Sleep Disruption Aggravates Focal Cerebral Ischemia in the Rat

Abstract

Study Objectives:

Sleep changes are frequent in stroke patients and predict a poor outcome. It remains unclear how sleep influences stroke evolution and recovery. We assessed effects of sleep disruption on brain damage and on the expression of axon sprouting genes after focal cerebral ischemia in rats.

Design:

12 h after ischemia induced by occlusion of the middle cerebral artery, rats were subjected to sleep disruption including sleep deprivation for 12h (SDpv12h) and sleep disturbances (SDis) by SDpv12h for consecutive 3 days. Control groups included ischemia without SDpv12h or SDis, sham surgery plus SDis and sham surgery without SDis. Sleep changes were evaluated based on EEG and EMG recordings.

Measurements and Results:

SDpv12h increased the infarct volume by 40% (SDpv12h 82.8 ± 10.9 vs. control 59.2 ± 13.9 mm³, P = 0.008) and SDis by 76% (SDis 58.8 ± 20.4 vs. control 33.8 ± 6.3 mm³, P = 0.017). SDpv12h also increased the number of damaged cells, visualized by TUNEL staining, by 137% (SDpv12h 46.8 ± 15 vs. control 19.7 ± 7.7/mm², P < 0.001) and SDis by 219% (SDis 32.9 ± 13.2 vs. control 10.3 ± 2.5/mm², P = 0.002). In addition, SDis significantly elevated the expression of the axonal extension inhibitory molecule neurocan (SDis 14.3 ± 0.4 vs. control 6.2 ± 0.1-fold of change, P < 0.001) in the injured hemisphere.

Conclusions:

These results provide the first direct evidence for a detrimental impact of sleep disruption on stroke evolution and suggest a potential role of sleep modulating treatments on stroke outcomes.

Citation:

Gao B; Cam E; Jaeger H; Zunzunegui C; Sarnthein J; Bassetti CL. Sleep disruption aggravates focal cerebral ischemia in the rat. SLEEP 2010;33(7):879-887.

ISCHEMIC STROKE REMAINS ONE OF LEADING CAUSES OF DEATH IN INDUSTRIALIZED COUNTRIES AND IS ONE OF THE MOST IMPORTANT CAUSES OF long-term disability. Sleep disturbances are frequently observed in post-stroke patients and appear to have detrimental consequences on neurological deficits, rehabilitation, cognitive functions and ultimately on the quality of life.^1–3 It is unknown, however, the impact of sleep disruption on ischemia-related pathophysiology and plasticity process that determine functional outcome.

Ischemic stroke, due to a sudden reduction of blood flow, triggers a cascade of events, including energy failure, excitotoxicity, free radical generation, and inflammation.^4,5 Evidence from sleep deprivation (SDpv) experiments under healthy conditions suggests also fundamental effects of sleep on brain energy metabolism, neurotransmitter activities and production of proinflammatory molecules.^6–11

After ischemic insults, the injured brain undergoes an extensive neuronal reconnection and reorganization.^12,13 Axonal sprouting in the peri-infarct area represents an important component of the post-stroke brain repair. A set of neuronal growth-associated genes has been identified¹⁴ to be involved in initiating, maintaining and terminating post-stroke axonal sprouting that occurs within days after ischemic insult.¹⁵ Our recent finding that the sleep stimulant gamma hydroxybutyrate alters expression profiles of neuroplasticity-related genes and accelerates stroke recovery¹⁶ suggest that sleep could modulate post-stroke brain plasticity at the molecular level. Since SDpv has been found to change expression of certain plasticity-related genes in healthy rodents,^17,18 it is likely that sleep disruption after stroke could also alter expression of axonal sprouting genes.

The aim of this study was to test the hypothesis in a rat model of focal cerebral ischemia that sleep disruption could exacerbate brain damage and change expression of genes associated with axonal sprouting. We focused in this study on acute effects of sleep disruption as an initial step to understand sleep's role in modulation of stroke recovery. Sleep disruption was induced by two protocols, i.e., a short-term SDpv and a prolonged sleep disturbance (SDis) over 3 days. For gene expression assay, we chose a small set of genes whose response to stroke is within the first 3 days after insult, such as growth-promoting genes Gap43 and c-jun, and growth-inhibiting genes neurocan, ephrinA5, and ephrinB1.¹⁴ In this study, assessments of brain damage and gene expression were conducted in the same brain.

