Glucocorticoid receptors in the locus coeruleus mediate sleep disorders caused by repeated corticosterone treatment

Stress induced constant increase of cortisol level may lead to sleep disorder, but the mechanism remains unclear. Here we described a novel model to investigate stress mimicked sleep disorders induced by repetitive administration of corticosterone (CORT). After 7 days treatment of CORT, rats showed significant sleep disturbance, meanwhile, the glucocorticoid receptor (GR) level was notably lowered in locus coeruleus (LC). We further discovered the activation of noradrenergic neuron in LC, the suppression of GABAergic neuron in ventrolateral preoptic area (VLPO), the remarkable elevation of norepinephrine in LC, VLPO and hypothalamus, as well as increase of tyrosine hydroxylase in LC and decrease of glutamic acid decarboxylase in VLPO after CORT treatment. Microinjection of GR antagonist RU486 into LC reversed the CORT-induced sleep changes. These results suggest that GR in LC may play a key role in stress-related sleep disorders and support the hypothesis that repeated CORT treatment may decrease GR levels and induce the activation of noradrenergic neurons in LC, consequently inhibit GABAergic neurons in VLPO and result in sleep disorders. Our findings provide novel insights into the effect of stress-inducing agent CORT on sleep and GRs' role in sleep regulation.

sleep changes were different. Thus, the unknown mechanism of repeated stress-induced sleep disorders needs further investigation.
To better understand the underlying mechanism of long-term treatment of glucocoritcoids induced sleep disorders, we repeatedly treated rats with CORT for 7 days to mimic stress, and these animals were then evaluated with regard to sleep parameters. To correlate such behavior with possible underlying mechanisms, we measured the activity of sleep-wake-regulating nuclei, major monoamine neurotransmitters in related brain areas, GR and MR expression in LC, glutamic acid decarboxylase (GAD) in VLPO, and tyrosine hydroxylase (TH) in LC. The levels of GRs decreased in LC after 7-day CORT treatment, prompting us to test whether administration of the GR antagonist RU486 can reverse stress-related sleep disturbances.

Results
Repeated administration of CORT induced sleep disorders in rats. Table 1 showed that there were significant reduction of total sleep time (TST, F 1, 16  Repeated administration of CORT augmented neuronal activity in LC and suppressed neuronal activity in VLPO. To identify alterations in sleep-wake-related nuclei, we performed doublestaining immunohistochemistry in the VLPO, Pef, TMN, PPT, DRN, LDT, and LC. After 7 days CORT treatment (40 mg/kg, s.c.), the ratio of Fos1 and GAD1 neurons in the VLPO (Fig. 1c) was significantly decreased (t 10 5 2.694, p 5 0.023), whereas the ratio of Fos1 and TH1 neurons in the LC (Fig. 1f) was significantly increased (t 9 5 22.803, p 5 0.021). We did not detect noticeable differences (Fig. 2) in the Pef (t 10  GR protein level in LC was lowered after repeated administration of CORT. To determine whether the increase in noradrenergic neuron activity in the LC was directly attributable to CORT administration, we analyzed total GR protein levels and GR distribution in the cytoplasm/nucleus and MR protein levels in the LC. Acute CORT administration (1-day CORT administration [CORT 1D group]; i.e., acute CORT administration on day 7 proceeded by 6 days of vehicle administration) was used as another control.
After 7 days of CORT treatment (40 mg/kg, s.c.), total GR protein levels in the LC (Fig. 3a) significantly decreased (F 2,11 5 8.489, p 5 0.019). Specifically, GR protein levels in the cytoplasm ( Fig. 3b; F 2,11 5 6.894, p 5 0.015) and nucleus ( Fig. 3c; F 2,11 5 10.760, p 5 0.004) significantly decreased. The normal distribution of GRs after acute exposure to CORT transferred from the cytoplasm to the nucleus, similar to the CORT 1D group ( Fig. 3d; F 2,11 5 4.180, p 5 0.052). However, no difference was found between the chronic CORTtreated group (CORT 7D) and vehicle group (Fig. 3d). No significant changes in MRs (F 2,8 5 0.226, p 5 0.804) were observed in the LC (Fig. 3f). These results suggest that GRs in the LC might be mainly involved in mediating the sleep disturbances caused by CORT. The method of separately extracting protein from the cytoplasm and nucleus was verified in Fig. 3e.
