Delineating community outbreaks of Salmonella enterica serovar Typhimurium by use of whole-genome sequencing: insights into genomic variability within an outbreak

Abstract

Whole-genome next-generation sequencing (NGS) was used to retrospectively examine 57 isolates from five epidemiologically confirmed community outbreaks (numbered 1 to 5) caused by Salmonella enterica serovar Typhimurium phage type DT170. Most of the human and environmental isolates confirmed epidemiologically to be involved in the outbreaks were either genomically identical or differed by one or two single nucleotide polymorphisms (SNPs), with the exception of those in outbreak 1. The isolates from outbreak 1 differed by up to 12 SNPs, which suggests that the food source of the outbreak was contaminated with more than one strain while each of the other four outbreaks was caused by a single strain. In addition, NGS analysis ruled in isolates that were initially not considered to be linked with the outbreak, which increased the total outbreak size by 107%. The mutation process was modeled by using known mutation rates to derive a cutoff value for the number of SNP difference to determine whether or not a case was part of an outbreak. For an outbreak with less than 1 month of ex vivo/in vivo evolution time, the maximum number of SNP differences between isolates is two or four using the lowest or highest mutation rate, respectively. NGS of S. Typhimurium significantly increases the resolution of investigations of community outbreaks. It can also inform a more targeted public health response by providing important supplementary evidence that cases of disease are or are not associated with food-borne outbreaks of S. Typhimurium.

FIG 1

MSTs of S. Typhimurium outbreak isolates based on the number of SNPs among isolates within each outbreak. The numbers in the circles are the isolate identities. The number on each branch is the number of SNP differences.

FIG 2

Maximum-parsimony tree of S. Typhimurium genomes based on SNPs identified by mapping to the reference chromosome of S. Typhimurium LT2. Only SNPs in the “core” genome were included (27). The number on each branch is the number of SNP differences. Isolates representing each outbreak are demarcated with curly brackets followed by the outbreak numbers. The isolate source, either human (orange, epidemiologically confirmed; green, unknown epidemiological link; yellow, no epidemiological link) or environmental (blue), is noted next to the isolate number. In parentheses are the GenBank accession numbers of the publicly available genomes. The unit of the scale bar is the number of SNPs.

FIG 3

The number of SNP differences (99th percentile) between isolates under a Poisson process of mutation. The expected number of SNP differences observable between two isolates from the same outbreak equals the mutation rate times twice the total time the pathogen spent in the food and in the host when isolates were isolated (ex vivo/in vivo evolution time). We used three mutation rates and up to 120 days of ex vivo/in vivo evolution time to model the expected number of SNPs. The ex vivo/in vivo evolution time is defined as the period of time from when the strain was introduced into (contaminated) the food to the time when an isolate was obtained from either the human or food source during the outbreak. The three mutation rates used were 1.9 × 10⁻⁷ (28), 3.4 × 10⁻⁷ (29), and 12 ×10⁻⁷ (11) substitutions per site per year, which generated the low, intermediate, and high upper limits of the number of SNP differences between a pair of isolates of the same lineage; these are shown as solid, dashed, and dotted lines, respectively. Note that the lines are ladder-like because of sampling from a discrete distribution.

FIG 4

MLE of SNP variation of S. Typhimurium outbreaks. (A) Likelihood of the population mutation parameter θ. The MLE is 1.37 per generation. (B) Range of expected numbers of SNPs under the coalescent model assuming a sample size of eight. The central 95% are orange. The expected umber of SNPs was 3.56. (C) Mean and central 95% of the expected number of SNPs for sample sizes ranging from 2 to 20. The 0.025 and 0.975 quantiles are shown as dotted and dashed lines, respectively. The mean is shown as a blue line.