METHODS

Animals

Male Sprague Dawley rats (n = 50), 8 weeks old and weighting 200-250g at the time of surgery, were used in this study. They were housed under 12-h light/dark cycle (light on 09:00-21:00) and ambient temperature at 22 ± 0.5°C. Food and water were provided ad libitum. All experiments were conducted with governmental approval according to local guidelines (Kantonales Veterinüaramt Zürich, Switzerland) for the care and use of laboratory animals. Effort was made to minimize the number of animals used.

Electrode Implantations, Data Acquisition, and Vigilance State Analysis

Rats were anesthetized with 2% isoflurane in 30% O₂ and 70% N₂O. Four gold-coated miniature screws (0.9 mm diameter) were inserted in the skull over both sides of the frontal cortex (Lateral to midline: 2 mm, Bregma: −2 and +2 mm) to record electroencephalogram (EEG). Two EEG electrodes on one side of the brain were paired to register EEG from each hemisphere. Two gold wires were inserted into neck muscles to record electromyogram (EMG). The electrodes were fixed to the skull with dental cement. After the surgery, rats were housed individually and allowed 7-10 days of recovery. They were then connected to recording system by a flexible cable and a swivel. Polygraphic recordings were performed with the Embla system and the SOMNOLOGICA software (Flaga, Iceland). EEG signals were sampled at 100 Hz with band-pass filter for low and high cut set at 0.3 and none, respectively, and EMG signals at 200 Hz with band-pass filter for low and high cut set at 10 and none, respectively. They were recorded for 8 h (11:00-19:00) in the light phase and 12 h (21:00-9:00) in the dark phase.

Vigilance states were scored by visual inspection of 8-s epochs. Wakefulness (W) was defined by the presence of a desynchronized EEG activity combined with elevated neck muscle tone; slow wave sleep (SWS) by high-amplitude slow wave EEG activities with a low-EMG and paradoxical sleep (PS) by dominant theta frequency activity (6–8 Hz) in EEG with low EMG signals. The EEG activity recorded on the contralateral hemisphere (i.e., undamaged) to ischemia was mainly used for scoring. Whenever there were too many artifacts, the ipsilateral channel was consulted to determine vigilance states. The percentage of each state was calculated for 8 h during the light (11:00-19:00) and 8 h during dark phase (21:00-5:00). The episode length for each vigilance states was also computed and used for assessing sleep fragmentation. To disregard brief interruptions of episodes and short episodes, we adopted minimum criteria of 12-s episode duration and 8-s episode interruption.¹⁹

EEG power spectra were computed from 8-s epochs recorded on the contralateral hemisphere by custom programming on the basis of standard mathematical and signal analysis functions in Matlab. We implemented the multi-taper method, which allows to trade resolution in the frequency domain for reduced variance. Spectra were calculated with a window length of 4 sec, overlap 2 sec, bandwidth parameter nw = 2 and k = 3 tapers, which offers optimal spectral smoothing. The power spectral density was given in units of 10*log 10 (μV²/Hz). Epochs with artifacts were excluded by a threshold procedure. The percentage of artifact-free epochs during wakefulness was 70 ± 24, 65 ± 18, 64 ± 27, and 71 ± 19 for the 4 experimental groups (see Experimental protocols) and 86 ± 15, 82 ± 18, 80 ± 30, and 90 ± 12 during SWS. Spectral band power was computed for the frequency bands: delta (1-4 Hz), theta (4-8 Hz), sigma (11-16 Hz), and beta (16-25 Hz). Delta power during SWS was averaged for the first 2 and 8 h for the light and the dark phase, and the relative change with respect to baseline was computed for corresponding phases.

