Gut microbes and the liver circadian clock partition glucose and lipid metabolism

Abstract

Circadian rhythms govern glucose homeostasis, and their dysregulation leads to complex metabolic diseases. Gut microbes exhibit diurnal rhythms that influence host circadian networks and metabolic processes, yet underlying mechanisms remain elusive. Here, we showed hierarchical, bidirectional communication among the liver circadian clock, gut microbes, and glucose homeostasis in mice. To assess this relationship, we utilized mice with liver-specific deletion of the core circadian clock gene Bmal1 via Albumin-cre maintained in either conventional or germ-free housing conditions. The liver clock, but not the forebrain clock, required gut microbes to drive glucose clearance and gluconeogenesis. Liver clock dysfunctionality expanded proportions and abundances of oscillating microbial features by 2-fold relative to that in controls. The liver clock was the primary driver of differential and rhythmic hepatic expression of glucose and fatty acid metabolic pathways. Absent the liver clock, gut microbes provided secondary cues that dampened these rhythms, resulting in reduced lipid fuel utilization relative to carbohydrates. All together, the liver clock transduced signals from gut microbes that were necessary for regulating glucose and lipid metabolism and meeting energy demands over 24 hours.

Keywords: Fatty acid oxidation; Gluconeogenesis; Metabolism; Mouse models.

Figure 1. Gut microbes are essential for liver circadian clock–mediated glucose metabolism.

(A) Resting blood glucose levels of SPF and GF WT and LKO male mice every 4 hours over 24 hours (n = 4-6/group/time point, SPF and GF groups also shown separately). CircWave statistics indicate significantly oscillating (P < 0.05) or not oscillating (P > 0.05) values. (B) GTT of SPF and GF WT and LKO male mice (n = 10–13/group). (C) Circulating insulin levels during GTT (n = 10–13/group). (D) PTT (n = 10–15/group) of SPF and GF WT and LKO male mice. Data are shown as the mean ± SEM. Lines in box plots represent the median, and whiskers represent the minimum and maximum, respectively. Two-tailed unpaired Welch’s t tests was performed between 2 groups; Brown-Forsythe and Welch’s ANOVA followed by Dunnett’s tests was performed between 3 or more groups. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, relative to SPF WT. Graphs represent AUC normalized to baseline glucose.

Figure 2. Modulation of gut microbes can both eliminate and restore liver clock–mediated GNG.

(A) PTT in WT and LKO male mice before (Pre-Abx) and after (Post-Abx) daily antibiotic treatment for 2 weeks (n = 12–13/group). The graph represents the AUC. (B) PTT in GF WT male mice conventionalized with fecal microbes from SPF WT or LKO male mice (n = 11–12/group). (C) PTT in GF WT and LKO male mice conventionalized with fecal microbes from SPF WT male mice (n = 15–16/group). Inset graphs represents AUC normalized to baseline glucose. Data points represent mean ± SEM. Lines in box plots represent the median, and whiskers represent the minimum and maximum, respectively. Two-tailed unpaired Welch’s t tests were performed between 2 groups; Brown-Forsythe and Welch’s ANOVA followed by Dunnett’s tests were performed between 3 or more groups. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, relative to Pre-Abx WT.

Figure 3. Liver circadian clock drives unique patterns of oscillations in microbial abundance.

16S rRNA gene sequencing of stool from SPF WT and LKO male mice every 6 hours over 48 hours via repeat collection (n = 7–8/group). (A) Proportion of nonoscillating (gray area) versus significantly oscillating (colored areas) amplicon sequence variants (ASVs) identified via eJTK (GammaBH < 0.05). Oscillating (Osc) ASVs were divided by taxonomic class. (B) Abundance counts of total versus oscillating ASVs within Bacteroidales and Clostridiales classes. (C) Number of oscillating Clostridiales ASVs at the family level in WT and LKO mice. (D) Abundance counts of total versus oscillating ASVs within Lachnospiraceae and Ruminococcaceae families. (E) R² values of nonzero base sinusoidal fits of log ratios at each time point relative to ZT2. Data represent mean ± SEM. Lines in box plots represent the median, and whiskers represent the minimum and maximum, respectively. Two-tailed paired t test was performed.

Figure 4. Metabolic pathway gene expression is downregulated in absence of a liver clock across time, regardless of microbial status.

