Impaired function of the suprachiasmatic nucleus rescues the loss of body temperature homeostasis caused by time-restricted feeding

Abstract

The suprachiasmatic nucleus (SCN) is the master circadian pacemaker that drives body temperature rhythm. Time-restricted feeding (TRF) has potential as a preventative or therapeutic approach against many diseases. The potential side effects of TRF remain unknown. Here we show that a 4-hour TRF stimulus in mice can severely impair body temperature homeostasis and can result in lethality. Nearly half of the mice died at 21 °C, and all mice died at 18 °C during 4-hour TRF. Moreover, this effect was modulated by the circadian clock and was associated with severe hypothermia due to loss of body temperature homeostasis, which is different from “torpor”, an adaptive response under food deprivation. Disrupting the circadian clock by the SCN lesions or a non-invasive method (constant light) which disrupts circadian clock rescued lethality during TRF. Analysis of circadian gene expression in the dorsomedial hypothalamus (DMH) demonstrated that TRF reprograms rhythmic transcriptome in DMH and suppresses expression of genes, such as Ccr5 and Calcrl, which are involved in thermoregulation. We demonstrate a side effect of 4-hour TRF on the homeostasis of body temperature and a rescue function by impairing the SCN function. Altogether, our results suggested that constructing a circadian arrhythmicity may have a beneficial effect on the host response to an acute stress.

1. Introduction

Circadian clocks regulate a large number of physiological processes and drive widespread rhythmic gene expression in mammals. At the molecular level, circadian autoregulatory negative feedback loops drive endogenous rhythmic gene expression in almost all cells ^[1] Both light input and non-photic cues (such as restricted feeding) can also act as zeitgebers to reset and reprogram circadian rhythms of gene expression in peripheral tissues ^[2–6] The suprachiasmatic nucleus (SCN) is the master circadian pacemaker in mammals; it coordinates and synchronizes slave oscillators in peripheral tissues to drive rhythmic events such as physical activity, sleep, immune functions and body temperature cycles ^[7–11]

Time-restricted feeding (TRF), which restricts food consumption to certain hours of the day, has been shown to have health benefits ^[12–20]. TRF therefore has been proposed to be a potential non-pharmacological intervention against obesity and metabolic associated disease. However, the potential side effects of different TRF schedules, if any ^[21], have not been carefully investigated.

In most mammals, body temperature oscillates daily within a narrow range that is governed by both homeostatic and circadian regulation despite wide variations in environmental conditions ^{[22, 23]}. The maintenance of body temperature (Tb) homeostasis is critical for animal survival. In humans and rodents, the hypothalamus acts as the temperature control center and works with other body temperature regulating systems, such as the muscle, skin, brown adipose tissue (BAT), liver, and blood vessels, to maintain the body temperature homeostasis so that body temperature is close to 37 °C ^{[23, 24]}. Within the hypothalamus, the dorsomedial hypothalamus (DMH) is an important region that regulates body temperatures ^{[23, 25, 26]}. The DMH is also critical for mouse behavior and the expression of food-entrainable circadian rhythms ^{[27, 28]}. On the other hand, the SCN projects directly or indirectly into the DMH and other hypothalamus regions to allow rhythmic control of physiological processes and behavioral. Ablation of the SCN abolishes behavior and body temperature rhythms ^[29–31].

Torpor, a highly controlled drop in body temperature (Tb) and metabolic rate, is an adaptive response that protects animals against shortage of food ^[32–34]. It was previously shown that the SCN controls the timing of torpor ^[35–37] and animals with ablation of the SCN are resistant to torpor ^[30]. Later, they found that the SCN is not necessary for the expression of torpor but critical for the temporal organization of torpor probably through metabolic or neuroendocrine process which is independent of its role as circadian pacemaker ^[38]. For cold and fasting stimuli, both AVP and VIP neurons in the SCN were activated, suggesting that the SCN may participate in the suppression of cold defense ^[39]. In contrast to torpor, pathogenic hypothermia is an uncontrolled state of low Tb induced by cool ambient temperature and food deprivation ^[40].

Here, we discovered that 4-hour TRF resulted in mouse lethality associated with loss of body temperature homeostasis and was different from torpor. Circadian clock gates the TRF response and the SCN lesion rescues the lethality under TRF. These results established an effect of TRF on body temperature homeostasis and the inhibitory function of the SCN in the homeostasis of body temperature control.

2. Materials and methods

2.1. Animals

C57BL/6J mice (male) were housed in specific pathogen-free animal facilities. All animal procedures were approved by the Animal Care and Use Committee of the CAM-SU Genomic Resource Center, Soochow University. Mice were fed a normal chow diet (ShooBree SPF Mice Diet, 28% protein, 13% fat, 57% carbohydrates). Each cage was covered with wood shavings for a tiny nest. The TRF experiments were repeated in at least 12 mice from each group for the survival analysis, and representative data sets are presented, animals were randomly allocated to experimental groups did not differ significantly in body weight. The incubator temperature controller is built based on high-precision temperature control module within 1 °C variation. Each layer of the incubator is tested by a mercury-in-glass thermometer to ensure that the temperature of each layer is the same. The Cry1^−/−/Cry2^−/− mice were generated as previously reported ^{[41, 42]} and only 6 or 7 mice subjected to TRF experiment because of their impaired reproductive, The Per1^−/−/Per2^−/− mice were generated as previously reported ^[43]. The investigators were not blinded to the group allocation during the experiments. All applicable institutional and/or national guidelines for the care and use of animals were followed.

2.2. Feeding schedule, food intake, and locomotor activity analyses

Six- to eight-week-old male or 18- to 26-week-old male C57BL/6J mice were individually housed in cages equipped with running wheels for 2 weeks to adapt to the incubator (PGX −350B, Ningbo Saifu) housing conditions under light dark cycle, with unrestricted access to food and water. The mice were maintained under a 12 h light: 12 h dark schedule (light intensity 200 lux), except when noted. After two weeks, eight- to ten-week old mice were either fed ad libitum or subjected to TRF for two weeks. Mice from each experimental group subjected to TRF were given access to food for 4 h (ZT4-8, ZT10-14, ZT16-20, or ZT22-2, where ZTO = lights on and ZT12 = lights off). Food access was regulated by transferring the mice from cages with food and water to corresponding cages with water only. In the control group, the mice fed ad libitum were transferred between cages at the same time. Daily food intake was measured by monitoring the weight of the remaining food. Wheel running was recorded and analyzed using ClockLab (Actimetrics, Evanston, IL). The percentage distributions of wheel-running activity in light phase vs. in the night phase were quantified. Data of food intake in Fig. 2e only including the survived mice, and the curve is similar to previous reports.

Fig. 2.

