The evolution of the type VI secretion system as a disintegration weapon

Abstract

The type VI secretion system (T6SS) is a nanomachine used by many bacteria to drive a toxin-laden needle into other bacterial cells. Although the potential to influence bacterial competition is clear, the fitness impacts of wielding a T6SS are not well understood. Here we present a new agent-based model that enables detailed study of the evolutionary costs and benefits of T6SS weaponry during competition with other bacteria. Our model identifies a key problem with the T6SS. Because of its short range, T6SS activity becomes self-limiting, as dead cells accumulate in its way, forming "corpse barriers" that block further attacks. However, further exploration with the model presented a solution to this problem: if injected toxins can quickly lyse target cells in addition to killing them, the T6SS becomes a much more effective weapon. We tested this prediction with single-cell analysis of combat between T6SS-wielding Acinetobacter baylyi and T6SS-sensitive Escherichia coli. As predicted, delivery of lytic toxins is highly effective, whereas nonlytic toxins leave large patches of E. coli alive. We then analyzed hundreds of bacterial species using published genomic data, which suggest that the great majority of T6SS-wielding species do indeed use lytic toxins, indicative of a general principle underlying weapon evolution. Our work suggests that, in the T6SS, bacteria have evolved a disintegration weapon whose effectiveness often rests upon the ability to break up competitors. Understanding the evolutionary function of bacterial weapons can help in the design of probiotics that can both establish well and eliminate problem species.

The evolution of tit-for-tat in bacteria via the type VI secretion system

“…We recently reported a new version of this model 44 , designed specifically for the study of T6SS competition, in which cells can intoxicate neighbors by firing T6SS needles. Here, we extend this model, such that different modes of T6SS firing (Supplementary Table 1) can now be represented and compared: cells can be programmed not to fire, or to fire constantly and in random directions, or to fire in more elaborate patterns.…”

“…We find that either competitor can invade the other by firing faster than it (Supplementary Movie 3 shows an example of this), but only up to a point. Beyond the yellow diagonal line, having a higher attack rate than one's competitor makes one vulnerable to invasion, since the increased costs of the higher attack rate outweigh any additional benefits conferred 44 . Figure 1e shows a similar pairwise invasion plot, this time computed for the case of global competition.…”

“…Where possible, parameters governing T6SS firing and response were estimated directly from previous experimental observations of T6SS competition 25,44 . We used a lower bound of N hits = 1 based on a previous study 44 . However, based on the observation that P. aeruginosa seems to withstand multiple T6SS hits before being killed 27 , we also considered larger values (N hits = 2, 3, and ∞ in cases where target cells possess immunity to oncoming T6SS attacks).…”

“…Numerical and mechanical parameters controlling cell mechanical interactions and movement were taken from previous publications 27 , 39 , 40 . Where possible, parameters governing T6SS firing and response were estimated directly from previous experimental observations of T6SS competition 25 , 44 . We used a lower bound of N hits = 1 based on a previous study 44 .…”

“…However, based on the observation that P. aeruginosa seems to withstand multiple T6SS hits before being killed 27 , we also considered larger values ( N hits = 2, 3, and ∞ in cases where target cells possess immunity to oncoming T6SS attacks). Throughout the study, we used the parameter values c = 0.001, c upfront = 0.05; we chose these values because they made the optimal firing rate in R vs. U competitions roughly consistent with the firing rates observed in random-firing bacteria like V. cholerae and A. baylyi (~50–100 firings cell −1 h −1 ) 27 , 44 . Since it was not possible to glean cost parameter values directly from experiments, we instead performed broad parameter sweeps to test the effects of different costs.…”