Transcriptome analysis of liver samples collected every 4 hours over 24 hours (ZT2, -6, -10, -14, -18, and -22) from SPF and GF WT and LKO male mice (n = 3/time point/group). (A) Data analysis workflow, demonstrating 3 arms of analysis: 1, differential expression; 2, oscillation; and 3, network coexpression. (B) Principal component analysis of transcriptome profiles; all samples within each group were pooled. (C) Differentially regulated KEGG pathways between WT and LKO mice within SPF and GF groups; all samples within each group were pooled. Metabolic pathways are colored (see key),and nonmetabolic pathways are colored gray. Bars to the right of the midline plot represent pathways downregulated in LKO mice compared with WT mice; bars to the left represent pathways upregulated in LKO mice compared with WT mice.

Figure 5. Gut microbes work through the liver clock to impart unique expression patterns of key gluconeogenic genes.

Diurnal transcriptome analysis of liver samples collected every 4 hours over 24 hours from SPF and GF WT and LKO male mice (n = 3/time point/group) maintained in 12:12 LD (ZT2, -6, -10, -14, -18, and -22). (A) WT-median-normalized expression of differentially expressed (DE) genes within identified KEGG pathways. (B–E) VST-normalized expression of leading-edge genes in FA metabolism (B), PPAR signaling (C), and GNG (D and E) D between SPF and GF WT and LKO mice, with the exception of Pck1 and Pck2 (E), which are only differentially expressed in the SPF group (SPF and GF groups shown separately). Data represent mean ± SEM.

Figure 6. Liver clock and gut microbes drive unique hepatic transcriptome oscillations.

Diurnal transcriptome analysis of liver samples collected every 4 hours over 24 hours from animals maintained in 12:12 LD (ZT2, -6, -10, -14, -18, -22) from SPF and GF WT and LKO male mice (n = 3/time point/group). (A) Venn diagram of significantly oscillating transcripts across each group; total number of oscillating transcripts are under each group title. Oscillating transcripts were identified via eJTK (GammaBH < 0.05). Bold numbers are visualized in Figure 5B. (B and C) Expression of significantly oscillating transcripts that are system driven, liver clock driven or independent, and microbe driven or independent (B). Expression was normalized by median, and transcripts were ordered by time of maximum expression and phase; the key indicates which transcripts are depicted with yellow (C).

Figure 7. Liver clock and gut microbes uniquely impact functional pathway enrichment of oscillating hepatic transcripts.

Diurnal transcriptome analysis of liver samples collected every 4 hours over 24 hours from SPF and GF WT and LKO male mice (n = 3/time point/group) maintained in 12:12 LD (ZT2, -6, -10, -14, -18, and -22). Reactome and KEGG pathways significantly enriched by oscillating transcripts within each group. A subset of pathways enriched in SPF WT oscillating genes (q < 0.05). A lack of bar indicates lack of significance for that group/pathway (q > 0.05). Pathways marked with red star are addressed in the text.

Figure 8. Hepatic transcriptome coexpression over time is differentially affected by the liver clock and gut microbes.

Network transcriptome analysis of liver samples collected every 4 hours over 24 hours from SPF and GF WT and LKO male mice (n = 3/time point/group) maintained in 12:12 LD (ZT2, -6, -10, -14, -18, and -22). Network coexpression analysis of correlating transcripts over time within each group (P < 0.001). Network visualization and the number of correlating transcripts (nodes) and connections (edges) in each group. Red dots represent nodes, and gray lines represent edges.

Figure 9. Liver clock and gut microbes differentially alter diurnal patterns of energy expenditure and fuel utilization.

Indirect calorimetry assessment of SPF and GF WT and LKO male mice, measured over 4 consecutive 12:12 LD cycles (n = 12–13). (A) Energy expenditure (EE) represented as VO₂. (B) EE divided into active (dark) and rest (light) periods, summarized by EC₅₀ values within each period. (C) Respiratory exchange ratio (RER) represented as VCO₂/VO₂. (D) RER during active (dark) and rest (light) phases, summarized by EC₅₀ values. Data points represent mean ± SEM. Lines in box plots represent the median, and whiskers represent the minimum and maximum, respectively. ANCOVA was performed between 2 groups. *P < 0.05.

Figure 10. Liver clock and gut microbes partition glucose and lipid metabolism.

Model figure. In SPF conditions, the liver circadian clock drives normal GNG and FA β-Ox, fecal microbial abundance oscillations, and hepatic transcriptome oscillations. Following hepatic Bmal1 deletion, GNG and FA β-Ox are reduced, oscillating microbiota increase, and oscillating hepatic transcripts are not enriched for metabolic pathways, including GNG and FA β-Ox. In GF conditions, GNG is reduced, and oscillating hepatic transcripts are not enriched for GNG and FA β-Ox metabolic pathways regardless of liver clock functionality. Green indicates upregulation, and red indicates downregulation. Solid arrows indicate intact communication, and dashed arrows indicate broken communication. The figure created using BioRender.