INDUCTION OF FOCAL CEREBRAL ISCHEMIA

After a 24-h baseline polygraphic recording, rats were anesthetized with 2% isoflurane (30% O₂, remainder N₂O) for the ischemia surgery that was carried out with occlusion of the distal middle cerebral artery (MCAo). The procedure was modified from Tamura et al.²⁰ Briefly, a 2 cm vertical skin incision is made midway between the ear and eye. The skull was exposed where the frontal bone joins the temporal bone. A 5×5 mm area of the bone overlying the middle cerebral artery (MCA) was removed and the dura retracted. The MCA and its three main branches dorsal to the rhinal fissure were occluded by bipolar electrocoagulation. The incision was closed with silk suture. Sham-control animals were subjected to the same procedure except for dura removal and vessel electrocoagulation. Rectal temperature was maintained between 36.5 ± 0.5°C by a warm lamp during the entire surgery. Animals were placed back into their home cages once they were awake from anesthesia.

Neurologic Assessments

The animal's motor behavior was carefully evaluated by a 4-point scale method described by Bederson et alcv²¹ at 12 h after stroke and at the end of experiments.

Experimental Protocols and Brain Tissue Collection (Figure 1)

Figure 1.

Schematic of the sleep deprivation (SDpv, A) and sleep disturbance (SDis, B) protocols. The white and black bars indicate the light and dark period, respectively. *SDpv was carried out for either 6 h (SDpv6h) or 12 h (SDpv12h) in separate experiments.

SDpv experiments (Figure 1A) SDpv was carried out 12 h after the ischemia during the light phase by gently knocking at the cage or providing the rats with new playing materials when they started showing signs of sleep. SDpv was carried out for either 6 h (SDpv 6h) or 12 h (SDpv12h). Control rats (ischemia/without SDpv) were subjected to the same ischemia surgery but left undisturbed afterward. Two groups in each experiment were used, i.e., ischemia/SDpv 6h (n = 4) and ischemia/without SDpv 6 h (n = 4); ischemia/SDpv 12h (n = 6) and ischemia/without SDpv12h (n = 6). At end of experiments, rats were decapitated after brief isoflurane anesthesia and brains removed, frozen immediately on dry ice and stored at −80°C. Control rats were decapitated at the same time point as SDpv rats.

SDis experiments (Figure 1B) SDpv12h was carried out as described above for 3 consecutive days during the light phase. Rats were undisturbed and allowed to sleep during the following 12-h dark phase. At the end of the experiment, i.e., 72 h after ischemia surgery and around 21:00, rats were decapitated, and brains harvested as described above. One ischemia/SDis group (n = 6) and 3 control groups (n = 6 per group) i.e., ischemic/without SDis, sham/SDis, sham/without SDis, were used.

Plasma Corticosterone Determination

The plasma corticosterone level was measured in SDpv12h and SDis experiments. Blood samples were collected from trunk blood at the time when rats were decapitated at the end of SDpv (20:30-21:00, at the end of the light phase, Figure 1) and centrifuged with 2500 xg for 10 min. The plasma corticosterone concentration was later determined with Radio Immuno Assay kit (MP Biomedicals, Orangeburg, NY, USA), according to the manufacture's instruction.

Brain Damage Analysis

For each brain, 20 μm sections at 6 brain levels [A = +2.7 (L1), +1.7 (L2), +0.7 (L3), −0.3 (L4), −1.3 (L5), −2.3 (L6) from the bregma²² were cut on a cryostat. Twelve to 15 sections at the each level were mounted on SuperFrost Plus slides (Menzel GmbH, Braunschweig, Germany) for histology assessments and remaining sections in each level were collected in RNase-free tubes for gene expression assay (see below). For assessing the infarct volume, sections were first fixed with phosphate-buffered saline (PBS, pH 7.4) containing 4% paraformaldehyde for 20 min at room temperature and then performed for standard cresyl violet staining. For assessing damaged cells, sections fixed with 4% paraformaldehyde were processed with terminal transferase biotinylated-dUTP nick end labeling (TUNEL) according to the manufacturer's instruction (Roche, Basel, Switzerland) to ascertain cells that contain fragmented DNA. On digitized cresyl violet sections, the infarct area were delineated and measured with the NIH imageJ software (NIH, Bethesda, MD, USA), and the infarct volume was converted with the known distance between each of the chosen levels and eventually corrected for edema by multiplying the ratio of the contralateral to ipsilateral volume.²³ Brain swelling was calculated with the formula as follows: [(ipsilateral – contralateral hemisphere)/contralateral hemisphere] × 100. The number of TUNEL positive cells was counted in the injured area measuring 20 mm² on sections collected at the level 6.

