Continuous exposure to new water-based stress disrupts daily activity and sleep patterns in mice

Male C3H-HeN mice purchased from Japan SLC Inc. (Hamamatsu, Japan) were housed under a 12-h light: 12-h dark cycle (lights on at 08:00) at a controlled ambient temperature of 24±1°C and provided with food (CE-2, CLEA Japan Inc., Tokyo, Japan) and water ad libitum throughout the study. Food intake during the study was determined by weighing the remains of a known daily excess of food at Zeitgeber time (ZT) 3. Chow spilled into the cages was not recovered under the conditions of perpetual avoidance of water on a wheel (PAWW) stress. New chow was exchanged for leftover food and we confirmed the absence of chow spilled in the cages every day to minimize the loss of spilled chow throughout the study. When we found a block of chow in the cage, we omitted the data from scoring. The mice were also weighed weekly when we measured the chow. Food intake is expressed as the average daily intake for 5 days per week (n = 6). The experiments were conducted in accordance with the guidelines for the Care and Use of Laboratory Animals at the National Institute of Advanced Industrial Science and Technology (AIST), and the Animal Care and Use Committee at AIST approved the study protocol (approval no. 2011 - 056b).

Mice (male, 8–15 weeks old) were maintained individually in plastic cages containing animal paper bedding and running wheels (SW-15; Melquest, Toyama, Japan) and they were provided with conventional food and water ad libitum for 1–2 weeks before exposure to stress and wheel activity recording. On the first day of exposure to stress, we exchanged the paper bedding with water to a depth of 1.5 cm at room temperature. The mice were averse to becoming wet and therefore chose to remain on the wheel almost all day and rarely entered the water under these conditions. The cages were placed under conventional light on ventilated shelves (6 cages on 1 shelf) and the water in the cages was exchanged 3 times weekly to keep the bedding clean. The mice accessed food from the wheels or directly from a stainless steel wire top feeder. Drinking water was accessible from the wheel. We named this model stressor PAWW. Activity was monitored even under exposure to the stressor. The lighting conditions were changed in some experiments from light/dark (L:D) to constant darkness (D:D). Wheel running activity in 1-min bins, which was monitored using a Chronobiology Kit (Stanford Software Systems, Santa Cruz, CA), is displayed as double-plot actograms, as described previously [16]. The free-running periods of individual mice were determined using a χ² periodogram. The wheel activity of each mouse was accumulated in 1-h bins for 7 days before, during, and after stress exposure, and then averaged to yield a single 24-h profile. Exposure to PAWW stress affected locomotor activity acutely on the first day, when all mice were active essentially throughout the entire day. Thus, we omitted the data collected on the first day of exposure to PAWW stress to create a 1-week stress profile (Figure 1D).

Core BT was continuously measured every 5 min as described previously [17] using an intraperitoneal transmitter (TA10TA-F20; Data Science International, St. Paul, MN) implanted under isoflurane anesthesia. Data from 24 h on the day before and the 6^th day of PAWW stress were averaged over 1-h intervals. Deviations in BT from the average temperature at each time point (every 5 min for 1 h) were measured for 6 h and are scored as the instability index of BT.

The liver and muscle were collected from each mouse, snap-frozen in liquid nitrogen, and stored at −80°C. Total RNA was extracted from the frozen tissues using RNAiso (Takara Bio Inc., Otsu, Japan) and cDNA was generated using PrimeScript RT Master Mix (Takara Bio Inc., Otsu, Japan). Quantitative real-time PCR proceeded using SYBR^® Premix Ex Taq™ II (Takara Bio Inc., Otsu, Japan) and a LightCycler™ (Roche Diagnostics, Mannheim, Germany). The PCR conditions comprised 10 s at 95°C followed by 45 cycles of 5 s at 95°C, 10 s at 58°C, and 10 s at 72°C. The endogenous quantity control for the liver was β-actin, and that for muscle cDNA was GAPDH, as described previously [18]. Each value was normalized to reference gene mRNA expression. The primer sequences were: Per2, 5′-GGCACATCTCGGGATCG-3′ and 5′-GAGCAGAGGTCCTCGCC-3′; Bmal1, 5′-GGAGAAGGTGGCCCAAA-3′ and 5′-AGGCGATGACCCTCTTA-3′; β-actin, 5′-CACACCTTCTACAATGAGCTGC-3′ and 5′-CATGATCTGGGTCATCTTTTCA-3′; and gapdh, 5′-GACCTCAACTACATGGTCTACA-3′ and 5′-ACTCCACGACATACTCAGCAC-3′.

