Changes in the vaginal microbiota across a gradient of urbanization

Abstract

The vaginal microbiota of healthy women typically has low diversity, which increases after perturbations. Among these, lifestyle associated with certain sexual and antimicrobial practices may be associated with higher diversity. To test this hypothesis, we characterized the vaginal microbiota in the cervicovaginal and introital sites in sexually active Amerindians (N = 82) spanning urbanization, and in urban mestizos (N = 29), in the Venezuelan Amazonas. HPV status was also considered. Sampling was performed in an urban gradient from remote villages to a town, and women were individually classified by the degree of urbanization (low, medium, and high). Amerindian cervicovaginal and introital microbiota diversity were not associated with major changes in urbanization or ethnicity. There was a non-significant trend of increased diversity with urbanization, with a few taxa found overrepresented in urban Amerindians (Brevibacterium linens and Peptoniphilus lacrimalis) or mestizos (Mobiluncus mulieris and Prevotella sp.). Among all women, cervicovaginal and introital samples clustered, respectively, in four and two community state types (CSTs), where most profiles were dominated by Lactobacillus iners, Gardnerella vaginalis or were highly diverse profiles. HPV status did not associate with microbial diversity. In conclusion, no association was found between urban level and the vaginal microbiome in Amerindian women, and little difference was found between ethnicities. L. iners and high diversity profiles, associated with vaginal health outcomes, prevail in these populations.

Figure 1

Study design. (a) Diagram of geographic locations (black points in the map) of the 7 villages and the urban town (Puerto Ayacucho, indicated with a star) of the study. Pictures depict the town (top) and Piaroa communities with intermediate and low access to urban services (middle and bottom pictures, respectively). (b) Numbers of women recruited, urban stratification, sampling, and sample analyses. In traditional communities, permission by the captain (Chief) preceded individual consent. In the urban town there was a public invitation to participate. From 228 women, who received a gynecological evaluation, 111 agreed to participate and complied with the inclusion criteria. A total of 82 Amerindians and 29 mestizos were included in the study. Surveys to assess urbanization status and clinical conditions were applied. A gynecologist sampled vaginal introitus and cervicovaginal (endo/ectocervix/posterior fornix) sites, using sterile swabs; Papanicolaou smear was performed and vaginal pH was taken. Additionally, blood for HIV, hemoglobin, hepatitis diagnostic, and fecal samples for parasite detection were also collected for ancillary studies. DNA was extracted from swabs, and used for human papillomavirus (HPV) detection and amplification of the regions V1-V3 of the 16S rRNA gene, which was then sequenced with Illumina MiSeq.

Figure 2

Cervicovaginal microbiota diversity. (a,b) Principal Coordinate Analysis (PCoA) of Bray–Curtis dissimilarity for Amerindians among urbanization groups (low n = 15, medium n = 20, high n = 20) (a), and between ethnicities: Amerindians n = 15, mestizo n = 27, (PERMANOVA) (b). Gray lines connect samples with the group centroid. Ellipses indicate one standard deviation. (c,d) Alpha diversity using Shannon index among Amerindian urbanization groups (linear mixed-effect model, LMM) (c), and between ethnicities (Kruskal–Wallis test) (b). (e) Discriminant taxa analysis between ethnicities (LEfSe).

Figure 3

Cervicovaginal microbiota composition and community state type (CST) clustering. (a) Individual vaginal taxa plots; the legend shows the 12 most abundant taxa. (b) Heatmap of the 30 most abundant taxa by cervicovaginal CSTs resulting from the hierarchical clustering analysis (see text for explanation). Right hand boxplot shows Shannon index for each cervicovaginal CSTs: CST-L. iners < CST-G.vaginalis < CST-div2 = CST-div1. Different letters over the boxplots indicate significant differences (Kruskal–Wallis, p < 0.001). Vaginal pH is also shown. (c) Discriminant taxa for each cervicovaginal CSTs (LEfSe, p < 0.01).

Figure 4

Bacterial community state types (CSTs) from cervicovaginal and introital sites in Amerindian urbanization groups and mestizos. (a,b) Heatmap of the 30 most abundant taxa in cervicovaginal (a) and introital (b) microbiota. There were no significant differences among groups for Shannon index and vaginal pH, shown at the bottom of the heatmaps. (c,d) Prevalence for each of the four cervicovaginal (c) and two introital (d) CSTs, by woman group. Confidence intervals of 95% are indicated in each bar. Although no association with urbanization was found for CST (log-linear model), comparison at the interior of each urbanization group showed significant differences (*) in the low urbanization groups (Fisher’s exact test); but not for the introital CSTs. NA indicates missing value.