Utility of Whole-Genome Sequencing of Escherichia coli O157 for Outbreak Detection and Epidemiological Surveillance

Abstract

Detailed laboratory characterization of Escherichia coli O157 is essential to inform epidemiological investigations. This study assessed the utility of whole-genome sequencing (WGS) for outbreak detection and epidemiological surveillance of E. coli O157, and the data were used to identify discernible associations between genotypes and clinical outcomes. One hundred five E. coli O157 strains isolated over a 5-year period from human fecal samples in Lothian, Scotland, were sequenced with the Ion Torrent Personal Genome Machine. A total of 8,721 variable sites in the core genome were identified among the 105 isolates; 47% of the single nucleotide polymorphisms (SNPs) were attributable to six "atypical" E. coli O157 strains and included recombinant regions. Phylogenetic analyses showed that WGS correlated well with the epidemiological data. Epidemiological links existed between cases whose isolates differed by three or fewer SNPs. WGS also correlated well with multilocus variable-number tandem repeat analysis (MLVA) typing data, with only three discordant results observed, all among isolates from cases not known to be epidemiologically related. WGS produced a better-supported, higher-resolution phylogeny than MLVA, confirming that the method is more suitable for epidemiological surveillance of E. coli O157. A combination of in silico analyses (VirulenceFinder, ResFinder, and local BLAST searches) were used to determine stx subtypes, multilocus sequence types (15 loci), and the presence of virulence and acquired antimicrobial resistance genes. There was a high level of correlation between the WGS data and our routine typing methods, although some discordant results were observed, mostly related to the limitation of short sequence read assembly. The data were used to identify sublineages and clades of E. coli O157, and when they were correlated with the clinical outcome data, they showed that one clade, Ic3, was significantly associated with severe disease. Together, the results show that WGS data can provide higher resolution of the relationships between E. coli O157 isolates than that provided by MLVA. The method has the potential to streamline the laboratory workflow and provide detailed information for the clinical management of patients and public health interventions.

FIG 1

Maximum-likelihood core genome phylogeny for 103 E. coli O157 isolates from Lothian, Scotland. Recombinant sequences XH18570E and XH22083W were excluded. The tree was constructed with RAxML by using a general time-reversible model of nucleotide substitution and gamma distributed rate heterogeneity across sites. Branch lengths are in numbers of substitutions per site. Isolates known to be epidemiologically related (Epi) are in the same color. MLVA types shared by more than one isolate are indicated by filled boxes of the same color; SLV MLVA types are indicated by checkered boxes of the same color. PTs, multilocus STs, LSPA-6 types, stx subtypes, and the number of antibiotics (out of 15 tested) to which an isolate was resistant (AB len.) are shown. RDNC indicates that the PT reaction did not conform to a recognized pattern. Isolates associated with recent travel are in blue (Travel). All Lothian isolates were associated with diarrhea; isolates associated with bloody diarrhea (BD), hospitalization (H), and/or HUS are in red. Sublineages, defined as described in the text, are indicated by vertical black bars. One thousand bootstrap replicates were conducted, and bootstrap values of >98%, for nodes corresponding to the major lineages and sublineages described in the text, are indicated by asterisks.