Evolutionary dynamics of Enterococcus faecium reveals complex genomic relationships between isolates with independent emergence of vancomycin resistance

Abstract

Enterococcus faecium, a major cause of hospital-acquired infections, remains problematic because of its propensity to acquire resistance to vancomycin, which currently is considered first-line therapy. Here, we assess the evolution and resistance acquisition dynamics of E. faecium in a clinical context using a series of 132 bloodstream infection isolates from a single hospital. All isolates, of which 49 (37 %) were vancomycin-resistant, underwent whole-genome sequencing. E. faecium was found to be subject to high rates of recombination with little evidence of sequence importation from outside the local E. faecium population. Apart from disrupting phylogenetic reconstruction, recombination was frequent enough to invalidate MLST typing in the identification of clonal expansion and transmission events, suggesting that, where available, whole-genome sequencing should be used in tracing the epidemiology of E. faecium nosocomial infections and establishing routes of transmission. Several forms of the Tn1549-like element-vanB gene cluster, which was exclusively responsible for vancomycin resistance, appeared and spread within the hospital during the study period. Several transposon gains and losses and instances of in situ evolution were inferred and, although usually chromosomal, the resistance element was also observed on a plasmid background. There was qualitative evidence for clonal expansions of both vancomycin-resistant and vancomycin-susceptible E. faecium with evidence of hospital-specific subclonal expansion. Our data are consistent with continuing evolution of this established hospital pathogen and confirm hospital vancomycin-susceptible and vancomycin-resistant E. faecium patient transmission events, underlining the need for careful consideration before modifying current E. faecium infection control strategies.

Keywords: Enterococcus faecium; infection control; multi-locus sequence typing; recombination; transposon; vancomycin resistance.

Fig. 1.

Clonal phylogeny and inferred recombination events. Midpoint-rooted clonal phylogeny of 132 E. faecium bloodstream isolates; metadata (centre) and recombination events (right), estimated jointly as described in Methods. The sizes and genomic locations of recombination fragments (dark blue line segments) and the positions of SNVs (white ticks) occurring along branches in the phylogeny are aligned with branches in the phylogeny. Vertical grey lines mask genomic fragments with incomplete data and vertical yellow lines show the positions of seven MLST loci. Despite the use of a plausible statistical model for recombination, marked clusters of substitutions remain in the estimated clonal phylogeny, suggesting additional undetected imports and emphasizing the analytical difficulty in distinguishing real divergence (long branches) from imports. Branch with asterisk indicates three divergent, non-CC17 isolates. Turquoise and red branches carry recombination events directly affecting ST, i.e. MLST loci leading to false negative and false positive MLST clustering, respectively. Isolate meta-data are depicted from left to right: sequence type, vanB vancomycin resistance status, year of isolation and epidemiological assignment to a putative outbreak (Out) by MLST. ‘Other’ STs use alternating green shades for convenience for successive distinct STs. Novel ST types ST990 (Number 1) and ST991 (Number 2) are indicated. Isolates assigned to putative MLST defined outbreaks are depicted by the grey bars with evidence of two definitive transmission events (red bars) when using WGS (see text and Table S3 for further details).

Fig. 2.

Tn1549-like van-carrying transposons. Gene arrangements of representative Tn1549-like transposons and their included vanB operons (inset). Labels show the numbers of isolates sharing insertion positions [with respect to the Aus0004 genome for comparison with Howden et al. (2013) – sharing is consistent with a single ancestral insertion at each position] and the insertion locus [gene, hypothetical ORF, or intergenic region (IGR)]. Transposon structure 2 was found in two different sites within a 32 bp region of an IGR. Where the transposon inserted into a mobile element, the name of the plasmid or IS is given; the nucleotide position of the insertion is given with respect to the reference plasmid sequence or the transposase gene of the IS. Coloured arrows indicate: red, mobilization genes; orange, transposon-related genes; green, vancomycin resistance genes (see inset); white, hypothetical ORFs; yellow, annotated genes not in one of these categories. IS elements (white pentagons) are shown with their insertion positions.

Fig. 3.

vanB Tn1549-acquisition/loss and transposon evolution events. Annotated circular phylogeny of 132 E. faecium isolates. Branch colours represent a parsimonious reconstruction of vanB transposon acquisition and loss events based on transposon structure and insertion site (green, gain; red, loss; blue, evolution of transposon in situ; from previous acquisition). Inner circle colours reflect Tn1549 structure (see Fig. 2 for more details). The number depicts the starting point of the tree relative to Fig. 1. Asterisks indicate plasmid-borne transposon. The triangle indicates two isolates that on the core genome level are identical and carry the same Tn1549 consistent cross transmission.

Fig. 4.

Australian phylogeny. (a) Circular clonal phylogeny of 177 E. faecium isolates comprising 132 bloodstream isolates from a single institution in Sydney (current study) and 45 isolates (42 from Melbourne, coloured red, and three VRE isolates from Perth, coloured yellow) from Howden et al. (2013). The same E. faecium reference strain DO (a VSE ST18 from the USA, 1998) was included. Note that the branch for strain DO has been shortened more than fourfold, as indicated by //. (b) Each isolate [coloured by city as in (a)] is plotted as the mutational distance in SNV to the most similar isolate from the same city (y-axis) or from a different city (x-axis). Points near the x-axis (below the diagonal) represent the majority of isolates, whose nearest neighbour is in the same population; points above the diagonal represent isolates with a nearest neighbour in another population; and points near to the diagonal (x = y) may be interpreted as isolates in lineages (at different scales of divergence) that are shared between cities. Points plotted as filled circles correspond to the same isolates marked with asterisks as in (a), and provide some evidence for short-term inter-city spread. Seven isolates that were only distantly related to other isolates are not plotted. The vertical dotted line represents approximately 1–2 years between isolates.