Prolonged Exposure to Social Stress Impairs Homeostatic Sleep Regulation

Abstract

Stress and sleep are tightly regulated as a result of the substantial overlap in neurotransmitter signaling and regulatory pathways between the neural centers that modulate mood and sleep-wake cycle. The chronicity of the stressor and variability in coping with it are major determinants of the psychiatric outcomes and subsequent effect on sleep. The regulation of sleep is mediated by the interaction of a homeostatic and a circadian process according to the two-process model. Chronic stress induces stress-related disorders which are associated with deficient sleep homeostasis. However, little is known about how chronic stress affects sleep homeostasis and whether the differences in adaptation to stress distinctively influence sleep. Therefore, we assessed sleep homeostasis in C57BL6/J mice following exposure to 15-d of chronic social defeat stress. We implemented wake:sleep ratio as a behavioral correlate of sleep pressure. Both stress-resilient and stress-susceptible mice displayed deficient sleep homeostasis in post-stress baseline sleep. This was due to poor temporal correlation between frontal slow wave activity (SWA) power and sleep pressure in the dark/active phase. Moreover, the buildup rate of sleep pressure in the dark was lower in susceptible mice in comparison to stress-naïve mice. Additionally, 4-h SD in the dark caused a deficient sleep recovery response in susceptible mice characterized by non-rapid eye movement (NREM) sleep loss. Our findings provide evidence of deficient homeostatic sleep process (S) in baseline sleep in stress-exposed mice, while impaired sleep recovery following a mild enforced wakefulness experienced during the dark was only detected in stress-susceptible mice. This alludes to the differential homeostatic adaptation to stress between susceptible and resilient mice and its effect on sleep regulation.

Keywords: NREM sleep; chronic social defeat; homeostatic sleep; process S; sleep pressure; slow wave amplitude.

FIGURE 1

Overview of the experimental design. (A) Chronic implantation of EEG electrodes: schematic illustration of the frontal and parietal EEG electrodes arrangement on the skull. (B) Timeline: mice were single-housed during the 7-d postoperative recovery period prior to CSD. After 15-d of CSD and a social interaction (SI) test, EEG and EMG baseline (BL) recordings were performed for 24-h after 48-h of habituation (Hab) in the sleep chambers. Post-CSD sleep homeostatic response, following 4-h of sleep deprivation (SD) starting at ZT14, was also acquired. (C) CSD paradigm: For 15 days, C57BL/6J mice were exposed to a daily 10-min physical aggression session with a novel retired CD1 breeder. Between sessions, mice were separated by a clear perforated plexiglass divider for the following 24-h allowing for psychological aggression. (D) SI test: CSD-exposed mice were classified into stress-resilient and susceptible phenotypes based on their SI score. (E) Tethered EEG and EMG recording: Mice were habituated to sleep chambers and connected to an EEG/EMG acquisition system that allowed for free movement in all three dimensions. Sleep deprivation (SD) was executed by sending electric pulses to a magnet placed under the platform which pushes it up and wakes up the mice. (F) There was a significant effect of ‘phenotype’ among SI scores using one-way ANOVA (F₂,₁₂ = 6.92, p = 0.01). The SI scores of the susceptible mice are lower than the scores of the stress-naïve mice (p < 0.01). There were no significant differences in (G) total travel distance or (H) time in corner zones between phenotypes during the SI test. n = 4–6. ^∗P < 0.05, ^∗∗P < 0.01.