The circadian clock affects the survival of mice subjected to time-restricted feeding, (a) Kaplan-Meier survival curves of mice subjected to TRF at the indicated time points at 21 °C; ZT10-14 vs. ZT22-2 (p= 0.0084), ZT10-14 vs. ZT4-8 (p=0.0006), ZT10-14 vs. ZT16-20 (p=0.0001), ZT22-2 vs. ZT4-8 (p=0.4688), ZT22-2 vs. ZT 16-20 (p=0.3281). (b) Representative double-plotted locomotor actograms shown for mice first fed ad libitum for two weeks and then subjected to TRF for two weeks at the indicated time points at 21 °C. The feeding time is indicated in the red-filled box and the top of actograms. Periods of darkness and light are indicated the black and white bars, respectively, (c, d) Kaplan-Meier survival curves of Per1/2 double knockout mice (c) and Cry1/2 double knockout mice (d) subjected to TRF treatment at ZT10-14 and ZT16-20, p values are indicated in the graphs, (e) Food intake of mice fed ad libitum was recorded for 7 days before the start of the indicated TRF treatment, which lasted for 12 days. Data of food intake only including the survival mice, and the curve is similar to previous reports. Prism two-way ANOVA with Bonferroni’s post hoc test was used to determine the significant interactions of food intake among the different groups: ZT10-14 vs. ZT16-20 at 21 °C (p=0.268), 25 °C vs. 21 °C (p=0.5796) or 19.5 °C (p=0.8447), 19.5 °C vs. 21 °C (p=0.7471) at ZT16-20. For the food intake of ad libitum fed period, 25 °C vs. 19.5 °C or 21 °C (p< 0.001), the first five days food intake of the TRF treatment between ZT10-14 and ZT16-20 at 21 °C (dayl, p<0.0001, day2, p=0.0277, day3, p=0.0015, day4, p=0.0377, day5, p=0.0486).

2.3. Core body temperature monitoring

The core body temperature was measured using implantable telemetry (F20-EET, Data Sciences International, New Brighton, MN). After the mice from each group were exposed to their corresponding experimental LD cycles for 1 week, a telemeter was inserted into the abdominal cavity of mice under sterile surgical conditions with isoflurane anesthesia. Following 10 days of convalescence and adaptation to the incubator, the core body temperatures of the mice fed ad libitum for 4 days and then maintained under the TRF scheme were continuously recorded by a receiver board (Data Sciences International, New Brighton, MN) beneath the cage. The data were recorded every minute and plotted as daily average temperatures at 1 h intervals for at least 3 mice per group.

2.4. DMH collection, RNA extraction, RNA-seq, Q-PCR, and mRNA expression analyses

The DMH were separated from C57BL/6J males on the second day of TRF at ZT0, 4, 8, 12, 16, and 20. Two mice were examined at each time point. RNA was extracted using TRIzol Reagent (Invitrogen, Carlsbad, CA, USA). For RNA-seq, the quality of the total RNA was determined using an Agilent 2100 Bioanalyzer, and RNA-seq was performed on an Illumina MiSeq platform with PE 150-bp reads at the BGI Genome Center, Shenzhen, China. We estimated the expression levels for all NCBI-annotated genes from different samples using RSEM1.2, and expression values were normalized by FPKM. For the quantitative real-time PCR (Q-PCR), the RNA concentration was determined by a NanoDrop, and cDNA was synthesized using a PrimeScript RT Reagent Kit (Takara, Takara, Otsu, Shiga, Japan). Q-PCR was performed using SYBR Premix Ex Taq with StepOne Plus (Applied Biosystems, Carlsbad, CA, USA). The relative levels of each transcript were normalized to GAPDH. The primer sequences are provided in Table S14 online.

2.5. Identification and functional enrichment analysis of rhythmic genes and differentially expressed genes

To identify genes with a daily rhythmic expression from the gene expression data in different experimental groups, the Jonckheere-Terpstra-Kendall (JTK) algorithm was used. For all array sets, a permutation-based p-value (ADJ.P) of less than or equal to 0.05 was considered significant. The expression levels for all rhythmic genes at each time point were calculated as the average expression of the biological replicates. The optimal phase (LAG), amplitude (AMP), and period (PER) estimates for each transcript of oscillating genes were extracted from the JTK algorithm. To identify differentially expressed genes, the R statistical package software DEseq2 was used. In addition, functional enrichment analysis with DAVID (https://david.ncifcrf.gov/home.jsp) was performed to significantly enriched GO terms and biological process at P values <0.01 compared with the whole-transcriptome background.

2.6. SCN lesions

Bilateral SCN lesions were performed stereotaxically in 2-month-old C57BL/6J mice under isoflurane anesthesia. A small hole was created in the skull using a dental drill bur (0.05 mm posterior to bregma and 0.1 mm lateral from the midline). A platinum-iridium alloy electrode (0.15-mm diameter; Kedou Brain-Computer Technology Co., Ltd.) coated entirely with polyimide except for the tip (0.2 mm in length) was inserted bilaterally into the SCN (5.85 mm depth from the surface of the skull). A direct electrical current of 0.4 mA for 40 s was given with a Ugo Basile lesion maker (model 53500). The electrode was left for 1 minute before being withdrawn from the brain. After the SCN lesion was induced, wheel-running activities were recorded under standard LD cycles for at least 1 week to confirm the loss of circadian rhythms. The sham operated mice underwent the same operation, but the electrode was inserted to a 5.4 mm depth from the skull surface, and no current was passed through the electrode.

2.7. Immunofluorescence

Mice were euthanized by CO2 asphyxiation and immediately perfused with saline, followed by 4% paraformaldehyde. The brains were dissected and post-fixed in 4% paraformaldehyde for 12 h. Coronal brain sections (30 μm-thick) were obtained using a Leica VT1000S Vibratome. The tissue sections were incubated in Coplin jars filled with 1 mM EDTA (pH 8.0) in a 95–99 9 water baths for 5 min, cooled at room temperature (RT) for 60 min, and washed three times with lx PBS for 5 min before blocking in 0.3% PBST for 30 min. The sections were incubated with anti c-Fos (Cell Signaling Tech, 9F6, 1:5000) antibodies overnight at 4 (, followed by incubation at room temperature for 2 h with the appropriate secondary antibodies Cy3 goat anti-rabbit IgG at 1:300 dilution (Jackson ImmunoResearch, 111-165-003). The cell nuclei were stained with DAPI (VECTASHIELD, H-1200, 1:5000) before the slides were washed with lx PBS for the third time for 10 min. Subsequently, the sections were mounted on slides with 50% glycerol and cover-slipped. The analysis was performed under an Olympus FluoView 1000 confocal microscope.

