Bmal1 knockdown suppresses wake and increases immobility without altering orexin A, corticotrophin‐releasing hormone, or glutamate decarboxylase

Summary

Objective

To determine the effect of Bmal1 knockdown (KD) on sleep, activity, immobility, hypothalamic levels of orexin, corticotrophin‐releasing hormone (CRH), and GABAergic glutamate decarboxylase (GAD).

Methods

We used Bmal1 siRNA, or control siRNA intracerebroventricular (ICV) injection to knock down Bmal1 in C57BL/6 mice. Sleep polysomnography, wheel‐running activity, and tail suspension test were performed. Polysomnographic (PSG) recordings in both groups were preceded by ICV injection made during both the light phase and the dark phase. We also measured brain orexin A and CRH using an ELISA and measured GAD using immunoblotting.

Results

Compared with control group, Bmal1 KD group had reduced wheel activity and increased immobility. Compared with control, the Bmal1 KD group had reduced wheel activity and increased immobility. During the first 24 hours after treatment, we observed that control siRNA induced a much greater increase in sleep during the dark phase, which was associated with lower orexin levels. However, beginning 24 hours after treatment, we observed an increase in sleep and a decrease in time spent awake during the dark phase in the Bmal1 KD group. These changes were not associated with changes in brain levels of orexin A, CRH, or GAD.

Conclusion

Bmal1 KD led to reduced activity, increased immobility, and dramatic reduction in time spent awake as well as an increase in sleep during the dark phase. Early after injection, there was a slight change in sleep but brain levels of orexin, CRH, and GAD remain unchanged. Control siRNA also affected sleep associated with changes in orexin levels.

1. INTRODUCTION

Sleep/wake cycles are regulated by both homeostasis and the circadian system. Despite significant advances made in understanding sleep homeostatic mechanisms, knowledge of circadian regulation of sleep/wake cycles remains very limited. The discovery of genes involved in circadian regulation, such as aryl hydrocarbon receptor nuclear translocator‐like protein 1 (Bmal1, Arntl or Mop3)1, 2 and circadian locomotor output cycles kaput (Clock),3 highlights the opportunity for further exploration of the relationship between homeostatic and circadian regulation of sleep/wake cycles. Quantitative alterations in genes, molecules, and sleep can be measured after modification of specific circadian genes. Significant increases in the expression of Bmal1 and Clock in suprachiasmatic nucleus (SCN)4 lead to protein synthesis of Bmal1 and Clock.5 These proteins form a heterodimer to trigger cellular, physiological, and behavioral activities via activation of E‐boxes, which are DNA response elements located in target gene promoters acting as protein‐binding sites,6 to initiate the “day” phase,5, 7 that is, the morning in humans and the dark period in nocturnal species including laboratory rats and mice. Deletion of the Bmal1 gene induces a global effect on behavior and biomarkers of aging,8 as well as a dramatic reduction in period of wakefulness, as demonstrated by an increase in both rapid eye movement (REM) and non‐REM (NREM) sleep primarily in the dark phase in mice.9 This evidence suggests that circadian activation of Bmal1 is necessary for the activation of the wake state or a strong trigger for the wake state. Despite a lack of evidence regarding the mechanism by which Bmal1 regulates sleep/wake cycles, one early study has presented evidence that SCN neurons project to the areas involved in sleep/wake regulation including orexin neurons in the lateral hypothalamus,10 GABA neurons in the ventrolateral preoptic nucleus (VLPO) and in the median preoptic nucleus (MnPO),11, 12 as well as corticotrophin‐releasing hormone (CRH) neurons in the paraventricular nucleus of the hypothalamus13 which mostly contribute to stress‐induced changes in wakefulness.14 Current evidence indicated that orexin neurons or orexin have wake‐promoting and/or sleep‐inhibitory effects. Brain infusion of either orexin A or B induces an increase in wakefulness and decreases in both REM and NREM sleep in mice and rats15, 16; a dual orexin receptor antagonist was shown to suppress wakefulness and promote sleep.17 Orexin gene deletion in mice increases REM sleep in the dark phase.18 Activation of orexin‐producing neurons results in an increased probability of transition to the wake state from either NREM or REM sleep19, 20 while silencing of orexin neurons induces NREM sleep.21 Activation of orexin‐producing neurons or an increase in brain orexin can be induced naturally by food restriction22 or stress.23 Acute stress also induces an immediate increase in CRH,14 a hormone that is considered a major initiator in the stress response system.24, 25 The CRH system innervates orexin‐producing neurons26 and depolarizes orexin‐producing neurons via the CRH‐R1 receptor.27 Corticotrophin‐releasing hormone increases the REM sleep response to inescapable shock28 and a CRH receptor antagonist significantly suppresses stress‐induced reactions.29 Brain injection of CRH induces an increase in wakefulness and a decrease in sleep in mice.30 This evidence suggests that the Bmal1 gene product may be able to regulate the wake state by affecting orexin and CRH neurons. While orexin and CRH neurons are highly involved in the regulation of wakefulness, GABA neurons are more involved in the regulation of sleep.31 GABA, the major inhibitory neurotransmitter in the brain,32 plays a key role in NREM sleep generation and maintenance33 and affects muscle tone.34, 35 The majority of preoptic sleep regulatory neurons are sleep‐active neurons and synthesize GABA,31 suggesting that Bmal1 gene product may also act via the SCN‐VLPO network to regulate sleep.