Because the highest density of GR expression was found in the LC and not in the VLPO 9 , we predicted that changes in GABAergic neurons in the VLPO may be attributable to the noradrenergic projection from the LC, which inhibited GABAergic neuron activity (see the HPLC results and discussion below).
Repeated administration of CORT increased TH protein in LC and decreased GAD protein in VLPO. To validate that repeated GR activation activates TH, Western blot was performed to further verify the changes in total TH and GAD protein levels in the LC and VLPO, respectively. A control group (CORT 1D) was included that received acute CORT administration on day 7 proceeded by 6 days of vehicle administration. After 7 days CORT treatment (40 mg/kg, s.c.), TH protein levels in the LC (Fig. 5a) significantly increased (F 2,11 5 5.771, p 5 0.024). The protein levels of GAD (Fig. 5b) were significantly decreased in the VLPO (F 2,8 5 8.077, p 5 0.020).
Levels of NE and DA in VLPO, LC, hypothalamus and prefrontal cortex after repeated administration of CORT. The results reported above indicate that the noradrenergic system was activated, and we further detected norepinephrine and dopamine levels in related brain areas. The results showed that the levels of norepine-     phrine in the VLPO (Fig. 6a) were significantly elevated in the CORT-treated group compared with the vehicle group (F 1,11 5 5.046, p 5 0.048), confirming our hypothesis that increases in norepinephrine in the VLPO inhibit GABAergic neurons. Corticosterone treatment did not alter dopamine levels in the VLPO (F 1,11 5 2.837, p 5 0.123). In the LC (Fig. 6b), norepinephrine levels were elevated in the CORT-treated group (F 1,10 5 8.175, p 5 0.019), whereas dopamine levels were unaltered (F 1,10 5 3.726, p 5 0.086). The LC is the major source of noradrenergic neurons that sustain a waking state. The noradrenergic projection that passes through the hypothalamus to the cortex is important in the sleep-wake cycle. To verify that noradrenergic neurons in the LC were activated, we detected norepinephrine and dopamine levels in the hypothalamus and prefrontal cortex. The results indicated that the levels of norepinephrine (F 1,11 5 10.004, p 5 0.010) and dopamine (F 1,11 5 13.597, p 5 0.004) in the hypothalamus were significantly increased ( Fig. 6c) after CORT treatment (40 mg/kg, s.c.). The increases in norepinephrine (F 1,9 5 1.569, p 5 0.246) and dopamine (F 1,9 5 0.815, p 5 0.393) in the PFC did not reach statistical significance (Fig. 6d).

Discussion
In the present study, we used repeated CORT treatment to induce relatively stable and high CORT levels (data not shown) to mimic stress-induced sleep disorders. Under this dosing regimen, we found that rats exhibited significant altered sleep parameters. Immunohistochemistry indicated that repeated CORT administration activated noradrenergic neurons in LC and suppressed GABAergic neurons in VLPO. The activity of other sleep-wake-regulating nuclei, such as the Pef, TMN, PPT, DRN and LDT, were unaffected. The sleep-wake-regulating pathway in the brain 12 is based on the hypothesis that VLPO is the primary sleep-promoting nucleus that inhibits the TMN, DRN and LC. Insomnia that results from acute stress is related to altered activity in VLPO, TMN and LC 15 . Consistent with this hypothesis, our results suggest that repeated CORT treatment induces insomnia by activating LC and suppressing VLPO.    Lussier et al 19 reported that 14 and 21 days injection of CORT (40 mg/kg) produced clear behavioral changes in the forced swimming test, but 7 days injection of CORT (40 mg/kg) did not show significant immobility compared to vehicle rats. They also proposed that treatment with CORT increases depression-like behavior gradually over a 21-day period and the development of depression-like behavior in CORT treated rats occurs in a linear manner. Our previous report also indicated that after 7 days treatment of CORT (40 mg/kg) rats showed higher level of immobility and lower levels of sucrose preference than did the vehicle rats, but the changes did not reach statistical significance 20 . Based on these reports, it could be presumed that administration of CORT (40 mg/kg) for 7 days may cause the stress-induced abnormal sleep pattern.
