Identification of genes influencing the evolution of Escherichia coli ST372 in dogs and humans

Abstract

ST372 are widely reported as the major Escherichia coli sequence type in dogs globally. They are also a sporadic cause of extraintestinal infections in humans. Despite this, it is unknown whether ST372 strains from dogs and humans represent shared or distinct populations. Furthermore, little is known about genomic traits that might explain the prominence of ST372 in dogs or presence in humans. To address this, we applied a variety of bioinformatics analyses to a global collection of 407 ST372 E. coli whole-genome sequences to characterize their epidemiological features, population structure and associated accessory genomes. We confirm that dogs are the dominant host of ST372 and that clusters within the population structure exhibit distinctive O:H types. One phylogenetic cluster, 'cluster M', comprised almost half of the sequences and showed the divergence of two human-restricted clades that carried different O:H types to the remainder of the cluster. We also present evidence supporting transmission between dogs and humans within different clusters of the phylogeny, including M. We show that multiple acquisitions of the pdu propanediol utilization operon have occurred in clusters dominated by isolates of canine source, possibly linked to diet, whereas loss of the pdu operon and acquisition of K antigen virulence genes characterize human-restricted lineages.

Keywords: E. coli; ExPEC; ST372; canine; genomic epidemiology; pathogen evolution; pdu operon.

Fig. 1.

Genomic epidemiological features of the ST372 genome collection. (a) Distribution of continents stratified by source, (b) distribution of phylogenetic clusters (see Fig. 2) stratified by source, (c) distribution of sources stratified by O:H type, (d) distribution of phylogenetic clusters stratified by O:H type, (e) distribution of sources stratified by fimH allele and (f) distribution of phylogenetic clusters stratified by fimH allele. Total number of observations for each variable are displayed above each bar. See Fig. S1 for source-stratified graphs of OH and fimH data.

Fig. 2.

Phylogeny of ST372. A maximum-likelihood core gene phylogeny based on alignment of 3493 genes identified in all ST372 sequences. Metadata displayed from inner to outermost rings display fastbaps cluster, source, continent, O:H type and fimH allele.

Fig. 3.

Low pairwise SNP distance heatmap between ST372 isolated from canine and non-identical sources. Metadata are represented for each sequence in coloured bars for each row and column. Each cell in the heatmap represents a pairwise SNP comparison between two sequences represented by white to purple gradient. Grey squares represent distances of >30 SNPs.

Fig. 4.

Map of cluster-associated genes aligned to the core gene phylogeny. Columns display metadata (cluster, source, O:H type) followed by genes found to be over-represented in clusters M, G, L and J. Gene presence/absence is indicated with the same colour as the associated cluster. Coloured dots next to gene names indicate where an alternative allele of the same gene is identified in another cluster. Bracketed labels indicate gene functions for select genes/operons.

Fig. 5.

Schematic representations of representative gene loci containing cluster-associated genes (see Fig. 4) and predicted genomic islands. (a) Cluster G-associated genes identified in proximity to pdu operon (cluster M-associated) represented by cluster G sequence MVC107, (b) cluster M-associated genes (pdu operon), as seen in M1 and M2 genotypes from Fig. 4. (represented by sequences MVC121 and MVC18, respectively), (c) cluster M-associated associated genes (kps operon, eps, gsp) represented by M2 genotype sequence MVC18 and (d) cluster M-associated genes (kps operon, eps, gsp) represented by M3 genotype sequence MVAST6839.