Niche and local geography shape the pangenome of wastewater- and livestock-associated Enterobacteriaceae

Abstract

Escherichia coli and other Enterobacteriaceae are diverse species with "open" pangenomes, where genes move intra- and interspecies via horizontal gene transfer. However, most analyses focus on clinical isolates. The pangenome dynamics of natural populations remain understudied, despite their suggested role as reservoirs for antimicrobial resistance (AMR) genes. Here, we analyze near-complete genomes for 827 Enterobacteriaceae (553 Escherichia and 274 non-Escherichia spp.) with 2292 circularized plasmids in total, collected from 19 locations (livestock farms and wastewater treatment works in the United Kingdom) within a 30-km radius at three time points over a year. We find different dynamics for chromosomal and plasmid-borne genes. Plasmids have a higher burden of AMR genes and insertion sequences, and AMR-gene-carrying plasmids show evidence of being under stronger selective pressure. Environmental niche and local geography both play a role in shaping plasmid dynamics. Our results highlight the importance of local strategies for controlling the spread of AMR.

Fig. 1. Overview of the diverse Escherichia coli isolates in this study.

(A) Relative sampling locations of the farms (cattle, pig, and sheep) and wastewater treatment plants (WwTWs) in this study, sampled at three different TPs. (B) Schematic illustration of the sampling, culture, and sequencing workflow, resulting in high-quality genome assemblies with a median of one circularized chromosome and two circularized plasmids per assembly. (C) Mid-point rooted core genome phylogeny of E. coli isolates (n = 488), with tips colored by phylogroup and ring colors showing sampling niche. Inset panel at center of phylogeny shows phylogroup abundances (as proportion of isolates) from different sampling niches.

Fig. 2. The plasmid-borne component of the pangenome is structured by niche and phylogeny, with greater variation than in the chromosomal component.

Plots are shown for isolates (A) across Enterobacteriaceae and (B) within E. coli, for both the chromosomal component of the pangenome and the plasmid-borne component analyzed separately. Color indicates (A) genus within Enterobacteriaceae and (B) phylogroup within E. coli. Stacked bar charts in the center of each show the variance in gene content explained by niche, phylogeny (genus or phylogroup), and their interaction. The plasmid-borne component has greater residual variance than the chromosomal component, with a comparatively stronger niche-phylogeny interaction (darkest shaded bar).

Fig. 3. Distinct plasmid lifestyles between AMR and non-AMR plasmids.

(A) Plasmid length (x axis) and inferred copy number (y axis) of all circularized plasmids (n = 2292), faceted by genus. Plasmids with ≥1 AMR gene (colored points) tended to be larger and present in lower copy numbers. (B) Relative GC content of all plasmids to their host chromosome for all circularized plasmids present in an assembly with a circularized chromosome (n = 1753 plasmids across 616 isolates), split by predicted plasmid mobility. Boxplots are additionally shown classifying plasmids within predicted mobility types by the number of AMR genes carried: those ≥1 AMR gene (red) or no AMR genes (black). Comparisons with P values are shown for all plasmids within a predicted mobility class. (C) Length distributions of plasmid clusters (see Materials and Methods).

Fig. 4. The interplay of phylogeny and niche in the E. coli pangenome.

(A) Pairwise comparisons of GRR for chromosomal genes show that chromosomal GRR falls off rapidly at small patristic distances, followed by an approximately linear decrease. Fits show intra-ST comparisons (thick black line), all comparisons (thin black line), and a linear model (dashed black line). Violin plots above show the distribution of patristic distances depending on the relative sample source of the two isolates in the pairwise comparison (white boxplot: median and IQR; black point: mean), showing that even isolates cultured from the same sample (same farm and same TP) span equivalent diversity to isolates cultured from different locations. (B) Coefficients from a linear model for chromosomal GRR with an interaction term with patristic distance (excluding intra-ST comparisons). (C) Variance explained by phylogeny and geography for chromosomal and plasmid GRR. (D) GRR for plasmid-borne genes with patristic distance. Fits show intra-ST comparisons (thick red line), all comparisons (thin red line), and a linear model (dashed red line). Inset panel shows left-hand region of the plot with only intra-ST comparisons, with chromosomal GRR relationship also shown (gray points, black line). (E) Plasmid GRR comparisons shown by isolate sources, excluding intra-ST comparisons. Colors on the x axis are the same as in (A). Plots include all E. coli isolates with a circularized chromosome (n = 363).