Repeated exposure to stress is usually regarded as the main cause of sleep disorders, such as insomnia 21 . Many human studies have linked prolonged stress to self-reported impairments in sleep. For example, sustained stress caused by academic demands 4 or work shifts 5 increases corticoid levels and induces sleep disorders, including poor sleep quality, prolonged sleep latency, a shorter sleep duration and disinhibition of REM sleep. In other situations, e.g., depression 22 , mania 23 , schizophrenia 24 , bipolar disorder 25 or who receive high-dose glucocorticoid treatment for multiple sclerosis 26,27 , the sleep disruptions are also observed. Notably, stress-induced sleep disorders are always associated with an increased risk for other psychiatric problems, such as depression 20,28 or panic disorder 29 . The aforementioned evidence indicates that a prolonged increase in corticoid levels caused by stress, disease, or high-dose corticoid treatment can result in sleep disorders. The sleep disorder characteristics of our model in the present study are consistent with previous reports that used other stress-inducing methods 30,31 . Thus, the present model may be used to study the underlying mechanism of sleep disorders caused by sustained elevations of corticoid levels.
Corticosteroids are important mediators of homeostasis and stress and exert their complex effects via two receptors, MRs and GRs. Both receptors are expressed in neurons and glia cells in the mammalian brain and contribute to the regulation of brain-cell properties in a slow and persistent manner by modifying the transcription of response genes and de novo protein synthesis 1 . The noradrenergic LC abundantly expresses GRs and MRs 9,10 . The LC has been found to be a highly stress-reactive nucleus that is capable of prolonged activation and modulating the central stress response 32,33 . To confirm whether GRs or MRs in LC participate in sleep disorders induced by repeated CORT administration, we performed two additional experiments. The levels of GRs but not MRs were decreased in LC after 7-day CORT treatment. These results are consistent with the study by Makino et al. 34 , which showed that systemic CORT treat- www.nature.com/scientificreports SCIENTIFIC REPORTS | 5 : 9442 | DOI: 10.1038/srep09442 ment for 1 or 2 weeks decreased GR mRNA in LC. We also directly tested the hypothesis that repeated CORT administration induces sleep disorders by activating GRs in LC. We injected the GR antagonist RU486 directly into LC and as predicted, this treatment resulted in the restoration of normal sleep pattern in CORT-treated rats. Our results are consistent with other studies reported that GR antagonist exerted a sleep-restoring effect in chronic insomnia patients 35 but not in healthy volunteers 36,37 .
Glucocorticoid receptors, after binding glucocorticoids in the cytoplasm, are translocated to the nucleus and bind glucocorticoid response elements (GREs) to activate or repress gene transcription 38,39 . Markey et al. 40 indicated that noradrenergic neurons in LC might be target cells for glucocorticoids, meaning that the effect of glucocorticoids on TH might be receptor-mediated. Indeed, excessive GR activation by repeated stress increases TH mRNA 34,41 and TH immunoreactivity 40,42 in LC. Additionally, other studies found a decrease in GAD protein levels after repeated CORT administration in the hippocampus and amygdala 43 . In the present study, we found that repeated CORT administration increased TH levels and TH immunoreactivity in LC as well as decreased GAD levels and GAD immunoreactivity in VLPO.
