Resilience to Chronic Stress Is Characterized by Circadian Brain-Liver Coordination

Abstract

Background: Chronic stress has a profound impact on circadian regulation of physiology. In turn, disruption of circadian rhythms increases the risk of developing both psychiatric and metabolic disorders. To explore the role of chronic stress in modulating the links between neural and metabolic rhythms, we characterized the circadian transcriptional regulation across different brain regions and the liver as well as serum metabolomics in mice exposed to chronic social defeat stress, a validated model for studying depressive-like behaviors.

Methods: Male C57BL/6J mice underwent chronic social defeat stress, and subsequent social interaction screening identified distinct behavioral phenotypes associated with stress resilience and susceptibility. Stressed mice and their control littermates were sacrificed every 4 hours over the circadian cycle for comprehensive analyses of the circadian transcriptome in the hypothalamus, hippocampus, prefrontal cortex, and liver together with assessments of the circadian circulatory metabolome.

Results: Our data demonstrate that stress adaptation was characterized by reprogramming of the brain as well as the hepatic circadian transcriptome. Stress resiliency was associated with an increase in cyclic transcription in the hypothalamus, hippocampus, and liver. Furthermore, cross-tissue analyses revealed that resilient mice had enhanced transcriptional coordination of circadian pathways between the brain and liver. Conversely, susceptibility to social stress resulted in a loss of cross-tissue coordination. Circadian serum metabolomic profiles corroborated the transcriptome data, highlighting that stress-resilient mice gained circadian rhythmicity of circulating metabolites, including bile acids and sphingomyelins.

Conclusions: This study reveals that resilience to stress is characterized by enhanced metabolic rhythms and circadian brain-liver transcriptional coordination.

Keywords: Chronic social stress; Circadian rhythms; Gene expression; Metabolism; Mouse model for depressive-like behaviors; Multitissue omics analysis.

Figure 1

Phenotyping of mice following chronic social defeat stress. (A) Schematic overview of the experimental design and the timeline. Days 1–10, chronic social defeat test; day 11, social interaction test; and day 12, sacrifice and tissue collection over the circadian cycle for RNA sequencing and metabolomics. (B) Social interaction test score showing the ratio in time spent in the CZ and IZ in the presence and absence of the aggressor. n = 30 in the control group, and n = 36 in the resilient and susceptible groups. (C) Body weight (g) on day 0 and day 10. n = 14 controls, n = 33 resilient, and n = 23 susceptible. (D) Average food intake (g) per day. n = 14 controls, n = 17 resilient, and n = 13 susceptible. (E) Average food intake (g) during light and dark periods. n = 10 per group. (F) A 10-day time course of food intake (g). n = 14 controls, n = 17 resilient, and n = 13 susceptible. Two-way analysis of variance followed by Tukey’s post hoc test. Data are presented as mean ± SEM. ∗p < .05, ∗∗p < .01, and ∗∗∗∗p < .0001. For (F), #p < .05 and ##p < .01 for the resilient vs. control group comparison. Control (gray), resilient (purple), and susceptible (blue) mice. CZ, corner zone; DP, dark period; IZ, interaction zone; LP, light period; ZT, zeitgeber time.

Figure 2

The circadian transcriptome of the brain is rewired by chronic social defeat stress. (A) Venn diagrams showing the overlap of rhythmic genes between control, resilient, and susceptible mice in the HT, HC, and PC. (B) Scatter plots representing the distribution of exclusively rhythmic genes from each group in the 3 brain regions together with the respective marginal histograms. The y-axis and x-axis represent the amplitude and phase, respectively. (C) Rhythmic expression of core clock genes in the brain. The y-axis represents the log₂-normalized counts, and the x-axis represents the time (zeitgeber time). Fitted lines represent significant rhythmicity, while flat lines represent non-rhythmicity. (D) Heatmaps of normalized rhythmic messenger RNA levels in the 3 brain regions. The box to the right of the heatmaps represent all models in the DryR analyses outputs. Model 1, nonrhythmic in all; model 2, rhythmic only in control; model 3, rhythmic only in resilient; model 4, rhythmic only in susceptible; models 5 and 6, rhythmic in control and resilient but not in susceptible; models 7 and 8, rhythmic in control and susceptible but not in resilient; models 9 and 10, rhythmic in resilient and susceptible but not in control; models 11–15, rhythmic in all. Higher and lower amplitudes are indicated by yellow and blue color, respectively. Flat lines represent no rhythms, and black waves represent significant circadian rhythmicity. White waves represent antiphasic rhythms in relation to the other group with rhythmic transcripts within the same model. Model 15 represents different rhythmic patterns between all 3 groups, hence the inclusion of a red wave line. (E) Circos plots showing the overlap and interconnection of rhythmic genes from each brain region in control, resilient, and susceptible mice. The outer layer of the circle represents the gene list of PC (purple), HC (green), and HT (yellow). The dark orange in the inner side of the circle represents the genes that are shared, and the light orange color shows the genes that are unique to certain brain regions. The purple lines connecting the genes show the same genes shared between the gene lists, and the blue lines link the genes that belong to the same ontology. (F) Stacked bar chart representing the cell type–specific hypothalamic rhythmic transcripts within each model. The cell-type specificity was extracted from previously published single-cell RNA sequencing data of mouse hypothalamus. n = 3 mice per time point and group. GABAergic, gamma-aminobutyric acidergic; HC, hippocampus; HT, hypothalamus; PC, prefrontal cortex.

