My host's enemy is my enemy: plasmids carrying CRISPR-Cas as a defence against phages

Abstract

Bacteria are infected by mobile genetic elements like plasmids and virulent phages, and those infections significantly impact bacterial ecology and evolution. Recent discoveries reveal that some plasmids carry anti-phage immune systems like CRISPR-Cas, suggesting that plasmids may participate in the coevolutionary arms race between virulent phages and bacteria. Intuitively, this seems reasonable as virulent phages kill the plasmid's obligate host. However, the efficiency of CRISPR-Cas systems carried by plasmids can be expected to be lower than those carried by the chromosome due to continuous segregation loss, creating susceptible cells for phage amplification. To evaluate the anti-phage protection efficiency of CRISPR-Cas on plasmids, we develop a stochastic model describing the dynamics of a virulent phage infection against which a conjugative plasmid defends using CRISPR-Cas. We show that CRISPR-Cas on plasmids provides robust protection, except in limited parameter sets. In these cases, high segregation loss favours phage outbreaks by generating a population of defenceless cells on which the phage can evolve and escape CRISPR-Cas immunity. We show that the phage's ability to exploit segregation loss depends strongly on the evolvability of both CRISPR-Cas and the phage itself.

Keywords: CRISPR-Cas; coevolution; evolutionary emergence; mobile genetic elements; phages; plasmids.

Figure 1.

Visualization of the modelled CRISPR-Cas, plasmid and phage dynamics and the corresponding parameters. (a1) Schematic diagram of plasmid and CRISPR-Cas dynamics illustrating the processes described by equations (2.1)–(2.3). (a2) Schematic diagram of phage infection and CRISPR-Cas processes described by equations (2.1)–(2.5). (b) Overview of model processes, associated parameters and simulation parameter values. The choice of model parameters is specified in the main text in §2(e).

Figure 2.

Anti-phage protection performance of plasmids using CRISPR-Cas. Percentage of all simulated plasmid parameter sets stratified according to CRISPR-Cas performance in comparison with chromosomal CRISPR-Cas. The mean RD between CRISPR-Cas efficiency on the chromosome and CRISPR-Cas efficiency on the plasmid is shown for each category.

Figure 3.

Importance of biological parameters to predict CRISPR-Cas performance on plasmids. (a) Global importance of parameters in classifying the CRISPR-Cas efficiency on plasmids into the three categories ‘worse’, ‘same’ and ‘better’. (b) The model’s most important parameters in determining when the CRISPR-Cas efficiency on plasmids is worse than the efficiency of chromosomal systems.

Figure 4.

Impact of varying acquisition rates, segregation loss and phage mutation on the probability of phage extinction (CRISPR-Cas efficiency). (a) The phage has the possibility to escape acquired spacers by mutating protospacers with probability μ = 3.4 × 10⁻⁷. (b) The phage has no possibility to escape acquired CRISPR-Cas spacers μ = 0. The conjugation rate, plasmid cost and phage burst sizes are fixed at γ = 10⁻¹⁴, c = 0.05, b = 190 and b_m = 179. The probability of phage extinction for CRISPR-Cas systems on the chromosome is indicated in pink. In the chromosomal simulations, the conjugation rate γ, the probability of losing the plasmid τ and plasmid fitness effect c are set to zero.

Figure 5.

Impact of segregation loss on the probability of phage extinction (CRISPR-Cas efficiency) depending on phage characteristics. For different phage protospacer mutation probability (a) or phage burst sizes (b), we show the distribution of CRISPR-Cas efficiency across all the tested probabilities of segregation loss (figure 1) for each probability of spacer acquisition. Hence, a wider boxplot indicates a greater impact of segregation loss on CRISPR-Cas efficiency. The conjugation rate and plasmid cost are fixed at γ = 10⁻¹⁴ and c = 0.05. The phage burst size is fixed in (a) to b = 190, b_m = 179 and the protospacer mutation probability in subplot (b) to μ = 3.4 × 10⁻⁷.