The social biology of quorum sensing in a naturalistic host pathogen system.

Many microorganisms cooperate by secreting products that are commonly available to neighboring cells. These "public goods" include autoinduced, quorum-sensing (QS) molecules and the virulence factors activated by these signals. Public goods cooperation is exploitable by cheaters, cells that avoid the costs of production but gain an advantage by freeloading on the products of others. QS signals and responses can be cooperative under artificial laboratory conditions, but it remains unclear whether QS is cooperative in nature: little is known about the frequency of cheaters in natural populations, and cheaters may do poorly because of the importance of QS in major transcriptional networks. Here, we investigate the cooperative nature of QS in a natural system: the Gram-positive insect pathogen Bacillus thuringiensis and the larvae of the diamondback moth, Plutella xylostella. Although we find evidence of cooperation, QS null mutants are not effective cheats in vivo and cannot outcompete wild-type strains. We show that spatial structure limits mutant fitness and that well-separated microcolonies occur in vivo because of the strong population bottlenecks occurring during natural infection. We argue that spatial structure and low densities are the norm in early-stage infections, and this can explain why QS cheaters are rare in B. thuringiensis and its relatives. These results contrast with earlier experiments describing the high fitness of Gram-negative QS cheaters and suggest that QS suppression ("quorum quenching") can be clinically effective without having negative impacts on the evolution of virulence.