Insights into the evolution of pathogenicity of Escherichia coli from genomic analysis of intestinal E. coli of Marmota himalayana in Qinghai-Tibet plateau of China

Abstract

Escherichia coli is both of a widespread harmless gut commensal and a versatile pathogen of humans. Domestic animals are a well-known reservoir for pathogenic E. coli. However, studies of E. coli populations from wild animals that have been separated from human activities had been very limited. Here we obtained 580 isolates from intestinal contents of 116 wild Marmot Marmota himalayana from Qinghai-Tibet plateau, China, with five isolates per animal. We selected 125 (hereinafter referred to as strains) from the 580 isolates for genome sequencing, based on unique pulse field gel electrophoresis patterns and at least one isolate per animal. Whole genome sequence analysis revealed that all 125 strains carried at least one and the majority (79.2%) carried multiple virulence genes based on the analysis of 22 selected virulence genes. In particular, the majority of the strains carried virulence genes from different pathovars as potential 'hybrid pathogens'. The alleles of eight virulence genes from the Marmot E. coli were found to have diverged earlier than all known alleles from human and other animal E. coli. Phylogenetic analysis of the 125 Marmot E. coli genomes and 355 genomes selected from 1622 human and other E. coli strains identified two new phylogroups, G and H, both of which diverged earlier than the other phylogroups. Eight of the 12 well-known pathogenic E. coli lineages were found to share a most recent common ancestor with one or more Marmot E. coli strains. Our results suggested that the intestinal E. coli of the Marmots contained a diverse virulence gene pool and is potentially pathogenic to humans. These findings provided a new understanding of the evolutionary origin of pathogenic E. coli.

Figure 1

Heat map of 22 virulence genes in 125 intestinal E. coli strains of M. himalayana. The presence of specific virulence gene was indicated by filled square with pathovar-specific colors as shown in the color legend. The strain numbers were listed at the bottom of panel.

Figure 2

Primordial virulence genes carried by intestinal commensal E. coli of M. himalayana. (A–H) Gene trees for virulence genes with early diverged alleles from Marmot E. coli. Each tree represents different virulence genes as indicated. Alleles from Marmot strains were highlighted in red color with strain names followed in brackets by number of strains and the phylogroup of the strain. Alleles on chromosome or plasmid were labeled with c or p, respectively. For lpfA, eaeA and papA, non-Marmot E. coli alleles were represented by allele type as classified in published studies. For other genes, a representative strain was used and in brackets following the strain name is pathovar type. The numbers above or below the branches represent percentage bootstrap support.

Figure 3

The phylogenetic relationship and virulence gene content of intestinal E. coli of M. himalayana and 54 representative E. coli genomes. The tree was constructed with neighbor-joining algorithm-based 540 core genes. The presence of pathovar-specific genes in each strain was indicated by colored boxes with the color scheme shown. A strain is positive for a specific pathovar if at least one pathovar-specific gene is present. White space represents absence of pathovar-specific genes for a given pathovar in that strain. Strains from Marmots were highlighted in red. Phylogroups were labeled in the inner circle as well as by coloring of the branches.

Figure 4

The evolution of major human pathogenic lineages of E. coli relative to their closely related Marmot E. coli strains. Key non-Marmot E. coli strains are given by name and Marmot E. coli strains were highlighted with a red dot. Outside the circle in the different shaded circles are a detailed depiction of the pathogenic E. coli lineages with gain and loss of genes. Gain of gains is marked by a black arrow, while loss of genes is marked by a vertical green bar. Escherichia Clades I–V strains were used as an outgroup.