The mice were sacrificed at 6- or 4-h intervals across the L:D cycle to obtain blood samples. Blood was collected by puncturing the inferior vena cava under inhalational anesthesia, and transferred into tubes coated with EDTA. Corticosterone was quantified in the first 100 µL of collected blood. Blood samples were separated by centrifugation at 4,500×g at 4°C for 10 min and the supernatant was stored at −80°C. The levels of plasma corticosterone (AssayPro, St. Charles, MO), melatonin, epinephrine, norepinephrine (IBL International, Hamburg, Germany), and leptin (Morinaga Institute of Biological Science Inc., Kanagawa, Japan) were determined by enzyme-linked immunosorbent assays according to the manufacturers' instructions.

Two electroencephalographic (EEG) electrodes were implanted into the skull and fixed with dental cement, and then 2 stainless steel wires were implanted in the neck muscles to collect electromyographic (EMG) signals. A telemetric device (TL11M2-F20-EET; Data Sciences International, St. Paul, MN) for sleep recording was then implanted subcutaneously into the backs of the mice according to the manufacturer's instructions under pentobarbital (50 mg/kg i.p.) anesthesia. After 10 days of recovery in plastic cages containing paper bedding and a wheel, polygraphic EEG, EMG, and subcutaneous BT were continuously recorded for 48 h during the pre-stress day and the first day of exposure to PAWW stress. We also collected EEG and EMG data for 24 h under L:D on the 7^th day of exposure to PAWW stress.

Cortical EEG and EMG signals were digitized at a sampling rate of 500 Hz and recorded using Dataquest A.R.T.^TM (Data Sciences International, St. Paul, MN). Polygraphic recordings were automatically scored offline in 10-s epochs into 3 stages of wake, rapid eye movement (REM) sleep, and non-rapid eye movement (NREM) sleep using SLEEPSIGN (Kissei Comtec, Nagano, Japan) according to the standard criteria described below. Wakefulness was characterized by low amplitude EEG signals with mixed frequency components and relatively high, often irregular, EMG activity. The amplitude of EEG activity during NREM sleep was high and dominated by 1–4-Hz slow-frequency waves, while EMG activity was low. Low-amplitude EEG signals dominated by 6–9-Hz theta waves and low EMG activity characterized REM sleep. Defined sleep-wake stages were examined visually and corrected if necessary. The EEG power during all NREM sleep epochs was calculated as the ratio of the slow wave or delta frequency range (1–4 Hz). To correct for interindividual differences in the strength of the EEG signals, the power values of all animals were normalized by expressing them relative to the average baseline power in the same frequency band.

All values are expressed as means ± SEM. Statistical analysis of the differences between groups in daily locomotor activity, body weight, food intake, BT, and wake/sleep length values were evaluated by two-way repeated measures (RM) analysis of variance (ANOVA). Univariate two-way ANOVA was used to evaluate the effect of PAWW stress on plasma leptin, corticosterone, and melatonin levels. When a statistical interaction was observed between factors, a comparison of all groups was performed using Tukey's post hoc test. Plasma corticosterone levels at ZT2 and catecholamine levels were evaluated using Student's or Welch's t-test. The effects of PAWW stress upon the circadian rhythmicity of plasma corticosterone levels and the expression of clock genes were analyzed by a cosinor method as described previously [19]. Comparisons of acrophase and amplitude were performed by Student's or Welch's t-test. The ratio of wake/REM sleep/NREM sleep duration between pre- and post-PAWW stress was evaluated using the G-square test. Statistical significance was established at P<0.05.

Locomotor activity was monitored as wheel running in cages containing paper bedding during a 12-h light/12-h dark cycle (L:D). The mice were more active under L:D during the night and less active during the day, which is normal for nocturnal mice. On the first day of exposure to PAWW stress, the bedding in each cage was replaced with water to a depth of 1.5 cm. Thereafter, the mice remained on the wheel throughout the day and accessed food and water from the wheel to avoid contact with the water. Figure 1A shows that exposure to PAWW stress significantly changed wheel-running behavior. Total locomotor activity was significantly enhanced on the first day of exposure to PAWW stress, when the mice ran on the wheels almost all day. However, total daily locomotor activity declined within 2 days of PAWW stress, but the timing of both the onset and offset of activity was imprecise and the active period continued until 6 h after lights on (Figure 1A). Furthermore, the mice often remained active even during the latter half of the daytime, when the control mice scarcely moved (Figure 1B).