We hypothesized that Bmal1 knockdown (KD) suppresses wakefulness and promotes sleep via hypothalamic orexin neurons, CRH neurons, and GABA neurons. To test our hypothesis, we conducted a study to measure the effect of Bmal1 KDon behavior, sleep, and brain levels of orexin, CRH, and glutamate decarboxylase (GAD), an enzyme that catalyzes the decarboxylation of glutamate to GABA. Due to our long‐standing interest in study depression and the circadian features of depressive disorders which are more severe in the morning,36 we also measured immobility, a commonly measured variable used in the study of depression.37

2. MATERIALS AND METHODS

2.1. Animals and experimental design

Because knockout (KO) of the Bmal1 gene has been shown to have severe global impacts on murine growth, including dramatically reduced wheel‐running activity,38 age‐dependent reduction in muscle and bone mass, reduced body size, early aging and a shorter life span,8 and significant reduction in mitochondrial volume,39 we opted to utilize Bmal1 KD in adult mice via brain administration of Bmal1 siRNA to create a mouse model of Bmal1 deficiency.

C57BL/6J (B6) mice of 12‐16 weeks of age were used in this study. To minimize individual variation that may be caused by unsynchronized sex hormone cycles, female mice were not used. Mice were purchased from The Jackson Laboratory (Sacramento, CA, USA) and housed in the Animal Recourses Facility (ARF) of the Louis Stokes Cleveland DVA Medical Center (LSCDVAMC) under standard housing conditions approved by the Standard Operation Procedures (SOP) of the ARF. This includes unlimited access to autoclaved water and food, housing at a room temperature of 20‐22°C, and a light‐dark cycle of 12:12 hours.

This study was designed in four parts. We first evaluated whether Bmal1 siRNA could knock down Bmal1 gene expression. Then, we studied the effect of Bmal1 KD on behavioral tests, specifically the forced swim test and the tail suspension test. In part 3, we studied sleep/wake cycles in mice injected with Bmal1 siRNA before the end of the light phase in one set of animals, and before the end of dark phase in another set of animals. In part 4, we measured brain levels of orexin and CRH in mice euthanized at 6 or 24 hours after intracerebroventricular (ICV) injection of Bmal1 siRNA or control RNA. All protocols were approved by the Institutional Animal Care and Use Committee (IACUC) of LSCDVAMC before the study was initiated.

2.2. ICV injection and Bmal1 KD

B6 mice were anesthetized using a cocktail solution of ketamine, xylazine, and acepromazine (100, 10 and 2 mg/kg, respectively), prepared for surgery and implanted with a guide cannula for ICV injection of Bmal1 siRNA. Brain ICV injection can deliver the Bmal1 siRNA in a broadly diffused manner because siRNA was directly introduced into the cerebrospinal fluid. Thereafter, the lateral ventricle was selected as the ICV injection site. A small hole was drilled at AP = 0.46, L = 1.0. A guide cannula (O.D. = 0.5 mm; hollow stainless steel tube; Cat# C315GS‐4; Plastic One, Inc., Roanoke, VA, USA) that guides an internal cannula (C315IS‐4; Plastics One, Inc., Roanoke, VA, USA) to the specific injection site (dorsoventral [DV] = 2.5) was stereotaxically implanted. A dummy cannula (Cat# C315dcs‐4, Plastics One, Inc., Roanoke, VA, USA) was inserted to cover the guide cannula. The guide cannula was secured with C&B Metabond bonding agent. This bonding agent adheres extremely well to bone, enamel, and dentin. The skin was then sutured together.

To determine whether the guide cannula functions in the correct location, we performed an angiotensin II test. It is known that angiotensin II induces a polydipsia response, and angiotensin II injection into the ventricle induces an immediate water drinking reaction.

Thus, the accuracy of the ICV injection site was verified by visually observing the water drinking behavior after ICV injection of 100 U angiotensin II 2 days after surgery. Only animals that had a water drinking response within 5‐minute postinjection were considered to have accurate guide cannula and were used for further study. To avoid clogging of the guide cannula and a corresponding reduced stress response, one injection of artificial cerebrospinal fluid (2 μL) was administered to all animals between angiotensin II testing and subsequent Bmal1 siRNA injection. These mice were called icv injection ready mice. At the time of Bmal1 gene knockdown, an ICV injection was performed using a 10 μL Hamilton microsyringe connected to a 33GA injection internal tube (Cat# C315IS‐4‐SPC, Plastics One, Inc., Roanoke, VA, USA) via PE 20 microtubing. Two microliters of either Bmal1 or negative control siRNA solution was injected for a period of 1 minute, and the injection tubing was held for another minute to avoid injectate backflow.

Bmal1 KD was accomplished by ICV injection of a pool of three target‐specific, 19‐25 nt siRNAs (BMAL1 siRNA (m): sc‐38166, Santa Cruz Biotechnologies, Santa Cruz, CA, USA). Negative Control siRNA (Cat# 1027310, Qiagen, Hilden, Germany) was used as a control; this siRNA contains a scrambled sequence that does not lead to the specific degradation of any known cellular mRNA. To achieve better siRNA‐mediated gene expression deficiency and fewer side effects then are caused by a lipid carrier, we used rabies virus glycoprotein peptide 9R (REG‐9R) to deliver Bmal1 siRNA to the mouse brain ventricle. siRNAs were solubilized in RNase‐free water at a concentration 2 μg/μL (0.145 nmol/L/μL). RVG‐9R (from Tafts University peptide core) was solubilized in RNase‐free water at a concentration 1.5 nmol/L/μL. The mixture of siRNA and RVG‐9R was made by combining 2 μL of siRNA and 2 μL RVG‐9R. In accordance with descriptions in previous publications, a volume of 3 μL or more was used for mouse ICV injection,38, 39 4 μL of the siRNA (4 μg) and peptide mixture were injected into each mouse. After 2 days of baseline recording, Bmal1 and negative control siRNA were injected through the ICV guide cannula.

