Analysis of the oral microbiome during hormonal cycle and its alterations in menopausal women: the "AMICA" project

Abstract

The maintenance of human health is dependent on a symbiotic relationship between humans and associated bacteria. The diversity and abundance of each habitat's signature microbes vary widely among body areas and among them the oral microbiome plays a key role. Significant changes in the oral cavity, predominantly at salivary and periodontal level, have been associated with changes in estrogen levels. However, whether the oral microbiome is affected by hormonal level alterations is understudied. Hence the main objective pursued by AMICA project was to characterize the oral microbiome (saliva) in healthy women through: profiling studies using "omics" technologies (NMR-based metabolomics, targeted lipidomics by LC-MS, metagenomics by NGS); SinglePlex ELISA assays; glycosidase activity analyses and bioinformatic analysis. For this purpose, thirty-nine medically healthy women aged 26-77 years (19 with menstrual cycle and 20 in menopause) were recruited. Participants completed questionnaires assessing detailed medical and medication history and demographic characteristics. Plasmatic and salivary levels of sexual hormones were assessed (FSH, estradiol, LH and progesteron) at day 3 and 14 for women with menstrual cycle and only once for women in menopause. Salivary microbiome composition was assessed through meta-taxonomic 16S sequencing and overall, the salivary microbiome of most women remained relatively stable throughout the menstrual cycle and in menopause. Targeted lipidomics and untargeted metabolomics profiling were assessed through the use of LC-MS and NMR spectroscopy technologies, respectively and significant changes in terms of metabolites were identified in saliva of post-menopausal women in comparison to cycle. Moreover, glycosyl hydrolase activities were screened and showed that the β-D-hexosaminidase activity was the most present among those analyzed. Although this study has not identified significant alterations in the composition of the oral microbiome, multiomics analysis have revealed a strong correlation between 2-AG and α-mannosidase. In conclusion, the use of a multidisciplinary approach to investigate the oral microbiome of healthy women provided some indication about microbiome-derived predictive biomarkers that could be used in the future for developing new strategies to help to re-establish the correct hormonal balance in post-menopausal women.

Figure 1

Plasmatic and salivary hormones in women with menstrual cycle at 3rd and 14th day (M 3rd day and M 14th day) and in menopause (MP). (A) Sex hormone levels in plasma in the three different group of women [follicle stimulating hormone (FSH), estradiol (E2), luteinizing hormone (LH), and progesterone (PGN)]. (B) Salivary FSH levels (not detectable, nd). ^* and ^° indicate significant differences compared to MP and M 14th day, respectively. P < 0.05 was considered statistically significant. Two-way ANOVA, followed by Tukey post-hoc.

Figure 2

Differences in abundances of taxa in salivary samples from women with menstrual cycle at 3rd and 14th day (M-3rd day and M-14th day) and in menopause (MP) (A, B, C) and alpha-and beta-diversity comparisons of the gut microbiomes of each group (D–I). (A) Relative abundances of genera, (B) families and specie (C). (D-F) Boxplot of the Shannon diversity of each group. (G-I) Principal coordinate analysis of Bray–Curtis distances. The colors of the boxplots and dots represent the different groups analyzed according to the legend.

Figure 3

Screening of Glycosyde Hydrolases Activities in saliva. ^* and ^° indicate significant differences compared to MP and M 14th day, respectively. P < 0.05 was considered statistically significant. Two-way ANOVA, followed by Tukey post-hoc.

Figure 4

Multivariate analysis of NMR-based metabolomics data. (A) Orthogonal Projections to Latent Structures Discriminant Analysis (OPLS-DA) displaying data projection (menopause MP, green squares; menstrual M-3rd day and M-14th, purple and red squares, respectively) and class separation along the predictive component t[1] and (B) loadings plot showing NMR spectral variables responsible for saliva samples distribution.

Figure 5

Bar plot showing normalized bin intensities of discriminant metabolites evaluated for menopause MP (green bar), menstrual M-3rd (purple bar) and M-14th (red bar) saliva classes. (A) Citrate, (B) Histidine, (C) Taurine, (D) Fucose, (E) Galactose and (F) n-Butyrate. Univariate test significance is reported as *p < 0.05 and **p < 0.001.

Figure 6

Correlation map based on Pearson correlation coefficients between selected metabolites and enzymes. Rows and columns are rearranged according to the WARD-based correlation matrix–based hierarchical clustering (CMBHC).

Figure 7

Targeted lipidomic analysis of salivary samples. (A) Anandamide (N arachidonoylethanolamine, AEA) levels expressed in pmol/ml of saliva; (B) 2-arachidonoylglycerol (2-AG) levels expressed in pmol/ml of saliva; (C) palmitoylethanolamide (PEA) levels expressed in pmol/ml of saliva and (D) oleoylethanolamide (OEA) levels expressed in pmol/ml of saliva. ^* and ^° indicate significant differences compared to MP and M-14th day, respectively. P < 0.05 was considered statistically significant. One-way ANOVA, followed by Tukey post-hoc.

Figure 8

Multiple factor analysis of saliva variables. (A) Multiple factor analysis including variable groups meta-taxonomic analysis, glycoside hydrolases activities, NMR metabolomics and targeted lipidomics through mass spectrometry with estradiol levels, and anthropometric data show separation between experimental groups. (B) Importance of variables in the MFA that included all data. (C) Multiple factor analysis without estradiol levels and anthropometric data. (D) Importance of variables in the MFA that did not include estradiol levels and anthropometric data.

Figure 9

Correlation network of all variables included in the study. Spearman correlations with FDR-corrected p < 0.001 are shown in the network. Positive correlation is indicated by red edges and negative correlation is indicated by blue edges. Color of nodes indicate the type of variables, as shown in legend.