Mutation-induced infections of phage-plasmids

Abstract

Phage-plasmids are extra-chromosomal elements that act both as plasmids and as phages, whose eco-evolutionary dynamics remain poorly constrained. Here, we show that segregational drift and loss-of-function mutations play key roles in the infection dynamics of a cosmopolitan phage-plasmid, allowing it to create continuous productive infections in a population of marine Roseobacter. Recurrent loss-of-function mutations in the phage repressor that controls prophage induction leads to constitutively lytic phage-plasmids that spread rapidly throughout the population. The entire phage-plasmid genome is packaged into virions, which were horizontally transferred by re-infecting lysogenized cells, leading to an increase in phage-plasmid copy number and to heterozygosity in a phage repressor locus in re-infected cells. However, the uneven distribution of phage-plasmids after cell division (i.e., segregational drift) leads to the production of offspring carrying only the constitutively lytic phage-plasmid, thus restarting the lysis-reinfection-segregation life cycle. Mathematical models and experiments show that these dynamics lead to a continuous productive infection of the bacterial population, in which lytic and lysogenic phage-plasmids coexist. Furthermore, analyses of marine bacterial genome sequences indicate that the plasmid backbone here can carry different phages and disseminates trans-continentally. Our study highlights how the interplay between phage infection and plasmid genetics provides a unique eco-evolutionary strategy for phage-plasmids.

Fig. 1. Mutations in a phage repressor region recurrently drives productive infection of the phage-plasmid.

a Productive switch of a phage-plasmid in Tritonibacter mobilis A3R06 was observed after ~40 generations of serial-dilution growth (red). A deletion mutation (11736: GA→G, purple bar) rapidly increased to ~50% relative genotypic frequency within one dilution cycle, before the increase slowed down in the next dilution cycle. A second mutation (11853: C→T, yellow bar) was observed in the last dilution cycle. Planktonic bacterial culture became highly clumpy after the productive switch, as indicated by the sharp decrease in OD600 (blue). For eco-evolutionary trajectories for the other 8 populations temporally-tracked with genomic sequencing, see Fig. S4 and Table S1. b Transmission electron microscope image of the phage-plasmid particle. Imaging was performed for seven times with biological triplicates, all yielding similar results. c Differential expression of phage-plasmid genes before and after observing the mutation. Genes related to phage production were significantly upregulated after the productive switch, in particular the phage structural genes and the phage lysozyme gene. Expression of genes that are housekeeping for plasmid replication and stability were only increased because of copy-number increase of genes. P values are calculated based on Wald test and are adjusted by the Benjamini-Hochberg (BH) procedure. d All 21 mutations identified in 15 independent lines of populations were within a short ~1000 bp region encoding a C1-type phage repressor (orange arrow). Most of mutations are insertions or deletions (purple diamond). Details of these mutations are listed in Table S1. Source data for Figs. 1a, 1c and 1d are provided in the Source Data file.

Fig. 2. Experimental confirmation of reinfection and segregational drift.

a Schematic illustration of reinfection, for which we hypothesize that mutated phages are able to infect hosts lysogenized by wild-type phages. Wild-type and mutated phage-plasmids were showed as green circles and blue/purple/red color circles. See Methods for full experimental details. b Mutated phages with the same genotype were observed across (I) the initial source host population carrying the mutated phage, (II) the victim host population re-infected by the mutated phage and (III) the descendant of (II), supporting our hypothesis of reinfection. Experiments were performed in biological duplicates with two source host populations carrying different genotypes of mutated phages (blue and purple). c Schematic illustration of segregational drift, for which we hypothesize that a host infected by mutated phages is able to generate offspring with only wild-type phages. See Methods for full experimental details. d Each of the 4 single-cell mother colonies (M1~M4) carrying a mixture of mutated phages and wild-type phages was able to generate descendants only carrying the wild-type phages, supporting our hypothesis of segregational drift. Source data of Fig. 2 are provided in the Source Data file.

Fig. 3. A minimal model for phage-plasmid hybrid reproduces the observed eco-evolutionary dynamics.

a Schematic illustration of the model simulation. A question mark is drawn under a host cell when it carries more than one copies of phage-plasmids with different genotypes (a heterozygote). In those cases, the phage-plasmid genotype in descendants becomes stochastic due to segregational drift (e.g., cells 9, 10, 11, and 12). b With reinfection and segregational drift as the only two components, the simulated eco-evolutionary dynamics well matches the observed patterns in the experiment. For the experimentally observed eco-evolutionary dynamics, see Fig. 1a and Fig. S4. See Methods for full details of the model simulation. Source data of Fig. 3 are provided in the Source Data file.

Fig. 4. The same plasmid backbone carrying different phages genes disseminate vast geographic distance.

Tritonibacter mobilis M41-2.2, isolated in Denmark, contains a phage-plasmid (triangle) whose plasmid-related genes are homologous and syntenic to those of Tritonibacter mobilis A3R06 phage-plasmid (square). However, their phage structural genes are very different from each other. Phage structural genes of Tritonibacter mobilis A3R06 phage-plasmid is both homologous and syntenic to that of a chromosome-integrated phage found in Roseobacter sp. SK209-2-6 (circle), which was isolated from deep water column in the Arabian Sea. Background map is from R package ggmap. Source data of Fig. 4 are provided in the Source Data file.