Taqman Gene Expression Assay

Brain tissues collected in RNase-free tubes (see above) were subjected to RNA extraction by the Trizol method (Life Technologies, Rockville, MD, USA) for individual animals. The total RNA was treated with RQ1 DNase (Promega, Madison, WI, USA) to digest genomic DNA. Oligo(dT)₁₅ primed the first-strand cDNA synthesis was synthesized by AMV reverse transcriptase (Promega). Group cDNA samples were synthesized by pooling together the RNA isolated from each animal. When necessary, cDNA from each rat was also synthesized to check the variation within group. The 5’-FAM labeled probes used in the Taqman real-time quantitative RT-PCR assay for GAPDH (endogenous control, Assay ID: Rn99999916_s1), GAP43 (Assay ID: Rn00567901_m1), c-JUN (Assay ID: Rn00572991_s1), neurocan (Assay ID: Rn00581331_m1), ephrinA5 (Assay ID: Rn005588118_s1), ephrinB1 (Assay ID: Rn00438666_s1) and glial fibrillary acidic protein (GFAP, Assay ID: Rn00566603_m1) were purchased from Applied Biosystems (Forster City, CA, USA). Reactions were performed in triplicates on AB 7900HT fast real time PCR system (Applied Biosystems). The relative level of mRNA expression in a given hemisphere was calculated as follows:

Immunoperoxidase Staining

Cryosections were fixed first with 4% paraformaldehyde in PBS for 20 min at room temperature and incubated overnight at 4°C with the primary antibodies against GFAP (raised in the rabbit, Dako, Glostrup, Denmark) and neurocan (raised in the mouse, Millipore, Billerica, MA, USA) diluted in PBS containing 2% normal goat serum (Jackson ImmunoResearch, West Grove, PA, USA) and 0.03% Triton X-100 (Sigma Chemicals, St. Louis, MO, USA). The concentration for the GFAP antibody was 1:6000; for the neurocan, it was 1:2000. Sections were washed with PBS, incubated with the biotin-conjugated secondary antibody (Jackson ImmunoResearch), diluted at 1:300 for 1 hour at room temperature, and processed with the ABC immunoperoxidase staining method using Vectastain Elite kits (Vector Laboratories, Burlingame, CA, USA) and diaminobenzidine (Sigma) as the chromogen. Stained sections were analyzed and photographed with Leica DM 6000B microscope (Leica Microsystems, Wetzlar, Germany).

Statistics

Data were presented as mean ± standard deviation (s.d.). The significance of differences in means was assessed by independent t-test, one-way analysis of variance (ANOVA), and repeated-measures ANOVA (SPSS, 12.01 for Windows) where appropriate, ANOVA was followed by post hoc comparisons to determine group differences. The significance level was set at P values < 0.05.

RESULTS

Neurological Score and Changes in Vigilance States

All animals subjected to MCAo did not exhibited neurological deficits according to Bederson's score,²¹ i.e., without forelimb flexion when lifted up by the tail and without cycling behavior, both at 12 h after stroke and the end of experiments. They walked normally but were somewhat drowsier than sham operated rats.

Changes in the amount of EEG-defined vigilance states and in the episode length under the SDis procedure are listed in Table 1 and 2, respectively. SDpv12h carried out during the light phase resulted in an increase in SWS and PS during the following dark phase in both the ischemia/SDis and sham/SDis groups. In order to evaluate how SDpv12h influenced the total amount of sleep on a daily base in the SDis experiment, values in the light phases were combined with those in the following dark phase (total in Table 1). The SWS amount in the ischemia/SDis group reached almost the baseline level during day 1 and even tended to overshoot, although not statistically significant, during day 2, whereas in the sham group it was continuously below the baseline level by 11% to 12% (or 1.6 h) for 2 days. The difference between the 2 SDis-treated groups, i.e., ischemia/SDis 39.7 ± 3.7 vs. sham/SDis 28.2 ± 2.0 during day 1 and ischemia/SDis 47.9 ± 3.9 vs. sham/SDis 27.0 ± 1.6 during day 2 (Table 1), was statistically significant (repeated-measures ANOVA, P < 0.001; independent t-test, P < 0.001), indicating an increased sleep propensity after ischemia. The total percentage of PS in the ischemia/SDis group decreased significantly from 10% of the baseline level (or 1.6 h) to 5% (or 0.8 h) during day 1 and recovered slightly during day 2. In the sham/SDis group PS decreased slightly (~0.5 h) for 2 days. As an effect of sleep deprivation, the episode length for SWS and PS was reduced during the light phase (L1, L2, and L3) but increased during the following dark phase for SWS (D1) (Table 2).

