- Top of page
- Materials and Methods
- LITERATURE CITED
Social interactions have been shown to play an important role in bacterial evolution and virulence. The majority of empirical studies conducted have only considered social traits in isolation, yet numerous social traits, such as the production of spiteful bacteriocins (anticompetitor toxins) and iron-scavenging siderophores (a public good) by the opportunistic pathogen Pseudomonas aeruginosa, are frequently expressed simultaneously. Crucially, both bacteriocin production and siderophore cheating can be favored under the same competitive conditions, and we develop theory and carry out experiments to determine how the success of a bacteriocin-producing genotype is influenced by social cheating of susceptible competitors and the resultant impact on disease severity (virulence). Consistent with our theoretical predictions, we find that the spiteful genotype is favored at higher local frequencies when competing against public good cheats. Furthermore, the relationship between spite frequency and virulence is significantly altered when the spiteful genotype is competed against cheats compared with cooperators. These results confirm the ecological and evolutionary importance of considering multiple social traits simultaneously. Moreover, our results are consistent with recent theory regarding the invasion conditions for strong reciprocity (helping cooperators and harming noncooperators).
Social interactions are widespread in microbes and include behaviors such as communication, cooperative public goods production, and spiteful toxin production (West et al. 2007; Brown and Buckling 2008). There has been a recent proliferation of empirical and theoretical studies investigating the selective forces acting on social traits in bacteria (Bremermann and Pickering 1983; Frank 1992, 1996; West and Buckling 2003; Gardner et al. 2004; Massey et al. 2004; Harrison et al. 2006; Brown et al. 2008; Buckling and Brockhurst 2008; Vigneux et al. 2008; Inglis et al. 2009, 2011; Köhler et al. 2009; Bashey et al. 2012). However, most studies to date consider social traits in isolation (but see Gardner et al. 2007; Lehmann et al. 2007; Harrison and Buckling 2009; Brown and Taylor 2010), yet multiple social traits are typically expressed simultaneously (Williams et al. 2007; Harrison and Buckling 2009). Here, we investigate how selection on two types of microbial social traits (spite: costly to both actor and recipient, and public goods cooperation: costly to the actor, beneficial to the recipient) can be altered by their interaction. Both spite and public goods cheating can be favored under the same competitive conditions, and we address how the success of a toxin-producing (spiteful) lineage is influenced by social cheating of toxin-susceptible competitors. We also investigate how competition between spiteful and cheating phenotypes affects disease severity (virulence). Both spite and public goods cooperation have been shown to be important factors in determining bacterial virulence as they both have an effect on population growth rate and density within a host (Harrison et al. 2006, Inglis et al. 2009, Köhler et al. 2009). Cooperative interactions have been shown to increase the virulence of infections (Harrison et al. 2006, Köhler et al. 2009) whereas spiteful interactions can attenuate virulence if the infecting population is also composed of individuals susceptible to the spiteful action (Massey et al. 2004, Vigneux et al. 2008, Inglis et al. 2009).
Spiteful behaviors appear to be common in microbes in the form of metabolically costly (and in some cases even suicidal) production of anticompetitor toxins (Riley and Wertz 2002; Hawlena et al. 2010). Spiteful behaviors can be favored by kin selection when nonrelatives are preferentially affected (negative relatedness between actor and recipient) (Hamilton 1970), and the adverse effect on the recipient is high relative to the cost to the actor. Differential effects of the toxins on nonkin versus kin is achieved by microbes via linkage between toxin and immunity genes, rendering all individuals with the same toxin gene resistant to the toxin; an example of a “green beard” (Hurst 1991; Gardner and West 2004, 2010; Massey et al. 2004; Greig and Travisano 2008; Wloch-Salamon et al. 2008; Inglis et al. 2009). Both theory and data suggest that selection for spite is maximized when the spiteful lineage is at an intermediate frequency in the population (Gardner and West 2004, Inglis et al. 2009). When the spiteful lineage is at low frequencies in the local interacting population, the benefit gained from the spiteful act (e.g., freeing up resources) will be more likely to be experienced by nonspiteful individuals. In contrast to this, at high local frequencies only a few competitors will be affected by the spiteful behavior, hence relative costs of spite will be high (Inglis et al. 2009).
