Fig 1.
The distribution of plasmid-encoded essential genes on plasmids and chromosomes.
A, The distribution of Escherichia protein families on plasmids in chromosomes (in grey). Essential genes (red) are found in most of the chromosomes and rarely on plasmids, with ssb as an exception. Transposable elements (blue) are frequent on both chromosomes and plasmids. IS66 is the most widely spread gene in Escherichia strains. IS66 was split into several protein families in our analysis, which are depicted by multiple data points. Antibiotic resistance (AMR) genes (yellow) are presented for comparison. The AMR genes can be roughly divided into two groups: the first group aligns along the y-axis hence it is more frequently found on chromosomes; those chromosome-encoded AMR genes are typically related to persistence and resilience functions. The second group aligns along the x-axis hence it is more frequently found on plasmids; those plasmid-encoded genes are typically related to antibiotics resistance functions (Wein et al. 2020). B, Distribution of plasmid size, mobility group and AMR for plasmids encoding an essential gene. Most of the ssb-coding plasmids are conjugative (red) while the majority of groEL/S coding plasmids are mobilizable (green). The remaining essential genes are encoded on plasmids that are often non-mobilizable (and non-AMR) (purple).
Table 1.
Data and phylogenetic analysis of essential genes encoded on plasmids.
Essential list indicates the source of essential genes: TKP is TraDIS-Keio-PEC, TP is TraDIS-PEC, T is TraDIS (Goodall et al. 2018); LB, M9 and GMM means this gene is essential in each of the three medias (Rousset et al. 2020). Relaxed selection shows evidence for relaxed selection on plasmid homologs’ branch was found using HyPhy-RELAX (Murrell et al. 2012). Tree topology shows the conclusions from the phylogenetic reconstruction of essential genes on plasmids and chromosomes: split between plasmids and chromosomes describe phylogenies that had a deep split between plasmid and chromosomal homologs (i.e., they are diverged); LGT events from chromosome to plasmid are labeled by transfer to plasmid; LGT events from plasmid to chromosome are labeled by transfer to chromosome; mixed is used to label phylogenies that contain deep divergence and LGT events. For gene trees with evidence for gene transfer (groEL, groES), we further tested the support in the plasmid/chromosome split while excluding the putatively transferred gene. In all tested cases the result showed that the constrained tree could not be rejected in a topology test, thus validating the gene transfer inference. All of the transfer event from plasmid to chromosome are better described as translocation of the plasmid homolog to chromosome.
Fig 2.
Phylogeny of plasmid-encoded essential genes in Escherichia.
A, Phylogeny of the single-stranded DNA-binding protein Ssb. Ssb homologs encoded on plasmid are shown in pink and encoded on chromosome are shown in blue. The plasmid ssb gene termed ssf is most abundant across the plasmid Ssb homologs. B, The conserved neighborhood (conserved syntenic block, CSB) of ssb and the chaperone groE encoded on plasmids (S3 and S4 Tables). C, Phylogeny of the chaperonin GroES (left) and GroEL (right). The triangle symbol marks the branch split that was constrained in the test for an alternative tree topology. D, Growth measurements of E. coli MGM100 and MGM100 carrying the plasmid-derived GroE (marked by *). The exceptional growth behavior observed in one E. coli MGM100 replicate should be considered an outlier since we found no evidence for other explanations to that result.
Fig 3.
Characterization of plasmids encoding an essential gene.
A, Genomic map of unstable pGroE and stable pGroE-S plasmids that were introduced into E. coli MG1655 (plasmid is non-essential) and E. coli MGM100 (plasmid is essential) hosts. B, Plasmid loss frequency of plasmids lacking (-) groE or encoding (+) groE in their genome in the host E. coli MG1655 (H0: L+>L-, P = 0.0247 using Wilcoxon test n = 6) C, Plasmid loss frequency of pGroE-S (stable) and pGroE (unstable) after incubation at 37°C or at 42°C (H0: L42>L37, P = 0.0091 using Wilcoxon test, n = 6). D, Growth rates of host strains E. coli MG1655 and E. coli MGM100 carrying pGroE (unstable) or pGroE-S (stable) after growth at 37°C and 42°C (H0: Gr42>Gr37, P = 0.051 using Wilcoxon test, n = 9).
Fig 4.
Evolution of plasmids encoding an essential gene.
A, Evolution experiment of pGroE (unstable) and pGroE-S (stable) in the host strains E. coli MG1655 (non-essential, left) and MGM100 (essential, right). B, Relative fitness of plasmid-carrying ancestral and evolved populations. Pairwise competition experiments between the plasmid-carrying strains and tagged wild-type (Tmr). A significant negative fitness effect could be observed for all ancestral populations (w: 0.69, P = 1.526x10-5 using Wilcoxon test, n = 4 per population) while the fitness of evolved populations increased compared to the ancestral state (w: 0.93, P = 9.05x10-7 using Wilcoxon test, n = 4 per population). C, Plasmid copy number (PCN) of plasmid-carrying ancestral and evolved populations. The median pGroE PCN is significantly different between the ancestral and evolved populations (PMG1655 = 0.03, PMGM100 = 0.0016, using Wilcoxon test, n = 9 per population), with the PCN being lower in the evolved populations in both hosts.