Fermented beetroot modulates gut microbial carbohydrate metabolism in prediabetes and prevents high-fat diet induced hyperglycemia in a prediabetic model

Abstract

The global increase in prevalence of (pre-)diabetes demands immediate intervention strategies. In our earlier work, we demonstrated in vitro antidiabetic potential of a fermented beetroot product (PN39). Here, we examined the impact of PN39 on glucose tolerance and gut microbiota in C57BL/6J male mice and on prediabetic (PD) subjects' stool microbiota. In mice, high-fat diet (HFD) consumption for 9 weeks resulted in hyperglycemia and impaired glucose tolerance (GT) while concomitant consumption of PN39 and HFD (PN39+HFD) prevented GT impairment. Meanwhile, feeding the mice with HFD for 5 weeks to induce PD and later administering them with PN39 for 4 weeks (PD + PN39) neither improved fasting blood glucose nor GT. Relative to control groups, the gut microbiota of both PD mice and humans were characterized by decreased Clostridia UCG-014 and Lactobacilli as well as significantly altered gut microbial carbohydrate metabolism. Feeding PN39 together with HFD preserved Clostridia UCG-014 and Lactobacilli, increased short chain fatty acid production relative to mice fed with HFD only. Treating gut microbiota of PD subjects with PN39 however increased Clostridia UCG-014 and Lactobacilli populations and increased short chain fatty acids concentrations in the stools. In both mice and humans, PN39 treatment rectified the altered microbial carbohydrate metabolism observed in their PD counterparts. This suggests that the gut microbial modulatory effects of PN39 coupled with its capacity to regulate gut microbial glucose metabolism, likely played a role in preventing PD in mice receiving PN39+HFD. Taken together, our results indicate that PN39 could act as a potent antidiabetic functional food for preventing diabetes and its associated dysbiosis.

Keywords: Dysbiosis; Functional food; Gut microbial metabolism; Gut microbiota; Hyperglycemia.

Graphical abstract

Fig. 1

Effects of PN39 consumption on A. body weight, B. fasting blood glucose on week nine after mice were fasted for 12 h, C. Intraperitoneal Glucose Tolerance Test measured for 120 min on the 9th week, D. area under the curve for blood glucose tolerance test on the 9th week, E. serum insulin levels on the 9th week of study, F. HOMA-IR (Homeostatic Model Assessment for Insulin Resistance) determined after the 9th week of study. Columns or time points with different alphabets (a, b) are significantly different (one-way ANOVA, Tukey's multiple comparisons tests, ∗ = p < 0.05). N = 8–10 mice and each data point in A. and C. represents mean ± SEM.

Fig. 2

Measures of alpha diversity for healthy, HFD (prediabetic/PD), PN39+ HFD and PD + PN39 mice. The number of observed OTUs, ACE index, Shannon index and Simpson's index were used to measure/analyze the alpha diversity of the mice cecal microbiota p < 0.05 indicates statistical significance using ANOVA. All these indices (ACE, Shannon, Simpson) are calculated based on the OTU data (equivalent to “genera detected” in ecology) not on taxa data. N = 10 mice randomly pooled into 3 groups.

Fig. 3

Representation of principal component analysis (PCoA) performed for healthy, prediabetic, PN39+ HFD and PD + PN39 mice based on unweighted UniFrac metric. The cecum contents of mice (n = 10) were randomly pooled into 3 groups (to make up for inter-individual variations).

Fig. 4

A. Top 20 most abundant genera in healthy and PD mice, B. Differential gut microbial composition in healthy, PN39+HFD and PD + HFD mice groups. LEfSe determines the biological relevance of significant enrichment of taxa and ranks them by effect size. LDA score indicates the magnitude of the effect size. Taxa enriched in the healthy mice, PN39+HFD and PD + HFD groups relative to PD group are described with a positive LDA score, C. Differential PICRUSt prediction based on KEGG pathways for healthy vs PD, PN39+HFD vs PD and PD + HFD vs PD mice groups by STAMP software. Blue bars represent healthy mice, yellow bars represent PD mice, red bars represent PN39+HFD and green bars represent PD + HFD group. An extended error bar plot indicates differences in the predicted functional profiles of gut microbiota between groups. Bar plots on the left side present the mean proportion of each KEGG pathway. Dot plots on the right display the differences in mean proportions between the two groups using p-values.

Fig. 5

A. Impact of PN39 consumption on mice cecal SCFAs. Data are expressed as Mean ± SEM for each group (n = 10). B. Heatmap showing Spearman's correlation coefficients between Lactobacillus and Clostridia UCG_014 and glucose tolerance as well as SCFAs levels in the cecum of PN39+HFD mice. Blue squares indicate positive linear correlation while white squares indicate no linear relationship. Significant correlations are indicated inside the heatmap with ∗ (p < 0.01).

Fig. 6

Impact of PN39 and inulin on in vitro fermentation of prediabetic human fecal microbiota. A. Relative abundance of top 20 genera in non-diabetic donors versus prediabetic donors, B. PD + PN39 versus prediabetic donors and C. PD + Inulin versus prediabetic donors. Differential PICRUSt was used to predict KEGG pathways that were significantly different among the sample groups after in vitro fermentation of PD stool samples with either PN39 or inulin. D. Differential KEGG pathways between non-diabetic volunteer gut microbiota and prediabetic volunteer gut microbiota, E. differential KEGG pathways PD + PN39 gut microbiota and prediabetic volunteer gut microbiota, F. differential KEGG pathways between PD + Inulin gut microbiota and prediabetic volunteer gut microbiota. For each comparison, the mean proportion of predicted KEGG pathways (left) and difference in mean proportions (right) were illustrated.

Fig. 7

Quantification of stool SCFAs and lactate in non-diabetic and prediabetic subjects as well as prediabetic stool samples co-incubated with PN39 or inulin for 24 h. For each group (n = 3), mean ± SEM is represented (one-way ANOVA, Tukey's multiple comparisons tests, ∗ = p < 0.05).