Bacterial warfare is associated with virulence and antimicrobial resistance

Abstract

Bacteria have evolved a diverse array of mechanisms to inhibit and kill competitors. However, why some bacteria carry such weapons while others do not remains poorly understood. Here we explore this question using the genomics of the bacteriocins of E. coli as a model system, which have large well-annotated bioinformatic resources. While bacteriocins occur widely, we find that carriage is particularly associated with pathogenic extra-intestinal (ExPEC) strains. These pathogens commonly carry large plasmids encoding bacteriocins alongside virulence factors and antimicrobial resistance mechanisms. Across all strains, we find many orphan immunity proteins, which protect against bacteriocins and suggest that these bacterial weapons are important in nature. We also present evidence that bacteriocin toxins readily move between strains via plasmid transfer and even between plasmids via transposons. Finally, we show that several E. coli bacteriocins are widely shared with the pathogen Salmonella enterica, further cementing the link to virulence. Our work suggests that the bacteriocins of E. coli are important antibacterial weapons for dangerous antimicrobial-resistant strains.

Fig. 1. The bacteriocins of E. coli are diverse and associated with particular pathotypes.

A Abundance of different bacteriocins found throughout 2601 E. coli genomes. B Core-genome phylogeny of 2601 E. coli strains with predicted pathotype (inner ring), presence of bacteriocin plasmids (middle ring) and chromosomal bacteriocins (outer ring). C Bacteriocin plasmids are not evenly distributed between non-pathogenic and pathogenic strains: bacterial genotypes that cause extraintestinal infections are particularly likely to carry a bacteriocin. D Bacteria with iron-uptake targeting bacteriocin plasmids are enriched in iron-uptake genes (orange) compared to strains without bacteriocin plasmids (blue) (MCMCglmm, posterior μ = 6.07, 95% CI = [5.500, 6.669], pMCMC < 3 × 10⁻⁴). Removing genes encoded on these plasmids (green) reveals that that the enrichment is driven by the fact that bacteriocin plasmids carry iron-uptake genes rather than the plasmids associating with genomes that carry high numbers of iron uptake genes (MCMCglmm, posterior μ = 0.4, 95% CI = [−0.1709, 0.9314]), pMCMC = 0.15).

Fig. 2. Bacteriocin plasmids carry a range of virulence factors and AMR genes and are enriched in iron-uptake genes.

A Bacteriocin plasmids are enriched in genes associated with iron-uptake genes as compared to other plasmids. B Bacteriocin plasmids are more likely to encode iron-uptake genes than non-bacteriocin plasmids, even when accounting for plasmid length (GLM, β = 0.06, p < 2 × 10⁻¹⁶). Shaded areas indicate 95% CIs. C Iron-targeting bacteriocins occur alongside diverse iron acquisition systems and other loci involved in extraintestinal, enterohaemorrhagic and enterotoxic pathotypes. Bacteriocin plasmids can also carry antibiotic resistance genes for a wide range of antibiotic classes. D Different E. coli pathotypes tend to carry different types of bacteriocin plasmid. E. coli strains that cause intestinal infections often carry plasmids with iron-targeting bacteriocin but without siderophores, while ExPEC strains are enriched for all types of iron-targeting bacteriocin plasmids.

Fig. 3. Bacteriocin plasmids are diverse in size, replication machinery and mobilisation, and orphan immunity genes occur widely.

A Bacteriocin genes occur alongside a wide range of replicon types. The 19 most abundant replicons out of a total of 31 are shown. B Bacteriocin plasmids vary in size and broadly divide into small and large plasmids. Each point represents a single plasmid. Plasmids with multiple colicins are represented in multiple columns, total colicins = 880. Lower and upper hinges indicate the 1st and 3rd quartiles. Central lines indicate the median. Whiskers indicate the furthest point within 1.5× the interquartile range. C Bacteriocin plasmids are commonly mobilisable with larger plasmids encoding conjugation machinery and smaller plasmids carrying the genes to mobilise via a helper plasmid. D Clustering of 5353 E. coli plasmids from PLSDB. Bacteriocin plasmids are diverse and spread throughout the E. coli plasmidome. Each circle represents an individual plasmid, edges indicate a Mash distance below the cut off (0.03). Only plasmid communities which contain ≥ 5 members are represented.

Fig. 4. Specific bacteriocin genes are mobilisable and found across diverse plasmids.

A Phylogeny of colicin Ia genes (cia-toxin and iia-immunity gene) with the plasmid communities they were identified with (left). Alignment of three ColIa plasmids shows a high level of diversity between plasmids, where either a small region of DNA or even just the ColIa genes (cia and iia) are shared between plasmids (Regions of similarity > 2000bp are shown in blue). B Percentage of bacteriocin (or immunity) genes predicted to occur within a composite transposon. C Example of an alignment of two colicin plasmids that share a shared predicted composite transposon that carries a colicin.

Fig. 5. Colicin plasmids are shared among multiple genera within Enterobacteriaceae.

A Louvian clustering of all plasmids in the genera Escherichia, Klebsiella, Salmonella, Citrobacter and Shigella. For clarity, only plasmid clusters with contain at least 1 colicin plasmid are shown. Each circle represents an individual plasmid and edges indicate a mash distance of < 0.03. Non-colicinogenic plasmids are shown as smaller circles. B All colicin plasmids identified in 5 different genera of Enterobacteriaceae are represented in a chord diagram. Connections show plasmids with ≥ 99% ANI shared between genera.