Comparison of whole genome sequencing typing tools for the typing of Belgian Legionella pneumophila outbreaks isolates

Abstract

Whole genome sequencing (WGS) marks a turning point for outbreak investigations for microorganisms related to public health matters, like Legionella pneumophila (Lp). Here, we evaluated the available Lp WGS typing tools for isolates of previously documented Belgian outbreaks, as well as small groups of related and non-related isolates. One reference strain and 77 clinical and environmental isolates were evaluated. Seven isolates belong to a Sequence Type (ST) 36 outbreak in 1999 and sixteen (ten clinical, two matching environmental and four non-related controls) belong to another ST1 outbreak in 1985-1987. The remaining isolates belong to small groups of related and non-related isolates of diverse ST's. WGS was performed and data were analysed using whole genome (wg) and core genome (cg) multilocus sequence typing (MLST) with "Ridom SeqSphere + " (cgMLST), "Applied Maths-Bionumerics" (wgMLST) and the 50 loci cgMLST (CDC/ESGLI_ESCMID). Results of the three tools were concordant with the traditional Sequence Based Typing (SBT). The known outbreaks and small clusters could be detected and clear discrimination of ST1 non-related isolates was obtained. In addition, the 50 loci cgMLST allowed to classify the isolates into subtypes because almost all the 50 genes could be called in all the analysed isolates, which was not achieved by the other tools. This is a big advantage in terms of standardisation and comparison between laboratories for future epidemiological investigations. WGS allowed to analyse a large volume of samples and generated more accurate conclusions for outbreak investigations compared to other typing methods due to its higher discriminatory power and throughput.

Keywords: Legionella pneumophila; Whole genome sequencing.

Fig. 1

Minimum spanning tree generated by cgMLST ridom seqsphere+ Legionella pneumophila scheme based on allelic profiles of 1412 genes present in all 78 analysed genomes. Circles correspond to the sequenced strains. Inside the circles, the strains are labelled by their ST. The number of allelic differences are indicated beside the branches. Strains with no allelic differences are classified in one circle. Strains classified in the same cluster are connected by a coloured background. For ST1 strains: Based on this MST, it could be shown that the ST1 environmental strain isolated from the hospital in 2011 was not the source of contamination of the patient hospitalised at that period (LEG759/LEG563)

Fig. 2

Minimum spanning tree generated by wgMLST Applied Maths, Bionumerics Legionella pneumophila scheme based on allelic profiles of all 78 analysed genomes. Circles correspond to the sequenced strains. Inside the circles, the strains are labelled by their strain ST. The number of allelic differences are indicated beside the branches. Strains with no allelic differences are classified in one circle. Strains classified in the same cluster are grouped in circles with coloured background

Fig. 3

MST of the 50 loci cgMLST for the 78 genomes. Minimum spanning tree generated by the 50 loci cgMLST Legionella pneumophila scheme based on allelic profiles of all 78 analysed genomes. Circles correspond to the sequenced strains. The scores beside the branches correspond to the normalized dissimilarity. Strains with no allelic differences are classified in one circle

Fig. 4

The 78 isolates belonging to 21 Sequence Types (ST) were classified into 34 subtypes as shown in this graphic (Genome count/subtype_ID) by using the 50cgMLST scheme. As follows, a list of the subtypes included in each ST: ST1: 5, 6, 7, 32, 33; ST6: 9, 10; ST22: 4; ST42: 25, 26; ST87: 16; ST110: 11, 12, 13, 14; ST188: 0, 1; ST196: 28; ST20: 24; ST23: 23; ST36: 27, 31; ST68:29; ST1333: 20; ST1362: 2, 3; ST1387: 17, 18; ST1443: 19; ST2458: ST 8; ST2461: 30; ST345: 15; ST146: 22; and one strain with unknown ST: 21