Tuning of liver circadian transcriptome rhythms by thyroid hormone state in male mice

Abstract

Thyroid hormones (THs) are important regulators of systemic energy metabolism. In the liver, they stimulate lipid and cholesterol turnover and increase systemic energy bioavailability. It is still unknown how the TH state interacts with the circadian clock, another important regulator of energy metabolism. We addressed this question using a mouse model of hypothyroidism and performed circadian analyses. Low TH levels decreased locomotor activity, food intake, and body temperature mostly in the active phase. Concurrently, liver transcriptome profiling showed only subtle effects compared to elevated TH conditions. Comparative circadian transcriptome profiling revealed alterations in mesor, amplitude, and phase of transcript levels in the livers of low-TH mice. Genes associated with cholesterol uptake, biosynthesis, and bile acid secretion showed reduced mesor. Increased and decreased cholesterol levels in the serum and liver were identified, respectively. Combining data from low- and high-TH conditions allowed the identification of 516 genes with mesor changes as molecular markers of the liver TH state. We explored these genes and created an expression panel that assesses liver TH state in a time-of-day dependent manner. Our findings suggest that the liver has a low TH action under physiological conditions. Circadian profiling reveals genes as potential markers of liver TH state.

Figure 1

Dose-dependent effects of thyroid hormone (TH) state on systemic energy metabolism. (A–F) Serum levels of T₃ and T₄, body weight across the experiment, 24-h profiles of locomotor activity, body temperature, and O₂ consumption are shown. Rhythmicity was assessed using CircaCompare algorithm. Presence (R) or absence (NR) of significant circadian rhythmicity is depicted. In the presence of significant 24-h rhythmicity, a sine curve was fit to the data. (G–J) 24 h average data for locomotor activity, food intake, body temperature, and O₂ consumption for MMI, CON, and T₃ groups. One-way ANOVA was performed (p value is shown) followed by Tukey’s post-test comparisons (depicted with asterisks). In (A) and (B), n = 4–6 animals per group or timepoint. In (C), n = 24. In (D), n = 4 and 5 for CON and MMI groups, respectively. In (E) and (F), n = 4 for each group. *, **, ***, **** represent a p value of < 0.05, 0.01, 0.001, and 0.0001, respectively.

Figure 2

Lowering thyroid hormone state has subtle effects on liver transcriptome rhythms. (A) Global DEG analysis (disregarding sampling time) is represented as a Venn diagram. (B) UpSet plots represent DEG analysis for each ZT separately. (C) Venn diagram represents all temporal DEGs (i.e., showing different expression levels of at least one ZT) identified in MMI and T₃ groups versus CON mice. (D) Selected examples of robust DEGs (37 in total) previously identified in T₃-treated mice. Absolute fold change comparison of all 37 DEGs in T₃ and MMI mice are shown. Absolute fold changes were used as some genes were up- or downregulated across the groups. (E) Selective TH output genes and fold change of these genes. Comparisons were performed using two-way ANOVA (main treatment effect, p < 0.05). n = 3–4 for all ZTs and groups. Pair-wise comparisons were performed by Student’s t test with Welch correction. Presence (R) or absence (NR) of circadian rhythm by JTK cycle (p value < 0.05). ***, **** represents a p value of < 0.001, and 0.0001, respectively.

Figure 3

Lowering thyroid hormone state does not affect the rhythmic expression of core clock genes and has a slight phase effect on robustly rhythmic genes. (A) Diurnal expression profile of core clock genes is shown. Presence (R) or absence (NR) of significant circadian rhythm by JTK cycle (p value < 0.05) is depicted. (B) Rose plot and heatmap of robustly rhythmic genes are shown. A minor phase delay of 0.24 h was identified in the MMI group compared to CON (test against zero, p < 0.001). n = 3–4 for all ZTs and groups.

Figure 4

Differential rhythm analysis reveals changes in liver transcriptome rhythms that affect lipid and cholesterol metabolism in MMI mice. (A) Differential rhythm analysis was performed using CircaCompare and is represented as Venn diagrams. (B) UpSet plots show alterations in diurnal rhythm parameters (mesor, amplitude, phase). (C) Gene set enrichment analysis (GSEA) of the genes with either mesor, amplitude, or phase alterations was performed. Top-5 biological processes for each category are shown. (D) In-depth diurnal lipid metabolism analysis in response to low thyroid hormone state. Heatmaps show genes with mesor changes. Volcano plot and rose plot show alterations in amplitude and phase, respectively. (E) Normalized gene expression of selected genes participating in cholesterol uptake, biosynthesis, and degradation (bile acid secretion) in CON and MMI mice. (F) Quantification of cholesterol and TAG in serum or liver. Presence (R) or absence (NR) of significant circadian rhythm by CircaCompare (p value < 0.05) is depicted. n = 3–4 for all ZTs and groups.

Figure 5

Identification of thyroid hormone-responsive genes by mesor comparison. (A) Heatmap shows the 516 TH-dose responsive genes with differential mesor expression among each group. (B) Gene set enrichment analysis (GSEA) of mesor-altered genes is shown for each condition. (C) Genes pertaining to similar biological pathways identified in B were normalized by CON mesor and plotted. FA biosynthesis pathway comprises lipid metabolic process, unsaturated fatty acid biosynthetic process, long-chain fatty-acyl-CoA biosynthetic process, diacylglycerol biosynthetic process, negative regulation of fatty acid biosynthetic process, and lipid storage. FA catabolism pathway is comprised of acyl-CoA metabolic process, lipid metabolic process, fatty acid metabolic process, very-low-density lipoprotein particle assembly processes. Cholesterol and bile acid metabolism pathways comprise cholesterol homeostasis, cholesterol metabolic process, steroid metabolism, and bile acid signaling pathway. Carbohydrate metabolism pathways are comprised of carbohydrate metabolic process, glycogen metabolic process, ATP metabolic process, and glucose homeostasis. (D) Selected biomarkers for TH state at all time points (additional genes can be found in Supplementary file 5). Presence (R) or absence (NR) of circadian rhythm by CircaCompare (p value < 0.05) is depicted. n = 3–4 for all ZTs and groups.