Gut microbiome-related effects of berberine and probiotics on type 2 diabetes (the PREMOTE study)

Abstract

Human gut microbiome is a promising target for managing type 2 diabetes (T2D). Measures altering gut microbiota like oral intake of probiotics or berberine (BBR), a bacteriostatic agent, merit metabolic homoeostasis. We hence conducted a randomized, double-blind, placebo-controlled trial with newly diagnosed T2D patients from 20 centres in China. Four-hundred-nine eligible participants were enroled, randomly assigned (1:1:1:1) and completed a 12-week treatment of either BBR-alone, probiotics+BBR, probiotics-alone, or placebo, after a one-week run-in of gentamycin pretreatment. The changes in glycated haemoglobin, as the primary outcome, in the probiotics+BBR (least-squares mean [95% CI], -1.04[-1.19, -0.89]%) and BBR-alone group (-0.99[-1.16, -0.83]%) were significantly greater than that in the placebo and probiotics-alone groups (-0.59[-0.75, -0.44]%, -0.53[-0.68, -0.37]%, P < 0.001). BBR treatment induced more gastrointestinal side effects. Further metagenomics and metabolomic studies found that the hypoglycaemic effect of BBR is mediated by the inhibition of DCA biotransformation by Ruminococcus bromii. Therefore, our study reports a human microbial related mechanism underlying the antidiabetic effect of BBR on T2D. (Clinicaltrial.gov Identifier: NCT02861261).

Fig. 1

Flow diagram of participant enrolment in the PREMOTE Trial.

Fig. 2. BBR significantly altered gut microbiome symbiosis after 13 weeks of treatment.

a Gene count (upper panel) and Shannon index (lower panel) of genes in different arms, baseline and post treatment; Plac, Placebo, n = 96; Prob, probiotics treatment, n = 98; BBR, berberine treatment, n = 85; Prob + BBR: berberine plus probiotics treatment, n = 102; *P < 0.05, **P < 0.01, ***P < 0.001, two-sided Kruskal–Wallis test. Dark lines in the boxes indicate medians, the width of the notches is the IQR, the lowest and highest values within 1.5 times the IQR from the first and third quartiles. b Distance-based redundancy analysis (dbRDA) plot based on Bray–Curtis distances of species in post-treatment samples was performed to assess the difference between the four treatment arms (Permanova P < 0.001). Projection of species-level gut microbiome samples constrained by treatment methods. Marginal box plots show the separation of the constrained projection coordinates (boxes show medians and quartiles, error bars extend to most the extreme value within 1.5 interquartile ranges), Plac, n = 96; Prob, n = 98; BBR, n = 85; Prob + BBR, n = 102. c Venn diagram showing the overlapping of microbial species among the four treatment arms that were altered from baseline to post treatment, two-sided Wilcoxon matched-pairs signed-rank test, q < 0.05. d Heatmap of gut microbial species that showed significantly changed their relative abundances (RAs) post treatment vs. baseline. Plac, n = 96; Prob, n = 98; BBR, n = 85; Prob + BBR: n = 102. The changes in nine species in probiotics formula ingested by participants were separately shown below. *q < 0.05, two-sided Wilcoxon match-pairs signed-rank test. The colour key represents the Z score. Bifidobacterium catenulatum–Bpc, B. catenulatum–Bifidobacterium pseudocatenulatum complex. Source data and exact P-value are provided in the Source Data file.

Fig. 3. BBR altered microbial BA metabolism and correlated with blood BAs and clinical outcomes.

a Changes in RAs of bile acid-inducible (Bai) genes induced by the treatments of four arms. hsdh, hydroxysteroid dehydrogenase; Bsh, gene encoding bile salt hydrolase. The Z-score was calculated with the two-sided Wilcoxon matched-pairs signed-rank test. A Z-score > 0 indicated an increase after treatment, while a z-score < 0 indicated a decrease after treatment. *P < 0.01, **P < 0.001, ***P < 0.0001; Plac, Placebo, n = 96; Prob, probiotics treatment, n = 98; BBR, berberine treatment, n = 85; Prob + BBR, berberine plus probiotics treatment, n = 102. b Comparisons of bile acid (BA) composition between baseline and post treatment in the four arms. CA, cholic acid; CDCA, chenodeoxycholic acid; DCA, deoxycholic acid; GCA, glycocholic acid; GCDCA, glycochenodeoxycholic acid; GDCA, glycodeoxycholic acid; GLCA, glycolithocholic acid; GUDCA, glycoursodeoxycholic acid; LCA, lithocholic acid; TCA, taurocholic acid; TCDCA, taurocholic chenodeoxycholic acid; TDCA, taurodeoxycholic acid; TLCA, taurolithocholic acid; TUDCA, tauroursodeoxycholic acid; UDCA, ursodeoxycholic acid. *q < 0.01, **q < 0.001, ***q < 0.0001, two-sided Wilcoxon match-pairs signed-rank test. c Correlations between microbial BA genes and blood BA compositions at the baseline (upper panel) vs. post treatment (lower panel) for all participants, Spearman correlation, colour key represented rho value, *q < 0.01. d Heatmap of correlations between the blood BAs and clinical outcomes. Multivariate GEE controlling for age, sex and BMI. The colour key represents the β-value, *q < 0.01. e Plasma FGF19 levels pre and post treatment, *P < 0.05, **P < 0.01, ***P < 0.001, two-sided Wilcoxon matched-pairs signed-rank test, dark lines in the boxes indicate medians, the width of the notches is the IQR, the lowest and highest values within 1.5 times the IQR from the first and third quartiles, Plac, n = 96; Prob, n = 98; BBR, n = 85; Prob + BBR: n = 102. 12a/nonBA, 12a-hydroxylated/non–12a-hydroxylated bile acids; 2hPPG, post-load plasma glucose; cp120, post-load serum C peptide; FPG, fasting plasma glucose; HbA1c, glycated haemoglobin; HOMA-IR, homoeostasis model assessment index for assessing insulin resistance; HOMA-β, homoeostasis model assessment index for assessing β-cell function; ins120, post-load serum insulin; TC, total cholesterol; Uncon/Con BA, unconjugated/conjugated bile acids. Baseline, baseline levels; post, post-treatment levels. Source data and exact P-value are provided in the Source Data file.

Fig. 4. R. bromii was inhibited by BBR to attenuate DCA transformation.

a The two-panel heatmap on the left shows the correlations between the key BBR responsive species and with major clinical outcomes and plasma levels of bile acid. The colour key shows Rho calculated by partial Spearman’s correlation with adjustment for age, sex and BMI. Δ of clinical parameters or BAs = 100% × (baseline value − post treatment value)/baseline value. Species in blue represent depleted species and species in orange represent enriched species after BBR treatments. *P < 0.05. b Bile acid transformation assay for R. bromii. The percentage composition of deoxycholic acid (DCA) and lithocholic acid (LCA) in the culture media with which R.bromii had grown for 24 h with primary bile acid (CA and CDCA) treatment were measured by LC/MS. n = 3, data are shown as the mean ± SD. c The growth curve of R. bromii with different concentrations of BBR in the in vitro culture experiment, demonstrated a significant inhibitory effect of BBR on R. bromii starting at a concentration of 25 μg ml⁻¹, n = 3, P < 0.001, determined by two-way repeated-measures ANOVA, data are shown as the mean ± SD. Bifidobacterium catenulatum − Bpc, B. catenulatum–Bifidobacterium pseudocatenulatum complex. Source data and exact P-value are provided in the Source Data file.