A citizen-science-enabled catalogue of the vaginal microbiome and associated factors

Abstract

Understanding the composition and function of the vaginal microbiome is crucial for reproductive and overall health. Here we established the Isala citizen-science project to analyse the vaginal microbiomes of 3,345 women in Belgium (18-98 years) through self-sampling, 16S amplicon sequencing and extensive questionnaires. The overall vaginal microbiome composition was strongly tied to age, childbirth and menstrual cycle phase. Lactobacillus species dominated 78% of the vaginal samples. Specific bacterial taxa also showed to co-occur in modules based on network correlation analysis. Notably, the module containing Lactobacillus crispatus, Lactobacillus jensenii and Limosilactobacillus taxa was positively linked to oestrogen levels and contraceptive use and negatively linked to childbirth and breastfeeding. Other modules, named after abundant taxa (Gardnerella, Prevotella and Bacteroides), correlated with multiple partners, menopause, menstrual hygiene and contraceptive use. With this resource-rich vaginal microbiome map and associated health, life-course, lifestyle and dietary factors, we provide unique data and insights for follow-up clinical and mechanistic research.

Fig. 1. Characterization of the Isala study cohort and key physiological, behavioural, lifestyle and environmental factors of the participating women.

a, The self-sampling kit sent to the participants via the national postal service. b, Geographical overview of the participants who sent in samples for this project, by overlaying their zip codes on a map of all Belgian municipalities. Darker colours represent higher numbers of participants with that specific zip code. As only participants who understood Dutch were allowed to participate, most participants were from the Northern part of Belgium. c, An overview of the population cohort that registered within 10 days after the first announcement, with their different citizen-science roles in the Isala project: minimal involvement by expressing online interest as potential donor via website and answering five questions on age, pregnancy, contraceptive use, country of living (for the past 3 years) and zip code (grey), partial involvement by filling out the extensive survey (blue) and full involvement as donors with 24-h follow-up survey (pink). d,e, The distribution of a selection of the survey variables: age and BMI (median and interquartile range (IQR) represented by the boxplot within the violin plot where the whiskers represent the distribution minima and maxima, n = 4,682) (d), reported contraceptive use of the whole cohort and a subset of the binary variables (e). The grey segments indicate participants who either did not know the answer to the question or did not respond to the second survey. The exact number of participants who answered ‘yes’ is included between brackets. NA, not applicable.

Fig. 2. Overview of the most abundant taxa in the vaginal microbiome of the Isala cohort with focus on the Lactobacillus taxa.

a, Maximum-likelihood phylogeny of species of the genus Lactobacillus inferred from the amino acid sequences of 100 single-copy core genes. Colours indicate the nine defined subgenera used in this study. Bold tip labels indicate representative species of the subgenera. Species names were taken from the Genome Taxonomy Database, which splits species that are very diverse, yielding, for example, L. delbrueckii_A and L. jensenii_A, the latter recently identified as L. mulieris. The size of the circles reflects the genome size (with the average genome size also shown between brackets). The underlined species names are the species names of the four typical vaginal Lactobacillus species, after which the subgenera are named. b, Validation of the 16S amplicon sequencing pipeline, including classification to Lactobacillus subgenera, with shotgun sequencing data (n = 18, Supplementary Fig. 2). For the typical four Lactobacillus subgenera, the Spearman correlations between their relative abundances in the amplicon and shotgun samples are shown. c, Stacked bar chart describing the microbiome composition of the Isala participants in terms of the nine taxa being first or second most abundant in at least 100 samples. d, Occurrence of the most dominant taxa based on the highest taxonomic resolution possible. Dominance was defined as the most abundant taxon that constituted at least 30% of the profile. ‘Other’ refers to the number of samples with a different (sub)genus dominant; ‘No dominance’ refers to the number of samples where no single (sub)genus reached at least 30% relative abundance.

Fig. 3. Vaginal microbiome structure of the Isala and VALENCIA cohorts.

a–d, t-SNE plot of microbiome samples in the Isala study; samples are coloured by the 13 CSTs as defined by VALENCIA (a). CST I—L. crispatus-dominated (A, high relative abundance; B, lower relative abundance), CST II—L. gasseri-dominated, CST III—L. iners-dominated (A, high relative abundance; B, lower relative abundance) and CST V—L. jensenii-dominated. CST IV-A—Candidatus Lachnocurva vaginae (BVAB1, identified up to the genus level as EU728721_g in the Isala data) with low abundance of G. vaginalis. CST IV-B—G. vaginalis with low relative abundance of Ca. L. vaginae. CST IV-C0—Prevotella, CST IV-C1—Streptococcus, CST IV-C2—Enterococcus-dominated, CST IV-C3—Bifidobacterium-dominated and CST IV-C4—Staphylococcus-dominated. Samples are coloured by the most abundant (sub)genus (b). Samples are coloured by the second most abundant (sub)genus (c). Samples are coloured by the largest relative abundance level in each sample (d). e–h, t-SNE plot of microbiome samples of the VALENCIA dataset (multi-temporal samples per participant included), coloured by the CSTs (e). Samples of the VALENCIA dataset coloured by the most dominant genus (f), by the second most dominant genus (g) and by the largest relative abundance level in each sample (h). In all plots, the grey dashed outlines indicate the ‘bridge’ regions samples dominated by L. iners and L. crispatus.

Fig. 4. Four main modules of interacting vaginal bacterial taxa as defined by a compositional correlation analysis and consisting of a minimum of three (sub)genera with a biologically plausible relationship.

Modules are enclosed in blue circles. Positive and negative correlations are represented in blue and red, respectively. Thickness of the line indicates the strength of the correlation. Quantitative correlations are given in Supplementary Fig. 11. Validation was done with the VALENCIA and VaHMP datasets. Taxa that are not part of modules, such as L. iners and L. gasseri, are not shown.

Fig. 5. Statistical analysis of the association of different personal, reproductive, lifestyle, health, environmental, hygiene and dietary factors with the vaginal microbiome space.

a–e, Each panel displays associations on different levels of the microbiome: the effect on beta diversity between the samples (Adonis test) (a), the effect on the alpha diversity (Shannon index) of the samples (linear model; centre is the effect size, error bars represent 95% confidence inteval, P values are indicated on the right) (b), the effect on the abundances on six VALENCIA CSTs, which have at least 120 samples in each group (CST I-A, I-B, III-A, III-B, IV-B and V) (c), the effect on the module eigentaxa (d), and associations of specific taxa (e). Asterisks represent significant associations (q < 0.05; colour difference between white and black asterisks is for visualization purposes). For the individual taxa tested, the number refers to the number of pipelines that indicate a significant association with the self-reported data: ALDeX2, ANCOM-BC, DESeq, limma, Maaslin2 and a linear regression on the CLR-transformed abundance data. The number of samples for each question was almost the entire study (n = 3,043 participants). Due to missing data or specific comparisons, this can deviate slightly: detailed counts and P values are provided in Supplementary Table 2. All tests were adjusted for multiple comparisons with Benjamini–Hochberg procedure.