2.3. Polysomnographic (PSG) recording

Mice used for the sleep study in part 3 were implanted with electrodes for recording electroencephalograms (EEGs) and electromyograms (EMGs) in addition to the ICV guide cannula, according to our previously described method.40 Due to ICV guide cannula implantation, the EEG (not EMG) electrode site locations were modified from those previously reported41 and used in our previous publication.42 In anesthetized B6 mice, three holes for electrodes E1 (in the frontal cortex, M1 area), E2 (in the parietal cortex), and E3 (in the visual cortex serving as a reference electrode) were drilled into the skull in the frontal and parietal regions to accommodate three stainless steel jewel screws serving as EEG electrodes (E1: AP = 2.0, L = 1.5; E2: AP = −1.5, L = 1.5; E3: AP = −3.5, L = 2.75) identified with a digital stereotaxic instrument (Model 940, David Kopf Instruments, Tujunga, CA, USA). The frontal cortical EEG was obtained from a combination of E1 and E2 reading, and parietal EEG was obtained from combination of E2 and E3 reading. Three Teflon‐coated stainless steel wires were stitched onto the surface of the nuchal muscles immediately posterior to the dorsal area of the skull and served as EMG electrodes. All electrodes were bonded to the dorsal surface of the skull with C&B Metabond bonding agent. This bonding agent adheres extremely well to bone, enamel, and dentin. The skin overlaying the skull and posterior muscles were reopposed, allowing for a cable to extend through the skin and connect to a 12‐channel commutator. Mice were monitored continuously until they recovered from anesthesia. Then, mice were moved to standard housing/recording chambers for continued recovery. During at least 10 days of postsurgical recovery, all mice were functionally verified for the correct location of the guide cannula, and were connected to the recording system (detailed in 41). The light/dark cycle was set as light on at 9:00 am and light off at 9:00 pm. Test recording was performed to verify recording quality. Mouse PSG data were scored as rapid eye movement (REM) sleep, non‐REM sleep (NREM sleep), and awake state in a 10‐second epoch using the Rodent Scoring Module of the Somnologica Science, a computer program that has been successfully employed in prior studies.40, 41 In the Rodent Scoring Module, the frontal cortical EEG was used to distinguish delta power and the parietal cortical EEG (or frontal EEG if the parietal EEG was not sufficient) was used to distinguish theta power. Scores were visually confirmed by two technicians. The percent of time in each state was calculated by dividing the total recording time by the cumulative minutes for each 4‐hour section. Differences in time spent in each state (REM sleep, NREM sleep, total sleep and wakefulness) in each 4‐hour section between control and Bmal1 KD groups were evaluated statistically using two‐way (treatment × time) ANOVA and all pairwise multiple comparison procedures (Bonferroni t test).

2.4. Verification of Bmal1 protein changes after Bmal1 KD by Western blotting

Before initiating the behavioral and sleep study, we evaluated the efficacy of Bmal1 siRNA on Bmal1 expression in separate groups of control and Bmal1 KD mice. To evaluate the efficacy of SCN Bmal1 KD by Bmal1 siRNA ICV injection, we measured Bmal1 protein levels using Western blot in SCN samples collected from animals that were euthanized 2 days after Bmal1 siRNA (n = 6) and control siRNA ICV (n = 6) injection. Beta‐actin was used as a control for protein loading in Western blots. Rabbit anti‐GAD (AB1511, Millipore), rabbit anti‐Bmal1 (SAB43006, sigma), and rabbit anti‐actin (sc‐1616r, Santa Cruz) were used as primary antibodies, and a corresponding secondary anti‐rabbit antibody coupled to horseradish peroxidase (HRP) was used in the protocol. After completion of standard incubation with primary and secondary antibodies and sufficient rinses, membranes were treated with a chemiluminescent HRP substrate and images were captured using a FOTO/Analyst^® Investigator/Eclipse System (FOTODYNE Incorporated, Hartland, WI). The optical density (OD) of Bmal1 (OD_B) and OD of 18S (OD₁₈) in RT‐PCR and OD_B and OD of actin (OD_A) in Western blot results were digitalized using ImageJ Software (National Institutes of Health, Bethesda, MD, USA). Relative optical density from the Bmal1 KD group was compared to that of the negative control group. This immunoblot method was also utilized for semiquantification of GAD in brain tissue dissected from the hypothalamus of a different set of animals.

2.5. Wheel‐running activity and immobility tests

As measurement of locomotor activity has been commonly used for the measurement of circadian rhythm in rodents, measurement of locomotor activity in this study helps to functionally evaluate the suppression of Bmal1 via Bmal1 siRNA. As immobility has been commonly used in the evaluation of depressive pathology, we performed wheel‐running activity and tail suspension tests to determine the effect of Bmal1 suppression on circadian amplitude and depressive symptoms of “despair.” All tests were conducted at 10 days after surgical implantation of the ICV guide cannula, which was verified for accuracy of placement. The aim of the wheel‐running test was not to determine whether Bmal1 KD affects circadian rhythm, but to determine the “amplitude” of the rhythm (ie the amount of activity). Thus, the test was only recorded for a few days and it was performed under 12‐hour‐light and 12‐hour‐dark cycles with an activity wheel chamber (Lafayette Model 80820). Activity sensors were attached to the wheel support frame in each chamber and connected to the Lafayette software program for continuous monitoring of animal activity. Calibration was performed before the start of the test to ensure that the apparatus was in good working condition. Using a MatLab program, the number of cycles/hour was calculated for each animal. Activity in 10 Bmal1 KD mice and 10 control mice was recorded for 3 baseline days and 5 days after Bmal1 siRNA was administered to the lateral ventricle. Data were processed using the program provided by the company and plotted as the mean + standard error (SE). Two‐way (treatment × time) ANOVA was used for statistical evaluation, and all pairwise multiple comparison procedures (Bonferroni t test) were used for comparing the differences in activity between the groups in three‐hour sections.

Immobility tests are commonly used to evaluate “despair” behavior in rodents for studying the pathology of depression43, 44, 45 and screening antidepressant drugs. In a separate set of animals, the tail suspension test was performed for 6 minutes using clean transparent buckets (8″ × 8″ × 9″) under a red light for clear viewing, similar to the method used in our previous publication.37 Video recording began before attaching each mouse individually—by its tail—to a Plexiglas rod, using adhesive tape. Offline video was subsequently rated for time spent moving and time of immobility during the 6 minutes. A t test was used to analyze the data. Longer immobile times and shorter struggling times were considered to be “depressive” behaviors.

