Phylum barrier and Escherichia coli intra-species phylogeny drive the acquisition of antibiotic-resistance genes

Abstract

Escherichia coli is a ubiquitous bacterium that has been widely exposed to antibiotics over the last 70 years. It has adapted by acquiring different antibiotic-resistance genes (ARGs), the census of which we aim to characterize here. To do so, we analysed 70 301 E. coli genomes obtained from the EnteroBase database and detected 1 027 651 ARGs using the AMRFinder, Mustard and ResfinderFG ARG databases. We observed a strong phylogroup and clonal lineage specific distribution of some ARGs, supporting the argument for epistasis between ARGs and the strain genetic background. However, each phylogroup had ARGs conferring a similar antibiotic class resistance pattern, indicating phenotypic adaptive convergence. The G+C content or the type of ARG was not associated with the frequency of the ARG in the database. In addition, we identified ARGs from anaerobic, non-Proteobacteria bacteria in four genomes of E. coli, supporting the hypothesis that the transfer between anaerobic bacteria and E. coli can spontaneously occur but remains exceptional. In conclusion, we showed that phylum barrier and intra-species phylogenetic history are major drivers of the acquisition of a resistome in E. coli.

Keywords: Escherichia coli; antibiotic-resistance genes.

Fig. 1.

Distribution of species/phylogroups in the Escherichia/Shigella EnteroBase. Among the 82 063 genomes of the Escherichia/Shigella EnteroBase (a), we identified 623 genomes from E. albertii , E. fergusonii and Escherichia clades (referred to as ‘others’) (b), and 70 301 genomes of E. coli distributed in seven phylogroups (c). Of note, the F phylogroup in the figure includes both F and G phylogroups [34]. Escherichia clade III and IV are two sub-species belonging to a unique species. Escherichia clade V corresponds to E. marmotae .

Fig. 2.

The 20 ARGs most frequently detected in E. coli from the EnteroBase database sharing 100 % identity with ARGs from AMRFinder. AME, aminoglycoside-modifying enzyme; Cat, chloramphenicol acetyltransferase; DfrA, dihydrofolate reductase type A; Mph, macrolide phosphotransferase; Sul, dihydropteroate synthase; Tet, tetracycline efflux pump.

Fig. 3.

Histogram of the 10 (when available) most representative ARGs in each E. coli phylogroup (referred to as the PPRGs). Only genes with at least 20 occurrences in the database are represented. For each phylogroup, genes are sorted by decreasing frequency.

Fig. 4.

Histogram of the distribution of the targeted antibiotic families in each E. coli phylogroup. The proportion of each targeted antibiotic family is represented as a percentage of each phylogroup.

Fig. 5.

Histogram of the most representative ARGs in each E. coli major ST. Only genes with at least 20 occurrences in the database were represented. For each ST, genes were sorted by decreasing frequency. ‘Others’ represents the 2430 STs with few occurrences of the indicated genes. Each gene was preferentially found in one ST with a P value less than 0.01 with Kruskal–Wallis test and Benjamini–Hochberg correction.

Fig. 6.

Scatter plot of the G+C deviation of ARGs according to their occurrence in EnteroBase and the type of resistance encoded. Each dot corresponds to one cluster of ARGs (90 % nucleotide identity) created using AMRFinder, Mustard and ResfinderFG databases. The G+C content deviation corresponds to the difference between the mean G+C content of E. coli [39] and the mean of the G+C content of each gene in the cluster.