Population genomics of hypervirulent Klebsiella pneumoniae clonal-group 23 reveals early emergence and rapid global dissemination

Abstract

Severe liver abscess infections caused by hypervirulent clonal-group CG23 Klebsiella pneumoniae have been increasingly reported since the mid-1980s. Strains typically possess several virulence factors including an integrative, conjugative element ICEKp encoding the siderophore yersiniabactin and genotoxin colibactin. Here we investigate CG23's evolutionary history, showing several deep-branching sublineages associated with distinct ICEKp acquisitions. Over 80% of liver abscess isolates belong to sublineage CG23-I, which emerged in ~1928 following acquisition of ICEKp10 (encoding yersiniabactin and colibactin), and then disseminated globally within the human population. CG23-I's distinguishing feature is the colibactin synthesis locus, which reportedly promotes gut colonisation and metastatic infection in murine models. These data show circulation of CG23 K. pneumoniae decades before the liver abscess epidemic was first recognised, and provide a framework for future epidemiological and experimental studies of hypervirulent K. pneumoniae. To support such studies we present an open access, completely sequenced CG23-I human liver abscess isolate, SGH10.

Fig. 1

Phylogenetic relationships within CG23. Time-calibrated Bayesian phylogeny (left) showing the relationships between 97 CG23 Kp, their sequences types (STs), the presence of chromosomal virulence loci (ybt, yersiniabactin; clb, colibactin; iro, salmochelin; rmpA, regulator of mucoid phenotype; mcc, microcin E492), plasmid virulence loci (iro salmochelin, iuc aerobactin, rmpA/rmpA2 regulator of mucoid phenotype), and the virulence plasmid backbone (pK2044-like, see Supplementary Fig. 6). Grey shading indicate partial mcc deletions (see Supplementary Fig. 7). Tips are coloured by region of isolate origin as indicated. In addition to SGH10, complete genomes for which chromosomes are shown in Fig. 2 are marked A-F as per inset legend. * indicates three human-associated multidrug resistant isolates. Subclades containing ICEKp are indicated by shading, labelled with the corresponding ICEKp and ybt lineages; the pink shaded clade carrying ICEKp10 (ybt 1) is the dominant lineage, CG23-I. Putative ICEKp acquisition events by a lineage or single strain are indicated by triangles at the corresponding node or next to a strain, and are shaded by chromosomal tRNA-Asn insertion site. Bayesian molecular dating estimates are shown for the most recent common ancestor of all isolates (i.e. the tree root), the CG23-I lineage, and the horse lineage (nodes as marked with white circles). The number of SNPs unique to CG23-I and the horse lineage are indicated on the relevant branches

Fig. 2

Chromosomal synteny and content comparisons between CG23 completed assemblies. a Homologous regions common to the chromosomes for six completed CG23 chromosomes are represented as blocks. Chromosomal inversions relative to the oldest genome NCTC9494 are indicated by blocks below the line, while blocks above the line indicate the same orientation as the reference. Unique acquisitions that were not common to all genomes are annotated. b Presence (+) or absence (−) of key chromosomal- and plasmid-encoded virulence loci in completed CG23 genomes

Fig. 3

Virulence and genomic properties of proposed CG23-I strain SGH10. a Bacterial burden (CFU/g) in organs and stool of n = 8 mice 48 h following oral infection, bars indicate mean values. b SGH10 chromosome and c plasmid. Tracks shown are (from inner to outer): GC skew (G − C/G + C), G+C content, key capsule and virulence features as labelled, coding sequences on reverse strand and coding sequences on forward strand