Dishonest Signaling in Microbial Conflicts

Abstract

Quorum sensing is a cell-cell communication system that bacteria use to express social phenotypes, such as the production of extracellular enzymes or toxins, at high cell densities when these phenotypes are most beneficial. However, many bacterial strains are known to lack a sensing mechanism for quorum signals, despite having the gene responsible for releasing the signals to the environment. The aim of this article is 2-fold. First, we utilize mathematical modeling and signaling theory to elucidate the advantage that a bacterial species can gain by releasing quorum signals, while not being able to sense them, in the context of ecological competition with a focal quorum sensing species, by reducing the focal species' ability to optimize the timing of expression of the quorum sensing regulated phenotype. Additionally, the consequences of such "dishonest signaling," signaling that has evolved to harm the signal's receiver, on the focal quorum sensing species are investigated. It is found that quorum sensing bacteria would have to incur an additional, strategic, signaling cost in order to not suffer a reduction in fitness against dishonest signaling strains. Also, the concept of the Least Expensive Reliable Signal is introduced and applied to study how the properties of the regulated phenotype affect the metabolic investment in signaling needed by the quorum sensing bacteria to withstand dishonest signaling.

Keywords: bacterial social interactions; individual based modeling; mathematical modeling; quorum sensing; signaling theory.

Figure 1

A schematic representation of the model: two bacterial strains, a toxin producer P and a sensitive strain S, are in a competition over nutrient, N. P uses QS regulation to regulate its toxin production; As P grows, it emits quorum signals into its environment. When the quorum sensing molecules concentration exceeds a certain threshold, indicated here by Q_th, P switches on its production of toxin T. S is a microbial strain which competes over the same nutrient, N, and is sensitive to the toxin T. In this model, S also produces the same quorum sensing molecules as P to the medium; however, in contrast to P, it does not sense the quorum signals concentration in its environment. The absence of a signal-sensing mechanism at S implies that the quorum signals produced do not act as a source of information for the said strain. However, these signals still interfere with the functioning of the QS system of the P strain. In that sense, the signals produced by S can be termed “dishonest signals,” from signaling theory perspective, as they do not serve to convey true information that is beneficial to the signal receiver.

Figure 2

An illustration of the role played by QS regulation of costly traits in deciding the fate of microbial conflicts. (A) The evolution of the density of the QS toxin producing strain and the sensitive strain in well-mixed environment, where toxin production is constitutive. (B) The evolution of the density of the QS toxin producing strain and the sensitive strain in well-mixed environment, where toxin production is regulated by QS to start at a high bacterial density. (C) The relationship between the time at which toxin production starts by the QS toxin producing strain and its fitness when competing with the sensitive strain, where the fitness of a given strain is defined as the proportion of the said strain in the population by the end of the simulation. Time zero corresponds to constitutive toxin production.

Figure 3

Producing dishonest signals by a sensitive strain could be an advantageous strategy when competing with a focal QS strain. However, this will depend on the metabolic cost of the signaling system. (A) The time evolution of the focal QS toxin producing strain in a competition with a sensitive strain, expressed as a fraction of the total population, under different levels of dishonest signals production from the sensitive strain, when the QS regulation system of the focal strain is metabolically cheap ($q_P=1\times 10^{-5}$, γ_th = 0.002mg/l). (B) The time evolution of the focal QS toxin producing strain in a competition with a sensitive strain, under different levels of dishonest signals production from the sensitive strain, when the QS regulation system of the focal strain is metabolically expensive (q_P = 0.004, γ_th = 0.8mg/l). (C) The time at which toxin production starts by a focal QS toxin producing strain, as a function of the ratio of the dishonest signals by produced by the sensitive strain to the self-signals produced by the focal strain.

Figure 4

The effect of the metabolic cost of a QS regulation system on its reliability when the focal QS bacteria is in a competition with a dishonest signals producer. (A) The final fitness of a focal QS strain under different levels of dishonest signals production by the sensitive strain, when the QS regulation system of the focal strain is metabolically cheap. (B) The final fitness of a focal QS strain under different levels of dishonest signals production by the sensitive strain, when the QS regulation system of the focal strain is metabolically expensive. (C) The time at which toxin production starts by the said strain the fraction of investment in quorum signals on behalf of the focal QS toxin producing strain and the optimal threshold for activation. There is a minimum metabolic investment in quorum signals production which gives rise to a reliable signaling system.

Figure 5

A linear relationship exists between the least expensive reliable signal of a QS toxin producing strain and its fraction of metabolic energy invested in toxin production, which can be explained by the increasing slope of the fitness-toxin production time curve under increasing levels of investment in toxin production. (A) The relationship between the least expensive reliable signal and the fraction of metabolic energy invested in toxin production by a focal QS toxin producing strain. (B) The final fitness of a QS toxin producing strain, expressed as its proportion in the population at the end of a simulation, vs. the time at which toxin production starts by the said strain under different values of the fraction of metabolic energy invested in toxin production.

Figure 6

The Least Expensive Reliable Signal to protect the QS regulation system of a QS toxin producing strain is insensitive to increasing toxin lethality. This is due to the little variation observed in the slope of the fitness-toxin production time curve under increasing levels of toxin lethality. (A) The relationship between the least expensive reliable signal and the lethality of the toxin produced by a QS toxin producing strain. (B) The final fitness of a QS toxin producing strain, expressed as its proportion in the population at the end of a simulation, vs. the time at which toxin production starts by the said strain under different values of toxin lethality.

Figure 7

The relationship between the time at which toxin production starts by the QS toxin producing strain and its fitness when competing with the sensitive strain, where the fitness of a given strain is defined as the proportion of the said strain in the population by the end of the simulation, in case of spatial competition in a biofilm. Time zero corresponds to constitutive toxin production.

Figure 8

Spatial competition, cheap signaling (lower than LERS). Similar to the observations in the well-mixed case, the sensitive strain gain advantage by engaging in dishonest signaling when competing with a focal QS toxin producing strain. (A) An individual-based model for a competition between a focal QS toxin producer (green) and a sensitive strain (red), under different dishonest signaling levels from the sensitive strain. (B) The total quantity of toxin produced as a function of time in the simulations. (C) The proportion of the focal toxin producing strain at the end of the experiments under different levels of dishonest signaling.

Figure 9

Spatial competition, expensive signaling (higher than LERS). As observed in the well-mixed scenario, while the production of dishonest signals can lead to the premature expression of toxin by the focal toxin producer, the cost of this strategy becomes too high for the dishonest signals producing sensitive strain. (A) An individual-based model for a competition between a focal QS toxin producer (green) and a sensitive strain (red), under different dishonest signaling levels from the sensitive strain. (B) The total quantity of toxin produced as a function of time in the simulations. (C) The proportion of the focal toxin producing strain at the end of the experiments under different levels of dishonest signaling.

Figure 10

The effect of having multiple QS circuits to regulate the expression of the toxin by the focal strain. The figure depicts the proportion of the focal toxin producing strain by the end of the simulation against increasing levels of dishonest signaling from the sensitive strain on one of the signals. The information from the two parallel QS circuits by the focal strain are combined either in an (A) “OR” gate: the expression of the trait happens when any of the two signals crosses a threshold concentration. (B) “AND” gate: the expression of the trait requires that both signals cross their corresponding threshold concentrations. (C) Weighted sum: both signals contribute equally to the expression of the trait.