Loss of Diurnal Oscillatory Rhythms in Gut Microbiota Correlates with Changes in Circulating Metabolites in Type 2 Diabetic db/db Mice

Abstract

Our hypothesis is that diabetes leads to loss of diurnal oscillatory rhythms in gut microbiota altering circulating metabolites. We performed an observational study where we compared diurnal changes of the gut microbiota with temporal changes of plasma metabolites. Metadata analysis from bacterial DNA from fecal pellets collected from 10-month old control (db/m) and type 2 diabetic (db/db) mice every 4 h for a 24-h period was used for prediction analysis. Blood plasma was collected at a day and night time points and was used for untargeted global metabolomic analysis. Feeding and activity behaviors were recorded. Our results show that while diabetic mice exhibited feeding and activity behavior similar to control mice, they exhibited a loss of diurnal oscillations in bacteria of the genus Akkermansia, Bifidobacterium, Allobaculum, Oscillospira and a phase shift in the oscillations of g.Prevotella, proteobacteria, and actinobacteria. Analysis of the circulating metabolites showed alterations in the diurnal pattern of metabolic pathways where bacteria have been implicated, such as the histidine, betaine, and methionine/cysteine pathway, mitochondrial function and the urea cycle. Functional analysis of the differential microbes revealed that during the day, when mice are asleep, the microbes of diabetic mice were enriched in processing carbon and pyruvate metabolic pathways instead of xenobiotic degradation as was observed for control mice. Altogether, our study suggests that diabetes led to loss of rhythmic oscillations of many gut microbiota with possible implications for temporal regulation of host metabolic pathways.

Keywords: TCA cycle; TMAO; circadian; histidine; metabolites; methionine/cysteine; microbiota; type 2 diabetes; urea cycle.

Figure 1

Diurnal oscillatory rhythms of the gut microbiota at the genus level in type 2 diabetes. Bacterial 16SrRNA sequence analysis was performed in fecal samples from 10 month old db/db (red) and aged-matched controls, db/m (black) collected every 4 h for a 24-h period at ZT0, ZT4, ZT8, ZT12, ZT16, ZT20. ZT indicates zeitgeber time, i.e., hours after the lights are on. (A–P) Identified OTUs belonging to each genus were summed up and cosinor analysis was performed for each genus. n = 3 per time point per group. Asterisks indicate the genera that exhibit significant oscillatory rhythmicity using the zero -amplitude test with a p-value of less than 0.05.

Figure 2

Global metabolomic analysis of plasma metabolites. Blood samples were collected at ZT5 (day) and at ZT17 (night) from 10-month-old db/db (diabetic) and aged-matched db/m (control) mice. Untargeted global metabolomics analysis was performed with the Metabolon Platform (Metabolon Inc). Statistically significant different metabolites were used for the analysis with MetaboAnalyst. (A) PCA analysis. (B) Metabolic pathway analysis of metabolites that were different in the transition from day to night in control and diabetic groups. (C) Venn graphs of metabolites that were different in the transition from day to night in control and diabetic groups. (D) Enrichment analysis of metabolites that similarly changed in the transition from day to night in control and diabetic groups. (E) Enrichment analysis of metabolites that changed in the transition from day to night only in control. (F) Enrichment analysis of metabolites that changed in the transition from day to night only in diabetes. n = 3 per time point per group.

Figure 3

Heatmap of plasma metabolites that changed during the transition from day to night. Blood samples were collected at day time (ZT5) and at night time (ZT17) from 10-month-old db/db (diabetic) and aged-matched db/m (control) mice. Untargeted global metabolomics analysis was performed with the Metabolon Platform (Metabolon Inc). MetaboAnalyst was used for the analysis. (A) Heatmap of statistically significant metabolites. We identified clusters that gained diurnal rhythmic patterns (increased (Cluster 1) or decreased (Cluster 3) in diabetes in the day but did not change in controls and clusters that lost diurnal rhythmic patterns in diabetes (increased (Cluster 2) or decreased (Cluster 4) in control in the day but did not change in diabetes. (B) Enrichment analysis of metabolites that gained rhythmic patterns in diabetes (Cluster 1 and 3). (C) Enrichment analysis of metabolites that lost rhythmic patterns in diabetes (Cluster 2 and 4). n = 3 per time point per group.

Figure 4

Pathways of metabolites that showed up in the cluster that lost diurnal patterns in type 2 diabetes (T2D). Blood samples were collected at day time (ZT5) and at night time (ZT17) from 10-month-old db/db (diabetes, red) and aged-matched db/m (control, black) mice. Untargeted global metabolomics analysis was performed with the Metabolon Platform (Metabolon Inc). (A) Histidine metabolism pathway. (B) Betaine metabolism pathway. (C) Methionine/cysteine pathway. n = 3 per time point per group. Two-way ANOVA. (*) indicates statistical difference p < 0.05, and (#) p < 0.01 in post-hoc comparisons. y-axis shows the volume and median = 1 normalized values of area under the curve values form each metabolite. Bars above each chart (grey vs. white) indicate night vs. day.

Figure 5

Pathways of metabolites that showed up in the cluster that lost diurnal patterns in type 2 diabetes (T2D). Blood samples were collected at day time (ZT5) and at night time (ZT17) from 10-month-old db/db (red) and aged-matched db/m (black) mice. Untargeted global metabolomics analysis was performed with the Metabolon Platform (Metabolon Inc). (A) Glycolysis, (B) TCA cycle and urea cycle, and (C) polyamine metabolism. n = 3 per time point per group. Two-way ANOVA. (*) indicate statistical difference p < 0.05, and (#) p < 0.01 in post-hoc comparisons. y-axis shows the volume and median = 1 normalized values of area under the curve values form each metabolite. Bars above each chart (grey vs. white) indicate night vs. day.

Figure 6

Diurnal activity, food consumption, and circadian gene expression in type 2 diabetic mice (T2D). 10-month-old db/db (red) and aged-matched controls, db/m (black). (A,B) All measurements were performed after a 48-h acclimation period followed by 48-h of data collection every 10 min. (A) Mean activity over the night period (19:00–07:00) and the day period (07:00–19:00). n = 8 per group. Two-way ANOVA. (*) indicate statistical significance at post-hoc multiple comparison test. (B) Food consumption during the dark period and the day. n = 8 per group. Two-way ANOVA. Asterisks indicate statistical significance at post-hoc multiple comparison test. (C) mRNA expression of Per-2 in colonic tissue (n = 3). Two-way ANOVA. (*) indicate statistical difference p < 0.05, and (#) p < 0.01 in post-hoc comparisons. (D) mRNA expression of Bmal-1 in colonic tissue. (n = 3). Two-way ANOVA. (*) indicate statistical difference p < 0.05, and (#) p < 0.01 in post-hoc comparisons.