Development and implementation of a core genome multilocus sequence typing scheme for Yersinia enterocolitica: a tool for surveillance and outbreak detection

Abstract

Yersinia enterocolitica (Y. enterocolitica) is the most frequent etiological agent of yersiniosis and has been responsible for several national outbreaks in Norway and elsewhere. A standardized high-resolution method, such as core genome Multilocus Sequence Typing (cgMLST), is needed for pathogen traceability at the national and international levels. In this study, we developed and implemented a cgMLST scheme for Y. enterocolitica. We designed a cgMLST scheme in SeqSphere + using high-quality genomes from different Y. enterocolitica biotype sublineages. The scheme was validated if more than 95% of targets were found across all tested Y. enterocolitica: 563 Norwegian genomes collected between 2012 and 2022 and 327 genomes from public data sets. We applied the scheme to known outbreaks to establish a threshold for identifying major complex types (CTs) based on the number of allelic differences. The final cgMLST scheme included 2,582 genes with a median of 97.9% (interquartile range 97.6%-98.8%) targets found across all tested genomes. Analysis of outbreaks identified all outbreak strains using single linkage clustering at four allelic differences. This threshold identified 311 unique CTs in Norway, of which CT18, CT12, and CT5 were identified as the most frequently associated with outbreaks. The cgMLST scheme showed a very good performance in typing Y. enterocolitica using diverse data sources and was able to identify outbreak clusters. We recommend the implementation of this scheme nationally and internationally to facilitate Y. enterocolitica surveillance and improve outbreak response in national and cross-border outbreaks.

Keywords: Yersinia enterocolitica; cgMLST; molecular typing; outbreak; yersiniosis.

Fig 1

Flowchart of the process of development of a stable cgMLST scheme for Y. enterocolitica.

Fig 2

Performance comparison of percentage good targets identified across the three developed cgMLST schemes (cgMLST_V1, cgMLST_QC, and cgMLST_Biotypes). Three outliers fall below the 85% mark in all schemes. The cgMLST_V1 scheme has 2,582 targets and a median of 97.9% (IQR 97.6%–98.8%) good targets; the cgMLST_Biotypes scheme has 2,277 targets and a median 96.9% (IQR 96.7%–97.7%) good targets; the cgMLST_QC scheme included 2,234 gene targets and a median 96.8% (IQR 96.6%–97.6%) good targets.

Fig 3

Performance comparison of percentage good targets identified across the three cgMLST schemes (cgMLST_V1, cgMLST_QC, and cgMLST_Biotypes) against the 13 biotype sublineages of Y. enterocolitica (1Aa, 1Ab, 1B, 2/3–5a, 2/3–5b, 2/3–9a, 2/3–9b, 3–3a, 3–3b, 3–3c, 3–3d, 4, and 5).

Fig 4

(A). Within outbreak pairwise allelic differences across nine identified outbreaks in Norway between 2018 and 2023. (B). Number of investigated outbreaks with only one outbreak cluster using different SLC threshold methods. Each integer after the SLC indicates the maximum allelic differences to cluster isolates.

Fig 5

Minimum-spanning tree of all Y. enterocolitica outbreak isolates from Norway between 2018 and 2023, using SLC. Numbers inside nodes represent the SLC cluster at a threshold of four allelic differences. Numbers on lines represent the number of allelic differences between isolates.

Fig 6

Minimum-spanning Tree of all Y. enterocolitica isolates sequenced at NRL in 2022. Nodes in red represent outbreaks. Numbers inside nodes represent the SLC cluster at four allelic differences. Numbers on lines represent the number of allelic differences.

Fig 7

Comparison with SLC cluster assignments between Enterobase SLC at thresholds varying between 0 and 13 allelic differences. At each Enterobase SLC definition, we map to the cgMLST_V1 SLC obtaining the maximum adjusted Rand index for that comparison. Values for the adjusted Rand index are indicated above the SLC comparison. The maximum value of the adjusted Rand index across all comparisons is marked in orange.