Table 1.

Changes in wakefulness (W), slow wave sleep (SWS), and paradoxical sleep (PS) in the SDis experiment

	Experiment groups (n = 6 for each group)
	Ischemia w SDis			Ischemia w/o SDis			Sham w SDis			Sham w/o SDis
	W§	SWS	PS§	W	SWS§	PS	W§	SWS§	PS§	W§	SWS	PS§
Baseline
L	38.1 ± 6	51.8 ± 4	10.1 ± 4	42.3 ± 6	48.9 ± 6	8.8 ± 2	39.3 ± 5	51.6 ± 4	9.1 ± 3	38.7 ± 9	50.7 ± 8	10.6 ± 1
D	56.5 ± 4	33.5 ± 3	10.0 ± 4	59.6 ± 5	30.6 ± 3	9.8 ± 4	61.0 ± 6	27.7 ± 4	11.3 ± 2	69.4 ± 10	23.8 ± 9	6.8 ± 2
total	47.1 ± 4	42.9 ± 1	10.1 ± 3	50.9 ± 4	39.7 ± 4	9.3 ± 2	50.1 ± 4	39.6 ± 3	10.2 ± 2	54.1 ± 7	37.3 ± 7	8.7 ± 1
Day 1
L	83.8 ± 6	15.8 ± 6	0.4 ± 1	39.2 ± 5	53.2 ± 5	7.6 ± 2	96.3 ± 1	3.7 ± 1	0.0 ± 0.0	33.5 ± 3	56.3 ± 4	10.1 ± 2
D	27.9 ± 8	61.9 ± 8	10.2 ± 4	49.6 ± 7	43.3 ± 7	7.1 ± 2	32.9 ± 4	52.3 ± 4	14.8 ± 2	53.9 ± 8	35.3 ± 6	10.8 ± 2
total	55.8 ± 4*	39.7 ± 3	5.2 ± 2*	44.8 ± 5	47.9 ± 3*	7.3 ± 2	64.6 ± 2*	28.2 ± 2*	7.4 ± 1*	42.2 ± 5*	46.3 ± 4	10.2 ± 1
Day 2
L	77.0 ± 5	22.8 ± 5	0.2 ± 0.3	39.1 ± 10	54.1 ± 8	6.8 ± 3	93.6 ± 2	6.4 ± 2	0.0 ± 0.0	18.6 ± 7	52.9 ± 6	15.3 ± 2
D	30.8 ± 3	57.8 ± 10	11.4 ± 5	55.1 ± 9	37.0 ± 9	7.9 ± 4	37.9 ± 6	47.5 ± 4	14.6 ± 2	58.0 ± 9	32.4 ± 7	9.5 ± 2
total	53.2 ± 5*	47.9 ± 4	7.3 ± 2	47.5 ± 8	45.2 ± 5	7.4 ± 3	65.7 ± 2*	27.0 ± 2*	7.3 ± 1*	42.8 ± 7*	42.8 ± 7	12.3 ± 2*
Day 3
L	85.1 ± 7	14.9 ± 7	0.0 ± 0	34.1 ± 9	59.8 ± 7	6.2 ± 3	94.0 ± 1	6.0 ± 1	0.0 ± 0.0	29.0 ± 2	57.4 ± 3	13.5 ± 2

Values are mean ± s.d. of percentages of recording hours (see Methods). L, the light phase (9:00-21:00, 8 h of recording time 11:00-19:00); D, the dark phase (21:00-9:00, 8 h of recording time 21:00-5:00); Total, both the light and dark phase combined (16h)

^§P < 0.05 in one-way ANOVA followed by Turkey post hoc comparison

^*P < 0.05 when compared with the baseline of the corresponding phase.

Table 2.

