Insights into evolution and coexistence of the colibactin- and yersiniabactin secondary metabolite determinants in enterobacterial populations

Abstract

The bacterial genotoxin colibactin interferes with the eukaryotic cell cycle by causing dsDNA breaks. It has been linked to bacterially induced colorectal cancer in humans. Colibactin is encoded by a 54 kb genomic region in Enterobacteriaceae. The colibactin genes commonly co-occur with the yersiniabactin biosynthetic determinant. Investigating the prevalence and sequence diversity of the colibactin determinant and its linkage to the yersiniabactin operon in prokaryotic genomes, we discovered mainly species-specific lineages of the colibactin determinant and classified three main structural settings of the colibactin-yersiniabactin genomic region in Enterobacteriaceae. The colibactin gene cluster has a similar but not identical evolutionary track to that of the yersiniabactin operon. Both determinants could have been acquired on several occasions and/or exchanged independently between enterobacteria by horizontal gene transfer. Integrative and conjugative elements play(ed) a central role in the evolution and structural diversity of the colibactin-yersiniabactin genomic region. Addition of an activating and regulating module (clbAR) to the biosynthesis and transport module (clbB-S) represents the most recent step in the evolution of the colibactin determinant. In a first attempt to correlate colibactin expression with individual lineages of colibactin determinants and different bacterial genetic backgrounds, we compared colibactin expression of selected enterobacterial isolates in vitro. Colibactin production in the tested Klebsiella species and Citrobacter koseri strains was more homogeneous and generally higher than that in most of the Escherichia coli isolates studied. Our results improve the understanding of the diversity of colibactin determinants and its expression level, and may contribute to risk assessment of colibactin-producing enterobacteria.

Keywords: Citrobacter; Escherichia coli; Klebsiella; cytopathic effect; high pathogenicity island; polyketide; secondary metabolite.

Fig. 1.

Schematic representation of the genomic architecture of the pks island (ca. 54 kb) present in Escherichia coli strain M1/5. The 19 genes within the island are coloured with respect to their function. The island codes for a phosphopantetheinyl transferase (clbA, in purple), a transcriptional autoactivator (clbR, in red), multiple core biosynthetic genes (clbB–clbL, clbN, clbO and clbQ in blue), a transporter (clbM, yellow), a peptidase (clbP, in green) and a resistance factor (clbS, orange).

Fig. 2.

Maximum-likelihood-based phylogenetic analysis of the colibactin and yersiniabactin determinants. (a) Phylogenetic tree of the colibactin gene cluster (collapsed), and (b) phylogenetic tree of the corresponding ybt determinants (collapsed) using the genetically distant K. michiganensis strains as an outgroup [38]. Additionally, the yersiniabactin sequence type (YbST) as defined by Lam and colleagues [38] associated with individual bacterial clades is indicated. The branch colours in both trees depict the prominent bacterial sequence type of the clade.

Fig. 3.

Maximum-likelihood-based phylogeny of the colibactin gene cluster detected in 2169 enterobacterial genomes. Every leaf represents a single sequence variant of the clb gene cluster, which can be allocated to different lineages and clades. From innermost to outermost, the first circle indicates the species harbouring the clb determinant; the second circle shows the Escherichia coli phylogroup, the third circle shows the presence/absence of the ybt operon; the fourth circle shows the yersiniabactin sequence types (YbST) of the ybt determinant (from Fig. 1b) that correspond to the pks island lineage present in the individual genome. The fifth circle shows the different colibactin sequence types (CbST) of the clb gene cluster. The branch colours in the centre of the tree depict the prominent bacterial sequence types (Fig. 1). The large conserved Escherichia coli phylogroup B2 clade is separated from the large Klebsiella clade with a faint broken line.

Fig. 4.

Structural variation of the colibactin and yersiniabactin-encoding chromosomal region in Escherichia coli, Enterobacter hormaechei, K. pneumoniae, K. aerogenes and C. koseri. The different genetic structures and chromosomal insertion sites of the colibactin and/or yersiniabactin determinants found within the three main structural classes are shown. The clb gene cluster (teal green), T4SS module (purple), ybt gene cluster (pink), integrase genes (green), the conserved sets of genes (Table S5) that are present up/downstream of the two polyketide determinants, classed into sets (blue boxes), and the Fe/Mn/Zn module (yellow) are shown. The number of genomes included in the tested set of genomes that harbour the different structural variants is indicated in parentheses. The colibactin–yersiniabactin chromosomal regions that do not conform to these major structures are as shown in Fig. S7.

Fig. 5.

The structural organization and GC profiles of the clb determinants in the five most genetically distant bacterial strains (according to Fig. 2a). The genes that make up the homologous pks gene cluster found in F. perrara and the most distant clb determinants present in Serratia marcescens, Erwinia oleae, K. michiganensis and Escherichia coli (phylogroup E) strain 14696-7 are depicted. The GC profile of the gene cluster in the different strains is shown alongside with the colours underlining the different species. SAM genes and clbA homologues (*) are shown downstream of the pks gene cluster in F. perrara and upstream of the clb determinant in K. michiganensis and Erwinia oleae. The gaps in assembly are shown with white spaces.

Fig. 6.

Comparison of colibactin production of different strains accessed by quantification of the precolibactin cleavage product N-Asn-d-myristol. This assay enabled us to compare the ability of different strains and different species to produce colibactin under controlled conditions in vitro. Measurements were conducted based on three biological replicates, and means with standard deviations are as shown.

Fig. 7.

Schematic representation of the predicted evolution of the colibactin–yersiniabactin genomic region in Enterobacteriaceae. The different elements of this region, i.e. the clb determinant (teal green), T4SS module (purple), ybt gene cluster (pink), integrase genes (green) and an invertible subset of genes (red arrow) are shown. Based on available genome sequence data, we suggest a development from single MGEs containing the clb determinant and the ybt gene cluster, respectively, towards the structural arrangement of both polyketide determinants, which is mainly found in enterobacterial populations. Black arrows (solid or dashed) indicate possible directions of development and DNA rearrangements. After the merge of the clb and ybt gene clusters into one MGE, represented by ICEKp10, there is evidence that three different structural variants have evolved from it: in Klebsiella species strains, the ICEKp10 has remained intact, whereas in C. koseri strains, a DNA rearrangement and re-localization of the ybt determinant to a different chromosomal position has taken place. In Escherichia coli, a gradual loss of the T4SS module and the inversion of a gene set between the two polyketide determinants led to immobilization or stabilization of the ICE, thus resulting the two PAIs known as pks island and HPI, respectively.