Microbial communication, cooperation and cheating: quorum sensing drives the evolution of cooperation in bacteria

Abstract

An increasing body of empirical evidence suggests that cooperation among clone-mates is common in bacteria. Bacterial cooperation may take the form of the excretion of "public goods": exoproducts such as virulence factors, exoenzymes or components of the matrix in biofilms, to yield significant benefit for individuals joining in the common effort of producing them. Supposedly in order to spare unnecessary costs when the population is too sparse to supply the sufficient exoproduct level, many bacteria have evolved a simple chemical communication system called quorum sensing (QS), to "measure" the population density of clone-mates in their close neighborhood. Cooperation genes are expressed only above a threshold rate of QS signal molecule re-capture, i.e., above the local quorum of cooperators. The cooperative population is exposed to exploitation by cheaters, i.e., mutants who contribute less or nil to the effort but fully enjoy the benefits of cooperation. The communication system is also vulnerable to a different type of cheaters ("Liars") who may produce the QS signal but not the exoproduct, thus ruining the reliability of the signal. Since there is no reason to assume that such cheaters cannot evolve and invade the populations of honestly signaling cooperators, the empirical fact of the existence of both bacterial cooperation and the associated QS communication system seems puzzling. Using a stochastic cellular automaton approach and allowing mutations in an initially non-cooperating, non-communicating strain we show that both cooperation and the associated communication system can evolve, spread and remain persistent. The QS genes help cooperative behavior to invade the population, and vice versa; cooperation and communication might have evolved synergistically in bacteria. Moreover, in good agreement with the empirical data recently available, this synergism opens up a remarkably rich repertoire of social interactions in which cheating and exploitation are commonplace.

Figure 1. Stationary genotype distributions of the QS-disabled set of simulations.

Fixed parameters: basic metabolic burden: M₀ = 100.0; metabolic cost of quorum signal production: m_s = 3.0; metabolic cost of quorum signal response: m_r = 1.0; fitness reward factor: r = 0.9; mutation rates: μ_s = μ_r = 0.0, μ_c = 10⁻⁴. Screened parameters: metabolic cost of cooperation (m_c), quorum threshold (n_e) and dispersal (D). Simulations lasted for 10.000 generations and they were initiated with the “All-Ignorant” (csr) state.

Figure 2. Details of a single QS-enabled simulation.

Parameters as in Fig. 1, except for μ_s = μ_r = 10⁻⁴; m_c = 30.0, n_e = 3 and D = 0.0. Time evolution of A.: genotype frequencies; B.: genotype distribution; C.: allele frequencies. D.: The spatial pattern of genotypes at T = 10.000.

Figure 3. Stationary genotype distributions of the QS-disabled set of simulations.

All parameters as in Fig. 1. except μ_s = μ_r = 10⁻⁴.

Figure 4. The evolution of QS in a population with the cooperating C allele fixed.

Parameters as in Fig. 3, with n_e = 6 and D = 0.2. The simulation was started from the “All-Blunt” initial state.