2.8. Statistics

The statistical analyses of food intake, locomotor activity, and body temperature were performed using Student’s t-tests or two-way ANOVAs using GraphPad Prism (GraphPad Software Inc., San Diego, CA, USA). For the survival analyses, Kaplan-Meier plots were drawn, and the significant differences were evaluated using a log-rank (Mantel-Cox) test. A P < 0.05 was considered statistically significant (*P < 0.05, ** P < 0.01, and *** P < 0.001).

3. Results

3.1. TRF causes lethality in mice near room temperature

The effect of ambient temperature has not been carefully examined in studies of TRF. To determine the effect of ambient temperature on mice subjected to TRF, C57BL/6J mice (6 – 8 weeks of age) were housed individually in cages in temperature-controlled incubators under standard light/dark (LD) cycles (12 h of light/12 h of darkness) at 25 °C for 2 weeks for adaptation in the incubator environment, and fed ad libitum. Then groups of mice (male, 8 – 10 weeks of age, average 23 – 25g) were housed individually at 18, 19.5, 21, or 25 °C for another two weeks of adaptation. Afterwards, these mice (10 – 12 weeks of age) were subjected to TRF treatment at different time windows (Fig. 1a). When mice were fed only during the day at Zeitgeber time 4-8 (ZT4-8), all mice survived at 25 °C. Surprisingly, all mice housed at 18 °C died within a few days of TRF, and only about 25% and 45% of the mice survived at 19.5 °C and 21 °C, respectively (Fig. 1b). Similar results were also observed when feeding was allowed during ZT16-20 (Fig. 1c). There was no significant difference in survival between ZT4-8 and ZT16-20 at 19.5 °C and 21 °C (p=0.14 and p=0.88) but mice fed at ZT4-8 had a faster rate of mortality than those fed at ZT16-20 at 18 °C (p= 0.0023) (Fig. 1b, c). These results demonstrate that TRF can have a major effect on mouse survival at near room temperature and its effect is dependent on ambient temperature (18, 19.5, 21 and 25 °C).

Fig. 1.

Survival of mice is reduced under TRF at temperatures only slightly below typical room temperatures, (a) A diagram showing the experimental design for TRF treatment. AL: ad libitum and thereafter, (b , c) Kaplan-Meier survival curves of mice fed at ZT4-8 (b) and ZT16-20 (c) at indicated ambient temperatures. Number of mice per group (n) is given in each graph. A log-rank test showed significant differences between different groups of mice: at ZT4-8: 25 °C vs. 21 °C (p=0.0019), 25 °C vs. 19.5 °C (p=0.0002), 21 °C vs. 19.5 °C (p=0.1014), 19.5 °C vs. 18 °C (p=0.0182); at ZT16-20: 25 °C vs. 21 °C (p=0.0006), 21 °C vs. 19.5 °C (p=0.437), 21 °C vs. 18 °C (p<0.0001), 19.5 °C vs. 18 °C (p=0.0015); and ZT4-8 vs. ZT16-20 at 18 °C (p=0.002). No significant differences between ZT4-8 and ZT 16-20 were observed at 19.5 °C and 21 °C. (d) Kaplan-Meier survival curves of mice fed at ZT 16-20 at indicated conditions. Number of mice per group (n) was indicated. A log-rank test showed significant differences between different groups of mice: group-housed mice (black curve) vs. individually-housed mice (green curve) at 18 °C (p<0.0001). The purple curve indicates the group that was subjected to TRF at 25 °C for 2 days before switching to 18 °C, this group vs. the 18 °C group (green curve) (p=0.0031). The experiments were ended at day 15.

To determine whether low temperature alone can result in mouse lethality, we individually housed mice at 4 °C and fed ad libitum. We found that all mice survived well (12/12) under the low temperature treatment for two weeks, indicating that mouse survival is dependent on the interaction between restricted feeding and ambient temperature. When mice were first subjected to TRF at 25 °C for 2 days and then transferred to 18 °C, survival was improved (purple vs. green, P= 0.0031; Fig. 1d). This result suggested that mice subjected to TRF are more susceptible to decreased ambient temperature and that adaptation to TRF prior to transfer to low temperature can improve survival, likely due to the development of a food anticipatory behavior ^{[44, 45]} also subjected 18- to 26-week-old mice (male, 25.6-30.6g) to TRF (ZT 16-20) at 21 °C and found that these mice did not have better survival advantage than the 8- to 10-week-old mice (blue vs black, p=0.82; Fig. S1a online), suggesting that age and body weight of adult mice do not influence the TRF-induced mortality in response to ambient temperature changes.

The ambient temperature-dependent lethality effect of TRF on WT mice was unexpected, even TRF during ZT 16-20, as such phenotype has only been reported before in Bmal1 knockout mice ^[44], and the temperatures used in our study were within the range normally suggested for mouse facilities. For example, Jackson laboratory suggests that mouse room temperatures are maintained between 18-23 °C (https://www.jax.org/iax-mice-and-services/customer-support/technical-support/breeding-and-husbandry-support/mouse-room-conditions). A major difference between our study and most others that have analyzed the effects of TRF is that mice were housed individually and in the temperature-controlled incubators in our study. To determine whether such a difference was responsible for the TRF-dependent lethality we observed, mice were housed in groups of five in each cage and subjected to the TRF treatment at 18 °C. All the mice survived (20/20) (Black; Fig. 1d). Therefore, the lethality phenotype can be reversed by cuddling among animals or increasing Ta, suggesting that lethality is probably caused by pathogenic hypothermia rather than a torpor state, which is transient and reversible ^[40].

3.2. The survival rate is modulated by the circadian clock in response to TRF

Because mouse body temperature is regulated by the circadian clock and feeding time can influence body temperature, metabolism, behavior, and circadian rhythms ^[46–48], we examined whether the timing of TRF influenced survival. We performed experiments at the different time points at 21 °C because about half of the mice survived the TRF treatment at this temperature. We found that survival during TRF was significantly improved when feeding was restricted to ZT10-14 compared to ZT4-8 (p=0.0001), ZT16-20 (p=0.0006), and compared to ZT22-2 (p= 0.0084) (Fig. 2a). The anticipatory activity before the restricted feeding has been reported previously that could increase survival ^{[44, 45]}. However, mice with TRF at ZT10-14 did not display an increase of anticipatory activity in the first few days of TRF compared at ZT4-8 (Fig. 2b). To directly determine the role of circadian clock, we examined the Per1/Per2 or Cry1/Cry2 double-knockout mice, which are deficient in behavioral and molecular circadian rhythms ^[49–51] We found that the TRF time-dependent survival differences between TRF at ZT10-14 and ZT16-20 were abolished in the Per1/Per2 double-knockout mice and decreased in the Cry1/Cry2 double-knockout mice (Fig. 2c, d). The morbidity rates of the Per1/Per2 and Cry1/Cry2 double-knockout mice during the TRF treatment were higher than that of the wild-type mice, most likely due to their known metabolism defects ^{[52, 53]} or defective adaptive thermogenesis ^[54]. These results suggest that the circadian clock affects the survival of mice subjected to TRF by gating the TRF response.

