Phase variable colony variants are conserved across Gardnerella spp. and exhibit different virulence-associated phenotypes

Abstract

The Gardnerella genus, comprising at least 13 species, is associated with the polymicrobial disorder bacterial vaginosis (BV). However, the details of BV pathogenesis are poorly defined, and the contributions made by individual species, including Gardnerella spp., are largely unknown. We report here that colony phenotypes characterized by size (large and small) and opacity (opaque and translucent) are phase variable and are conserved among all tested Gardnerella strains, representing at least 10 different species. With the hypothesis that these different variants could be an important missing piece to the enigma of how BV develops in vivo, we characterized their phenotypic, proteomic, and genomic differences. Beyond increased colony size, large colony variants showed reduced vaginolysin secretion and faster growth rate relative to small colony variants. The ability to inhibit the growth of Neisseria gonorrhoeae and commensal Lactobacillus species varied by strain and, in some instances, differed between variants. Proteomics analyses indicated that 127-173 proteins were differentially expressed between variants. Proteins with increased expression in large variants of both strains were associated with amino acid and protein synthesis and protein folding, whereas those increased in small variants were related to nucleotide synthesis, phosphate transport, ABC transport, and glycogen breakdown. Furthermore, whole genome sequencing analyses revealed an abundance of genes associated with variable homopolymer tracts, implicating slipped strand mispairing in Gardnerella phase variation and illuminating the potential for previously unrecognized heterogeneity within clonal populations. Collectively, these results suggest that phase variants may be primed to serve different roles in BV pathogenesis.IMPORTANCEBacterial vaginosis is the most common gynecological disorder in women of childbearing age. Gardnerella species are crucial to the development of this dysbiosis, but the mechanisms involved in the infection are not understood. We discovered that Gardnerella species vary between two different forms, reflected in bacterial colony size. A slow-growing form makes large amounts of the toxin vaginolysin and is better able to survive in human cervix tissue. A fast-growing form is likely the one that proliferates to high numbers just prior to symptom onset and forms the biofilm that serves as a scaffold for multiple BV-associated anaerobic bacteria. Identification of the proteins that vary between different forms of the bacteria as well as those that vary randomly provides insight into the factors important for Gardnerella infection and immune avoidance.

Keywords: Gardnerella; phase variation; phenotypic variation; vaginosis.

Fig 1

cpn60 alignment to determine species designation of Gardnerella strains of interest. Neighbor-joining phylogenetic tree of full-length cpn60 nucleotide sequences (1629 bp) of 91 Gardnerella spp. available genomes from GenBank. Clade distribution as described by (29). AKK101 most closely groups with clade C and strain 3336 with clade B. Strain 2492 is closest to clade D but may be outside of the original four clade designation. Regardless, each strain of interest is a distinct Gardnerella species. Arrows mark strains characterized in this study.

Fig 2

Gardnerella spp. exhibit variation in colony size and opacity. (A) (Left) Strains from three different Gardnerella species assembled into large (Lg) and small (Sm) colonies on BHIFF agar medium. (Right) Opaque (Op) and translucent (Tr) colonies were observed on sBHI plates. Black arrows indicate Op, and white arrows indicate Tr. (B) Colonies from strain AKK101 show a spectrum of opacity intensities. Images were taken with a Leica S8 APO stereo microscope.

Fig 3

Phenotypic switching of colony size occurs at appreciable frequencies in both directions. Data are presented as an average of three independent experiments. Error bars represent the standard deviation between independent experiments.

Fig 4

Large colony variants grow faster than small colony variants. Growth of each variant was quantified by measuring OD₆₀₀ over a period of 8 h in BHIFF medium. Data are presented as an average of three independent experiments. Error bars represent the standard deviation across independent replicates.

Fig 5

Supernatants from small variants elicit greater hemolysis. Bacteria were grown in BHI_YDS medium, and the culture supernatant was collected after 24 h. (A) Concentrated supernatant from triplicate wells was assayed for VLY production by percent hemolysis of human erythrocytes. (B) Strain growth was measured by the change in OD₆₀₀ between inoculation and supernatant harvesting. Data from A and B are presented as an average of three independent experiments. Error bars represent standard deviation. *P < 0.05,**P < 0.01, ***P < 0.001. (C) Western blots of VLY in Gardnerella spp. supernatants. Reported VLY molecular weight is 57 kDa for the full-length protein (with signal sequence). It is unclear why multiple bands are present for some strains.

Fig 6

Colony variants differ in their antagonism of urogenital bacteria. Gardnerella colony variants from strain AKK101 were streaked onto BHI_YDS medium + 10% FBS (fetal; bovine serum) and grown for 24 h before adding a spot of diluted bacterial culture from N. gonorrhoeae, L. crispatus, or L. gasseri . After 24 h of co-culture, plates were assessed for a zone of clearing (red brackets) within the spot suggestive of growth inhibition. The results from a representative experiment are shown, but three independent experiments were performed.

Fig 7

Colony variants differ in protein production. Volcano plots of all proteins differentially produced (P < 0.05) between colony variants in strains 14018 (top) and 2492 (bottom). Each circle represents one protein. Proteins accumulate in a linear fashion at a −log10(P-value) of 4 due to P-values lower than 0.00010 being assigned a value of 0.00010 (−log10(P-value) of 4). The gray-dashed line represents a P-value of 0.05; proteins above the line have a P-value less than 0.05. The red line represents a fold change of 2 for the Sm variant; proteins to the left of the line are increased in Sm by 2-fold or greater. The green line represents a fold change of 2 for the Lg variant; proteins to the right of the line are increased in Lg by 2-fold or greater.

Fig 8

Small colony variants survive longer during experimental cervix infection. Survival over time of Lg and Sm variants from strain AKK101 in (A) CTCM (cervical tissue culture medium) alone or (B) in co-culture with human ectocervical tissue. Data are presented as geometric mean of at least three independent experiments. Independent experiments in tissue co-culture use tissue from different donors. *P < 0.05

Fig 9

Homopolymer tract distribution in Gardnerella genomes. (A) Characteristics of homopolymer tracts in strains of interest. (B) Median tract number grouped by Gardnerella cpn60 clade. The line within each box represents the median; the “X” represents the mean. Bars extending from boxes represent maximum and minimum values. Data points outside of bars are outliers. **P < 0.01, ***P < 0.001.

Fig 10

Gardnerella colony variant infection cycle hypothesis. (A) The BV biofilm encasing both Lg and Sm Gardnerella variants and other BV-associated bacteria enters the vaginal canal. (B) Upon biofilm dispersal, the Sm variant withstands the host innate immune defenses. (C) Increased expression of GA-module-containing proteins enhances autoaggregation and subsequent Sm adhesion to the host epithelium. (D) Secretion of vaginolysin toxin releases glycogen stored in epithelial cells, which Sm uses as a nutrient source through increased PulA and MalQ production. (E) An environmental cue triggers phenotype switching from the Sm to Lg variant. (F) The Lg variant produces sugars and amino acids that support the growth of Prevotella and other anaerobes. The presence of Lg is inhibitory toward Lactobacilli. (G) Lg initiates biofilm production and other BV-associated bacteria become incorporated. (H) Maturation of the BV biofilm gives rise to Sm variants. Biofilm aggregates, either themselves or attached to shed epithelial cells (clue cells), are transmitted to the next host.