Stress-induced increases in arousal are related to the noradrenergic system in the brain 44 . The noradrenergic projection from LC is known to promote the waking state in the hypothalamus 45 . Reinforced noradrenergic activation via a1-adrenoceptors in the prefrontal cortex in rats has been found to enhance cognitive performance, and this was interpreted as resulting from an increase in arousal 46 . In the present study, the increase in noradrenergic activity in LC significantly elevated norepinephrine levels in hypothalamus and VLPO. These two areas are critical for sleep-wake states and neuropsychological regulation 46 . Norepinephrine and dopamine levels were elevated in VLPO, suggesting that our hypothesis of decreased activity in VLPO might be secondary to the activation of noradrenergic neurons in LC.
Presumed sleep-promoting neurons are GABAergic, galaninergic, and multi-polar triangular-shaped and have a potent low-threshold calcium potential 47 . These neurons are always inhibited by norepinephrine. Other studies have reported that postsynaptic a 2 -adrenoceptors mediate this effect 48 . When VLPO neurons begin to fire at sleep onset, they inhibit wake-promoting neurons, allowing for their own disinhibition and reinforced firing. Conversely, during arousal, wake-promoting neurons fire at a high rate, thus inhibiting VLPO neurons. The reciprocal inhibitory interaction between these systems provides a mechanism for the maintenance of one of two stable configurations. Accordingly, the disruption of wake-and sleep-promoting pathways results in behavioral instability caused by destabilization of reciprocal inhibitory interactions. The functional properties of the ''reciprocal inhibitory interaction'' model support the production of a stable waking state by a simple neuronal network and important resistance to behavioral switching by limiting inappropriate changes when inputs to VLPO or wake-promoting areas fluctuation. Our results showed that increased LC activity suppressed VLPO activation, and eventually inhibited the transition from wake to sleep, which was represented by prolonged the sleep latency and decreased sleep time.
The present study showed that repeated CORT treatment significantly increased REM sleep. Previous reports found no changes 18 or reduction of REM sleep 7,49 produced by acute administration of glucocorticoids. These discrepancies may due to the different doses and duration of glucocorticoids treatment, as well as different sleep recording period and species.
In conclusion, our findings indicate that GRs in LC may play an important role in chronically elevated CORT-induced sleep disorders through the activation of noradrenergic LC neurons. Increased TH and norepinephrine levels led to the suppression of GABAergic VLPO neurons and consequently sleep disorders. A GR antagonist, which has been previously reported to have antidepressant effect in animal studies 50 , also has important effect on sleep-wake regulation, particularly when circulating corticoids were above physiological levels, and may prompt the development of therapeutic strategies for stress-related sleep disorders. However, the underlying mechanism of the ability of GRs to regulate sleep requires further investigation.

Methods
Animals and treatment. Male Sprague-Dawley rats (270-300 g, Grade I, purchased from the Animal Center of Peking University, Beijing, China) were used. All experiments were conducted in accordance with the European Community guidelines for the use of experimental animals and approved by the Peking University Committee on Animal Care and Use. The rats were individually housed in acrylfiber cages under a 12 h/12 h light/dark cycle (lights on at 9:00 AM) and had ad libitum access to food and water. The ambient temperature was 23 6 1uC, and the relative humidity was 50 6 10%.
Drugs and drug administration. According to the previous studies 51,52 , rats were injected subcutaneously with 40 mg/kg CORT (corticosterone 21-acetate, Sigma, St. Louis, MO, USA) dissolved in saline with 2% Tween 80 (Sigma, St. Louis, MO, USA) continuously for 7 days at 9:00 AM. In the microinjection experiment, RU486 (Sigma, St. Louis, MO, USA) was used as GR antagonist as previously reported 53 and microinjected into LC in a volume of 0.25 ml/side 30 min before CORT administration every day at doses of 100 ng/side and 250 ng/side by dissolving in saline with 50% ethanol.
