Insertion sequence-driven diversification creates a globally dispersed emerging multiresistant subspecies of E. faecium

Abstract

Enterococcus faecium, an ubiquous colonizer of humans and animals, has evolved in the last 15 years from an avirulent commensal to the third most frequently isolated nosocomial pathogen among intensive care unit patients in the United States. E. faecium combines multidrug resistance with the potential of horizontal resistance gene transfer to even more pathogenic bacteria. Little is known about the evolution and virulence of E. faecium, and genomic studies are hampered by the absence of a completely annotated genome sequence. To further unravel its evolution, we used a mixed whole-genome microarray and hybridized 97 E. faecium isolates from different backgrounds (hospital outbreaks (n = 18), documented infections (n = 34) and asymptomatic carriage of hospitalized patients (n = 15), and healthy persons (n = 15) and animals (n = 21)). Supported by Bayesian posterior probabilities (PP = 1.0), a specific clade containing all outbreak-associated strains and 63% of clinical isolates was identified. Sequencing of 146 of 437 clade-specific inserts revealed mobile elements (n = 74), including insertion sequence (IS) elements (n = 42), phage genes (n = 6) and plasmid sequences (n = 26), hypothetical (n = 58) and membrane proteins (n = 10), and antibiotic resistance (n = 9) and regulatory genes (n = 11), mainly located on two contigs of the unfinished E. faecium DO genome. Split decomposition analysis, varying guanine cytosine content, and aberrant codon adaptation indices all supported acquisition of these genes through horizontal gene transfer with IS16 as the predicted most prominent insert (98% sensitive, 100% specific). These findings suggest that acquisition of IS elements has facilitated niche adaptation of a distinct E. faecium subpopulation by increasing its genome plasticity. Increased genome plasticity was supported by higher diversity indices (ratio of average genetic similarities of pulsed-field gel electrophoresis and multi locus sequence typing) for clade-specific isolates. Interestingly, the previously described multi locus sequence typing-based clonal complex 17 largely overlapped with this clade. The present data imply that the global emergence of E. faecium, as observed since 1990, represents the evolution of a subspecies with a presumably better adaptation than other E. faecium isolates to the constraints of a hospital environment.

Figure 1. A Bayesian Phylogenomic Relationship of Strains Associated with Different Ecological Niches

The Bayesian posterior probability (PP) supports internal branch robustness. PP = 1.0 represents 100% of all phylogenies showing a given topology. Strains are designated at the end of branches and are colored according to the ecological niche from which the E. faecium strain was isolated. Strains in bold indicated with an asterisk are part of CC17. AUS, Australia; AUT, Austria; BEL, Belgium; BRA, Brazil; CHE, Switzerland; DEU, Germany; DNK, Denmark; ESP, Spain; FRA, France; GBR, Great Britain; GRC, Greece; IRL, Ireland; ISR, Israel; ITA, Italy; NLD, Netherlands; NOR, Norway; POL, Poland; PRT, Portugal; TZA, Tanzania; USA, United States of America; ZAF, Republic of South Africa.

Figure 2. Complete Linkage Hierarchical Clustering of E. faecium Inserts with Pearson Correlation and Euclidian Distance, Zoom on Accessory Insert

The hierarchical clustering of inserts visualizes composition of the microbial pangenome: the core genome and accessory genes. The lower panel of the figure represents a zoomed-in accessory genome, as indicated by the vertical and diagonal arrows in the middle of the figure. Presence of an insert is indicated in red and absence of an insert in green. Inserts are clustered on the x-axis, and strains are presented on the y-axis (indicated by a vertical arrow left of the lower panel). Horizontal straight lines separate the hospital clade strains (H) from the non-hospital clade strains. The core genome and accessory genes are indicated by a blue and purple bar, respectively. Hospital clade–associated clustered inserts are indicated with pink bars.

Figure 3. SDA of Hospital Clade–Specific Genes on Contigs 656 and 658 among E. faecium Strains

SDA of hospital clade–specific genes on contig 656 of 13 hospital clade strains (A) and hospital clade–specific genes on contig 658 of ten hospital clade strains (B). The nodes represent strains (E numbers correspond to Table 1) and are depicted as blue circles. The scale bar represents Hamming distance. Numbers at the edges represent the percent bootstrap support of the split obtained after 1,000 replicates. Paralellograms indicate recombinatory events. Fit in both graphs is 100%.

Figure 4. Presence and Absence of Hospital Clade–Specific Genes on Contigs 656 and 658 among E. faecium Strains

Presence and absence of hospital clade–specific genes on contig 656 (A) and 658 (B) among E. faecium strains. Strains are given on the x-axis (E numbers correspond to Table 1). Hospital clade strains are shown on the left panel and are separated from non–hospital clade strains by a vertical line. Hospital clade–specific genes with similarity to E. faecium DO genes, gene numbers, and GC content are given on the y-axis on the right side, in presumed order. Spot number is given on the y-axis on the left side; more than one gene can be located in one insert. A highly certain present gene (plus 0.5) is indicated with a red square, an absent or highly diverging gene (minus 0.5) with absence of a square. Slightly deviating genes (equal to or larger than 0, and smaller than 0.5) are indicated with a pink square, deviating genes (larger than minus 0.5, and smaller than 0) with a green square.

Figure 5. Distribution of IS16 among E. faecium Strains

Maximum likelihood–based gene analysis for determining the distribution of individual IS16 throughout the phylogenetic tree is shown. Strains in which IS16 is present are colored red; strains in which IS16 is absent are colored blue. Strains in the hospital clade all contain IS16 with one exception. IS16 is absent in all strains in the non–hospital clade.