2.6. Orexin and CRH measurement

Additional ICV injection ready mice were used to quantify brain orexin and CRH. All mice were implanted with ICV injection guide cannula only. Then, one group of mice was injected with control siRNA and the other group of mice was injected with Bmal1 siRNA. One set of mice (including six control and six Bmal1 KD mice) were euthanized at 6 hours after the ICV injection was administered. The other set of mice (six control and six Bmal1 KD mice) were euthanized at 24 hours after ICV injection. According to a mouse atlas,46 a hypothalamus tissue block (not including the SCN) was dissected from the ventricle side of the brain at AP: Bregma 0.0 to −3.0, lateral: left 1.0 to right 1.0, and deep = 1.0. Tissue was processed for ELISA quantification of Orexin A using a standard kit (Cat# EK‐003‐30), and CRH (Cat# EK 019‐06; Phoenix Pharmaceuticals, Belmont, CA, USA) according to previously established protocols.37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50

2.7. Data analysis and statistical evaluation

Imaging data of protein levels by Western blot were digitalized using ImageJ Software (National Institutes of Health, Bethesda, MD, USA). Data are expressed as relative optical density ratios. A Student's t test was used to evaluate statistical significance. Concentrations of orexin A and CRH were detected using an ELISA kit and quantified according to the instructions provided by the manufacturer. Data were evaluated using a Student's t test. All t tests were two‐tailed. For all statistical analyses, P ≤ 0.05 was considered significant. Data were presented as the mean + standard error (SE).

3. RESULTS

3.1. Evaluation of Bmal1 siRNA on Bmal1 protein levels

The injection of Bmal1 siRNA into the cerebral ventricle was intended to affect Bmal1 expression in most brain regions. As SCN is a region known as the master control of circadian rhythm, we used samples of SCN for Western blot semiquantification of Bmal1 protein. The ratio of Bmal1 protein in Western blot was calculated using the ratio: OD_B (Western blot)/OD_A. As shown in Figure 1, the ratio of Bmal1 to actin was dramatically and significantly lower in mice treated with Bmal1 siRNA (n = 6) as compared to controls (n = 6). The relative level of Bmal1 protein vs β‐actin in the Bmal1 siRNA group was approximately 40% of that of the control group (t = 2.801 and P < 0.023), indicating that Bmal1 siRNA effectively reduced Bmal1 expression. These results confirm that ICV injections of Bmal1 siRNA can successfully knock down Bmal1 gene expression.

Figure 1.

Bmal1 protein levels in mice injected with Bmal1 siRNA or negative control siRNA. (A) The imaging of Western blot exhibited actin (upper band) and Bmal1 (lower band) proteins. Six samples on the left side were from control group and six on the right side were from Bmal1 siRNA group. (B) The ratio of Bmal1/actin protein levels from digitalized optical density in control mice and Bmal1 siRNA mice. Bmal1 protein level in the Bmal1 siRNA mice was significantly lower compared with that in the control mice. *: P < 0.05

3.2. Behavioral effects of Bmal1 KD

To evaluate the effect of Bmal1 KD on behavior, we performed wheel‐running activity, forced swim, and tail suspension tests. The purpose of performing the wheel‐running test was not to determine whether Bmal1 KD affects circadian rhythm, rather, to determine the “amplitude” of the rhythm, that is, the amount of activity. Thus, wheel‐running activity was recorded for only a few days under 12:12 hours light/dark cycles. In the baseline recording, activities in both groups exhibited similar activity levels in both the light and dark phases. However, activity in the Bmal1 KD group was substantially and significantly reduced in the dark phase immediately postinjection (Figure 2A). Two‐way ANOVA showed that group differences were significant (F = 83.599, P < 0.001). All pairwise multiple comparison procedures (Bonferroni's t test) showed that group differences were significant (t = 9.143, P < 0.05). As shown in Figure 2A, group differences were found in most dark phase data sections, and the mean of activity of each data section in the Bmal1 KD group in the dark phase was also significantly reduced as compared to the same data section at baseline (B1 and B2). The reduction in wheel‐running activity is further evidence that Bmal1 suppression was successful.

Figure 2.

Behavioral tests in mice injected with Bmal1 or negative control siRNA. (A) Effect of Bmal1 knockdown (KD) on wheel‐running activity. The Y‐axis depicts total activity and the x‐axis indicates testing day. T0‐T4 are treatment days. The arrow behind T0 indicates the time of siRNA injection. Running activity was dramatically decreased in mice treated with Bmal1 siRNA after baseline as compared to either the control group or the baseline value of the same group. (B) Moving and immobile times in the tail suspension test. Mice given Bmal1 siRNA had significant decreases in moving time and significant increases in immobility. *: P < 0.05, **: P < 0.01

The tail suspension test was conducted in a separate set of animals. The mean mobility time in the Bmal1 KD group was 59.27% less than in the control group; the mean time of immobility in the Bmal1 KD group was 94.57% longer than in the control group (Figure 2B). A t test showed that group differences were significant both for immobility (t = −3.521, two‐tailed P = 0.0034) and for time of mobility as well (t = 3.613 and two‐tailed P = 0.0028) indicating that Bmal1 KD increased immobility.

