Bacteria have evolved a wide range of mechanisms to harm and kill their competitors, including chemical, mechanical and biological weapons. Here we review the incredible diversity of bacterial weapon systems, which comprise antibiotics, toxic proteins, mechanical weapons that stab and pierce, viruses, and more. The evolution of bacterial weapons is shaped by many factors, including cell density and nutrient abundance, and how strains are arranged in space. Bacteria also employ a diverse range of combat behaviours, including pre-emptive attacks, suicidal attacks, and reciprocation (tit-for-tat). However, why bacteria carry so many weapons, and why they are so often used, remains poorly understood. By comparison with animals, we argue that the way that bacteria live -often in dense and genetically diverse communities -is likely to be key to their aggression as it encourages them to dig in and fight alongside their clonemates. The intensity of bacterial aggression is such that it can strongly affect communities, via complex coevolutionary and ecoevolutionary dynamics, which influence species over space and time. Bacterial warfare is a fascinating topic for ecology and evolution, as well as one of increasing relevance. Understanding how bacteria win wars is important for the goal of manipulating the human microbiome and other important microbial systems. StrategyRule(s) defining the decisions of an actor in response to prevailing conditions (for example, fight back if attacked). TacticsBehaviours displayed in a given contest (the realised output of a strategy). ToxinSubstance that disrupts cell physiology. WarfareConflict involving weapons between two or more groups. WeaponCompetitive phenotype that damages and/or disrupts the physiology of recipients, and that evolved, at least in part, for that purpose (for example, bacteriocin production).

The production of beneficial public goods is common in the microbial world, and so is cheating -the exploitation of public goods by nonproducing mutants. Here, we examine co-evolutionary dynamics between cooperators and cheats and ask whether cooperators can evolve strategies to reduce the burden of exploitation, and whether cheats in turn can improve their exploitation abilities. We evolved cooperators of the bacterium Pseudomonas aeruginosa, producing the shareable iron-scavenging siderophore pyoverdine, together with cheats, defective in pyoverdine production but proficient in uptake. We found that cooperators managed to co-exist with cheats in 56% of all replicates over approximately 150 generations of experimental evolution. Growth and competition assays revealed that co-existence was fostered by a combination of general adaptions to the media and specific adaptions to the co-evolving opponent. Phenotypic screening and whole-genome resequencing of evolved clones confirmed this pattern, and suggest that cooperators became less exploitable by cheats because they significantly reduced their pyoverdine investment. Cheats, meanwhile, improved exploitation efficiency through mutations blocking the costly pyoverdine-signalling pathway. Moreover, cooperators and cheats evolved reduced motility, a pattern that likely represents adaptation to laboratory conditions, but at the same time also affects social interactions by reducing strain mixing and pyoverdine sharing. Overall, we observed parallel evolution, where co-existence of cooperators and cheats was enabled by a combination of adaptations to the abiotic and social environment and their interactions.

Bacterial opportunistic pathogens are feared for their difficult-to-treat nosocomial infections and for causing morbidity in immunocompromised patients. Here, we study how such a versatile opportunist, Pseudomonas aeruginosa, adapts to conditions inside and outside its model host Caenorhabditis elegans, and use phenotypic and genotypic screens to identify the mechanistic basis of virulence evolution. We found that virulence significantly dropped in unstructured environments both in the presence and absence of the host, but remained unchanged in spatially structured environments. Reduction of virulence was either driven by a substantial decline in the production of siderophores (in treatments without hosts) or toxins and proteases (in treatments with hosts). Whole-genome sequencing of evolved clones revealed positive selection and parallel evolution across replicates, and showed an accumulation of mutations in regulator genes controlling virulence factor expression. Our study identifies the spatial structure of the non-host environment as a key driver of virulence evolution in an opportunistic pathogen.