To determine whether the temperature-dependent TRF-induced lethality difference was due to differences in food intake, we recorded the amount of food consumption. As expected, mice ate more when housed at lower temperatures (19.5-21 °C) than when housed at 25 °C when fed ad libitum (day 0-7, p<0.001; Fig. 2e). After the start of TRF on day 7, food intake of all tested mice was initially reduced but gradually recovered to reach a constant level after one week. The food intake of mice was not monitored at 18 °C since they were all dead during first week of the treatment, and was also not monitored at 4 °C because the amount of food intake is not accurately weighed due to moisture. But we acknowledged that the amount of food intake is definitely more consumed at 4 °C. When mice were fed during ZT16-20, there were no significant differences in food intake at different temperatures (19.5, 21 and 25 °C, p>0.4; Fig. 2e). Interestingly, when fed during ZT10-14, which resulted in higher survival than feeding during ZT16-20, mice ate less than groups fed during ZT16-20 for the first five days of the TRF treatment, a critical time window for survival (p<0.05; Fig. 2a, e). This result indicates that food intake alone cannot explain the lethality phenotype under TRF treatment. To further evaluate the effect of caloric uptake on the lethality during TRF treatment, mice were fed with a high fat diet (60%Kcal%fat, D12492, Research Diets, Inc.) during ZT16-20 at 18 °C. The high fat diet appeared to improve mouse survival compared to normal chow diet at 18 °C but the statistical difference was not achieved (yellow vs green, Fig. S1b, p=0.09). When extending to 6-hour TRF at 21 °C, the survival significantly improved (Fig. S1a online). Together, these results suggest that TRF time window, TRF duration and ambient temperature modulate the side effect caused by TRF intervention.

3.3. Mice subjected to TRF develop severe hypothermia

The ambient temperature dependence of TRF-induced lethality raised the possibility that pathogenic hypothermia is the cause of mouse mortality. Thus, we determined body temperature of mice before and after TRF treatment at 25 °C and 21 °C. C57BL/6J mice were surgically implanted with intraperitoneal transmitters to record body temperature over time. Similar to previous reports ^{[47, 48, 55]}, body temperature oscillated with a deviation of less than 1 °C in mice fed ad libitum with a trough during the day and bimodal peaks at night at 25 °C and 21 °C (Fig. 3a–d; Fig. S2a, b, g online). At 25 °C, TRF treatment (mice fed ZT16-20) decreased daily average body temperature from approximately 37 °C to about 36 °C (Fig. 3b). In addition, the daily deviation increased to about 1.4 °C from ~1 °C in mice fed ad libitum (Fig. 3c), and the lowest body temperature reached 34.5 °C (Fig. 3a). These data suggest that homeostasis of body temperature was modestly impaired by TRF at 25 °C. In contrast, the TRF treatment at 21 °C resulted in a dramatic decrease in daily average body temperature and an increase of oscillation range (Fig. 3d–g; Fig. S2c–f, h, i online). The daily average body temperature was approximately 25 °C and the lowest was slightly below 23 °C. In addition, the TRF treatment severely disrupted or abolished the normal daily rhythm of body temperature which was masked by food intake time. Fig. 3d shows a representative temperature trace of a mouse that eventually died during TRF treatment. There was a gradual decrease of body temperature after initiation of TRF, and body temperature failed to rise from the trough on the fourth day (Fig. S2d, f online). Of note, the TRF may have a greater disruption of body temperature than that of 48hr of fasting which maintains the homeostasis of body temperature ^[56]. These results suggest that the combination of TRF and a decrease in ambient temperature results in severe hypothermia and the loss of body temperature homeostasis, which is the cause of TRF-induced mortality.

Fig. 3.

Restricted feeding impairs body temperature homeostasis, (a) Core body temperatures were measured by intra-abdominally implanted thermometers. The representative recording of the body temperatures at each time of day of mice fed ad libitum for four days and then subjected to food consumption at ZT16-20 at Ta of 25 °C (See also Fig. S2a, b, g online). The feeding time is indicated in the red-filled box. Periods of darkness and light are indicated the black and white bars, respectively, (b, c) The body average temperature (b) and body temperature deviation (c) at fed ad libitum and time-restricted feeding day 1 to 4 were averaged separately as ad, TRF1, TRF2, TRF3 and TRF4 and plotted as the mean ± s.e.m at 25 °C. (d) Core body temperatures of mice subjected to food consumption at ZT 16-20 under 21 °C (See also Fig. S2c, d, h online), (e) Core body temperatures of mice subjected to food consumption at ZT10-14 under 21 °C (See also Fig. S2e, f, i online), (f, g) Comparison of the daily body average temperature (f) and body temperature deviation (g) between mice subjected to food consumption at ZT 16-20 and ZT10-14, at fed ad libitum and time-restricted feeding day 1 to 4 were averaged separately as ad libitum, TRF1, TRF2, TRF3 and TRF4 and plotted as the mean ± s.e.m at 21 °C. (h, i) Diurnal locomotion changed in response to restricted feeding. Representative double-plotted locomotion actograms of mice fed ad libitum for two weeks and then subjected to TRF for two weeks at 25 °C (h) or 21 °C (i) (See also Fig. 2b; Fig. S3 online), (j) Percentage distributions of wheel-running activity by time of day were analyzed (See also Fig. S4 online, values represent the average ± s.e.m., * : p < 0.05, * * * : p < 0.001, Error bars: ± s.e.m.).

Comparison of body temperature after TRF treatment at 21 °C between ZT10-14 and ZT16-20 showed that daily averages and daily fluctuations were significantly higher for mice fed during ZT10-14 than during ZT16-20 (Fig. 3d, e; Fig. S2c–f, h, i online). These results are consistent with the higher rate of survival of mice fed during ZT10-14 (Fig. 2a) and suggest that the circadian clock plays an important role in modulating body temperature homeostasis in response to TRF.

Mice are nocturnal and exhibit a robust daily rhythm of activity (Fig. 3h). Animals alter their behavioral to protect themselves from hypothermia ^[57]. When mice were fed at ZT16-20 at 25 °C, during which they normally seek food, the robust diurnal activity rhythm was maintained but their activity during the daytime increased (Fig. 3h, j; Figs. S3a, S4a online). This result is similar to previous observations ^[58] suggesting that mice modulate behavior in response to TRF. In contrast, for the mice that survived TRF at 21 °C, the nocturnal activity rhythm was abolished and much higher levels of daytime activity were observed (Fig. 3i, j; Figs. S3b, S4b online). These results indicate that the combination of TRF and a small decrease of ambient temperature resulted in dramatic changes in behavior, likely to counter impaired body temperature homeostasis.