Social environments that influence selection for spiteful behaviors are also likely to affect selection on indiscriminate cooperation. Indiscriminate cooperation (individually costly behaviors that benefit all others in the vicinity) is extremely common in microbes and includes the public goods production of extracellular enzymes and nutrient-scavenging molecules, such as siderophores (Buckling and Brockhurst 2008). Maintaining cooperation in a population relies on individuals interacting with their relatives, which are more likely to possess the same cooperation gene, due either to kin discrimination or population viscosity (Hamilton 1963; Smith 1964). Otherwise “social cheats,” individuals that do not pay the cost of the cooperative behavior but reap all the rewards, are able to invade and displace the cooperators. Note that spiteful behaviors can also be viewed as a form of indirect, discriminating altruism, in that unaffected individuals with the spite allele benefit from the behavior through removal of competitors (Lehmann et al. 2006).
In this study, we take a joint theoretical and empirical approach, using the pathogenic bacterium Pseudomonas aeruginosa, to determine how the success of spiteful cooperators is affected by whether the susceptible competitors are themselves public goods cheats or cooperators. Both strategies exist in natural populations of P. aeruginosa (e.g., Bodilis et al. 2009), yet to date selection on spite and selection on cooperation have only been studied in isolation (i.e., spite vs. nonspite; cooperators vs. cheats). We focus on the production of pyocin S2 and type 1 pyoverdine by P. aeruginosa, a spiteful and a public goods cooperative trait, respectively. Pyocin S2 is a protein-based toxin that inhibits phospholipid synthesis through DNA breakdown, whereas type 1 pyoverdine is a fluorescent yellow–green, high-affinity iron chelator (a siderophore). We also investigate the impact of competition between spiteful and cheating genotypes on P. aeruginosa virulence in an insect model system.
- Top of page
- Materials and Methods
- LITERATURE CITED
We investigated how the success of a siderophore- and bacteriocin-producing P. aeruginosa genotype is altered by whether its conspecific competitor is itself a siderophore producer or cheat, and the resulting impact on virulence in insects. We show in vitro that the unimodal relationship between the fitness of spiteful, toxin-producing bacteria and their frequency (Fig. 4) is altered when the spiteful genotype is more cooperative than its competitor, resulting in a systematic shift in peak fitness toward higher frequencies (Fig. 5). Furthermore, we demonstrate that the virulence of infections of insects where there is competition between cooperating and cheating bacteria is significantly altered depending on the starting frequencies of the genotypes, and whether the cooperating bacteria are also spiteful (Fig. 6A, B). We found positive monotonic relationships between virulence and cooperator frequency for both spiteful and nonspiteful genotypes, with greater, overall virulence in infections with nonspiteful bacteria. Importantly, previous work has shown that in the absence of a cooperator–cheat conflict, virulence is minimized at intermediate frequencies of spiteful genotypes (Inglis et al. 2009), demonstrating that competing against social cheats fundamentally alters the relationship between virulence and spite. All these results are qualitatively consistent with the predictions from our mathematical model (Figs. 1–3, Appendix).
Our model gives us some very simple insights as to why spite is more beneficial at high frequencies when competing with public goods cheats. First, the model shows that simultaneous investments in cooperation can change the relative benefit/cost ratio of investments into spite, with the result that selection for spite is systematically enhanced when spiteful cooperators are at high frequencies. This is because high frequencies of cooperation will increase the growth rate of the focal strain, and so lessen the relative growth rate costs of a fixed investment into spite. Second, if we additionally allow for cheats to initially increase locally in frequency, due to their competitive advantage with respect to the public good trait, then investments in spite can be further favored if the initial frequency of the spiteful cooperators is sufficiently high. This is because spiteful behaviors are maximally favored at intermediate frequencies, and the invasion of the cheats can push the frequency of the spiteful cooperators from high to intermediate frequencies. The net result of both processes is to shift the maximal value of spite to higher frequencies. There may of course be other mechanisms operating that help to explain our results that are not captured in our model: for example, if these traits are expressed at different times. However, it is not necessary to invoke such specific biological details to explain the results.
We wish to reiterate that that our model is not a demographically explicit or “closed” model of evolutionary change, instead we take a simplifying “black box” approach by summarizing numerous demographic effects via the parameters a and p. Although the heuristic “open model” approach sacrifices ecological specificity, it gains in terms of generality (allowing comparisons across diverse demographies) and tractability (see e.g., Gardner and West 2006; Brown and Taylor 2010). Clearly, the two approaches are highly complementary, and it remains to be seen what further insights can be generated via closed models of multiple social trait evolution under specific demographic regimes.
