Stability of multispecies bacterial communities: signaling networks may stabilize microbiomes

Abstract

Multispecies bacterial communities can be remarkably stable and resilient even though they consist of cells and species that compete for environmental resources. In silico models suggest that common signals released into the environment may help selected bacterial species cluster at common locations and that sharing of public goods (i.e. molecules produced and released for mutual benefit) can stabilize this coexistence. In contrast, unilateral eavesdropping on signals produced by a potentially invading species may protect a community by keeping invaders away from limited resources. Shared bacterial signals, such as those found in quorum sensing systems, may thus play a key role in fine tuning competition and cooperation within multi-bacterial communities. We suggest that in addition to metabolic complementarity, signaling dynamics may be important in further understanding complex bacterial communities such as the human, animal as well as plant microbiomes.

Figure 1. Experimentally observed sharing of bacterial signals and public goods in olive knot disease.

Pseudomonas savastanoi and Pantotea aggolomerans produce and perceive the same acyl-homoserine lactone signals, C6-3-oxo-HSL (C63O) and C8-3oxo-HSL (C83O), which is an example of symmetrical sharing. On the other hand, Pantotea agglomerans uses two different signals, C6-HSL (C6) and C4-HSL (C4), one or both of which are perceived (“exploited”) by P. savastanoi. All three species produce indolacetic acid (IAA), which is a public good that causes the plant host mobilize nutrients for the bacteria (based on .).

Figure 2. Scenarios for sharing signals and public goods in quorum sensing.

Scheme 1: Symmetrical sharing. The two species, A and B, can both utilize the signals and public goods of the other species. Scheme 2: Asymmetrical sharing. Species B can utilize the signals and public goods of Species A, but not vice versa. The circular arrows indicate that each species is capable of utilizing its own signals and public goods.

Figure 3. Competition outcomes observed with two competing QS agent populations (filled and non-filled circles).

A) Stable, mixed community of two species (colocalization). Both types of cells are in the active, swarming state. B) Winning. The winner population forms a stable, swarming community (filled cells on top) while the loosing species (non-filled cells, near the starting position) will form a small community that will either stagnate in the solitary state, or die out, depending on the nutrients available. C1) Segregating populations. The species indicated with filled-dots is nearer to the resources, i.e. to the region of intact nutrients. C2) Patch-wise (mosaic-like) segregation. In the dfferent patches, either one or the other species is nearer to the resources.

Figure 4. Competition of agent populations without QS.

These systems lack signals and public goods, so the parameter space has only one variable, nutient sharing (denoted c in Methods). Relative fitness is defined in relation to the growth of the same species growing alone in the same conditions (eqn. 4, Methods). At lower nutrient sharing values the populations segregate. At higher nutrient sharing values, one of the populations goes extinct in less than 500 generations. When segregation and exclusion are stochastic, either species can be the winner or the loser with equal probabilities. Symmetrical sharing of nutrients (bottom curve) means that the two populations are equivalent, and their fitness decreases as nutrient sharing increases. Asymmetrical sharing of nutrients means that the exploiter species (top curve) can consume the nutrients of the exploited species (middle curve) but not vice versa. Note that the curve of the exploited species in asymmetrical sharing is virtually identical with the curve of the symmetrically sharing species. The values are the average of 10 calculations, error bars represent the standard deviation of the mean.

Figure 5. Sharing.

Competition of species A and B that can utilize each other’s signals, public goods and nutrients to a varying extent. a = signal sharing, b = public goods sharing, c = nutrient sharing. Left: regions of co-colocalizing communities (i.e. segregation coefficient is below 0.5, see Methods). Right: Relative fitness of the mixed communities (shaded area on the left) as a function of food sharing (top curve). RF>1 indicates that both species grow faster in a community than alone. Bottom curve: relative fitness of non-colocalizing communities. The values are the average of 10 calculations, error bars represent the standard deviation of the mean.

Figure 6. Exploitation.

Species B exploits the QS system (signals, public goods) and nutrients of species A. This provides a fitness advantage to the exploiter species B in the entire parameter range. Left: Regions of the parameter space represent either competitive exclusion or competitive segregation. Right: Fitness of the two species relative to growing alone, as a function of nutrient sharing. Relative fitness = 1 in the top curve indicates that the growth of species B is not hampered by the competition. The values are the average of 10 calculations, error bars represent the standard deviation of the mean.

Figure 7. Principle of the dendrit growth model , .

The dendrite is modeled as a longitudinal, infinite 2D surface covered with a nutrient. Cell agents (black dots) placed at the start will begin to consume the nutrients and migrate. In the environment of the cell agents (the active zone) there are signals and public goods (indicated as grey area) sufficient to keep the cells in an activated state.

Figure 8. A heat map of segregation as a function of signal and public goods sharing.

The black area indicates the parameter range wherein the two competing populations form a mixed community i.e. segregation coefficient is below 0.5. The data are from a simulation of asymmetrical sharing of signals and public goods at intermediate sharing of nutrients (c = 0.6).