Experimental evidence that network topology can accelerate the spread of beneficial mutations

Abstract

Whether and how the spatial arrangement of a population influences adaptive evolution has puzzled evolutionary biologists. Theoretical models make conflicting predictions about the probability that a beneficial mutation will become fixed in a population for certain topologies like stars, in which “leaf” populations are connected through a central “hub.” To date, these predictions have not been evaluated under realistic experimental conditions. Here, we test the prediction that topology can change the dynamics of fixation both in vitro and in silico by tracking the frequency of a beneficial mutant under positive selection as it spreads through networks of different topologies. Our results provide empirical support that meta-population topology can increase the likelihood that a beneficial mutation spreads, broaden the conditions under which this phenomenon is thought to occur, and points the way toward using network topology to amplify the effects of weakly favored mutations under directed evolution in industrial applications.

Mutant fate in spatially structured populations on graphs: connecting models to experiments

“…Modern evolution experiments [27][28][29][30][31][32][33][34][35] allow to investigate evolution in the lab. In particular, some experiments have investigated the impact of spatial structure on evolution, implementing spatial structure through transfers between well-mixed cultures [30-32, 34, 35].…”

“…In particular, some experiments have investigated the impact of spatial structure on evolution, implementing spatial structure through transfers between well-mixed cultures [30-32, 34, 35]. A recent seminal study explicitly considered different graph structures, including the star graph [34]. In this exciting context, there is a strong disconnect between theoretical predictions and experiments.…”

“…In this exciting context, there is a strong disconnect between theoretical predictions and experiments. Experiments allow for complex structures and asymmetric migrations [34], thus going beyond traditional population genetics models [7][8][9][10]. Evolutionary graph theory assumes that the size of each deme is strictly constant (generally set to one, or to a fixed number in generalizations).…”

“…Evolutionary graph theory assumes that the size of each deme is strictly constant (generally set to one, or to a fixed number in generalizations). This is not the case in experiments, which generally employ batch culture with serial passages (also called serial transfers), alternating phases of growth and bottlenecks [27][28][29][30][31][32][33][34][35]. Furthermore, these experiments do not have update rules or strict couplings between birth, death and migration events, in contrast to evolutionary graph theory.…”

“…We recently developed a new model of spatially structured populations on graphs, which explicitly models serial passages with migrations, allows asymmetric migrations, and can handle large population sizes [36]. This makes it ideal to model current experiments, such as those performed in [34], and to make quantitative predictions for future experiments. Formally, our model is close to a structured Wright-Fisher model, thus bridging classical population genetics models [7][8][9][10][11] and evolutionary graph theory [16,18,19].…”