Role of intraspecies recombination in the spread of pathogenicity islands within the Escherichia coli species

Abstract

Horizontal gene transfer is a key step in the evolution of bacterial pathogens. Besides phages and plasmids, pathogenicity islands (PAIs) are subjected to horizontal transfer. The transfer mechanisms of PAIs within a certain bacterial species or between different species are still not well understood. This study is focused on the High-Pathogenicity Island (HPI), which is a PAI widely spread among extraintestinal pathogenic Escherichia coli and serves as a model for horizontal transfer of PAIs in general. We applied a phylogenetic approach using multilocus sequence typing on HPI-positive and -negative natural E. coli isolates representative of the species diversity to infer the mechanism of horizontal HPI transfer within the E. coli species. In each strain, the partial nucleotide sequences of 6 HPI-encoded genes and 6 housekeeping genes of the genomic backbone, as well as DNA fragments immediately upstream and downstream of the HPI were compared. This revealed that the HPI is not solely vertically transmitted, but that recombination of large DNA fragments beyond the HPI plays a major role in the spread of the HPI within E. coli species. In support of the results of the phylogenetic analyses, we experimentally demonstrated that HPI can be transferred between different E. coli strains by F-plasmid mediated mobilization. Sequencing of the chromosomal DNA regions immediately upstream and downstream of the HPI in the recipient strain indicated that the HPI was transferred and integrated together with HPI-flanking DNA regions of the donor strain. The results of this study demonstrate for the first time that conjugative transfer and homologous DNA recombination play a major role in horizontal transfer of a pathogenicity island within the species E. coli.

Figure 1. Phylogenetic unrooted tree using maximum likelihood procedure of the 72 E. coli strains of the ECOR collection .

The B2 strains are arbitrarily grouped apart and the star indicates the midpoint rooting. The tree is based on the simultaneous analysis of six chromosomal housekeeping genes (trpA, trpB, pabB, putP, icd, and polB) and represents the strain evolutionary history ,. Bootstrap values, calculated on 1,000 replicated trees, are shown if higher than 70%. The ECOR strains are indicated by EC following by their number and the phylogenetic group to which they belong (A, B1, D, B2, and UG for ungrouped) . Strains given in red are HPI positive.

Figure 2. Phylogenetic unrooted trees using maximum likelihood procedure of the 37 HPI positive ECOR strains used in this study.

The B2 strains are arbitrarily grouped apart and the star indicates the midpoint rooting. (A) The E. coli phylogenetic tree is based on the simultaneous analysis of six chromosomal housekeeping genes (trpA, trpB, pabB, putP, icd, and polB) and represents the strains' evolutionary history ,. (B) The HPI–phylogenetic tree is based on the simultaneous analysis of six genes of the HPI (int, ybtQ, ybtA, irp2, irp1, and fyuA). Strains that do group together in the HPI-phylogenetic tree, but not in the E. coli phylogeny tree (A) are boxed. These strains reveal identical grouping in the trees of the upstream region of the HPI (C) and the downstream region of the HPI (Figure S4). (C) The tree is based on the region upstream the HPI (UR). Strains EC31 UG and EC72 B1 are indicated in grey boxes, as these strains carry a distinct type of HPI , (Figure S3). Bootstrap values, calculated on 1,000 replicated trees, are shown if higher than 70%. The ECOR strains are indicated by EC following by their number and the phylogenetic group to which they belong (A, B1, D, B2, and UG for ungrouped) .

Figure 3. Percentage of non-synonymous (Ka) and synonymous (Ks) mutations of the studied genes in the 30 HPI positive E. coli strains, for which a DR region is complete.

The genes are ranked according to an increase in the Ks.

Figure 4. Tree representations of the distance matrix between gene tree structures.

The trees represent the path length difference distance (pld) for (A) 30 E. coli HPI–positive strains, for which a DR region is complete and (B) 13 E. coli HPI–negative strains. Numbers are the bootstrap percentages (see text). In (A), the node, with its bootstrap value, delineating on one hand the six housekeeping genes and the strain MLST trees and on the other hand the HPI, UR, and DR trees are in bold.

Figure 5. Conjugative F-Plasmid mediated transfer of the HPI.

(A) Physical map and (B) partial sequence of the region upstream the HPI (UR) as found in donor (NU14 HPI-Cm F'), the transconjugant (AB1157 HPI-Cm F'), and the recipient (AB1157) strains. The partial sequence of yeeI gene given in the box reveals identical sequences in the donor and transconjugant strains, but not in the recipient strain indicating a transfer of the HPI together with the flanking upstream DNA region.