Figure 3

Social stress impacts the liver circadian transcriptome. (A) Venn diagram showing the overlap of rhythmic genes between control, resilient, and susceptible mice in the liver. (B) Rhythmic expression of core clock genes in the liver. The y-axis represents the log₂-normalized counts, and the x-axis represents the time. (C) Scatter plot representing the distribution of exclusively rhythmic genes from each group in the liver together with the respective marginal histograms. The y-axis represents the amplitude, and the x-axis represents the phase. (D) Heatmaps of normalized rhythmic messenger RNA levels in the liver in control, resilient, and susceptible groups from model 11, rhythmic in all; model 2, rhythmic only in control; model 3, rhythmic only in resilient; model 4, rhythmic only in susceptible; model 5, rhythmic in control and resilient but not in susceptible; and model 9, rhythmic in resilient and susceptible but not in control. Higher and lower amplitudes are indicated by yellow and blue, respectively. Explanation for the models is visualized in Figure 2D. (E–G) Tree plot presenting a hierarchical clustering of enriched terms from input gene lists from model 9 [(E) resilient- and susceptible specific], 5 [(F) control- and resilient specific], and model 3 [(G) resilient specific] together with the expression patterns of selected genes belonging to chosen pathways from each model. The background colors indicate a different hierarchical group of enriched pathways. The bubble size shows the enrichment size, and the color of the bubble shows the significance (adjusted p value cutoff < .05). n = 3 mice per time point and group. ECM, extracellular matrix; ncRNA, noncoding RNA; NF, nuclear factor.

Figure 4

Brain-liver coordination of circadian transcriptional pathways. (A–C) Circos plots representing the overlap and interconnection of rhythmic genes between the liver and the different brain regions in (A) model 2, rhythmic only in control; (B) model 3, rhythmic only in resilient; and (C) model 4, rhythmic only in susceptible. The outer layer of the circle represents the gene list from each tissue: liver (red), PC (purple), HC (green), and HT (yellow). The dark orange in the inner side of the circle represents the genes that are shared between the tissues, and the light orange color shows the genes that are unique in each tissue. The purple lines connecting the genes show identical genes shared between the tissues, and the blue lines link the genes that belong to the same ontology. n = 3 mice per time point and group. HC, hippocampus; HT, hypothalamus; PC, prefrontal cortex.

Figure 5

Resilience to social stress is characterized by enhanced rhythmicity of circulatory metabolites. (A) Venn diagram showing the overlap of rhythmic metabolites between the control, resilient, and susceptible mice in the serum. (B, C) Chord plot showing the (B) number and (C) percentage of rhythmic metabolites belonging to super pathways in control, resilient, and susceptible groups. (D) Heatmaps of log-normalized metabolite abundancy in the serum of the control, resilient, and susceptible mice from model 11, rhythmic in all; model 2, rhythmic only in control; model 3, rhythmic only in resilient; model 4, rhythmic only in susceptible; and model 5, rhythmic in control and resilient but not in susceptible. Explanation for the models is visualized in Figure 2D. Higher and lower amplitudes are indicated by orange and blue, respectively. Row annotations show the super pathways of each metabolite. (E) Scatter plots representing the distribution of metabolites from model 11 (rhythmic in all), model 2 (control specific), model 3 (resilient specific), and model 4 (susceptible specific) with the respective marginal histograms. The y-axis represents the amplitude, and the x-axis represents the phase. n = 4–5 mice per time point and group.

Figure 6

Circadian metabolic pathways in response to chronic social defeat stress. (A–F) Relative levels of metabolites in (A) disaccharide and oligosaccharide metabolism, (B) sphingomyelin metabolism, (C) corticosterone metabolism, (D) sterol metabolism, (E) primary bile acids, and (F) secondary bile acids in the serum collected from the control, RES, and SUS mice. The y-axis represents the log-transformed values, and the x-axis represents the time. n = 4–5 mice per time point and group. RES, resilient; SUC, susceptible.