Comparative genomic analyses of 17 clinical isolates of Gardnerella vaginalis provide evidence of multiple genetically isolated clades consistent with subspeciation into genovars

Abstract

Gardnerella vaginalis is associated with a spectrum of clinical conditions, suggesting high degrees of genetic heterogeneity among stains. Seventeen G. vaginalis isolates were subjected to a battery of comparative genomic analyses to determine their level of relatedness. For each measure, the degree of difference among the G. vaginalis strains was the highest observed among 23 pathogenic bacterial species for which at least eight genomes are available. Genome sizes ranged from 1.491 to 1.716 Mb; GC contents ranged from 41.18% to 43.40%; and the core genome, consisting of only 746 genes, makes up only 51.6% of each strain's genome on average and accounts for only 27% of the species supragenome. Neighbor-grouping analyses, using both distributed gene possession data and core gene allelic data, each identified two major sets of strains, each of which is composed of two groups. Each of the four groups has its own characteristic genome size, GC ratio, and greatly expanded core gene content, making the genomic diversity of each group within the range for other bacterial species. To test whether these 4 groups corresponded to genetically isolated clades, we inferred the phylogeny of each distributed gene that was present in at least two strains and absent in at least two strains; this analysis identified frequent homologous recombination within groups but not between groups or sets. G. vaginalis appears to include four nonrecombining groups/clades of organisms with distinct gene pools and genomic properties, which may confer distinct ecological properties. Consequently, it may be appropriate to treat these four groups as separate species.

Fig 1

NeighborNet of 473 protein-encoding core genes, found in all 17 G. vaginalis genomes, aligned with high confidence. Two major sets of strains are apparent, A and B, each of which is composed of two groups, indicated by the coloring of the labels identifying each strain: group 1, green; group 2, blue; group 4, red; group 3, orange.

Fig 2

Histograms showing the dispersion of gene clusters among genomes, as a function of the clustering identity threshold, ranging from 50% to 70%. Horizontal axes depict the number of genomes for which the cluster has a representative, and the vertical axes count the number of clusters that are dispersed among the given number of genomes. The standard species threshold of 70% is represented by the black bars. (a) Gardnerella vaginalis. (b) Staphylococcus aureus. Additional species are described in Fig. S1 in the supplemental material.

Fig 3

Neighbor-joining tree of 17 G. vaginalis genomes based on phylogenetically informative gene content; the midpoint is rooted at the left. The four neighbor-grouping complexes are colored differently. Node labels indicate the gene clusters found among the set of genomes associated with each node; the numerator is the number of gene clusters that are unique to that set, and the denominator is the total number of gene clusters with members in that set.

Fig 4

Maximum-likelihood phylogeny calculated for the concatenated alignment of 332 genes shared among all 17 G. vaginalis strains. All nodes have SH-like confidence values of 1. Node labels indicate the total number of gene clusters that confidently supported the clade (bootstrap > 190/200) relative to the total number that could confidently support or reject the clade.

Fig 5

NeighborNet diagram derived from the vaginolysin (vly) gene sequences.