Although our experimental results highlight how public goods cooperation enhances the value of spite, our theory also illustrates that the converse can also hold true—spite can enhance the value of public goods investment by increasing the local frequency of producers (eq. A10). Together, these results outline a potential mechanistic synergy between spiteful and cooperative traits, which suggests that the long-term joint evolutionary dynamics will be more complex than predicted by analysis of either trait alone (Brown and Taylor 2010).
Qualitative patterns in our virulence data were consistent with previous studies examining social traits in isolation (Brown et al. 2002; West and Buckling 2003; Gardner et al. 2004; Harrison et al. 2006; Inglis et al. 2009). First, virulence, as determined by the growth rate of the whole infecting population, increased with the frequency of siderophore cooperators. This is unsurprising, as siderophore concentration increases growth rate by allowing iron to be scavenged from the caterpillar host. Second, the presence of spiteful genotypes reduced mean in vivo growth rate and virulence, because bacteriocin production is both costly to the actors (cell lysis) and the victims (arrests cell replication by inhibiting phospholipid synthesis). However, in contrast to previous work investigating spite and virulence in P. aeruginosa (Inglis et al. 2009) in the absence of a cooperator–cheat conflict, virulence was not minimized at intermediate frequencies of spiteful genotypes, but rather was minimal at both low and intermediate frequencies and increased at high frequencies. This is the pattern expected from our theoretical work, and is presumably the result of increased conflict between spiteful and cheating strains at low and intermediate frequencies. When the spiteful lineage is at low frequency in the competing population, there will be a reduction in growth rate as siderophore levels are low, thereby reducing virulence. At intermediate frequencies of the spiteful genotype, siderophore production will also be reduced to some extent, and total bacterial density reductions resulting from killing the sensitive cheats will also be at their greatest. The time to death data were largely consistent with the in vivo growth data, with variation likely to be attributable to the genetic heterogeneity of the caterpillars.
In the current study, cheating genotypes are always susceptible to pyocins, and hence spite can be considered as a form of punishment in this experimental context; an association noted in some recent theoretical studies (Nakamaru and Iwasa 2006; Lehmann et al. 2007). The concept of strong reciprocity—the punishment of noncooperators by cooperators—is suggested to be an important process in the evolution of cooperation (e.g., Bowles and Gintis 2004). Our theoretical results suggest that punishment can be favored over nonpunishment when at intermediate to high frequencies in structured populations, but cannot invade from rare. This result is broadly consistent with recent theory under conditions where cooperation and punishment are not in linkage; an assumption appropriate for our theoretical results, given that we are measuring fitness of cooperating punishers invading cooperating nonpunishers. By contrast, punishment can theoretically invade from rare when these traits are in linkage (Lehmann et al. 2007). It is interesting to note that there is an association between pyocin and pyoverdine outside of this experimental context. Specifically, P. aeruginosa pyocin S2 is taken up through the primary pyoverdine receptor, FvpA type 1 (Denayer et al. 2007). Genotypes that evolve to exploit a strain's pyoverdine type (of which there are a number of alleles [Tummler and Cornelis 2005]) may, therefore, become vulnerable to the competitor's pyocin.
Here, we have shown that spiteful behaviors, or more specifically bacteriocin (pyocin) production, are critically affected by simultaneous investments in another social trait, in this case indiscriminate public goods cooperation (pyoverdine production). We have shown that this, in turn, can have important repercussions when considering virulence in P. aeruginosa infections. Both pyocin (spite) and pyoverdine (cooperation) are likely to be of importance in clinical settings, especially cystic fibrosis, where pyocin-producing and pyoverdine-cheating strains are commonly found during the course of an infection (Govan 1986; De Vos et al. 2001). This study may, therefore, help to explain patterns of virulence in natural infections, and could even have practical implications in terms of manipulating the competitive arena through some type of directed therapy (i.e., the introduction of pyocin-producing, public goods cheats), thereby reducing overall virulence. More generally, spiteful and cooperative behaviors are almost ubiquitous among bacteria (and may be common among other microbes, both pathogenic and nonpathogenic), hence an understanding of the evolution of these behaviors may be key to understanding microbial community dynamics and functions.