Total daily activity was reduced gradually in the mice exposed to PAWW stress over a period of 3 weeks, but gradually increased in the control mice (F(1,20) = 11.01, P = 0.003, Figure 1C). However, this reduction in locomotor activity was remarkable even during the first week of exposure to stress and it was not expressed equally across all times of the day and night (F(1,10) = 4.01, P = 0.073, Figure 1D). The distribution of activity was biphasic, with a burst of activity that was coincident with lights on and a second more profound and longer bout of activity just after lights off, when a general active phase was evident in the mice before exposure to PAWW stress. Both pre-stress and stressed mice exhibited consolidated locomotor activity that started at lights off; however, this activity began to diverge between the pre-stress and stressed mice at 3 h into the dark period at ZT15. The wheel running activity of the stressed mice began to decline to half-maximal where it remained for 10 h, whereas that in the pre-stress mice was sustained at a high level for the first 6 h and declined gradually during the last 6 h of darkness during the same period (Figure 1D). Exposure to PAWW stress obviously reduced the amplitude of behavioral activity over 24 h compared with control conditions. Accordingly, >30% and <10% of the total daily activity occurred during the light period in the PAWW-stressed and control mice, respectively (F(1,21) = 66.77, P<0.001, Figure 1E). These effects persisted for up to 3 weeks without adaptation. Locomotor activity was similarly abnormal in almost all mice tested, but varied intra- and inter-individually during the daytime in the PAWW-stressed mice (Figure 1A).

Locomotor activity recovered when the mice were transferred from stressful to normal cages.–C shows that 3 of the 4 tested mice did not fully recover locomotor activity in the dark phase during the first week after exposure to PAWW stress. However, the other mouse immediately regained a similar locomotor profile to that before exposure to PAWW stress, namely rest during the light period and locomotion during the dark period (). Figure S1A Figure S1D

Chronic stress influences food intake and body weight [20]. The mice were weighed weekly from 1 week before until 3 weeks after starting exposure to PAWW stress. The average weekly rate of weight gain in the control mice was 2.6% (Figure 2A). In contrast, the mice lost a significant amount of weight during the first week of exposure to PAWW stress and then gained a little after the second week, but their weight gain was more moderate than in the control mice (F(1,21) = 7.01, P = 0.015, Figure 2A).

Food consumption increased notably within 2 weeks of PAWW stress and persisted over the time course of the experiment (F(1,21) = 21.98, P = 0.0001, Figure 2B). Accompanying this hyperphagia, daytime plasma leptin levels became significantly downregulated in the mice stressed for 1 week (Figure 2C, ZT6, P = 0.019 and ZT10, P = 0.046), while univariate two-way ANOVA test showed no significant difference between both groups (F(1,70) = 3.43, P = 0.068). Exposure to PAWW stress affected not only daily locomotor rhythmicity but also metabolic balance after 2 weeks. To minimize the metabolic and psychiatric effects of continuous stress on locomotor rhythmicity, we analyzed the homeostatic effects of abnormal locomotor rhythms caused by PAWW stress after 1 week, when the change in mouse locomotor activity was conspicuous.

The activity of the hypothalamic-pituitary-adrenal (HPA) axis is tightly controlled in a clock-dependent manner under normal conditions. The activity of the HPA axis is often acutely activated by exposure to conventional stress [9]. We measured plasma corticosterone levels in mice exposed to PAWW stress. Although corticosterone secretion increased acutely on the first day of PAWW stress (by 5.8- and 2-fold after 2 and 14 h of stress, respectively), the levels did not increase significantly after 1 week of continuous PAWW stress (F(1, 34) = 0.33, P = 0.56, Figure 3A). Cosinor analysis revealed that PAWW stress slightly reduced the amplitude of the corticosterone rhythm, but the levels and acrophase did not change significantly (Table S1). Corticosterone levels were significantly higher at ZT2 (P = 0.035).

The circadian rhythm of plasma melatonin is robust and closely associated with sleep homeostasis [21]. We found that continuous PAWW stress did not affect either plasma melatonin levels or its circadian rhythmicity (Figure 3B).

The enhanced locomotor activity in the daytime encourages speculation as to whether elevated sympathetic tone in the stressed animals contributes to sleep disturbance. We determined plasma epinephrine and norepinephrine levels because elevated sympathetic nerve activity often results in increased systemic levels of catecholamines. The levels of norepinephrine were significantly elevated (P = 0.031, Figure 3C), whereas those of epinephrine were only moderately and non-significantly increased in mice exposed to PAWW stress for 1 week (P = 0.072, Figure 3D).

