Systematic longitudinal survey of invasive Escherichia coli in England demonstrates a stable population structure only transiently disturbed by the emergence of ST131

Abstract

Escherichia coli associated with urinary tract infections and bacteremia has been intensively investigated, including recent work focusing on the virulent, globally disseminated, multidrug-resistant lineage ST131. To contextualize ST131 within the broader E. coli population associated with disease, we used genomics to analyze a systematic 11-yr hospital-based survey of E. coli associated with bacteremia using isolates collected from across England by the British Society for Antimicrobial Chemotherapy and from the Cambridge University Hospitals NHS Foundation Trust. Population dynamics analysis of the most successful lineages identified the emergence of ST131 and ST69 and their establishment as two of the five most common lineages along with ST73, ST95, and ST12. The most frequently identified lineage was ST73. Compared to ST131, ST73 was susceptible to most antibiotics, indicating that multidrug resistance was not the dominant reason for prevalence of E. coli lineages in this population. Temporal phylogenetic analysis of the emergence of ST69 and ST131 identified differences in the dynamics of emergence and showed that expansion of ST131 in this population was not driven by sequential emergence of increasingly resistant subclades. We showed that over time, the E. coli population was only transiently disturbed by the introduction of new lineages before a new equilibrium was rapidly achieved. Together, these findings suggest that the frequency of E. coli lineages in invasive disease is driven by negative frequency-dependent selection occurring outside of the hospital, most probably in the commensal niche, and that drug resistance is not a primary determinant of success in this niche.

Figure 1.

Maximum-likelihood core genome phylogeny of E. coli associated with bacteremia in England. The columns on the right show, from left to right, phylogroup, STs containing more than 10 isolates, and hierBAPS clusters. Phylogroups are also presented by background shading and STs labeled on the right. Black represents ST designation not shown due to these having fewer than 10 isolates. The root has been placed according to previous understanding of Enterobacteriaceae phylogeny.

Figure 2.

Proportions of STs during the 11-yr sampling framework. The percentage of each ST has been plotted by year ordered by the frequency at the start of the study (most common at the bottom). The emergence of ST131 and ST69 can be observed in 2003 and 2002, respectively.

Figure 3.

ST131 maximum-likelihood phylogenetic tree based on SNPs called against the reference EC958. Columns to the right of the tree show the in silico predicted serotype (O16-H5 or O25-H4); phenotypic resistance to ciprofloxacin (CIP res.); SNP-based definition of fimH, gyrA, and parC genotypes; and the presence of bla_CTX-M and the type. NA:H4 in the serotype indicates that we were unable to assign a definite O type for the isolate. It has not been counted as a new serotype. Clades assigned based on the markers and clade-specific SNPs are shown on the right. The only isolate with missing data (black) is the reference strain EC958. The tree is mid-point rooted.

Figure 4.

Multidrug-resistance plasmids present in ST131. Phylogeny of the whole collection with columns to the right representing phylogroup, the five most frequent STs, phenotypic antibiotic-resistance data linked to the plasmid (ceftazidime, ciprofloxacin, cefotaxime, cefuroxime- and gentamicin), the presence of incFIA and incFIB, and antibiotic-resistance genes carried by the plasmid (aac(6′)-Ib, bla_CTX-M, bla_OXA, dfrA, mphA, sul1, and tetA). (Black) Missing; (color) present. The phylogenetic tree is the same as in Figure 1.

Figure 5.

ST73 maximum-likelihood phylogenetic tree based on SNPs called against the reference CFT073. The in silico predicted serotype is shown to the right of the tree. For serotype NA:H1 the O type could not be assigned. This is not counted as a new serotype. The clades are labeled on the right. The tree is mid-point rooted.

Figure 6.

Temporal analysis on ST131 using BEAST. Figure shows the serotype, resistance to ciprofloxacin (CIP res.), assignment to clade C2 (H30-Rx), and the gene alleles of fimH, gyrA, and parC. (*) MRCA; (**) emergence of clade C; (***) emergence of CIP-resistant clade C1.

Figure 7.

Comparison of antibiotic resistance of ST131, ST73, and the whole collection. Phenotypic antibiotic-resistance data are represented by the percentage of nonsusceptible (resistant + intermediate) isolates per year. Each subfigure represents one antibiotic class. Carbapenems (imipemen, meropenem, and ertapenem) and tigecycline are not shown due to the lack of resistance against these classes in this collection. (AMK) Amikacin; (GEN) gentamicin; (TOB) tobramycin; (TMP) trimethoprim; (CEF) cefalotin; (FOX) cefoxitin; (CXM) cefuroxime; (FEP) cefepime; (CTX) cefotaxime; (CAZ) ceftazidime; (AMX) amoxicillin; (AMP) ampicillin; (CIP) ciprofloxacin; (ATM) aztreonam; (AMC) amoxicillin-clavulanic acid; (TZP) piperacillin-tazobactam.