Diverse phage communities are maintained stably on a clonal bacterial host

Abstract

Bacteriophages are the most abundant and phylogenetically diverse biological entities on Earth, yet the ecological mechanisms that sustain this extraordinary diversity remain unclear. In this study, we discovered that phage diversity consistently outstripped the diversity of their bacterial hosts under simple experimental conditions. We assembled and passaged dozens of diverse phage communities on a single, nonevolving strain of Escherichia coli until the phage communities reached equilibrium. In all cases, we found that two or more phage species coexisted stably, despite competition for a single, clonal host population. Phage coexistence was supported through host phenotypic heterogeneity, whereby bacterial cells adopting different growth phenotypes served as niches for different phage species. Our experiments reveal that a rich community ecology of bacteriophages can emerge on a single bacterial host.

Fig. 1. Phage communities were diverse, common, and predictable on a single host.

(A) Illustration of the experimental workflow. (1) Isolate and genome sequence 27 E. coli phage species from natural samples. (2) Infect E. coli with 10 different starting communities each containing a random set of 10 species from our collection. (3) Filter, dilute, and passage the phage community onto fresh bacteria every 24 hours. (4) At day 12 determine the composition through deep sequencing and plaque morphologies. (B) Diverse plaque morphologies of the species in our collection, with each species given a different alphabetical name. (C) Phylogenetic tree including all 27 phage species in our collection and model phages T4, T5 and lambda as references. The numbers above the branches are GBDP pseudo-bootstrap support values from 100 replications and the branch length is scaled by the D0 formula. The branches were colored based on family classification (see fig. S1 or table 1 for family names). (D) Every community contained multiple diverse phage species at the final passage, as determined through deep sequencing (N = 3 biological replicates). “No” indicated that the species was present in the starting community but was not detectable at the final passage in any biological replicate. “Sometimes” indicated that the species was present in some replicates, while “Yes” indicated that the species was present in all replicates. (E) Communities differed in their composition and relative abundance of species, but were similar across the three biological replicates for the same community.

Fig. 2. Diversity was stable throughout community passaging.

Phage species abundances over time for communities 2, 3, 6, and 7 (top to bottom), quantified using plaque morphologies. (A) The three biological replicates (labelled above the charts) of four communities from Fig. 1D showed the early dominance of final community members. The average limit of detection for each replicate is indicated by the black dotted line or is otherwise indicated by the x-axis. Phage species shown with dashed lines were absent from the final community. Phage abundances were quantified on passage 0 and then every other passage starting at passage 1 until passage 11 and 12. (B) Reconstituted 2-member communities maintained both phage species irrespective of starting proportion. The initial relative abundances of each phage species varied from ∼1% to ∼99% of the total phage population and are listed above the charts. From top to bottom these represent the ratio of phage N:S, N:Y, M:U, and E:R. For simplicity we reconstituted community 7 with only phages E and R, although the original communities also had phage W. The limit of detection is only shown for replicates where one of the two members was undetectable at least once during passaging. Phage abundances were measured through plaquing on passage 0 and then every day starting at passage 2 until passage 9 (communities 2, 3, 6) or passage 8 (community 7).

Fig. 3. Phage-phage interactions were primarily negative.

Pairwise interactions for 2-species and 3-species phage communities. To determine the direction of interactions, we measured a species’ abundance when infecting alone versus when co-cultured with another phage species. All phage samples were passaged for 1-3 days to allow populations to reach equilibrium before measuring each species abundances through top agar plating. Outlined dots represent the abundance of each phage in a competition experiment in each biological replicate (N=3-6 replicates); colored bars are the mean ±SE. For statistical comparisons: ‘ns’ is a non-significant interaction and asterisk indicates significant p<0.05 using a two-sample two-tailed student’s t-test, after adjusting for multiple comparisons using the Benjamini-Hochberg procedure. For each community, we plot a graphical representation of the significant interactions, with green arrows for positive interactions, and red for negative interactions. (A) Example results showing how ecological interactions are determined through changes in the pink phage species’ abundance. During coinfection, the yellow phage species either had no impact, increased, or decreased the abundance of the pink phage species, relative to when the pink phage species was infecting alone. We classify these interactions as neutral, positive, or negative, respectively. (B) Phage interactions for 2-species communities 2, 3, 6 and 7 (left to right). From left to right, adjusted p-values are: 0.006; 0.004; 0.033; 0.027; 0.003; 0.147; <1e-04; 0.012. (C) Same as (B), but showing pairwise interactions between 3-member community 8 and (D) community 9. From left to right for (C) and (D), adjusted p-values are: 0.131, <1e-03, 0.968, 0.025, 0.357, 0.182, <1e-03, 0.047, 0.432, 0.089, 0.001, 0.003.

