Bacteriophages limit the existence conditions for conjugative plasmids

Abstract

Bacteriophages are a major cause of bacterial mortality and impose strong selection on natural bacterial populations, yet their effects on the dynamics of conjugative plasmids have rarely been tested. We combined experimental evolution, mathematical modeling, and individual-based simulations to explain how the ecological and population genetics effects of bacteriophages upon bacteria interact to determine the dynamics of conjugative plasmids and their persistence. The ecological effects of bacteriophages on bacteria are predicted to limit the existence conditions for conjugative plasmids, preventing persistence under weak selection for plasmid accessory traits. Experiments showed that phages drove faster extinction of plasmids in environments where the plasmid conferred no benefit, but they also revealed more complex effects of phages on plasmid dynamics under these conditions, specifically, the temporary maintenance of plasmids at fixation followed by rapid loss. We hypothesized that the population genetic effects of bacteriophages, specifically, selection for phage resistance mutations, may have caused this. Further mathematical modeling and individual-based simulations supported our hypothesis, showing that conjugative plasmids may hitchhike with phage resistance mutations in the bacterial chromosome.

Importance: Conjugative plasmids are infectious loops of DNA capable of transmitting DNA between bacterial cells and between species. Because plasmids often carry extra genes that allow bacteria to live in otherwise-inhospitable environments, their dynamics are central to understanding bacterial adaptive evolution. The plasmid-bacterium interaction has typically been studied in isolation, but in natural bacterial communities, bacteriophages, viruses that infect bacteria, are ubiquitous. Using experiments, mathematical models, and computer simulations we show that bacteriophages drive plasmid dynamics through their ecological and evolutionary effects on bacteria and ultimately limit the conditions allowing plasmid existence. These results advance our understanding of bacterial adaptation and show that bacteriophages could be used to select against plasmids carrying undesirable traits, such as antibiotic resistance.

FIG 1

Prevalence of the pQBR103 plasmid over time in six replicate populations of each treatment. (a and b) The 32 µM HgCl₂ environment without (a) or with (b) coevolving phages. (c and d) The 0 µM HgCl₂ environment without (c) or with (d) coevolving phages. Colors distinguish individual replicate populations. Axes show the proportions of plasmid-carrying cells in the population (y axes) measured through time (x axes).

FIG 2

Plot of the stable fixed points of the mathematical model (equation set 1). The axes are the parameters μ and η, which are the mortalities associated with phage and mercury toxicity, respectively. Shading indicates the proportion of plasmid-carrying bacteria in the population. At high mercury concentrations and/or low phage mortality rates, plasmids are maintained at an interior fixed point, i.e., they coexist with plasmid-free cells (red; shading indicates the proportion of plasmid-containing cells, from high [dark] to low [light]). With a sufficiently low mercury concentration and under high phage pressure, plasmids are lost (blue). There is a line of transcritical bifurcations which separates these two regions. In the vicinity of the transcritical bifurcation, the convergence to the fixed point is slow enough that other factors (e.g., compensatory mutations [7, 12, 39–41]) are likely to occur prior to the model’s prediction of the loss or retention of the plasmid. The four corners of the plot correspond to the empirical treatments in Fig. 1 and the phase planes in Fig. 3. Parameter values are from Table 1.

FIG 3

Mathematical phase planes for the population dynamics. Axes show total population values for plasmid-free (F; horizontal axes) and plasmid-containing (P; vertical axes) populations, scaled to the carrying capacity in the system without phage or mercury. The panels are arranged to correspond to both the overall diagram in the empirical data (Fig. 1) and the μ and η space (Fig. 2). The top panels show the 32 μM HgCl₂ environment without phage (left; rapid fixation of P) and with phage (right; rapid fixation of P, reduced population density). The bottom panels show the HgCl₂-free environment without phage (left; slow elimination of P) and with phage (right; rapid elimination of P, reduced population density). The blue curves show the trajectories from a grid of initial conditions, the green arrows indicate the global flow field, and the red circles show the stable fixed points to which trajectories are attracted.

FIG 4

The individual-based model. The IBM captures the bistability of the system when phage is present. (a) The proportion of plasmid-containing cells after 250 h for 12 implementations across a range of mercury toxicity values, demonstrating that prevalence is predominantly fixed near either 1 or 0. (b) Plasmid frequencies through time for 12 replicate implementations under the poison = 0 condition (comparable to the results shown in Fig. 1d for the empirical data). (c) The positive frequency dependence appearing naturally in the IBM. Values denote the difference in the mean defense value for plasmid-carrying and plasmid-free cells, plotted against plasmid prevalence. Plasmid-carrying cells had a higher mean resistance to phage than plasmid-free cells when common and lower mean resistance when rare.

FIG 5

The dynamics of plasmid loss in the IBM simulation. (a) Plasmid dynamics in 12 iterations of the IBM. (b and c) Two iterations shown in detail (highlighted in panel a as coarse dashed [panel b] and fine dashed [panel c] lines) to demonstrate the link between phage resistance evolution and plasmid loss. In b and c, plasmid prevalence is shown by gray shading, and colored lines represent the frequencies of different phage resistance alleles present in the plasmid-containing (fixed) and plasmid-free (dashed) portions of the population. Plasmids are transiently maintained in the population by hitchhiking on sweeps of phage resistance mutations.