3.3. Wake/sleep changes after Bmal1 siRNA injection at ZT8

As the circadian cycle begins with the accumulation of Bmal1 protein in the cytoplasm, and both Bmal1 mRNA and protein levels rapidly increase between circadian time (CT, CT0 is defined as the onset of activity for a diurnal animal) 15‐CT3 in human and other diurnal species and between CT3 and CT15 in nocturnal species,4 the optimal time to suppress Bmal1 expression is during the upward phase of gene expression, that is, between CT3 and CT15 in mouse. According to the manufacturer's instructions, the effect of Bmal1 siRNA should occur within a few hours of brain delivery by ICV injection of RVG‐9R and should persist for several days. Thus, we performed ICV injection of siRNA in the middle of the upward slope, that is, at Zeitgeber Time (ZT) 8 (4 hours before the start of the dark phase) in B6 mice after 2 days of baseline PSG recording. Then, PSG recording continued for three more days. Control mice (n = 8) exhibited overall active behavior after a few minutes of rest. The PSG of this group exhibited normal sleep/wake features including low amplitude EEG and high amplitude EMG in wake, high amplitude EEG and low amplitude EMG in NREM sleep and low amplitude EEG (dominated by theta frequency) and low amplitude EMG during REM sleep.40 Interestingly, approximately 20‐30 minutes postinjection, the bodies of mice in the Bmal1 KD group (n = 7) were very soft, similar to muscle tone generally observed during REM sleep (muscle atonia); however, mice in the Bmal1 siRNA group were not sleeping, as we confirmed that their eyes were open and EMG amplitude was low. We observed this period of muscle atonia in most of the Bmal1 KD mice which lasted 10‐30 minutes. Therefore, we compared the features of the wake, REM sleep, and NREM sleep states at the baseline (Figure 3) with those after Bmal1 siRNA injection (Figure 4) from the same mouse (R3S2 DRC) in the same circadian phase. Both REM sleep and NREM sleep states did not greatly differ as measured by the EEG and EMG; however, the EMG during wakefulness was much lower in the post‐Bmal1 siRNA injection period. We termed this state “atypical wake” and scored it as wakefulness because (i) the mice's eyes were opening, (ii) EEG power during the post‐Bmal1 siRNA injection period was similar to EEG power in the baseline, and (iii) the muscle tone recovery recorded in the same epoch did not occur with large changes in the EEG recording (Figure 4). Not surprisingly, injection of both Bmal1 siRNA and control RNA induced changes in sleep/wake states. Compared with the same section on the baseline day, wakefulness and sleep in both groups were significantly altered on the first postinjection day. The overall circadian pattern of wake/sleep states was interrupted on the first postinjection day, but returned to normal on the second postinjection day (Figures 5 and 6). As indicated by a black “+” (change when compared with the same circadian time of the baseline day), wakefulness was significantly decreased in both the Bmal1 KD and the control group during the second 4‐hour section and only in the control group during the third 4‐hour section indicating that the effect was stronger in control mice than in the Bmal1 KD group (Figure 5: Wake). A similar but opposite trend was observed for total sleep (Figure 5: Total sleep). Changes in NREM sleep were similar to that of total sleep in the control group; however, NREM sleep in the Bmal1 KD group was significantly decreased in the second 4‐hour section postinjection, but increased during the next 4‐hour section. The subsequent increase was significant more compared to the baseline in the fourth 4‐hour section (Figure 5: NREM sleep). However, in the control group, REM sleep was immediately decreased after injection but recovered during the next 4‐hour section and surpassed baseline in the third 4‐hour section in the control group (Figure 5: REM sleep). Overall, ICV injection of both Bmal1 siRNA and control RNA induced significant changes in sleep/wake states. The effect was stronger in the control group than in Bmal1 KD group. One difference was that the effect on sleep in the Bmal1 KD group quickly reversed. Interestingly, the effect of injection resulted in significantly more wakefulness for approximately 6‐8 hours primarily during the dark phase in the Bmal1 KD group. The increase in wake was mostly compensated for decreases in NREM sleep (Figure 5). Then, the plotted lines of either REM sleep or NREM sleep in the Bmal1 siRNA group in Figure 5 crossed over that in control group; sleep was significantly increased during the dark phase on the following 2 days. Bmal1 KD mice displayed a reduced percentages of time spent awake during the dark period, but no overall change in overall percentage of time spent awake during the light period. The treated group also exhibited changes in REM and NREM sleep (data not shown). Statistical analysis revealed that significant differences existed between groups at all time points in REM, NREM, and total sleep as well as during the dark phase. Significant effects on sleep were not seen during the light phase, indicating that Bmal1 is a dark phase wake promoter. In summary, ICV injection of either Bmal1 siRNA or control siRNA induced dramatic changes in sleep/wake behavior during the first 16 hours and resulted in an interruption of the circadian pattern. Thereafter, Bmal1 KD enhanced sleep during the dark phase. Injection produced a short‐term decrease in total sleep, primarily in NREM sleep, within the first few hours; however, the consistent effect was a dramatic increase in total sleep or a decrease in wakefulness during the dark phase beginning day 2 following injection. Due to termination of the study, this effect remained on day 3.

Figure 3.

Screenshots of baseline polysomnographic (PSG) from mouse R3S2 DRC. Electroencephalogram (EEG) power spectrum and EEG and electromyogram (EMG) recording in computer scored Wake, nonrapid eye movement (NREM) sleep and rapid eye movement (REM) sleep were displayed in upper, middle, and lower section, respectively. The power spectrum displayed power frequency of the scored EEG channel. Both the EEG and EMG waveforms displayed typical features in wake (Upper section), NREM sleep (middle section), and REM sleep (lower section) states. During awake, EEG amplitude is low and the EMG amplitude is very high. During NREM sleep, the EMG amplitude is very low but EEG amplitude is high. During REM sleep, theta frequency is dominated in EEG and with the EMG, amplitude is also very low

Figure 4.