Fig. 4. Host phenotypic heterogeneity supported phage diversity.

Specifically, a mixture of fast- and slow-growing cells supported phage coexistence. (A) We infected bacterial cultures at different ages with community 2 that had been passaged to equilibrium and then determined the relative abundance of each phage species through plaquing. Culture age determined which species was dominant: In the 3-hour culture, phage N was dominant; in the 24-hour culture, the proportions were relatively equal; and in the 72-hour culture, phage N was in the minority [95% confidence intervals = 0.91 to 0.97, 0.53 to 0.64, and 0.15 to 0.28, respectively]. Shifting from the youngest culture (3-hour) to the oldest (72-hour) therefore reversed the dominant species (P < 2 x 10−16; two-sample, two-tailed z test for equality of proportions) (n = 3). (B) Infections with phage N or phage S alone qualitatively agreed with the pattern in (A). Phage N achieved a higher abundance in the 3-hour versus the 24- or 72-hour cultures (P = 0.005 and P < 1 x 10−6, respectively, with one- tailed student’s t tests using the Benjamini-Hochberg correction for multiple comparisons) and a higher abundance in the 24-hour versus the 72-hour culture (P < 1 x 10−6). By contrast, phage S achieved a higher abundance on the 72-hour versus both the 24- and 3-hour cultures (P = 0.004 and 0.002, respectively). Phage S achieved a similar abundance in the younger cultures (3-hour versus 24-hour) (P = 0.56) (n = 3). (C) Schematic for determining the relative fitness of a phage species on small and large cells from the same 24-hour E. coli culture: Cells were separated by size, reconcentrated, and infected with community 2. Each species’ abundance was determined after incubation through top agar plating. (D) Phage N and S abundances after infecting small and large subpopulations (n = 3). (E) The relative abundance of each phage species from (D) with the dotted line indicating the starting proportion of phage N. (F) Phage N decreased in relative abundance on small cells and increased on large cells (P = 0.033, P < 1 x 10−4, respectively, using a one-tailed one-sided t test, null hypothesis true mean = 0). The relative fitness of phage N is its final relative abundance divided by its starting relative abundance of 0.66. Gray dots indicate n = 7 total replicates from (D) and (E) and an additional flow cytometry run (fig. S17). The gray-shaded regions indicate ± SE and are bisected by a black line indicating the mean. (G) To investigate the molecular basis of coexistence, we coinfected cultures of different ages with phage U and either phage T7 or T7Dgp5.7 (i.e., T7D). T7 generally reached a higher abundance than phage U after coinfection (fig. S19A), so we added T7 and T7Dgp5.7 at a lower MOI (0.5) than phage U (5.0) to more easily detect growth differences. Each bar plot shows the final relative abundance of each phage after 24 hours of incubation (n = 3). (H) We calculated the log10 fold change in absolute abundance of phage T7 or T7Dgp5.7, in competition with phage U, on different cultures from (G). Each dot reflects the log10 fold change relative to each species’ starting abundance ± SE (n = 3). When in community with phage U, T7Dgp5.7 did worse than T7 on the older cultures (24-hour culture: P = 0.003; 72-hour culture: P = 0.001, using one-tailed t test), but there was no significant difference on the 3-hour culture, suggesting that the phage protein Gp5.7 is important in competition for slower-growing cells. Asterisks indicate significant comparisons.