3.4. Lesions of the SCN rescue the lethal phenotype induced by TRF

Our results suggest that the circadian clock is an important regulator of body temperature homeostasis in response to TRF. The SCN is the master oscillator of mammalian circadian clock; its ablation abolishes circadian rhythms of behavioral and many physiological processes ^[59]. To determine the function of circadian clock in this response, we performed a bilateral ablation of the SCN ^[59] and evaluated the effect of TRF on these SCNx mice. We used Chi-square periodogram analysis of locomotor rhythms (Fig. S5 online) to screen for mice with arrhythmic locomotor activity, and ~40 % lesioned animals can be used for further experiments (Fig. 4a, b). At the end of the experiment, histological analyses of the brains in the control and the SCNx mice were performed to confirm the SCN lesion (Fig. S6a–d online). Remarkably, all SCNx mice subjected to TRF during ZT16-20 or ZT4-8 at 21 °C survived (Fig. 4c, d). Their body temperature rhythmicities were completely abolished under ad libitum and were driven by the food availability after TRF (Fig. 4f; Fig. S7 online). In addition, in contrast to the dramatic decrease of average body temperature in the SCN-sham animals after TRF treatment, average body temperature of the SCNx mice was similar before and after TRF treatment (Fig. 4e–g). Furthermore, body temperature oscillated within a much smaller range after TRF treatment in the SCNx mice than SCN-sham group, and body temperature of the SCNx mice rarely dropped below 34 °C (Fig. 4f–h; Fig. S7 online). These results indicate that bilateral ablation of the SCN almost completely rescued the loss of body temperature homeostasis phenotype induced by TRF at 21 °C. This function by SCN lesions is likely to through metabolic or neuroendocrine process which is independent of its role as circadian pacemaker because this phenotype cannot be rescued by Per1/Per2 or Cry1/Cry2 double knockout.

Fig. 4.

Mice with disruption of the SCN function rescued hypothermia and survival. (a, b) SCN sham and lesioned mice were subjected to time-restricted feeding at ZT16-20 at 21 °C. Arrhythmic mice confirmed by locomotor assay (Fig. S5 online) and histological analyses (Fig. S6 online). Restricted feeding region was labeled in red filled bar (TRF). (c, d) Kaplan-Meier survival curves of SCN sham and SCNx mice fed at ZT16-20 (c) and ZT4-8 (d). Log-rank test verified the statistically significant differences, (e, f) The effects of bilateral SCN lesions on the body temperatures of mice subjected to restricted feeding at ZT 16-20. The representative recordings of the body temperatures from the SCN sham (e) and the SCN lesioned mice (f) (See also Fig. S7 online), (g, h) Comparison of daily body average temperature (g) and daily temperature deviation (h) after restricted feeding (ad libitum (AL), TRF1, TRF2, TRF3 and TRF4) in the SCN sham (black bar) and the SCN lesioned mice (red bar). Values represent the average + s.e.m., Error bars: +/− s.e.m., * * : p < 0.01). (i) Kaplan-Meier survival curves of restricted feeding mice under constant light condition (LL) versus constant dark condition (DD). Log-rank test verified the statistically significant differences. Number of mice in each group was indicated in the graph.

To further examine the role of the circadian clock in the regulation of body temperature homeostasis and to find an intervention approach without the need of surgery, we performed the TRF experiment at 21 °C in constant light (LL), which is a condition that desynchronizes the SCN neurons and causes arrhythmicity ^[60]. As expected, all mice exposed to LL also survived under the TRF treatment, whereas only about 50% of animals survived in constant darkness (DD) or under standard LD cycles (Figs. 4i, 1c). Together, these results demonstrate a negative role of the SCN in maintaining body temperature homeostasis in response to TRF and that LL can suppress such an effect of the SCN.

3.5. TRF reprograms rhythmic transcriptome in the (DMH)

The dorsomedial hypothalamus (DMH) is a major direct anatomical output region of the SCN in the hypothalamus. The DMH regulates feeding behavior, locomotor activity, and body temperature ^[61–64]. To assess whether the DMH was functionally associated with TRF response at 21 °C, we examined the activation of the immediate-early gene c-fos in the DMH on the second day following TRF treatment at ZT12-before feeding and ZT18-during feeding compared with mice fed ad libitum. We found that c-fos positive neurons in the DMH were markedly increased when mice were exposed to the TRF treatment compared to the control mice (Fig. S8 online), suggesting an activation of the DMH under TRF treatment at 21 °C.

To determine how the interaction of TRF and ambient temperature (21 °C) regulates circadian gene expression in the DMH in response to the TRF treatment, we obtained DMH tissues of mice at different time points over 24 hours at 21 °C under standard LD cycles that were either fed ad libitum or subjected to TRF treatment from the second day. Whole-transcriptome analyses were performed by RNA sequencing. Using the Jonckheere-Terpstra-Kendall (JTK) algorithm we identified 1533, 2480, and 1436 transcripts that were rhythmically expressed with periods of approximately 24 hours in mice fed ad libitum and TRF at ZT4-8 and at ZT16-20, respectively (p < 0.05; Fig. 5a). A large increase of rhythmic transcripts was found in the TRF ZT4-8 group compared to the ad libitum or the ZT16-20 group, indicating the sensitivity of the DMH to conflicting zeitgebers of LD cycle and timing of food consumption and demonstrating the induction of de novo rhythmic transcripts by TRF. Surprisingly, despite the large number of rhythmically expressed genes in these three conditions, only 52 rhythmic genes were shared among them (Fig. 5a; Fig. S9a, Table S1 online). These shared genes included circadian rhythm genes Per2, Chrono and Nr1d1 and genes involved in chromatin modifications Brd4, Hdac4, Kmt2c, Kmt2d, Tet2, and Ubn1, suggesting that the expression of these rhythmic transcripts are resistant to TRF. There were 152 rhythmic transcripts were shared between the ad libitum and TRF ZT16-20 groups (Fig. 5a; Fig. S9b, Table S2 online) and 287 transcripts shared between the ad libitum and TRF ZT4-8 groups (Fig. 5a; Fig. S9c, Table S3 online). Between the TRF ZT4-8 and ZT16-20 groups, 393 (445-52) rhythmic transcripts were shared that were not rhythmic in the ad libitum group (Fig. 5a; Fig. S9d, Table S4 online). Comparison of expression levels of rhythmically expressed transcripts in each group to those in the two other groups further demonstrated the lack of significant overlap of rhythmically expressed genes among the three groups (Fig. 5b). Thus, this reconstitution of rhythmic transcripts was specifically induced by TRF and functional clock independently. Furthermore, for most known circadian-associated transcriptional repressors such as Nr1d1, Bhlhe40, Chrono, Cry1, Cry2, Hif, Btrc, Sirt1, Kdm5b, Per2, and Per1, amplitudes of rhythmic expression were markedly increased in TRF groups compared with ad libitum, and their phases were dependent on the timing of TRF (Fig. S10 online). In contrast, the rhythmicity of Bmal1 was significantly reduced during TRF (Fig. S10 online). In addition, rhythmic expression of clock-controlled genes such as Dbp and Nfil3 was abolished in both TRF groups (Fig. 5c). These data suggested the molecular clock mechanism was attenuated in response to the acute TRF stimuli. It was previously shown that Per1 and Per2 mRNA levels in the DMH are induced after 10-14 days in the restricted feeding ^{[28, 65}], suggesting a different role of functional clock in response to acute TRF and chronic adaptation to TRF. Together, these results demonstrate that both TRF and its timing result in global changes of the rhythmic DMH transcriptome, consistent with changes of the body temperature rhythms driven by food intake time (Fig. 3d, g) and arrhythmic locomotor activity (Fig. 3i). These results indicate the disruption of normal circadian clock functions and reconstruction of novel rhythmic transcriptome in response to TRF treatment.

