Bacteria-phage coevolution with a seed bank

Abstract

Dormancy is an adaptation to living in fluctuating environments. It allows individuals to enter a reversible state of reduced metabolic activity when challenged by unfavorable conditions. Dormancy can also influence species interactions by providing organisms with a refuge from predators and parasites. Here we test the hypothesis that, by generating a seed bank of protected individuals, dormancy can modify the patterns and processes of antagonistic coevolution. We conducted a factorially designed experiment where we passaged a bacterial host (Bacillus subtilis) and its phage (SPO1) in the presence versus absence of a seed bank consisting of dormant endospores. Owing in part to the inability of phages to attach to spores, seed banks stabilized population dynamics and resulted in minimum host densities that were 30-fold higher compared to bacteria that were unable to engage in dormancy. By supplying a refuge to phage-sensitive strains, we show that seed banks retained phenotypic diversity that was otherwise lost to selection. Dormancy also stored genetic diversity. After characterizing allelic variation with pooled population sequencing, we found that seed banks retained twice as many host genes with mutations, whether phages were present or not. Based on mutational trajectories over the course of the experiment, we demonstrate that seed banks can dampen bacteria-phage coevolution. Not only does dormancy create structure and memory that buffers populations against environmental fluctuations, it also modifies species interactions in ways that can feed back onto the eco-evolutionary dynamics of microbial communities.

Fig. 1. Illustration of seed bank manipulation.

In the + seed bank treatment, we used a strain of Bacillus subtilis that was capable of forming endospores after resources were exhausted by growth (black arrows). In addition, we established an external seed bank (shown in blue) to extend endospore residence time. The first step of this process involved purifying endospores through heat treatment (flame = 80 °C, 20 min), which eliminated phages and vegetative cells from a sample taken from the focal population contained in a flask. Then, we mixed these endospores with endospores preserved from previous transfers that were obtained in the same fashion. This spore mixture (i.e., external seed bank) and an untreated sample taken from a focal population were used to inoculate fresh medium and establish the next transfer. In the - seed bank treatment, serial transfers (black arrows) were conducted with a mutant strain of B. subtilis that was not capable of producing endospores in rich medium after resource exhaustion owing to an engineered mutation in a gene that is essential for sporulation (spoIIE). After establishing the seed bank with an initial serial transfer (t_-1), we began the experiment at t₀ by infecting half of the populations with phage SPO1. For simplicity, non-infected controls are not shown. See methods for further details.

Fig. 2. Endosporulation provided bacteria with a refuge from phages.

We demonstrate that phage SPO1 cannot attach to endospores of wild type Bacillus subtilis. Percent adsorption was calculated from the decline in free phages over 5 min when mixed with either purified endospores or vegetative cells. Mean (○) and standard deviation of four biological replicates (●) are shown. Grey bars indicate the lower limit of the 95% confidence interval of a one-sided t-test for each of the host cell types.

Fig. 3. Seed banks altered host-phage population dynamics.

Bacteria and phage dynamics were tracked in replicate (n = 3) populations that were propagated by serial transfer every two days (see Fig. 1). In the + seed bank treatment, the host could sporulate. In the - seed bank treatment, the host had an engineered mutation that prevented sporulation. Phage SPO1 was added to all populations (flasks) in the + phage treatment on day 0. See Fig. S5 to compare population dynamics of the different host strains in the - phage and + phage treatments. Data represented as mean ± SEM.

Fig. 4. Retention of susceptible hosts with a seed bank.

We quantified susceptibility by challenging clones of Bacillus subtilis against the ancestral phage. In both the + seed bank and - seed bank treatments, we performed this assay on clones (n = 22) that were isolated from each focal population contained in a flask (Fig. 1) just prior to serial transfer (). In addition, we revived clones (n = 22) from the external seed bank (Fig. 1) at each time point and challenged those against the ancestral phage (). The expected percentage of susceptible clones () is based on losses to dilution caused by serial transfer of clones originating from the pre-infection external seed bank (see Supplementary Information). Data represented as mean ± SEM of the replicate populations (n = 3).

Fig. 5. Rank-abundance distribution of mutations in host populations with and without a seed bank.

a Mutations were identified by sequencing and mapped to the genes which they affect. The multiplicity of a gene reflects the number of mutations observed in a gene given its length and was weighed by the frequency of those mutations in the population. Given genes of equal length (genes A, B and C) high multiplicity can arise from high mutation frequency in the population (gene A), multiple mutated sites (gene B), or a combination of the two. b For comparison among populations, we calculated the relative multiplicity, by normalizing the sum of multiplicities in each population to equal one. Each curve represents the relative gene multiplicity ranked by decreasing multiplicity values for a single population. Solid lines represent populations from the + phage treatment (n = 3) while dashed lines represent populations from the – phage treatment (n = 3). The effect of seed banks on the distribution of multiplicity was determined using a permutational Kolmogorov-Smirnov test.

Fig. 6. Seed banks dampened molecular coevolutionary dynamics between bacteria and phage.

The frequency trajectories of mutant alleles in phage-infected communities (a) without a seed bank and (b) with a seed bank. Each row shows the data for the host and phage of a single community (numbers on right side). Non-synonymous mutations that reached a frequency >0.3 are colored by the gene in which it occurred. The names of genes with high-frequency mutations are provided for each population. See Table S4 for details on genes. c In the - seed bank treatment, the distribution of correlation coefficients between host and phage mutation trajectories skewed negative relative to a null distribution obtained by permuting time labels. In the + seed bank treatment, the distribution resembled that of the null with a slight overabundance of low correlation pairs, consistent with seed bank buffering of host-phage coevolutionary dynamics.