Legionella pneumophila strain associated with the first evidence of person-to-person transmission of Legionnaires' disease: a unique mosaic genetic backbone

Abstract

A first strong evidence of person-to-person transmission of Legionnaires' Disease (LD) was recently reported. Here, we characterize the genetic backbone of this case-related Legionella pneumophila strain ("PtVFX/2014"), which also caused a large outbreak of LD. PtVFX/2014 is phylogenetically divergent from the most worldwide studied outbreak-associated L. pneumophila subspecies pneumophila serogroup 1 strains. In fact, this strain is also from serogroup 1, but belongs to the L. pneumophila subspecies fraseri. Its genomic mosaic backbone reveals eight horizontally transferred regions encompassing genes, for instance, involved in lipopolysaccharide biosynthesis or encoding virulence-associated Dot/Icm type IVB secretion system (T4BSS) substrates. PtVFX/2014 also inherited a rare ~65 kb pathogenicity island carrying virulence factors and detoxifying enzymes believed to contribute to the emergence of best-fitted strains in water reservoirs and in human macrophages, as well as a inter-species transferred (from L. oakridgensis) ~37.5 kb genomic island (harboring a lvh/lvr T4ASS cluster) that had never been found intact within L. pneumophila species. PtVFX/2014 encodes another lvh/lvr cluster near to CRISPR-associated genes, which may boost L. pneumophila transition from an environmental bacterium to a human pathogen. Overall, this unique genomic make-up may impact PtVFX/2014 ability to adapt to diverse environments, and, ultimately, to be transmitted and cause human disease.

Figure 1. SBT-based representation of the worldwide genetic diversity among L. pneumophila isolates.

Relationships between allelic profiles of 10085 isolates (from the EWGLI SBT database) are shown in a goeBURST full minimum spanning tree (MST) constructed using PHYLOViZ software. The MST connects the ST profiles in a way that the summed distance of all links of the tree is the minimum. STs are indicated by numbers within circles, where connecting lines are labeled with the number of allelic differences between STs. The cluster including the PtVFX/2014 ST1905 strain (in red) is zoomed, where STs found in more than one country are labeled with an asterisk. The matrix describes the ST profiles differing by a single allele from the ST1905.

Figure 2. Phylogenomic analysis of PtVFX/2014 strain within the L. pneumophila species.

(A) A phylogenetic reconstruction enrolling genome sequences from 24 L. pneumophila strains and from 11 outbreak-associated PtVFX/2014 clones (ten clinical isolates plus one environmental isolate; in dark blue) is shown paired with the corresponding SNP density plot across the length of the genome. Serogroup (Sg) classification is also shown (see Supplementary Table S1 for details). The tree is drawn to scale, with branch lengths reflecting number of base substitutions per site. SNP density in the core-genome (~64% of the genome) is depicted proportionally to the color intensity and reflects SNPs against the genome of the L. pneumophila subsp. fraseri ATCC33216 strain (surrounded with a light blue box). Areas in grey refer to regions in the accessory genome. Genome orientation is according to that of the L. pneumophila Philadelphia-1 strain (GenBank accession number AE017354). (B) Detailed pairwise SNP density plot between PtVFX/2014 and the most close related L. pneumophila subsp. fraseri ATCC33216 strain showing putative HGT-inherited regions (labeled from A to H; see details in Table 1). Calculations of SNP density were performed across the genome alignment over a sliding window (window size = 25000 bp; step size = 25000 bp). Vertical dashed lines delimiting each region make the correspondence with the SNP density plot of panel (A).

Figure 3. Genetic basis underlying the serogroup classification of PtVFX/2014.

Schematic non-scaled condensed trees showing the genetic relatedness among 25 L. pneumophila strains (Supplementary Table S1) based on the core-genome alignment from Fig. 2A (panel A), as well as on the individual alignments of two genes that have been applied for genetic serogroup classification: lpg0773/wzt (panel B) and lpg0772/wzm (panel C). For simplification purposes, only one clone of the serogroup 1 PtVFX/2014 strain is presented in all trees. The core-genome tree topology matches the one of Fig. 2A, while the lpg0773/wzt and lpg0772/wzm trees topologies were inferred in MEGA5 using the Neighbor-Joining method, with Kimura 2-parameter model for estimating evolutionary distances (Bootstrapping = 1000 replicates). Branches are colored according to the serogroup classification.

Figure 4. Relevant putative pathogenicity islands inherited by PtVFX/2014.

(A) The graph shows the SNP density across a core-genome alignment (encompassing about 39–49% of each whole-genome sequence) between PtVFX/2014 and the L. oakridgensis ATCC 33761 (or DSM 21215) strain over a sliding window (window size = 38000 bp; step size = 38000 bp). Both gene content and organization of the ~37.5 kb genomic island is highlighted above the graph, where the lvh/lvr cluster is shown in black arrows, while the gene cluster homolog to a type I-like restriction-modification system is displayed in white arrows. (B) The graph shows the SNP density across a core-genome alignment (encompassing about 86–88% of each whole-genome sequence) between PtVFX/2014 and the L. pneumophila Philadelphia-1 strain over a slinding window (window size = 20000 bp; step size = 20000 bp). Gene organization of the ~65 kb genomic island is highlighted above the graph. For both graphs, core-genome alignments were extracted (using MAUVE software) by keeping and concatenating regions where genomes aligned over at least 500 bp, x-axis do not reflect any genome orientation. Genes delimiting each genomic island are labeled in each panel.

Figure 5. Repertoire of Dot/Icm T4BSS substrates in PtVFX/2014 strain.

(A) Venn diagram [constructed using jvenn] showing the distribution of 303 known Dot/Icm substrates among six L. pneumophila strains. Given that any counting of strain-specific substrates would be an underestimation (as deep functional studies have not been systematically performed for all strains), those substrates were labeled with a question mark, except for the Philadelphia-1 strain as it was used as a query in the pairwise comparisons. Corby strain is not represented in the Venn diagram as it possesses a Dot/Icm repertoire almost identical to that of Alcoy. The only exceptions occur in the genes lpg0080 and lpg1965, which are absent in Corby (lpg0080 is also absent in Lens, whereas lpg1965 is present in 130b and Philadelphia-1 besides 2300/99 Alcoy). Details regarding the Dot/Icm make-up of each strain are given in Supplementary Table S2. (B) goeBURST full minimum spanning tree showing that PtVFX/2014 (in red) carries a unique the Dot/Icm repertoire. Relationships based on the Dot/Icm make-up among strains were estimated using the goeBURST algorithm implemented in the PHYLOViZ software. Connecting lines are labeled with the number of Dot/Icm substrates differences between each strain. (C) Pairwise matrix displaying the number of Dot/Icm substrates that are shared (light red) or divergent (light blue) among strains. Diagonal values (in grey) represent the total number of Dot/Icm substrates inferred for each strain.