A Multi-Omic Analysis for Low Bone Mineral Density in Postmenopausal Women Suggests a RELATIONSHIP between Diet, Metabolites, and Microbiota

Abstract

The effect of microbiota composition and its health on bone tissue is a novel field for research. However, their associations with bone mineral density (BMD) have not been established in postmenopausal women. The present study investigates the relation of diet, the microbiota composition, and the serum metabolic profile in postmenopausal women with normal-BMD or with low-BMD. Ninety-two Mexican postmenopausal women were classified into normal-BMD (n = 34) and low-BMD (n = 58). The V4 hypervariable region was sequenced using the Miseq platform. Serum vitamin D was determined by chemiluminescence immunoassay. Serum concentrations of acyl-carnitines and amino acids were determined by electrospray tandem mass spectrometry. Diet was assessed by a food frequency questionnaire. The low-BMD group had fewer observed species, higher abundance of γ-Proteobacteria, lower consumption of lycopene, and lower concentrations of leucine, valine, and tyrosine compared with the normal-BMD group. These amino acids correlated positively with the abundance of Bacteroides. Lycopene consumption positively correlated with Oscillospira and negatively correlated with Pantoea genus abundance. Finally, the intestinal microbiota of women with vitamin D deficiency was related to Erysipelotrichaceae and Veillonellaceae abundance compared to the vitamin D non-deficient group. Associations mediated by the gut microbiota between diet and circulating metabolites with low-BMD were identified.

Keywords: bone health; bone mineral density; estrogen deficiency; inflammation; intestinal microbiota; lycopene; postmenopausal women; vitamin D deficiency.

Figure 1

Alpha microbiota diversity in normal-BMD and low-BMD postmenopausal women. (A) Alpha rarefaction curves representing the observed number of species in the two study groups. The y-axis indicates less observed species in the low-BMD group than the normal-BMD group. No significant differences were found between the groups in richness (Chao index) (B) or diversity (Shannon index) (C). ** p < 0.01 compared to the normal-BMD group.

Figure 2

Bacterial taxonomy in normal-BMD and low-BMD postmenopausal women. A total of seven phyla dominated the microbiota composition in both groups, the predominant phyla were Bacteroidetes, colored green, followed by Firmicutes, colored orange (A). About the Firmicutes/Bacteroidetes (F/B) ratio, no statistical difference between the groups was found (B) (p = 0.22). (C) Heat tree for pair-wise comparison. The color of each taxon indicates the log-2 ratio of proportions observed in each condition. Only those taxa that were statistically significant using the Wilcox rank sum test followed by a Benjamin–Hochberg correction (FDR) for multiple comparisons are shown colored. γ-Proteobacteria colored green, specifically the Klebsiella and Erwinia genus, were enriched in the low-BMD group; meanwhile, the Bilophila and Bacteroidales group colored yellow were enriched in the normal-BMD group. (D) Bacterial taxa differentially represented between groups identified by the linear discriminant analysis (LDA) effect size. Only taxa with an alpha value of 0.05 and with an LDA score of at least 2 are shown. Lachnospira, Anaeroplasmataceae, and Yersinia were enriched in the low-BMD group, while Akkermansia and Actinobacteria were overrepresented in the normal-BMD group.

Figure 3

Effects of age-related BMD on bacterial community composition. Heat tree for pair-wise comparison, divided by BMD-age, (A) middle-age women (ages 45–59 years), (B) middle-old women (60–74 years), and (C) old women (>74 years). The color of each taxon indicates the log-2 ratio of proportions observed in each condition. Only those taxa that were statistically significant using the Wilcox rank sum test followed by a Benjamin–Hochberg correction (FDR) for multiple comparisons are shown as colored. Taxa colored green are enriched in the low-BMD group and those colored yellow are enriched in the normal-BMD group.

Figure 4

Contribution of dietary components to the composition of gut microbiota. (A) VIP analysis represents the relative contribution of nutrient consumption to the variance between normal-BMD and low-BMD. The high value of the VIP score indicates the great contribution of the nutrient consumption to the group separation. The blue and red boxes on the right indicate whether the consumption is increased (blue) or decreased (red). The low-BMD group diet was characterized by higher fiber intake, starch, and manganese consumption. The normal-BMD group was characterized by lycopene and vitamin D intake. (B) Correlation heatmap. Blue and red color indicates an increase and decreased correlation, respectively. The size of each dot was associated with the p-value, where a big circle represents a smaller p-value. Lachnospiraceae abundance was positively correlated with fiber consumption and in a negative manner with vitamin D intake. Calcium consumption was negatively correlated with Butyricimonas. Lycopene consumption was decreased in the low-BMD group, and positively correlated with Oscillospira and negatively correlated with Pantoea genus abundance.

Figure 5

Association between serum metabolites with bone mineral density and vitamin D deficiency. (A) Amino acid concentration heatmap in each group; red and green indicate an increased and decreased concentration, respectively; the low-BMD group had a lower valine and leucine concentration than the normal-BMD group. (B) Amino acid concentration heatmap in each group, red and green indicate increase and decreased concentration, respectively. Postmenopausal no vitamin D-deficient women (VDND) had higher serum concentrations of valine and leucine, tyrosine, alanine, and proline than postmenopausal vitamin D-deficient women (VDD).

Figure 6

Association between vitamin D deficiency and bacterial community composition. (A) A total of seven phyla dominated the microbiota composition in postmenopausal no vitamin D-deficient women (VND) and postmenopausal vitamin D-deficient women (VDD); the predominant phyla were Bacteroidetes, colored purple, followed by Firmicutes, colored green. (B) At the genus level, the most abundant taxa were Akkermansia, Clostridium, Faecalibacterium, Lachnospira, Phascolarctobacterium, Ruminococcus, Bacteroides, Parabacteroides, Paraprevotella, Prevotella, and Methanobrevibacter. (C) Bacterial taxa differentially represented between groups identified by linear discriminant analysis (LDA) effect size. Only taxa with an alpha value of 0.05 and with an LDA score of at least 2 are shown. The intestinal microbiota of women in the vitamin D-deficient group (VDD) had significantly higher proportions of Erysipelotrichaceae and Veillonellaceae compared to the vitamin D non-deficient group (VDND). (D) No significant differences were found between the groups in observed species, richness (Chao), or diversity (Shannon index). (E) Vitamin D levels negatively correlates with Enterobacteriaceae (r = −0.262; 95% CI (−0.444-0.060), p = 0.011) and Erwinia (r = −0.212; 95% CI (-0.406−0.0008), p = 0.043).