Surgery and sleep analysis. At least 1 week prior to sleep recordings, the animals (9 for vehicle and 8 for CORT group) underwent surgery using previously described standard procedures 54 . Briefly, under chloral hydrate (300 mg/kg) anesthesia, two stainless steel screws attached to insulated wire were implanted in the skull over the frontal-parietal cortex for electroencephalography (EEG). One of the screws was placed approximately 2 mm anterior and 2 mm to the right of bregma, and the other screw was placed approximately 3 mm posterior and 2 mm to the left of bregma. A ground electrode was placed between the two screws, 3 mm lateral to the midline. For electromyography (EMG), a pair of wire electrodes was threaded through the nuchal muscles. These electrodes were attached to a miniature connector that was fixed to the skull with dental acrylic. After surgery, the rats were injected with penicillin for 3 days and allowed to recover for 7 days prior to the experiments. For habituation, they were connected to the recording apparatus at least 1 day before the sleep recording.
For the electrophysiological recordings, all rats were placed in an electrically shielded box in a noise-attenuated environment with a light-weight shielded cable plugged into the connector on the rat's head and attached to a counterbalanced swivel. The signals were routed to an electroencephalograph (model MP 150, BIOPAC Systems, CA, USA). Recordings were performed for 6 h, beginning at 9:00 AM, immediately after the last CORT injection. The signals were amplified and filtered (EEG, 0.5-30 Hz; EMG, 10-100 Hz) and then digitized at a sampling rate of 128 Hz and recorded using AcqKnowledge software (BIOPAC Systems). The EEG/EMG recordings were subjected to manual scoring using SleepSign 2.0 software (BIOPAC Systems), with the following criteria 55 : wakefulness (low-amplitude EEG activity and high-voltage EMG activity), rapid-eye-movement (REM) sleep (Fast-fourier transform [FFT] theta ratio of EEG $ 60%, desynchronized EEG, absence of tonic EMG, and occasional body twitches while maintaining a recumbent sleep posture), slowwave sleep (SWS; FFT delta ratio of EEG $ 70%, large-amplitude, synchronous EEG with sleep spindles present, greatly diminished tonic EMG, eyes closed and recumbent posture), and light sleep (FFT delta ratio of EEG , 70%, high-amplitude slow or spindle EEG activity, and low-amplitude EMG activity). Non REM (NREM) sleep time was equal to SWS time 1 light sleep time. The total sleep time was equal to NREM sleep time 1 REM sleep time.
Tissue preparation and immunostaining. Another set of animals were perfused through the ascending aorta with 0.9% NaCl solution 1 h after the last injection of CORT on day 7, followed by 4% paraformaldehyde in 0.1 M phosphate buffer 56 . Whole brains were immediately removed and post-fixed in the same fixative at 4uC for 24 h and then immersed in 30% sucrose in 0.01 M phosphate-buffered saline (PBS; pH 7.4) at 4uC for cryoprotection. Prior to tissue sectioning, the brains were blocked at approximately 11.0 mm with reference to bregma, rapidly frozen in liquid n-hexane that was cooled with a mixture of solid carbon dioxide and ethanol, and stored at 220uC.
Coronal sections that encompassed VLPO, Pef, TMN, PPT, DRN, LDT, and LC were freeze-cut at 20 mm thicknesses. The anatomical nomenclature (based on Paxinos and Watson, 1998) and protocol are summarized in Table 2. Each section was first immunostained for nucleus-specific neurotransmitter markers (Table 2) and then processed for Fos staining. double-stained cell ratio (i.e., the number of Fos1/neuron1 cells divided by the number Fos6/neuron1 cells, multiplied by 100%) was compared in the different treatment groups. Black immunostaining was counted as Fos1 nuclei, and brown staining throughout the cytoplasm was counted as neuron1 cells. Cells with brown staining of the cytoplasm and black staining of nuclei were counted as doubleimmunostained cells.