Screenshots of polysomnographic (PSG) from mouse R3S2 DRC in the post‐Bmal1 siRNA injection period. Results were similar to those described in Figure 3. On the upper section, we considered it as an atypical awakening, likely occurring because the amplitude of electromyogram (EMG) was much lower than the wake state EMG at baseline in the same animal (see the upper section of Figure 3). However, the amplitude and the power spectrum of the electroencephalogram (EEG) were similar to baseline (Figure 3). In addition, the EMG amplitude was significantly increased at later time point (indicated by arrow) indicating the termination of the effect of Bmal1 siRNA injection. Neither the features of EEG and EMG during nonrapid eye movement (NREM) sleep nor in rapid eye movement (REM) sleep were significantly different from those seen during the same states at the baseline of the same animal (Figure 3)

Figure 5.

The percentage of time spent awake vs sleep at baseline and after treatment with Bmal1 siRNA or control siRNA injected just before the light phase. Percentage of wake/sleep was averaged in each 3‐h section. The arrow indicates the time of injection 4 h prior to the dark phase. Grids with thick blue bars on the x‐axis indicate the dark period. Statistics result for group differences were *: P < 0.05; **: P < 0.01; ***: P < 0.001; for comparison with the same section of the baseline, +: P < 0.05; ++: P < 0.01; +++: P < 0.001. The upper level shows the percentage of wake, which increased beginning at just a few hours after injection. This was followed by a subsequent decrease in the last section of the dark phase and was consistently similar or below the level of the control group until the end of the experiment. Correspondingly, nonrapid eye movement (NREM) sleep was decreased in the Bmal1 siRNA group a few hours postinjection in the dark phase. Thereafter, the amount of NREM sleep in the Bmal1 knockdown (KD) group was below that of the control group at almost all time points. Significant differences were observed mostly during the dark phase and not the light phase, indicating that Bmal1 KD affects sleep during the dark phase but not the light phase. The Bmal1 KD group exhibited similar pattern of rapid eye movement (REM) sleep alteration as was observed tfor NREM sleep (initial increase a few hours postinjection but decreased thereafter until recording concluded). Thus, total sleep in the lower graph displayed a similar pattern for both REM and NREM sleep. Notably, all significant differences were recorded in the dark phase but not the light phase

Figure 6.

The percentage of time spent awake vs sleep at baseline and after treatment with Bmal1 or control siRNA during dark phase. The percentage of time spent awake or sleep was averaged in each 4‐h section. The arrow indicates the time of injection 4 h prior to light phase. Grids with thick blue bars in the x‐axis indicate the dark period. Statistical results for group differences were *: P < 0.05; **: P < 0.01; ***: P < 0.001; for comparison with the same section of the baseline, +: P < 0.05; ++: P < 0.01; +++: P < 0.001. The upper level shows that time spent awake was increased immediately after Bmal1 siRNA intracerebroventricular (ICV) injection, which was administered during the dark phase. Similar to Figure 3, time spent awake then fell below the level of the control group at about 6 h following injection. Subsequently, the percentage of time spent awake was consistently similar to or below the level of the control group. Correspondingly, nonrapid eye movement (NREM) sleep in the Bmal1 siRNA group exhibited an opposite pattern. Significant differences were seen mostly during the dark but not the light phase, indicating that Bmal1 knockdown (KD) affects sleep during the dark but not the light period. In the middle, the Bmal1 KD group exhibited dramatic reduction in rapid eye movement (REM) sleep immediately postinjection combined with a subsequent long‐lasting increase. Thereafter, time spent in REM sleep in the Bmal1 KD group was always greater than in the control group; thus, significance was achieved mostly during the dark period. The pattern of total sleep was similar to that shown in Figure 4. Differences included the following: (i) the decrease in total sleep occurred much sooner in this experiment when the injection was administered 4 h prior to the light phase; and (ii) group differences in total sleep on the last day were no longer significant

3.4. Effect of Bmal1 siRNA injection at ZT20

As Bmal1 expression is dynamically different in light and dark phases, we hypothesized that suppression of Bmal1 in a different phase would produce a different effect on sleep/wake patterns. Thus, we conducted similar study in a separate set of mice and performed ICVinjection of Bmal1 siRNA (n = 10) or vehicle (n = 8) 12‐hour offset from the previous part of the study, that is, at ZT20 (4 hours to light off). Similar to the ICV injection made in ZT8, ICV injection affected sleep in the control group. This may be because the injection was made during the dark phase, the effect of ICV injection on sleep occurred immediately as indicated by increases in NREM sleep and total sleep along with a decrease in wakefulness. However, the effect disappeared in control animals during the subsequent three 4‐hour sections, probably because these sections occurred during the light phase. Then, as soon as the dark phase resumed, a similar effect (increases in NREM sleep and total sleep and decrease in wakefulness in the control group) was observed for another 4 hours (Figure 6). In the Bmal1 KD group, REM sleep decreased only during the light phase followed by an increase in NREM sleep and total sleep and decrease in wakefulness during the subsequent dark phase (Figure 6). Because ICV injection induced a significant increase in NREM sleep in the control group along with a later decrease in both NREM and REM sleep in the Bmal1 KD group, an oval loop curve appeared in the first three 4‐hour sections across all four plotted variables (Figure 6). Compared with the control group, the data indicated that the Bmal1 KD group had significantly less NREM, REM, and total sleep and more wakefulness during the loop. Thereafter, the pattern of change in sleep/wake state was similar to that of Bmal1 KD at ZT8, that is, sleep was dramatically increased during the dark phase but absolutely no effect on sleep was observed in the light phase (Figure 6). This type of group difference was significant in multiple time sections during the first, second, and third dark phases and during a single time section in the fourth dark phase; no changes were observed during the last (fifth) dark phase, indicating that the effect of siRNA injection lasted only 4‐5 days. These data indicate that the effect of injection made in ZT20 on sleep/wake states differed slightly from effects of injections made in ZT8 during the first 24 hours, but did not differ in terms of later effects, suggesting that the later effects were due solely to Bmal1 siRNA knockdown. These data further support the conclusion that deficiency in Bmal1 is necessary for maintaining normal amount of wakefulness during the active phase in mice.