Exposure to PAWW stress induced blunted circadian locomotor rhythms, decreased body weight, and increased food consumption. These findings suggest that exposure to continuous PAWW stress was closely linked to a negative energy balance in mice, meaning that caloric intake was lower than caloric expenditure, which reflects thermoregulation. Therefore, we compared the rhythms of core BT in the mice between the day before and after 6 days of PAWW stress to determine whether or not circadian thermoregulation was altered. The shapes of the pre-stress and stressed BT rhythms were significantly different. Nocturnal elevation was organized and core BT was reduced during the daytime in pre-stressed mice. Although the BT of the stressed mice also became nocturnally elevated, it was not maintained throughout the nighttime (Figure 4). In addition, the decrease in BT during the latter half of the dark period and at the start of the light period in the pre-stressed mice, which might induce falling asleep, disappeared in the continuously stressed mice. The BT of each mouse plotted for 10 consecutive days changed remarkably at each time point during continuous PAWW stress compared with the pre-stressed state (Figure S2A). Furthermore, the deviation of each BT in 5-min bins from a corresponding average of 1 h was higher in the stressed mice compared to the control mice (F(1,10) = 21.88, P = 0.001, Figure S2B). These findings suggest that PAWW stress destabilizes thermoregulation or that it dysregulates locomotion, which in turn induces thermogeneration.

Circadian sleep-wake rhythms were precise under baseline conditions, with more sleep during the light period than during the dark period, as reported previously [22]. However, sleep architecture and sleep-wake rhythms were remarkably affected in the stressed mice in a similar manner to their circadian locomotor activity. Seven days of exposure to PAWW stress obviously affected the daily sleep-wake cycles (Figure 5). Continuous stress increased wakefulness for 6 h after lights on (when mouse activity is usually masked and they fall asleep), whereas NREM sleep was significantly reduced (Figure 5A, B, and D). These findings correspond to the difficulties of falling asleep associated with insomnia. Conversely, the duration of wakefulness and NREM sleep was reduced in mice under continuous PAWW stress during the first half of the nighttime, when normal mice are fully awake and highly active (Figure 5A, B, and D). These results suggest that continuous stress induced sleepiness in the mice during the active phase, which is the typical phenotype of insomnia. The duration of REM sleep in the stressed mice was reduced during the light period and increased during the dark period, although the difference did not reach significance (Figure 5C and D).

The sleep/wake rhythms differed between days 1 (acute phase) and 7 (chronic phase) of exposure to PAWW stress.shows that acute stress significantly reduced the total duration of NREM and REM sleep, which was accompanied by an increase in wakefulness (P<0.05). Continuous stress exerted a treatment effect upon sleep-wake time, but significant differences were scant (). The total duration of wakefulness, NREM sleep, and REM sleep did not completely recover to pre-stress levels. Figure S3A Figure S3A

We compared hypnograms between the mice before and during continuous PAWW stress to further confirm whether continuous stress affects sleep architecture. Figure S3B shows that the continuously stressed mice underwent far more state transitions during both the light and dark periods than the pre-stressed mice. Representative hypnograms show that NREM sleep was interrupted by numerous awakenings during the daytime, and that bouts of activity during the nighttime were often interrupted by short periods of NREM sleep. Thus, the frequency of sleep-wake transitions increased during the day and the daily sleep-wake cycles resembled those of wheel-running locomotor activity (Figure 1A). As shown in Figure S3, the averaged EEG delta power during NREM sleep was reduced under continuous stress in 6 of the 9 tested mice compared with the pre-stress period to 87.65±3.8% for 24 h (normalized by the baseline period), whereas the levels were similar to the pre-stress state in the remaining 3 mice (107±3.5%). Averaged EEG in NREM sleep during L1 (68.8±10.1%) and D2 (63.2±3.4%) was more severely disrupted than during L2 (93.7%±12.5%) and D1 (87.4±21.1%) in the stressed mice. These data suggested that continuous stress affects not only the duration of sleep but also its quality.

Rhythmic activity was entrained to the L:D cycle in mice even under continuous PAWW stress (Figures 1 and 6). When the mice were kept in D:D under continuous PAWW stress, endogenous circadian rhythmicity was sustained with a tau value of 23.66±0.04 h, and activity onset was predicted by the phase of their entrained rhythms (Figure 6A). These results suggested that PAWW stress did not affect the circadian clock system of the central pacemaker oscillator that resides in the SCN. Furthermore, the hyperactivation induced by the stressor at circadian time 1–4 was also evident even under D:D. This indicated that locomotor activation during the daytime in the stressed mice was not a result of their enhanced susceptibility to light. When the mice were released from PAWW stress under D:D, circadian rhythmicity was sustained with a tau value of 23.62±0.11 h, whereas locomotor activity completely disappeared during the subjective day (Figure 6A).

Corresponding to the integrity of the circadian locomotor rhythm, circadian oscillations in the gene expression of Per2 and Bmal1 in peripheral tissues, including liver and muscle, were robust in both the stressed and control mice under L: D conditions with similar phases and amplitudes (Figure 6B and C, Table S1), except for Per2 in muscle, which peaked at approximately 2.6 h earlier in the stressed mice.