No abstract

Summary Behaviors that cause the death of an actor are typically strongly disfavored by natural selection, and yet many bacteria undergo cell lysis to release anti-competitor toxins [ 1 , 2 , 3 , 4 , 5 ]. This behavior is most easily explained if only a small proportion of cells die to release toxins and help their clonemates, but the frequency of cells that actually lyse during bacterial warfare is unknown. The challenge is finding a way to distinguish cells that have undergone programmed suicide from those that were simply killed by a competitor’s toxin. We developed a two-color fluorescence reporter assay in Escherichia coli to overcome this problem. This revealed conditions where nearly all cells undergo programmed lysis. Specifically, adding a DNA-damaging toxin (DNase colicin) from another strain induced mass cell suicide where ∼85% of cells lysed to release their own toxins. Time-lapse 3D confocal microscopy showed that self-lysis occurs locally at even higher frequencies (∼94%) at the interface between toxin-producing colonies. By exposing E. coli that do not perform lysis to the DNase colicin, we found that mass lysis occurs when cells are going to die anyway from toxin exposure. From an evolutionary perspective, this renders the behavior cost-free as these cells have zero reproductive potential. This helps to explain how mass cell suicide can evolve, as any small benefit to surviving clonemates can lead to this retaliatory strategy being favored by natural selection. Our findings have parallels to the suicidal attacks of social insects [ 6 , 7 , 8 , 9 ], which are also performed by individuals with low reproductive potential.

Pathogenic bacteria engage in social interactions to colonize hosts, which include quorum-sensing-mediated communication and the secretion of virulence factors that can be shared as "public goods" between individuals. While in-vitro studies demonstrated that cooperative individuals can be displaced by "cheating" mutants freeriding on social acts, we know less about social interactions in infections. Here, we developed a live imaging system to track virulence factor expression and social strain interactions in the human pathogen Pseudomonas aeruginosa colonizing the gut of Caenorhabditis elegans. We found that shareable siderophores and quorum-sensing systems are expressed during infections, affect host gut colonization, and benefit non-producers. However, non-producers were unable to successfully cheat and outcompete producers. Our results indicate that the limited success of cheats is due to a combination of the down-regulation of virulence factors over the course of the infection, the fact that each virulence factor examined contributed to but was not essential for host colonization, and the potential for negative frequency-dependent selection. Our findings shed new light on bacterial social interactions in infections and reveal potential limits of therapeutic approaches that aim to capitalize on social dynamics between strains for infection control.

Bacterial traits that contribute to disease are termed “virulence factors” and there is much interest in therapeutic approaches that disrupt such traits. What remains less clear is whether a virulence factor identified as such in a particular context is also important in infections involving different host and pathogen types. Here, we address this question using a meta-analytic approach. We statistically analyzed the infection outcomes of 81 experiments associated with one well-studied virulence factor—pyoverdine, an iron-scavenging compound secreted by the opportunistic pathogen Pseudomonas aeruginosa. We found that this factor is consistently involved with virulence across different infection contexts. However, the magnitude of the effect of pyoverdine on virulence varied considerably. Moreover, its effect on virulence was relatively minor in many cases, suggesting that pyoverdine is not indispensable in infections. Our works supports theoretical models from ecology predicting that disease severity is multifactorial and context dependent, a fact that might complicate our efforts to identify the most important virulence factors. More generally, our study highlights how comparative approaches can be used to quantify the magnitude and general importance of virulence factors, key knowledge informing future anti-virulence treatment strategies.

BackgroundA common form of cooperation in bacteria is based on the secretion of beneficial metabolites, shareable as public good among cells within a group. Because cooperation can be exploited by “cheating” mutants, which contribute less or nothing to the public good, there has been great interest in understanding the conditions required for cooperation to remain evolutionarily stable. In contrast, much less is known about whether cheats, once fixed in the population, are able to revert back to cooperation when conditions change. Here, we tackle this question by subjecting experimentally evolved cheats of Pseudomonas aeruginosa, partly deficient for the production of the iron-scavenging public good pyoverdine, to conditions previously shown to favor cooperation.ResultsFollowing approximately 200 generations of experimental evolution, we screened 720 evolved clones for changes in their pyoverdine production levels. We found no evidence for the re-evolution of full cooperation, even in environments with increased spatial structure, and reduced costs of public good production – two conditions that have previously been shown to maintain cooperation. In contrast, we observed selection for complete abolishment of pyoverdine production. The patterns of complete trait degradation were likely driven by “cheating on cheats” in unstructured, iron-limited environments where pyoverdine is important for growth, and selection against a maladaptive trait in iron-rich environments where pyoverdine is superfluous.ConclusionsOur study shows that the path to re-evolve public-goods cooperation can be constrained. While a limitation of the number of mutational targets potentially leading to reversion might be one reason for the observed pattern, an alternative explanation is that the selective conditions required for revertants to spread from rarity are much more stringent than those needed to maintain cooperation.Electronic supplementary materialThe online version of this article (10.1186/s12862-017-1060-6) contains supplementary material, which is available to authorized users.