3.5. Brain orexin A but not CRH was increased 6 hours after Bmal1 KD

Given the data above indicating that REM and NREM sleep were dramatically and significantly altered between treatment groups for either a few hours (ZT8 groups) or immediately after treatment (ZT20 groups), we concluded that the change observed was either an increase in time awake or a decrease in sleep, and that wake promoting or stress regulation molecules might be involved. Thus, we quantified tissue content of orexin A and CRH in the hypothalamus in a separate set of mice treated with Bmal1 siRNA or control siRNA. The treatment was completed in ZT8 and mice were euthanized 6‐hour postinjection, that is, in the middle of the period of increased wakefulness. Interestingly, orexin A levels were significantly higher in the Bmal1 KD group (57.285 ± 5.596 pg/mg, n = 7) as compared to the control group (41.664 ± 2.476 pg/mg, n = 8). A t test showed that this difference was statistically significant (t = 2.427, P = 0.0305), indicating that changes in orexin A may be involved in the response to Bmal1 siRNA ICV injection. By contrast, the difference in CRH expression between groups was not significant (t = 0.687, P = 0.507), suggesting that CRH is not involved in the increase of wakefulness/decrease in sleep induced by Bmal1 siRNA injection (Figure 7A‐B). Although sleep reduction in the control group was the major change observed during this period, differences in orexin levels might be interpreted as reduction of orexin A in the control group instead of increased orexin levels in the Bmal1 KD group. However, this interpretation needs further support by comparison with baseline orexin level.

Figure 7.

Hypothalamic content of orexin A (A) and corticotrophin‐releasing hormone (CRH) (B) in mice euthanized 6 h after Bmal1 or control siRNA injection. Mean orexin A level was significantly higher in Bmal1 siRNA‐treated mice as compared with control mice; however, the difference in CRH was not significant. Hypothalamic content of orexin A (C) and CRH (D) in mice euthanized 24 h after Bmal1 or control siRNA injection. Neither orexin A nor CRH level was significantly different between groups. *: P < 0.05

3.6. Orexin A and CRH remained unchanged 24 hours after Bmal1 KD

Following the first type of wake/sleep alteration on the first day after injection, the pattern of sleep/wake progressed to an increase in both REM and NREM sleep during the dark phase, but not in the light phase indicating suppression of time spent awake. We again analyzed hypothalamic tissue content for orexin A and CRH levels in animals euthanized 24 hours after ICV injection of Bmal1 siRNA and control siRNA. Tissue collection was performed when mice exhibited a large change in time awake according to sleep data in Figure 5. Interestingly, mean orexin A levels were very similar between the two groups (74.542 ± 8.641 pg/mg in the Bmal1 KD group and 78.753 ± 17.178 pg/mg in the control group); these were not statistically significant (t = −0.237, P = 0.817). Similarly, the mean CRH level was 15.855 ± 2.512 pg/mg in the Bmal1 KD group and 14.894 ± 3.402 pg/mg in the control group (Figure 7C‐D). A Student's t test showed that the difference between these means was not significant (t = 0.232, P = 0.820). These results indicate that neither orexin A nor CRH is likely to be involved in the alteration of sleep/wake changes induced by Bmal1 KD.

3.7. GAD was unchanged in the hypothalamus by Bmal1 KD

As GABA is a major inhibitory neurotransmitter, and GABAergic neurons in the hypothalamus play an important role in the regulation of sleep generation and maintenance [51], we expected an alteration in GABAergic transmission in Bmal1 KD mice because the total sleep and total time awake was altered significantly and consistently for 24 hours following Bmal1 siRNA injection. GAD is an enzyme that synthesizes GABA and alteration of GAD levels often indicates a change in GABA levels; therefore, we performed Western blotting on samples from the hypothalamus in mice euthanized 24 hours following brain injection of either control siRNA (n = 10) or Bmal1 siRNA(n = 10). Surprisingly, the t test indicated that the power of the comparison was 0.744, below the desired power of 0.800. In addition, GAD levels were not significantly altered, as shown in Figure 8, indicating that GABA in the hypothalamus was unaffected. These data suggest that GABA is not involved in the alteration of sleep induced by Bmal1 siRNA.

Figure 8.

Hypothalamic levels of glutamate decarboxylase (GAD) in mice treated with control siRNA and Bmal1 siRNA. (A) Western blot of GAD and actin in the hypothalamus from Bmal1 knockdown (KD) or control mice. Visually, optical density in each individual band was not obviously different. (B) ratios of GAD/actin in digitalized optical density in each sample (no significant differences)

4. DISCUSSION

Our work investigated the effect of adult suppression of Bmal1 expression via Bmal1 siRNA on circadian amplitude, depressive signs, sleep/wake states, and brain levels of orexin A and GAD. We first confirmed that Bmal1 protein levels in the SCN were significantly reduced 24 hours after Bmal1 siRNA was administered. The reduction in Bmal1 was also supported by a reduction in wheel‐running activity as a functional evaluation. AS both the measurement of wheel‐running activity and tissue collection for Western blot analysis occurred 24 hours after Bmal1 siRNA administration, these data indicate that Bmal1 expression was reduced at early as 24 hours after injection but could have occurred even earlier.

We hypothesized that the mechanisms of Bmal1 deficiency‐induced wake/sleep alterations might be mediated through alterations in orexin neurons, CRH neurons, and GABA neurons. As Bmal1 KO induced severe global effects, including physical size and longevity,47, 48 we introduced Bmal1 deficiency via adult injection of Bmal1 siRNA. Our siRNA treatment was shown to be efficient because the Bmal1 KD group exhibited reduced Bmal1 expression at both the mRNA (data were not shown) and protein level in the SCN (Figure 1). We also observed a dramatic and significant decrease in wheel‐running activity, an indication of reduced circadian amplitude, which was further evidence for a reduction of Bmal1 as well as increased immobility in a tail suspension test in the Bmal1 KD group as compared to the control group. These data are consistent with previous findings that Bmal1 KO mice exhibit a significant reduction in skeletal muscle force with an accompanying reduction in mitochondrial volume and disrupted myofilament architecture associated with decreased expression of actin, myosin, and Myo D target genes.39 Because brain ICV of Bmal1 siRNA affects only those tissues that can be reached by CSH, muscle and other type of cells in the peripheral organs should not be affected. However, results from the wheel‐running test support the view that Bmal1 KD in this study had effects sufficiently impacted Bmal1's regulation of sleep patterns.