Fig. 5.

Global reprogramming of the rhythmic transcriptome in the DMH by TRF. (a) Venn diagram displays the total number of rhythmic genes in ad libitum (1533 genes), TRF ZT4-8 (2480 genes) and TRF ZT 16-20 (1436 genes), including common genes, sorted every 4 hr from ZT0. Extracted RNA was processed for RNA-seq, the number of samples in each group (n = 2). The numbers indicated by black lines are the overlapping gene numbers between two groups. Rhythmic genes in each group were determined based on the JTK algorithm (p < 0.05). See also Fig. S9, Tables S1, S2, S3, S4 online. (b) Comparison of the rhythmic genes among ad libitum, TRF ZT4-8, and TRF ZT16-20. Heatmaps display genes exclusively rhythmic in ad libitum (1146) (left panel), TRF ZT4-8 (1800) (middle panel) and TRF ZT16-20 (891) (right panel) (p < 0.05). (c) Expression levels (RPKMs) of circadian output genes of Dbp and Nfil3 in ad libitum, TRF ZT4-8 and TRF ZT16-20 (error bars: mean ± s.e.m., p < 0.001, by Prism two-way ANOVA with Bonferroni’s post hoc test). See also Fig. S10 online. (d) Biological process signatures of common rhythmic genes in the TRF16-20 and TRF4-8 groups. Numbers within the pie charts indicate number of rhythmic genes identified within each biological process based on P value cutoff of 0.05. Two biological replicates per time point were subjected to the RNA sequencing. See also Table S5, S6 and S7 online for each group and Table S8 for enrichment pathways of common rhythmic genes in the TRF16-20 and TRF4-8 groups. (e) Biological process signatures of common rhythmic genes in the ad libitum and TRF ZT4-8 groups. (f) Biological process signatures of common rhythmic genes in the ad libitum and TRF16-20 groups.

Gene ontology (GO) analysis of the unique rhythmically expressed genes in each group showed enrichment in a number of different pathways (Fig. S11, Tables S5, S6, S7 online). Genes involved in protein dephosphorylation, protein ubiquitination, cellular response to DNA damage, miRNA-mediated inhibition of translation, and DNA repair were found to be enriched in the genes induced by TRF ZT4-8 and ZT16-20 (Fig. 5d; Table S8 online) and transcription, chromatin modification etc. induced by ad libitum and TRF ZT4-8 (Fig. 5e) or ad libitum and TRF ZT16-20 (Fig. 5f). These findings suggest that an acute TRF at the cool ambient temperature might impair genome integrity in the DMH. Further, as high death rate occurred within a short time period of TRF, these rapid reprogramming of genes involved in DNA damage may play a significant role in the morbidity.

To identify gene expression that are potentially responsible for the rescue of loss of body temperature homeostasis in response to TRF by the SCN lesions, we determined the DMH gene expression profiles of control mice not subjected to an operation and fed ad libitum, the SCN sham, and SCNx animals fed ZT16-20 at two time points (ZT8 and ZT16) at 21 °C by RNA-seq. By comparing these results, we sought to identify genes impacted by TRF treatment that were rescued by the SCN lesions. There were 361 and 513 transcripts differentially expressed (DE) (fold change ≥ 2, p < 0.05 up- or down-regulated) between ad libitum and TRF-SCN sham-treated mice at ZT8 and ZT16, respectively (Fig. 6a, b; Tables S9, S10 online). Among these DE genes, the expression levels of 136 genes (37.7%) at ZT8 and 202 genes (39.4%) at ZT16 in the DMH of SCNx mice were rescued to at least 50% level (middle panel of Fig. 6a, b; Tables S11, S12 online). These rescued genes are involved in different signaling pathways (Fig. 6a, b, bottom columns). Among them, 36 genes were shared between the two time points (Fig. 6c; Table S13 online), suggesting that the rescue of expression of these genes might be involved in the restoration of body temperature homeostasis. These genes are enriched in genes involved in regulation of cellular signaling transduction, response to steroid hormones, and induction of immune responses and include Ccr5, Calcrl, Gpr17, and P2ry12.

Fig. 6.

Genes and signal pathways are partially restored in the DMH from lesioned SCN mice. (a, b) Rescue the disturbances of TRF by the SCNx at ZT8 and ZT16. Upper tables represent differential expression (DE) genes between ad libitum and TRF (DE > 2-fold change (<50% or >200%), See also Tables S9, S10 online. Genes rescued (>50%) by the SCNx as DE between TRF and the SCNx (DE > 2-fold change (<50% or >200%) at ZT8 in (a) and ZT16 in (b), See also Tables S11 and S12 online. Heatmaps display the expression levels in ad libitum, TRF, TRF-SCNx at ZT8 (136) and ZT16 (202) (middle lanes). GO enrichment analysis of rescued genes in ZT8 and ZT16 (bottom lanes), (p < 0.05). (c) Common rescued genes in both ZT8 and ZT16 groups. See also Table S13 online. (d, e) Expression levels of genes related to thermoregulation in DMH from ad libitum, TRF and the SCNx-TRF at ZT8 and ZT16. (f) Q-PCR analysis result showing the mRNA levels of genes related to thermoregulation in the DMH from ad libitum, TRF ZT4-8 and TRF ZT16-20 over 24 hr. (g) Q-PCR analysis result showing the mRNA levels of Ccr5 and Calcrl in the hypothalamus under indicated conditions, up: 21 °C AL vs 21°C TRF (p=0.0483), 25°C AL vs 21°C AL (p=0.044), 25°C AL vs 21°C TRF (p=0.0201); down: 25°C AL vs 25°CTRF (p=0.0063), 21°C AL vs 21°C TRF (p=0.0138), 25°C AL vs 21°C TRF (p=0.0044). Values represent the average + s.e.m., Error bars: +/− s.e.m., * : p < 0.05, * * : p < 0.01, Error bars: ± s.e.m..