Changes in the episode length (s) of wakefulness (W), slow wave sleep (SWS), and paradoxical sleep (PS) in the SDis experiment

	Experiment groups (n = 6 for each group)
	Ischemia w SDis			Ischemia w/o SDis			Sham w SDis			Sham w/o SDis
	W§	SWS§	PS§	W	SWS§	PS	W§	SWS§	PS§	W	SWS	PS
L
Baseline	128 ± 19	141 ± 33	80 ± 13	152 ± 44	119 ± 28	88 ± 20	143 ± 19	120 ± 16	69 ± 7	131 ± 21	125 ± 26	94 ± 15
L1	351* ± 82	69* ± 12	20* ± 22	182 ± 26	147 ± 24	93 ± 23	1044* ± 416	48* ± 3	0.0* ± 0	133 ± 12	144 ± 9	95 ± 16
L2	285* ± 93	96* ± 12	17* ± 29	196 ± 30	186* ± 44	101 ± 20	618* ± 150	47* ± 5	0.0* ± 0	122 ± 25	130 ± 31	110 ± 15
L3	378* ± 104	73* ± 11	0.0* ± 0	168 ± 40	215* ± 39	86 ± 27	604* ± 136	51* ± 6	0.0* ± 0	122 ± 12	155 ± 16	103 ± 16
D	W	SWS§	PS	W§	SWS§	PS	W§	SWS§	PS	W	SWS	PS
Baseline	347 ± 68	141 ± 17	104 ± 30	308 ± 39	114 ± 16	95 ± 24	414 ± 124	116 ± 12	116 ± 10	468 ± 162	120 ± 32	98 ± 17
D1	315 ± 174	297* ± 54	110 ± 17	320 ± 51	200* ± 40	108 ± 24	225* ± 36	206* ± 21	112 ± 12	337 ± 26	144 ± 21	114 ± 19
D2	287 ± 62	332 ± 110	111 ± 35	389* ± 55	167* ± 20	96 ± 33	239 ± 59	174* ± 19	113 ± 8	318 ± 72	120 ± 14	97 ± 10

Values are means (s) ± s.d. of the episode length (see Methods). L, the light phase (9:00-21:00, 8 h of recording time 11:00-19:00); D, the dark phase (21:00-9:00, 8 h of recording time 21:00-5:00)

^§P < 0.05 in One-way ANOVA followed by post hoc comparison

^*P < 0.05 when compared with the baseline of the corresponding phase.

Delta power during SWS, a parameter for sleep pressure under normal conditions, increased in the sham/SDis group at the first 2 h following SDpv (Figure 2). In contrast, there was no change in the ischemia/SDis group (Figure 2). The difference between the two SDis groups suggests that the delta activity be altered after brain injury. When data were averaged for 8 h in the light or dark phase, there were no significant changes for all groups (not shown).

Figure 2.

Effects of SDis on delta power (1-4 Hz) during slow wave sleep. Delta power was averaged for the first 2 h of each recording phase, and relative changes are plotted against the corresponding baseline (100%). Refer to Figure 1 for legends of L1, L2, L3, D1, and D2.

Effects of SDpv and SDis on Brain Damage

Occlusion of the distal MCA resulted in an infarct located in the primary somatosensory cortex. While SDpv6h barely (P = 0.9) influenced the infarct volume, SDpv12h increased it by 40% (SDpv12h 82.8 ± 10.9 vs. control 59.2+13.9 mm³, P = 0.008) and SDis by 76% (SDis 58.8 ± 20.4 vs. control 33.8 ± 6.3 mm³, P = 0.017), respectively (Figure 3A). The increased infarct area was found mostly at the level 5 and 6 (Figure 4), where a less arterial anastomosis may account for the vulnerability. In parallel to the change in the infarct size, SDpv12h increased the number of damaged cells, visualized by the TUNEL positive staining, by 137% (SDpv12h 46.8 ± 15 vs. control 19.7 ± 7.7/mm², P = 0.003) and SDis by 219% (SDis 32.9 ± 13.2 vs. control 10.3 ± 2.5/mm², P = 0.008), respectively (Figure 3B). There was no significant change in brain swelling in all experiments (Figure 3C).

Figure 3.

Scatter plots showing effects of SDpv and SDis on the infarct volume (A), the number of TUNEL positive cells (B), and brain swelling (C) in individual animals. Horizontal lines present the mean values and analyzed by independent t-test for each experiment. a, ischemia with sleep manipulation. b, ischemia without sleep manipulation.