FIGURE 2

Deficient sleep homeostasis during baseline sleep in stress-exposed mice. (A) Percent time in sleep-wake state in stress-naïve, resilient, and susceptible mice. (B) Sleep pressure quantified as wake:sleep ratio (top) and frontal and parietal baseline SWA power (0.5–4.5 Hz, bottom). SWA value was normalized to the 24-h baseline median value of SWA. In the dark, the sleep pressure across all phenotypes peaked by ZT20, while frontal SWA peaked at ZT20 in stress-naïve only (p = 0.006). (C) Pearson correlation between mean SWA power and mean sleep pressure (wake:sleep ratio). Dark: There was a positive correlation between mean frontal SWA power and mean sleep pressure in the stress-naïve mice (r₆ = 0.87, p = 0.026). Light: There was a trend of significant positive correlation between mean parietal SWA power and mean wake:sleep ratio in resilient and susceptible mice (r₆ = 0.81, p = 0.051 and r₆ = 0.79, p = 0.064 respectively). (D) Baseline SWA build-up rate. Each bar represents the coefficient of regression between log SWA (dependent variable) and time (independent variable) in the dark and light separately. Dark:There was a significant increase in the buildup of frontal and parietal SWA in stress-naïve mice in the dark (p < 0.001 and p = 0.002). The buildup of parietal SWA was significant in resilient (p = 0.006) and the buildup of frontal SWA was significant in susceptible mice (p = 0.04). Light: The parietal SWA power decayed in resilient mice (p < 0.001) while both frontal and parietal SWA power decayed in susceptible mice (p = 0.006 and p = 0.01 respectively). (E) The buildup rate of sleep pressure (wake:sleep ratio) from the beginning of the dark period (ZT12) to maximum wake:sleep ratio, reversal point at ZT20, was significant across all phenotypes (stress-naïve: p = 0.003, resilient: p = 0.003, susceptible: p = 0.007). The buildup rate of stress-naïve is significantly higher than the buildup rate of susceptible mice (p < 0.05). (F) Only stress-naïve mice displayed a significant buildup rate of frontal and parietal SWA within ZT12 to ZT20 (p = 0.001 and p = 0.005). Values are expressed as mean ± sem across 2-h intervals (A,B). n = 4–6 for each group. ^*,#P < 0.05, ^∗∗P < 0.01, and ^∗∗∗P < 0.001.

FIGURE 3

Frontal SWA power is negatively correlated with sleep fragmentation. (A) Quantification of sleep fragmentation based on previously published method (Karamihalev et al., 2019) displayed a trend of a phenotypic effect in the dark and an effect in the light (p = 0.0534 and p < 0.01 respectively). Dark: Trend showing greater fragmentation in susceptible mice versus resilient and stress-naïve mice (p = 0.07 and p = 0.08 respectively). Light: Sleep fragmentation in resilient mice was lower than in susceptible and stress-naïve mice (p < 0.01 and p < 0.05 respectively). (B) Correlation between SWA power and sleep fragmentation: There was a significant negative association between sleep fragmentation and frontal SWA in stress-naïve and susceptible mice in the dark (Stress-naive: r₃₆ = -0.51, p = 0.002; Susceptible: r₂₃ = -0.50, p = 0.02). Resilient mice exhibited significant negative association between sleep fragmentation and parietal SWA (r₃₀ = -0.38, p = 0.04) in the light. Values are expressed as mean ± sem across 2-h intervals of sleep fragmentation and log SWA power in the correlation analyses. n = 4–6 for each group. ^∗P < 0.05, ^∗∗P < 0.01.

FIGURE 4

Deficient recovery sleep response post SD applied in the dark in susceptible mice. (A) Recovery sleep-wake response: Percent time of sleep and wake in stress-naïve, resilient, and susceptible mice post SD. (B) Homeostatic SWA (0.5–4.5 Hz): Frontal and parietal SWA power of stress-naïve, resilient, and susceptible mice. Homeostatic SWA value was normalized to the 24-h pre-SD baseline median value of SWA. 4-h SD elicited mild non-significant increase in frontal SWA power in stress-naïve and resilient mice. Frontal and parietal SWA power were blunted and less variable in susceptible mice. (C) Cumulative difference in duration of vigilance states was calculated by subtracting the cumulative baseline duration from the cumulative recovery duration. Left: Change in cumulative duration of wake post SD. Susceptible mice spent a significantly greater amount of time in wake (one-sample t tests; ZT23: p = 0.044, ZT02 p = 0.030, ZT05: p = 0.057, ZT10: p = 0.038). Middle: Change in cumulative duration of NREM sleep post SD. Susceptible mice lost a significant amount of NREM sleep (one-sample t tests; ZT23 and ZT01–ZT10: p < 0.05; ZT11: p < 0.01). Right: No change in cumulative duration of REM post SD in stress-exposed. However, there was a negative change in the cumulative duration of REM in stress-naïve mice post SD (one sample t-tests, ZT04–ZT07: 0.021 < p-values < 0.0345; ZT08: p = 0.0042, ZT10: p = 0.0137, ZT11:p = 0.0336). (D) Percent change in cumulative slow wave energy (SWE) post SD in the frontal (left) and parietal (right) leads. Two-way repeated measures ANOVA in the light and in the dark showed no difference in the intensity of the sleep in either frontal or parietal SWE between the three phenotypes. Values are expressed as mean ± sem across 2-h intervals (A,B) and 1-h intervals (C,D). n = 4–6 for each group. ^∗P < 0.05, ^∗∗P < 0.01.