In our sleep study, two types of sleep alterations were observed and were clearly distinguishable from the circadian pattern displayed in Figures 5 and 6. First, ICV injection of both Bmal1 KD and control siRNA induced early sleep/wake changes that occurred on the first day or within 24 hours of treatment. When the injection occurred in the last 4 hours of the light phase, the immediate effect was an increase in REM sleep followed by an increase in NREM sleep in both groups; however, the increase was greater in the control group. The difference in increasing sleep levels resulting in less total sleep and more time spent awake in the Bmal1 KD group (Figure 5). When the injection of both Bmal1 siRNA and control RNA was made at ZT20, that is, 4 hours under light conditions, the control group exhibited an immediate and dramatic increase in NREM sleep (in the dark phase). This was followed by a dramatic decrease in REM sleep in the Bmal1 KD group (in all three 4‐hour sections in the light phase) and subsequently, produced group differences as if an increase in time spent awake had just occurred (Figure 6). Compared with the measured variables in the same section of baseline, sleep changes were much stronger in the control group than in the Bmal1 KD group (Figures 5 and 6). Interestingly, when an injection was made in the light phase (4 hours before the light was turned off), NREM sleep increased in the control group 4 hours later (during the dark phase). However, when the injection was made in the dark phase (4 hours before the light was turned on), an increase in NREM sleep was observed immediately. These data indicate that the effect of control siRNA on sleep requires the physiological environment provided by the dark phase and suggests that this reaction may require neuronal systems involved in stress and wake regulation. Thus, we measured brain levels of orexin A and CRH in separate set of mice treated with either control siRNA or Bmal1 siRNA. All animals were euthanized in the middle of the period of increased wakefulness (6‐hour postinjection). Interestingly, we found that orexin A levels were significantly higher in the Bmal1 KD group, but CRH remained unchanged. This observation indicates that the observed sleep/wake decrease in the early phase is associated with a change in orexin but not CRH. We suspect that control siRNA induced reduction of brain orexins, but the mechanism remains to be further discovered.

A second type of sleep alteration occurring after ICV injection of Bmal1 siRNA or control RNA was an increase in sleep and decrease in wakefulness. These effects occurred only in the dark phase and were not accompanied by alterations of either orexin A or CRH (Figure 7). This effect was secondary and occurred after the first effect on wakefulness occurred completion of the first type of effect and persisted 3‐4 days (until study conclusion). This effect on sleep is consistent with previously reported sleep data recorded in Bmal1 KO mice,9 suggesting that the long‐lasting effect of Bmal1 gene deficiency on sleep/wake regulation is actually a dysregulation of wake initiation and maintenance of the dark phase in nocturnal rodents. This result strongly suggests that Bmal1 regulates wakefulness rather than sleep aside from any effect on basic physiology and circadian rhythm. Therefore, we measured hypothalamic orexin and CRH levels in mice treated with control and Bmal1 siRNA. Surprisingly, neither orexin A nor CRH expression levels were changed in the Bmal1 KD group as compared to the control group. This indicates that a deficiency in Bmal1 induces wake/sleep changes through a mechanism that does not include orexin regulation or the CRH system. As GABA plays an important role in the generation and maintenance of NREM sleep31, 49; in fact, GABA release is significantly higher during sleep.17 As an increase in sleep need corresponds with an increase the expression level of GAD,46 an enzyme of GABA synthesis, we measured the tissue levels of GAD. We found that GAD levels were unaltered in the hypothalamus, suggesting that the GABAergic system was not involved in the dramatic increase in total sleep beginning at 24 hours after Bmal1 siRNA injection.

Circadian rhythm‐associated changes in synaptic density in orexinergic axons have been demonstrated in living fish.50 However, findings from hamsters showed that Per1 expression in the area of orexin‐producing neurons did not differ following constant dark treatment and that neither the total number of orexin cells or the total number of orexin cells expressing Per1 in the lateral hypothalamus/dorsal lateral hypothalamus show circadian differences in hamsters [52]. As both mRNA and protein levels of the circadian triggers Bmal1 and Clock have regular circadian rhythm‐influenced expression, expression of orexin suggests that orexinergic neurons do not participate in circadian regulation of wake/sleep rhythms. Our data demonstrating that Bmal1 KD does not affect brain levels of orexin, CRH, or GAD indicates that neither orexinergic/CRH nor the GABAergic system is involved in Bmal1 KD‐induced wake reduction during the dark phase.

In summary, Bmal1 KD reduced wheel‐running activity and increased immobility. In PSG recordings, injection of both Bmal1 siRNA and control RNA induced short‐term sleep/wake changes, but the control siRNA induced effect was much stronger. The short‐term effect was associated with brain orexin level changes. At 24 hours following injection, the major effect was an increase in sleep during the dark phase in the Bmal1 KD group. This type of change was consistent with previous findings from the Bmal1 KO model and was associated with no changes in brain levels of orexin, CRH and GAD, suggesting that the mechanism of the circadian regulation of wake/sleep does not include these types of neurons. The limitation of the paper is that the mechanism of Bmal1 KD induced both acute wake increase and later sleep increase remains to be discovered.

Akladious A, Azzam S, Hu Y, Feng P. Bmal1 knockdown suppresses wake and increases immobility without altering orexin A, corticotrophin‐releasing hormone, or glutamate decarboxylase. CNS Neurosci Ther. 2018;24:549–563. 10.1111/cns.12815