Bioavailable Microbial Metabolites of Flavanols Demonstrate Highly Individualized Bioactivity on In Vitro β-Cell Functions Critical for Metabolic Health

Abstract

Dietary flavanols are known for disease preventative properties but are often poorly absorbed. Gut microbiome flavanol metabolites are more bioavailable and may exert protective activities. Using metabolite mixtures extracted from the urine of rats supplemented with flavanols and treated with or without antibiotics, we investigated their effects on INS-1 832/13 β-cell glucose stimulated insulin secretion (GSIS) capacity. We measured insulin secretion under non-stimulatory (low) and stimulatory (high) glucose levels, insulin secretion fold induction, and total insulin content. We conducted treatment-level comparisons, individual-level dose responses, and a responder vs. non-responder predictive analysis of metabolite composition. While the first two analyses did not elucidate treatment effects, metabolites from 9 of the 28 animals demonstrated significant dose responses, regardless of treatment. Differentiation of responders vs. non-responder revealed that levels of native flavanols and valerolactones approached significance for predicting enhanced GSIS, regardless of treatment. Although treatment-level patterns were not discernable, we conclude that the high inter-individual variability shows that metabolite bioactivity on GSIS capacity is less related to flavanol supplementation or antibiotic treatment and may be more associated with the unique microbiome or metabolome of each animal. These findings suggest flavanol metabolite activities are individualized and point to the need for personalized nutrition practices.

Keywords: flavanol metabolites; glucose sensitivity; gut microbiome; personalized nutrition; phytochemicals; responder analysis.

Figure 1

Treatment-level Comparisons show Urine Flavanol Metabolites Increase β-cell Insulin Secretion during Low Glucose Condition. Low glucose (2.8 mM) insulin secretion results of INS-1 832/13 β-cells following 24-h culture with metabolites from rats fed the vehicle (white bars), catechin/epicatechin (C/EC) (light gray bars), or grape seed extract (GSE) (dark gray bars) and treated with (striped bars) or without antibiotics (Abx) (solid bars). Urine flavanol metabolites were diluted in media at 1% (A), 2.5% (B), 5% (C), 7.5% (D), and 10% (E) final concentrations. Values are reported relative to the control β-cells cultured with water (Supplemental Figure S2). Data represent the average of 3 β-cell culture triplicates for each animal (n = 4–5 animals). * Represent 2-way ANOVA with Šidák’s post hoc test results of flavanol effects, Abx effects, interaction effects, and significance compared to the vehicle control. * p < 0.05, ** p < 0.01, *** p < 0.001 or not significant (ns).

Figure 2

Metabolites Do Not Affect Insulin Secretion during High Glucose Condition. High glucose (12 mM) insulin secretion results of INS-1 832/13 β-cells following 24-h culture with metabolites from rats fed the vehicle (white bars), catechin/epicatechin (C/EC) (light gray bars), or grape seed extract (GSE) (dark gray bars) and treated with (striped bars) or without antibiotics (Abx) (solid bars). Metabolites were diluted in media at 1% (A), 2.5% (B), 5% (C), 7.5% (D), and 10% (E) final concentrations. Values are reported relative to the control β-cells cultured with water (Supplemental Figure S2). Data represent the average of 3 β-cell culture triplicates for each animal (n = 4–5 animals). Represent 2-way ANOVA with Šidák’s post hoc test results of flavanol effects, Abx effects, interaction effects, and significance compared to the vehicle control. Not significant (ns).

Figure 3

Metabolites Do Not Affect Insulin Secretion Fold Induction. Insulin secretion fold induction difference between high (Figure 2) and low (Figure 1) glucose stimulation results of INS-1 832/13 β-cells following 24-h culture with metabolites from rats fed the vehicle (white bars), catechin/epicatechin (C/EC) (light gray bars), or grape seed extract (GSE) (dark gray bars) and treated with (striped bars) or without antibiotics (Abx) (solid bars). Metabolites were diluted in media at 1% (A), 2.5% (B), 5% (C), 7.5% (D), and 10% (E) final concentrations. Values are reported relative to the control β-cells cultured with water (Supplemental Figure S2). Data represent the average of 3 β-cell culture triplicates for each animal (n = 4–5 animals). Represent 2-way ANOVA with Šidák’s post hoc test results of flavanol effects, Abx effects, interaction effects, and significance compared to the vehicle control. Not significant (ns).

Figure 4

Metabolites Do Not Affect Insulin Content. Insulin content values of INS-1 832/13 β-cells following 24-h culture with metabolites from rats fed the vehicle (white bars), catechin/epicatechin (C/EC) (light gray bars), or grape seed extract (GSE) (dark gray bars) and treated with (striped bars) or without antibiotics (Abx) (solid bars). Reconstituted urine metabolites containing metabolites were diluted in media at 1% (A), 2.5% (B), 5% (C), 7.5% (D), and 10% (E) final concentrations. Values are reported relative to the control β-cells cultured with water (Supplemental Figure S2). Data represent the average of 3 β-cell culture triplicates for each animal (n = 4–5 animals). Represent 2-way ANOVA with Šidák’s post hoc test results of flavanol effects, Abx effects, interaction effects, and significance compared to the vehicle control. Not significant (ns).

Figure 5

Responder vs. Non-responder Predictive Analyses of GSIS based on Metabolite Profiles. (A) Differentiation of urine sample extracts (5% dose only) by highest vs. lowest quartiles for GSIS parameters. GSIS was compared between the highest (black bars) vs. lowest quartiles (gray bars) by t-tests, with the Holm-Sidak method to control for multiple comparisons (1 family, 4 t-tests: 1 per GSIS response measure), with the overall family α = 0.05. Each GSIS measure was analyzed individually, without assuming a consistent SD. Comparison of urine compositions between the most effective vs. least effective samples for each GSIS measure including insulin secretion during low glucose (B), during high glucose (C), insulin secretion fold induction (D), and insulin content (E). Within each GSIS measure, levels of each compound class were compared between the highest vs. lowest quartiles by t-tests, with the Holm-Sidak method to control for multiple comparisons (1 family per GSIS measure, 6 t-tests: 1 per compound class). For all tests: **** p < 0.0001.

Figure 6

Predictive Value of Metabolite Profiles on GSIS activity. Comparison of GSIS response between urine derived samples with the highest and lowest quartiles of concentrations of various classes of compounds, including total (A) native flavanols, (B) microbial metabolites, (C) valerolactones, (D) phenylalkyl acids, (E) cinnamic acids and (F) hippuric acids. Values are presented as mean ± SEM. Within each compound class, response was compared between the highest vs. lowest quartiles by t-tests, with the Holm-Sidak method to control for multiple comparisons (1 family per compound class, 4 t-tests: 1 per response measure). * p < 0.05.