A phylogenomic analysis of Escherichia coli / Shigella group: implications of genomic features associated with pathogenicity and ecological adaptation

Abstract

Background: The Escherichia coli species contains a variety of commensal and pathogenic strains, and its intraspecific diversity is extraordinarily high. With the availability of an increasing number of E. coli strain genomes, a more comprehensive concept of their evolutionary history and ecological adaptation can be developed using phylogenomic analyses. In this study, we constructed two types of whole-genome phylogenies based on 34 E. coli strains using collinear genomic segments. The first phylogeny was based on the concatenated collinear regions shared by all of the studied genomes, and the second phylogeny was based on the variable collinear regions that are absent from at least one genome. Intuitively, the first phylogeny is likely to reveal the lineal evolutionary history among these strains (i.e., an evolutionary phylogeny), whereas the latter phylogeny is likely to reflect the whole-genome similarities of extant strains (i.e., a similarity phylogeny).

Results: Within the evolutionary phylogeny, the strains were clustered in accordance with known phylogenetic groups and phenotypes. When comparing evolutionary and similarity phylogenies, a concept emerges that Shigella may have originated from at least three distinct ancestors and evolved into a single clade. By scrutinizing the properties that are shared amongst Shigella strains but missing in other E. coli genomes, we found that the common regions of the Shigella genomes were mainly influenced by mobile genetic elements, implying that they may have experienced convergent evolution via horizontal gene transfer. Based on an inspection of certain key branches of interest, we identified several collinear regions that may be associated with the pathogenicity of specific strains. Moreover, by examining the annotated genes within these regions, further detailed evidence associated with pathogenicity was revealed.

Conclusions: Collinear regions are reliable genomic features used for phylogenomic analysis among closely related genomes while linking the genomic diversity with phenotypic differences in a meaningful way. The pathogenicity of a strain may be associated with both the arrival of virulence factors and the modification of genomes via mutations. Such phylogenomic studies that compare collinear regions of whole genomes will help to better understand the evolution and adaptation of closely related microbes and E. coli in particular.

Figure 1

The distribution of the number of LCBs shared by the strains. Before being filtered, 528 and 618 LCBs were shared by 2 and 36 strains, respectively (orange). After being filtered using a cutoff of 1.01, 412 and 35 LCBs shared by 2 and 36 strains, respectively, remained (blue). A cutoff value of 1.01 means that the length of the gaps in the LCBs is less than 1% of the length of the non-gap regions.

Figure 2

The evolutionary and similarity whole-genome phylogenies ofE. coli/Shigella. (A) The maximum parsimony tree was constructed using the concatenated LCBs that are shared by all of the 36 sequenced strains. The reliability of the topology was assessed by bootstrapping with 1000 pseudo-replicates. The phylogroup (A, B1, B2, D, E, F, S1, S3, SS or SD1) and pathotype (Commensal, InPEc, ExPEc, or Shigellosis) of each strain is indicated on the right. (B) The neighbor-joining phylogeny was constructed using the Jaccard distances. Only the LCBs that were absent in at least one strain and present in at least two strains were analyzed. The reliability of the topology was assessed by re-sampling the collinear regions 1000 times.

Figure 3

Substitution variants revealed by the multiple sequence alignment of the seven DNA segments (label 1047). The red inset highlights the LexA binding site of ydjM, and the orange inset indicates the start and stop codons of ydjM. Yellow and green shading highlight the synonymous and non-synonymous nucleotide substitutions, respectively, within ydjM; red shading highlights nucleotide substitutions within the 3’-region of ydjM; and blue shading shows nucleotide substitutions within the LexA binding site of ydjM.

Figure 4

The number and length distributions of the LCBs. (A) Distributions of the unfiltered LCBs. (B) Distributions of the filtered LCBs with a cutoff value of 1.01.