Antagonism between killer yeast strains as an experimental model for biological nucleation dynamics

Abstract

Antagonistic interactions are widespread in the microbial world and affect microbial evolutionary dynamics. Natural microbial communities often display spatial structure, which affects biological interactions, but much of what we know about microbial warfare comes from laboratory studies of well-mixed communities. To overcome this limitation, we manipulated two killer strains of the budding yeast Saccharomyces cerevisiae, expressing different toxins, to independently control the rate at which they released their toxins. We developed mathematical models that predict the experimental dynamics of competition between toxin-producing strains in both well-mixed and spatially structured populations. In both situations, we experimentally verified theory's prediction that a stronger antagonist can invade a weaker one only if the initial invading population exceeds a critical frequency or size. Finally, we found that toxin-resistant cells and weaker killers arose in spatially structured competitions between toxin-producing strains, suggesting that adaptive evolution can affect the outcome of microbial antagonism in spatial settings.

The evolution of spectrum in antibiotics and bacteriocins

“…Our model predicts narrow-spectrum toxins often carry a major advantage, but broad-spectrum toxins can become effective when a strain is abundant enough to compete effectively with multiple species. This density dependence for broad-spectrum toxins is consistent with the general prediction from recent mathematical models and experiments that show that toxin efficacy can depend strongly on producer abundance ( 33 ). However, there are other factors that may select for broad-spectrum toxins that are not captured by our model.…”

“…There, it has been typical to assume that toxins are lost at a constant rate, simply proportional to their concentration, and we followed this assumption with our first model. However, an alternative assumption is that toxins will be lost from the system in a cell density–dependent fashion, where loss rate is linked to the number of target cells in the system ( 13 , 16 , 33 ). While this results in a more complex model, there is a good reason to believe that this is the more realistic assumption.…”

“…We modify this model to incorporate a density-dependent toxin loss term. While the majority of previous models take the form of our first model, there are some that have density-dependent toxin loss, and we follow these in this second model ( 13 , 16 , 33 ). This modification means the model now captures the potential for target cells to drive loss of the toxin from the system, such as would occur if they degrade or bind the toxin.…”

Bows and swords: why bacteria carry short and long-range weapons

“…Our modelling predicts that a long-range weapon will perform poorly at low attacker frequency, which is consistent with the findings of several previous studies -both theoretical and empirical -showing that toxin production is most effective when attackers are abundant 34,[36][37][38][39][40] . However, to test our predictions on the relative benefits of short versus long-range weapons, a wellcontrolled comparison of the two types of weapons is required.…”

Phase separation in fluids with many interacting components

“…Other prominent examples include microbial ecosystems that organize into fluid-like communities (9)(10)(11), self-assembling colloidal systems (12,13), and synthetic multi-phase materials derived from biomolecules (14,15). Despite their extensive prevalence, our understanding of how microscopic interaction networks between individual constituents encode emergent multi-phase behavior remains limited.…”