Western blot. Based on the previous studies 57, 58 , VLPO and LC tissue (with some of their surrounding tissue) were punched (1 mm diameter for VLPO and 2 mm diameter for LC) bilaterally using brain matrix guided by the rat brain atlas of Paxinos and Watson (1998). VLPO and LC proteins were pooled, extracted, boiled in 1% sodium dodecyl sulfate (SDS) solution, and quantified using a BCA assay kit (Pierce, Rockford, IL, USA) with bovine serum albumin as the standard. Equal amounts of protein (25 mg) were separated by 10% SDS-polyacrylamide gel electrophoresis (PAGE) and transferred to polyvinylidene difluoride membranes (Millipore, MA, USA). The membranes were blocked with 5% non-fat milk for 1 h at room temperature and incubated with primary antibodies, including anti-b-actin (151000; Abmart, Shanghai, China), anti-histone H3 (152000; Cell Signaling Technology, MA, USA), anti-GAD (151000; Chemicon Millipore, MA, USA), anti-TH (151000; Santa Cruz Biotechnology, Santa Cruz, CA, USA), anti-GR (151000; Santa Cruz Biotechnology, Santa Cruz, CA, USA), and anti-MR (15100; Santa Cruz Biotechnology, Santa Cruz, CA, USA) in TBS-T buffer (Tris-buffered saline 10.1% Tween-20) at 4uC overnight. After 3 3 10 min TBS-T washes, the blots were incubated with horseradish peroxidase-conjugated secondary antibodies (151000; Abmart Shanghai, China) for 2 h at 37uC and then washed with TBS-T buffer for 3 3 10 min. The blots were then treated with an enhanced chemiluminescence detection kit (CoWin Bioscience, Beijing, China). Western blot bands were scanned with a GelDoc XR System (Bio-Rad, Hercules, CA, USA) and subsequently analyzed densitometrically using Image Lab software. The results were normalized to the protein expression level of b-actin or histone in each sample.
Microinjection into locus coeruleus. Together with the EEG/EMG surgery described above, additional steps were performed as described below. Rats were anesthetized and positioned in a stereotaxic instrument. A double guide cannula (26 gauge, C/C dist. 2.4 mm, Plastics One, Roanoke, VA) was implanted with the tip 1 mm above the LC (anterior/posterior, 29.8 mm; lateral, 61.2 mm; dorsal/ventral, 26.0 mm below the brain surface). The microinjections were performed with a Hamilton syringes that were connected to the 33 gauge injection cannula (Plastics One, Roanoke, VA). Drug or vehicle was delivered through the injection cannula that extended 1 mm beyond the guide in a 0.25 ml volume over a 2-min period. The injection cannula was kept in place for another 2 min to allow the drug to completely diffuse from the tip. At the end of the experiments, cannula placements were defined histologically using Nissl staining with thickness of 20 mm under light microscopy. The correctness of the cannula/injection sites was assessed according to the atlas of Paxinos and Watson (1998). All the data presented in this report are derived from animals whose injection site was within the limits of the LC. The location of injection tip is showed in Fig. 4a.
High-performance liquid chromatography analtsis. According to previous report 59 , the VLPO, LC, hypothalamus, and prefrontal cortex were dissected and extracted with 0.2 M perchloric acid by ultrasonic homogenation. The mixture was centrifuged twice at 14,000 rpm for 50 min at 4uC. The levels of norepinephrine and dopamine were detected by high-performance liquid chromatography with electrochemical detection (HPLC-ECD; Dionex) using a C18 reverse-phase column (3.0 mm inner diameter, 75 mm length; Capcell Pak C18 MG S3, Shiseido). The mobile phase of the HPLC system consisted of 0.85 mM sodium octylsulfate, 0.5 mM ethylenediamine tetraacetic acid, and 0.1 mM NaH 2 PO 4 in 11% methanol, pH 3.25. The flow-rate was set at 0.6 ml/min. The electrochemical detector was set with an oxidizing potential of 200 mV and reducing potential of 2175 mV.
Statistical analysis. The data were analyzed using SPSS software 17.0 (SPSS Inc., Chicago, IL, USA) and are expressed as mean 6 SEM. The biochemical data for comparisons between the vehicle-and CORT-treated groups were analyzed using unpaired Student's t-test. One-way analysis of variance followed by the Student-Newman-Keuls post hoc test was used for multiple comparisons. In all of the tests, p , 0.05 was considered statistically significant.