Towards Real-Time and Affordable Strain-Level Metagenomics-Based Foodborne Outbreak Investigations Using Oxford Nanopore Sequencing Technologies

Abstract

The current routine laboratory practices to investigate food samples in case of foodborne outbreaks still rely on attempts to isolate the pathogen in order to characterize it. We present in this study a proof of concept using Shiga toxin-producing Escherichia coli spiked food samples for a strain-level metagenomics foodborne outbreak investigation method using the MinION and Flongle flow cells from Oxford Nanopore Technologies, and we compared this to Illumina short-read-based metagenomics. After 12 h of MinION sequencing, strain-level characterization could be achieved, linking the food containing a pathogen to the related human isolate of the affected patient, by means of a single-nucleotide polymorphism (SNP)-based phylogeny. The inferred strain harbored the same virulence genes as the spiked isolate and could be serotyped. This was achieved by applying a bioinformatics method on the long reads using reference-based classification. The same result could be obtained after 24-h sequencing on the more recent lower output Flongle flow cell, on an extract treated with eukaryotic host DNA removal. Moreover, an alternative approach based on in silico DNA walking allowed to obtain rapid confirmation of the presence of a putative pathogen in the food sample. The DNA fragment harboring characteristic virulence genes could be matched to the E. coli genus after sequencing only 1 h with the MinION, 1 h with the Flongle if using a host DNA removal extraction, or 5 h with the Flongle with a classical DNA extraction. This paves the way towards the use of metagenomics as a rapid, simple, one-step method for foodborne pathogen detection and for fast outbreak investigation that can be implemented in routine laboratories on samples prepared with the current standard practices.

Keywords: Flongle; SNP analysis; STEC; food surveillance; metagenomics; nanopore; outbreak; strain-level.

FIGURE 1

Percentages of reads classified to the genus level using Kraken2 (taxonomic classification tool) from blank and spiked beef samples extracted with two DNA extraction kits (one involving host removal, HZ) and sequenced on Illumina (MiSeq), MinION, or Flongle, with in-house databases of mammals, archaea, bacteria, fungi, human, protozoa, and viruses. The data for Illumina sequencing (*) was published in Buytaers et al. (2020). Light blue represents the proportion of “Bos” corresponding to beef reads. Yellow indicates the presence of “Escherichia” in the sample. The reads that could not be classified to the genus level for mammals, archaea, bacteria, fungi, human, protozoa, or viruses are represented in gray.

FIGURE 2

In silico DNA walking results, presenting the genera in the inner circle (following the color scheme specified in the legend) and the genes detected for each taxon in the outer circle for MinION sequencing of the Shiga toxin-producing E. coli (STEC)-spiked beef sample after 1 h of sequencing.

FIGURE 3

Single-nucleotide polymorphism (SNP)-based phylogenetic tree of STEC strains inferred from metagenomics sequencing (beef) and of sequenced isolates with percentage of the reference genome covered (i.e., percentage of reference genome that is useful for SNP analysis, see section “Materials and Methods”) and gene detection (O-type and H-type and genes eae, stx1, and stx2; green shaded blocks representing the query coverage) in each strain represented on the side of the branch. Isolates TIAC 1151, 1152, 1153, and 1638 are from food origin. Isolates TIAC 1165 and 1169 are from human origin. Reference: E. coli O157:H7 str. Sakai (BA000007.3). Green: closely related strains from the outbreak cluster. Black: sporadic cases outside the outbreak cluster. The scale bar represents nucleotide substitution per 100 nucleotide sites. Node values represent bootstrap support values.

FIGURE 4

In silico DNA walking results, presenting the genera in the inner circle (following the color scheme specified in the legend) and the genes detected for each taxon in the outer circle of the STEC-spiked beef sample after DNA extraction with or without host removal. (A) Flongle sequencing after 1 h of sequencing, DNA extract with host removal. (B) Flongle sequencing after 5 h of sequencing, DNA extract without host removal.

FIGURE 5

Integrated metagenomics-based strategy for microbiological foodborne outbreak investigation. As first optional steps, food samples can be screened for the presence of bacterial pathogens using metagenomics Flongle sequencing, taxonomic classification, and in silico DNA walking (based on BLAST to the nucleotide database and virulence genes database) in parallel with the ongoing attempt to isolate the pathogen, followed by WGS of the obtained isolates. A strain-level characterization can be attempted from the Flongle sequencing or conducted after using an Illumina strategy (more cost-effective for multiple samples) or a MinION strategy (fast response for one sample) in food samples for which the presence of a pathogen is confirmed. The strain-level data analysis for Illumina sequencing was previously presented Buytaers et al. (2020). The strain-level data analysis workflow for MinION sequencing is based on classification using Metamaps, a gene detection with BLAST, and phylogenetics with a SNP-calling pipeline. The asterisk (*) indicates that WGS isolate data is interesting to feed to reference genome databases for the classification of the metagenomics reads for future analyses.