CCR5 is a G protein-coupled chemokine receptor that has been shown to be involved in LPS- and RANTES-induced fever, and the injection of CCR5 antagonist results in a decrease of body temperature ^{[66, 67]}. CALCRL is a calcitonin receptor-like protein that mediates daily fluctuation of body temperature in mice during the active phase of the LD cycle ^[47]. The RNA-seq results showed that the levels of both Ccr5 and Calcrl were significantly repressed by the TRF treatment and were rescued in the SCNx mice (Fig. 6d, e). In addition, we found that the levels of Ccr5 and Calcrl in the DMH were significantly repressed over 24 h in both TRF ZT4-8 and ZT16-20 groups (Fig. 6f). We then examined whether the expression of Ccr5 and Calcrl is regulated by TRF and ambient temperature changes. We found that mRNA levels of Ccr5 and Calcrl were both repressed by TRF treatment (Fig. 6g). For Ccr5, both TRF and 21 °C treatments resulted in a reduction of its mRNA levels, and their combination led to its further decrease (Fig. 6g). These results suggest the possibility that Ccr5 and Calcrl might contribute to the loss of body temperature homeostasis after TRF treatment. The other genes identified here may also be involved in body temperature regulation. It should be acknowledged that the impact of SCN lesions on mouse survival should not be restricted to those genes, and other brain regions are also involved in thermoregulation ^{[22, 68]}.

4. Discussion

The body temperature homeostasis system and the circadian clock control the oscillation of body temperature within a small range in endothermic homeotherms. Here we report that TRF of WT mice causes loss of body temperature homeostasis and lethality near room temperature, this phenomenon only reported in Bmal1^−/− mice due to its selective vulnerability ^[69]. TRF at 18-21 °C, within the temperature range commonly used in mouse facilities, resulted in severe hypothermia, disruption of body temperature homeostasis, and death of mice tested. Such a TRF effect is due to combinatory influence of restricted feeding and ambient temperature fluctuations and is not solely due to changes in calorie uptake. Fasting at low temperature was previously shown to induce torpor, a controlled and temporary drop in body temperature in mice ^{[70, 71]}. Under the TRF conditions used here, however, all TRF mice died within 7 days at 18 °C, and half of the TRF mice were dead at 21 °C, suggesting that the treatment results in pathogenic hypothermia and the loss of temperature homeostasis ^[40]. Thus, ambient temperature is very critical for survival when food is limited. In addition, the TRF response is gated by the circadian clock as the timing of TRF influences survival. When TRF occurred at ZT10-14, which spans the light to dark transition time when mice start to become active, an increased chance of survival was observed. In arrhythmic mice (Per1/2 and Cry1/2 double knockout mice), the impact of the timing of TRF is abolished. We did not observe an increased survival in these arrhythmic mice as SCN lesioned mice do, likely due to known metabolic defects in these knockout mice ^[52–54] or a non-circadian pacemaker function of the SCN.

The primary function of the SCN is to synchronize cell autonomous circadian oscillators in peripheral tissues to drive many rhythmic processes, including behavior, gene expression, and body temperature at the organismal level ^[7–9]. The coupling between the SCN and the body temperature homeostasis system allows optimal adaptation of body temperature to daily environmental and behavior changes under normal conditions. However, under TRF conditions, this coupling mechanism results in the loss of body temperature homeostasis, as indicated by the rescue of body temperature homeostasis by the SCN lesions. Our conclusion was further supported by survival of mice subjected to TRF in LL, a condition when the SCN intercoupling is impaired. Therefore, although the couplings between the SCN and peripheral circadian oscillators are advantageous for adaptation of mammals to daily environmental cycles, it can be detrimental to animal physiology under certain stress conditions. Our results further suggest that methods such as LL treatment which can disrupt SCN-associated clock functions can have potential beneficial effects under certain stress conditions (Fig. S12).

In human and rodents, the median preoptic nucleus (MnPO) in the hypothalamus acts as a temperature control center and works with other body temperature regulating systems, such as the muscle, skin, and blood vessels, to maintain the body temperature homeostasis so that body temperature is close to 37 °C ^{[22–24, 72]}. The DMH is an important output area for the MnPO for body temperature regulation ^{[23, 25, 26]}. In addition, DMH is involved in the regulation of locomotor activity and food-intake of mouse ^{[27, 73, 74}]. On the other hand, the SCN projects directly or indirectly into the DMH and functions to impose a temporal organization on physiological processes and behavior. Our results here demonstrated that TRF results in extensive reprogramming of gene expression in the DMH. TRF not only drastically alters rhythmic gene expression profiles in DMH but also repressed expression of genes such as Ccr5 and Calcrl that are known to be involved in body temperature control [^{47, 66]}.–We propose that the SCN mediates the repression of Ccr5, Calcrl and other genes in the DMH and other brain regions to promote the hypothermia in response to TRF under low ambient temperature. This function of the DMH should be different from food-entrainable function the DMH ^{[75, 76]} and other nucleus also play critical roles in body temperature regulation, thus, more in-depth research is needed in the future. Reprogramming of circadian gene expression is a response to allow animals to adapt to different conditions ^{[4, 77, 78]}. Our results suggest that both reprograming of DMH gene expression and the SCN function contribute to body temperature response to TRF under cool ambient temperature. In addition, our results suggest that the impaired function of circadian clocks in the SCN can also provide potential benefits for animal survival under certain stress conditions.

Supplementary Material

2

Figure S1. Survival of mice during TRF under various conditions.

(A-B) Kaplan-Meier survival curves of mice subjected to TRF at indicated conditions. Number of mice per group (n) was indicated. A log-rank test showed significant differences between different groups of mice: (A) 18-26-week-old mice (blue curve) vs. 8-10-week-old mice (black curve) at 21 °C (p=0.05106), TRF restricted to 6h (ZT16-22) mice (red curve) vs. TRF restricted to 4h (ZT16-20) mice (black curve) at 21 °C (p=0.0048); (B) The yellow curve indicates mice fed with a high fat diet during TRF at 18 °C, and this curve vs. green curve that mice at 18 °C with normal chow during TRF (p=0.09). The experiments were ended at day 15.

