Lineage specific recombination rates and microevolution in Listeria monocytogenes

Abstract

Background: The bacterium Listeria monocytogenes is a saprotroph as well as an opportunistic human foodborne pathogen, which has previously been shown to consist of at least two widespread lineages (termed lineages I and II) and an uncommon lineage (lineage III). While some L. monocytogenes strains show evidence for considerable diversification by homologous recombination, our understanding of the contribution of recombination to L. monocytogenes evolution is still limited. We therefore used STRUCTURE and ClonalFrame, two programs that model the effect of recombination, to make inferences about the population structure and different aspects of the recombination process in L. monocytogenes. Analyses were performed using sequences for seven loci (including the house-keeping genes gap, prs, purM and ribC, the stress response gene sigB, and the virulence genes actA and inlA) for 195 L. monocytogenes isolates.

Results: Sequence analyses with ClonalFrame and the Sawyer's test showed that recombination is more prevalent in lineage II than lineage I and is most frequent in two house-keeping genes (ribC and purM) and the two virulence genes (actA and inlA). The relative occurrence of recombination versus point mutation is about six times higher in lineage II than in lineage I, which causes a higher genetic variability in lineage II. Unlike lineage I, lineage II represents a genetically heterogeneous population with a relatively high proportion (30% average) of genetic material imported from external sources. Phylograms, constructed with correcting for recombination, as well as Tajima's D data suggest that both lineages I and II have suffered a population bottleneck.

Conclusion: Our study shows that evolutionary lineages within a single bacterial species can differ considerably in the relative contributions of recombination to genetic diversification. Accounting for recombination in phylogenetic studies is critical, and new evolutionary models that account for the possibility of changes in the rate of recombination would be required. While previous studies suggested that only L. monocytogenes lineage I has experienced a recent bottleneck, our analyses clearly show that lineage II experienced a bottleneck at about the same time, which was subsequently obscured by abundant homologous recombination after the lineage II bottleneck. While lineage I and lineage II should be considered separate species from an evolutionary viewpoint, maintaining single species name may be warranted since both lineages cause the same type of human disease.

Figure 1

Unrooted consensus network based on 10,000 phylograms obtained from ClonalFrame analyses of the individual sequence types. Genealogical inference was performed for all 92 unique STs using ClonalFrame as described in the Methods. A consensus network was built using Splitstree [23]. Reticulate relationships were found in at least 20% of the trees and indicate phylogenetic uncertainty. Branches supported by a posterior probability of more than 95% are colored in red. Leaves are labeled with ST designations.

Figure 2

Mixture of ancestry of the different STs as inferred by STRUCTURE. Proportions of ancestry from ancestral lineage I (purple), ancestral lineage II (green), ancestral lineage IIIA (red) and ancestral IIIB subpopulations (yellow) as inferred by STRUCTURE assuming K = 4 ancestral subpopulations. The asterisk marks a lineage IIIC isolate. Each vertical line represents an individual sequence type and is colored according to the inferred proportion of single nucleotide alleles that were derived from one of the ancestral subpopulations. This bar plot was created with the DISTRUCT software [52].

Figure 3

Distribution of the Interior/exterior branch length ratio of trees resulting from ClonalFrame analysis of lineage I (A) and II (B) as compared to trees simulated under the coalescent model. Both lineages show a higher internal/external branch length ratio (lineage I 1.50, P = 0.005; lineage II 1.64, P = 0.00002) than expected under the coalescent model, which is indicative of a bottleneck event during the population history.

Figure 4

Phylogenies inferred by ClonalFrame without (A) and with (B) correction for recombination. The phylogram (A) shows a 50% majority-rule consensus tree based on ClonalFrame output (see Methods section) for all 92 unique STs and ignoring the role of recombination. The phylogram (B) is the same, but recombination was taken into account in the model of genetic diversification. The rulers indicate the time in coalescent units. Dashed grey lines show the estimated time to the most recent common ancestors of lineage I and II.