both mild inducers of colicin expression. ColK, the producer of a slightly more potent colicin, challenged ColA at various initial frequencies. At the higher starting frequencies ColK outcompeted ColA (note that ColA fluorescence did not increase over time), while at lower starting frequencies ColA outcompeted ColK. Data points are the average of three independent measurements. (B) Time evolution is illustrated by the bacteriocin producers strains A (blue line) and B (red line); both strains are mild inducers but bacteriocin B is slightly more potent. The strains were simulated to compete at equal initial frequencies and the more potent strain prevailed.

Figure 3. Community dynamics of bacteriocin producers in an unstractured environment.

Competitions between bacteriocin producers in an unstructured environment were tested (A) empirically and (B) numerically. Both simulations demonstrate that competition between unequal bacteriocinogenic strains (one triggering mild and the other strong bacteriocin expression) resolves in the mild bacteriocinogenic strain always prevailing. (A) We followed the fluorescently labeled ColA over time to illustrate the competition between a pore former (ColA) and a DNase (ColE7) a mild and strong inducers of colicin expression, respectively. ColE7, the producer of a potent colicin, challenged ColA at various initial frequencies and was always outcompeted. At the higher starting frequencies ColA’s fluorescence was halted for a while (6 hr) but then it increased to its maximum. At lower starting frequencies, ColA outcompeted ColE7 at the onset. Data points are the average of three independent measurements. (B) Time evolution is illustrated by the bacteriocin producers strains A (blue line) and B (red line), mild and strong inducers, respectively, while strain B also produces a potent toxin. The strains were simulated to compete at equal initial frequencies and the weak inducer strain A prevailed.

Figure 4. Frequency-dependent community dynamics of bacteriocin producers in an unstructured environment.

We initialized the system such that the sum of the densities was constant, , and the initial relative frequency varied _. We observed a critical frequency, ρc, above which u₁ eventually dominates. ρc was plotted for a range of toxicity (γ_A) and induction ability (l _A) values. As expected, ρc increased with both γ_A and l _A. Parameters: γ_B = 2, l _A = 40.

Figure 5. Community dynamics of bacteriocin producers in a structured environment.

Competitions between bacteriocin producers in a structured environment were tested (A, B) numerically and (C) empirically. (A) Snapshot of local competition between two bacteriocinogenic strains. One, u_A, dominates in some regions in space, while its competitor, u_B, dominates in others. These regions are separated by fronts that are moving further away from the dominant strain, u_B. Bacteriocin concentrations are higher near the fronts due to mutual induction. In (B), the winner of the pairwise competition was plotted for various values of the test strain induction, l _A, and sensitivity, γ₁. Blue indicates domination of strain A, u_A, and red indicates domination of its’ competitor u_B. (C) A static plate environment was initiated by randomly depositing 24 droplets from pure cultures of the colcinogenic strains ColA and ColK. The changing spatial pattern of the community was documented and the mean area of each strain’s coverage of the plate surface calculated. The aerial coverage of the strains was shown to remain invariable over time. Data points are average of two independent measurements and the bars represent the deviation from the average.