Large-scale genomic sequencing of extraintestinal pathogenic Escherichia coli strains

Abstract

Large-scale bacterial genome sequencing efforts to date have provided limited information on the most prevalent category of disease: sporadically acquired infections caused by common pathogenic bacteria. Here, we performed whole-genome sequencing and de novo assembly of 312 blood- or urine-derived isolates of extraintestinal pathogenic (ExPEC) Escherichia coli, a common agent of sepsis and community-acquired urinary tract infections, obtained during the course of routine clinical care at a single institution. We find that ExPEC E. coli are highly genomically heterogeneous, consistent with pan-genome analyses encompassing the larger species. Investigation of differential virulence factor content and antibiotic resistance phenotypes reveals markedly different profiles among lineages and among strains infecting different body sites. We use high-resolution molecular epidemiology to explore the dynamics of infections at the level of individual patients, including identification of possible person-to-person transmission. Notably, a limited number of discrete lineages caused the majority of bloodstream infections, including one subclone (ST131-H30) responsible for 28% of bacteremic E. coli infections over a 3-yr period. We additionally use a microbial genome-wide-association study (GWAS) approach to identify individual genes responsible for antibiotic resistance, successfully recovering known genes but notably not identifying any novel factors. We anticipate that in the near future, whole-genome sequencing of microorganisms associated with clinical disease will become routine. Our study reveals what kind of information can be obtained from sequencing clinical isolates on a large scale, even well-characterized organisms such as E. coli, and provides insight into how this information might be utilized in a healthcare setting.

Figure 1.

Whole-genome phylogenetic tree of ExPEC E. coli isolates. (A) Approximate maximum likelihood phylogeny showing the population structure of ExPEC E. coli. Isolates cultured from blood are represented as red terminal nodes and those cultured from urine are shown in blue. Colored ring denotes annotation of major E. coli phylogroups. Seven isolates assigned to phylogroups that are inconsistent with their phylogenomic placement are indicated with colored bars internal to this ring. The outermost ring (black) indicates groups of MLST sequence types. Sequence types with at least two representatives are numbered. The group corresponding to subclone S131-H30 is indicated. (B) Approximate maximum likelihood phylogeny of blood isolates only. Isolates are labeled according to the patient of origin and the relative day of collection (in red, ranging from day 0 for patient 43 to day 1184 for patient isolate 3_2). In instances where multiple isolates were obtained from the same patient, the order in which specimens were recovered is indicated by an underscore and a number. Patients for which multiple, genomically distinct strains were identified are highlighted. Isolates from patients 1 and 29 are indicated by blue text. The group corresponding to subclone ST131-H30 is indicated. Colored ring as in A. Scale bars are expressed in changes per site for both panels.

Figure 2.

Whole-genome phylogenetic tree of ST131 isolates. The ST131-H30 subgroup is indicated and is marked by a high prevalence of fluoroquinolone resistance and extended-spectrum β-lactamase (ESBL) activity. Node color indicates the relevant drug resistance phenotype (white circle indicates missing data). Nodes supported by log likelihood values below 0.8 are marked with a black circle. Scale bar is expressed in changes per site.

Figure 3.

Proportion and relative enrichment of virulence factors and antibiotic resistance phenotypes carried by isolates in distinct groups. Rows correspond to individual VFs (top) or antibiotic resistance phenotypes (Abx, bottom). VFs are grouped by class. (Tox) toxin, (Prot) protectin, (Iron) iron metabolism, (Tr) transporter, (Cap) capsule, (Inv) invasion, (Adh) adhesion, (Mi) miscellaneous. Columns correspond to categories of isolates grouped according to different classification schemes. Prevalence of factors within each category is shown at left for each panel (blue heatmap). Raw P-values from all possible pairwise comparisons of factor prevalence between them is shown at right for each panel (green heatmap), with the specific pairwise comparison indicated above each column. P-values were obtained after correcting for inferred population structure (left, labeled in red with “Corrected” or “C”) or without such correction (right, labeled in red “Uncorrected” or “U”). (A) Comparison of E. coli phylogroups. (B) Comparison of isolates obtained from blood (B) and urine (U). (C) Comparison of the six most prevalent MLST groups (sequence type numbers are indicated) and a seventh category encompassing all other MLST groups (“O”).