3

Figure S2. Representative recordings of core body temperatures from mice subjected to time-restricted feeding.

(A-B) Representative recordings of core body temperatures of mice subjected to food consumption at ZT16-20 at 25 °C.

(C-D) Representative recordings of core body temperatures of mice subjected to food consumption at ZT16-20 at 21 °C.

(E-F) Representative recordings of core body temperatures of mice subjected to TRF at ZT10-14 at 21 °C. Feeding time window is indicated by the filled red box. Periods of darkness and light are indicated by black and white bars. Values represent the average + s.e.m., Error bars: +/− s.e.m.

(G-I) Body temperature comparison between ad lib and TRF phase of the entire group of mice under TRF ZT16-20 at 25 °C (G), TRF ZT16-20 at 21 °C (H) and TRF ZT10-14 at 21 °C (I), respectively. Periods of darkness and light are indicated by black and white bars. Values represent the average + s.e.m., Error bars: +/− s.e.m.

4

Figure S3. Representative double-plotted locomotor actograms of mice during ad libitum and TRF.

(A) Representative double-plotted locomotor actograms of mice subjected to TRF ZT16-20 at 25 °C.

(B) Representative double-plotted locomotor actograms of mice subjected to TRF ZT16-20 at 21 °C. The feeding time is indicated by the filled red box. Periods of darkness and light are indicated the black and white bars, respectively.

5

Figure S4. Low ambient temperature and TRF result in diurnal activity in mice.

(A-B) Graphic representations of the entire group of mice subjected to TRF ZT16-20 at 25 °C (A) and 21 °C (B). Values represent the average + s.e.m., Error bars: +/− s.e.m.

6

Figure S5. Screening of arrhythmic mice after the SCN ablation.

Representative actograms of wild type (A-B), the SCN sham mice (C-D), and the SCN lesioned mice (E-F). Chi-square periodograms confirm robust rhythms for wild type (G-H), the SCN sham (I-J) and the SCN lesioned mice (K-L) under the LD cycle.

7

Figure S6. The SCN ablation confirmed by histology using Nissl-staining.

Representative bilateral SCN-sham (A-B) and the SCN lesioned (C-D). The scale is indicated.

8

Figure S7. Representative recordings of core body temperatures from the SCN-sham mice and the SCN lesioned mice subjected to TRF treatment.

(A-B) Representative recordings of core body temperatures of the SCN-sham mice subjected to food consumption at ZT16-20 at 21 °C. Feeding time is indicated by the filled red boxes. Periods of darkness and light are indicated by black and white bars.

(C-E) Representative recordings of core body temperatures of the SCN lesioned mice subjected to TRF at ZT16-20 at 21 °C.

(F-G) Body temperature comparison between ad lib and TRF phase of the entire group of the SCN-sham mice (F) and the SCN lesioned mice (G) at 21 °C. Values represent the average + s.e.m., Error bars: +/− s.e.m.

9

Figure S8. The DMH neuron of mice is activated by TRF at 21 °C.

Double-staining of c-Fos positive neurons and DAPI nuclear staining (c-Fos in red, DAPI in blue) of the DMH neurons in mice fed ad libitum (A-B) and subjected to TRF ZT16-20 (C-D) at indicated time points at 21 °C. TRF increased c-Fos positive neurons in the DMH neuron of mice at 21 °C (C-D). Scale bars: 100 μm.

10

Figure S9. Heatmaps display common rhythmic genes in the ad libitum, TRF ZT4-8 and TRF ZT16-20 groups.

(A) Common rhythmic genes in the ad libitum, TRF ZT4-8 and TRF ZT16-20. JTK < 0.05.

(B) Common rhythmic genes in the ad libitum and TRF ZT 16-20. JTK < 0.05.

(C) Common rhythmic genes in the ad libitum and TRF ZT4-8. JTK < 0.05.

(D) Common rhythmic genes in TRF ZT4-8 and TRF ZT16-20. JTK < 0.05.

11

Figure S10. Expression levels (RPKMs) of clock-related genes in ad libitum, TRF ZT4-8 and TRF ZT16-20 groups.

Expression levels (RPKMs) of clock-related genes in the ad libitum, TRF ZT4-8 and 965 TRF ZT16-20 by RNA sequencing. FPKM value calculated as normalization of the 966 raw sequencing results. Error bars: mean ± s.e.m.

12

Figure S11. Unique rhythmic genes in the ad libitum, TRF ZT4-8 and TRF 16-20 groups.

Numbers within the pie charts indicate number of rhythmic genes identified within each biological process based on P value cutoff of 0.05. Two biological replicates per time point were subjected to the RNA sequencing.

13

Figure S12. A diagram depicting the mechanism of body temperature is controlled by the SCN and temperature homeostasis in response to both TRF and a decrease of ambient temperature.

Upper: tight control of daily temperature cycles by body temperature homeostasis with circadian changes under ad libitum; middle: under TRF and a decrease of ambient temperature, the function of the SCN impairs body temperature homeostasis; bottom: impaired clock function in the SCN promotes body temperature homeostasis under TRF.

14

Table S1. Common rhythmic genes among ad libitum, TRF ZT4-8 and TRF 16-20 groups.

Table S2. Shared rhythmic transcripts between the ad libitum and TRF16-20 groups.

Table S3. Shared rhythmic transcripts between the ad libitum and TRF4-8 groups.

Table S4. Shared de novo rhythmic transcripts between the TRF16-20 and TRF4-8 groups.

Table S5. Significant GO enrichment pathways of uniquely rhythmical genes in ad libitum groups (p<0.01). Related to Fig. 5d.

Table S6. Significant GO enrichment pathways of uniquely rhythmical genes in TRF4-8 groups (p<0.01).

Table S7. Significant GO enrichment pathways of uniquely rhythmical genes in TRF16-20 groups (p<0.01).

Table S8. Significant GO enrichment pathways of common rhythmical genes between TRF4-8 and TRF16-20 groups.

Table S9. Differentially expressed transcripts between ad libitum and TRF treatment mice at ZT16-20, sampling at ZT8 (blue for down- , red for up-regulated).

Table S10. Differentially expressed transcripts between ad libitum and TRF treatment mice at ZT16-20, sampling at ZT16 (blue for down- , red for up-regulated).

Table S11. Rescued genes at ZT8 after the SCN lesions.

Table S12. Rescued genes at ZT16 after the SCN lesions.

Table S13. Shared common rescue genes between ZT8 and ZT16 after the SCN lesions.

Table S14. Primer list in Q-PCR.