Figure 4.

Effects of SDpv and SDis on the infarct area at different brain levels. A, Representative sets of brain sections from a rat subjected to ischemia/SDis (upper panel) and a rat to ischemia without (w/o) SDis (lower panel). The infarct areas are delineated by a black thin line. L1 is at 2.7 mm anterior to bregma, and the interval between each level is 1 mm (Methods). B, Values are presented as mean ± s.d. (n = 6 per group) and analyzed by independent t-test. *P < 0.05.

To investigate how the exacerbated brain damage was influenced by changes in vigilance states, correlation of the infarct volume with altered W, SWS and PS was computed (Table 3). The results showed that the amount of W during day 1 was positively (Figure 5 and Table 3), and SWS negatively, correlated with the increased infarct size (Table 3). In addition, the episode length for SWS was also negatively correlated with the infarct size during three days of SDpv (L1 - L3) and for PS during the first two days (L1 and L2). These results suggest that both reduced and fragmented sleep contribute to the exacerbated stroke.

Table 3.

Correlaton of infarct size with wakefulness (W), slow wave sleep (SWS), and paradoxical sleep (PS) in the SDis experiment.

Infarct (mm3)	Day	R Sq	Pearson	P
W (%)	1	0.587	0.766**	0.004
	2	0.201	0.449	0.193
SWS (%)	1	0.592	−0.770**	0.003
	2	0.246	−0.496	0.144
PS (%)	1	0.007	−0.082	0.8
	2	0.07	−0.264	0.461
W episode length (s)	L1	0.413	0.642*	0.024
	L2	0.513	0.716**	0.009
	L3	0.637	0.798**	0.002
SWS episode length (s)	L1	0.339	−0.582*	0.047
	L2	0.352	−0.593*	0.042
	L3	0.413	−0.642*	0.024
PS episode length (s)	L1	0.453	−0.673*	0.016
	L2	0.689	−0.830**	0.001
	L3	0.361	−0.601	0.039

The amount of each vigilance states is presented as percentage of total recording time for each day. The episode length values are from the sleep deprivation period and refer to Figure 1 for L1, L2, and L3.

^*P < 0.05

^**P < 0.01.

Figure 5.

Correlation of the infarct volume with the amount of wakefulness (W) during the first day of the sleep disturbance. W values are the percentage of the total recording time including both the light and dark phase (Table 1, total).

An additional experiment was carried out to determine whether sleep rebound following SDpv would alleviate the exacerbated brain damage induced by sleep disruption. In this experiment rats (n = 6) were subjected to the same procedure as these for SDpv12h but allowed to sleep for 24 h following SDpv before sacrificed. There was no difference (P = 0.91) in the infarct volume between this group (81.5 ± 24.7 mm³) and the SDpv12h group (82.8 ± 10.9 mm³), indicating that the SDpv12h-induced brain damage during the acute phase of stroke was not reversible.

Effects of SDis on Gene Expression

Changes in expression of neuroplasticity-related genes are summarized in Figure 6. The most striking change was the dramatic increase (P < 0.001) of the growth-inhibiting gene neurocan in the ischemic (ipsilateral) hemisphere in the ischemia/ SDis group (14.3 ± 0.4-fold), compared with the ischemia/ without SDis (6.1 ± 0.1-fold) and both sham groups (~1-fold). Additional assays with individual RNA samples were carried out to assess the variation within groups and the result confirmed the SDis-induced massive increase in neurocan expression (P < 0.003), with 11.03 ± 5.9-fold change in the ischemia/SDis group. SDis also induced a significant increase, although at smaller scale, in expression of another growth-inhibiting gene ephrinB1 in the injured hemisphere. In comparison, there was no SDis-induced further change in expression of GAP43, c-jun, and ephrinA5 (Figure 6A). In the contralateral hemisphere (Figure 6B), SDis slightly increased (P < 0.05) expression of GAP43, c-jun, neurocan, and ephrinB1, when compared with the ischemia/ without SDis.

Figure 6.

Effects of SDis on expression of neuroplasticity-related genes. Values are presented as mean ± s. d. ^aP < 0.05 when compared with the ischemia w/o SDis group. ^bP < 0.05 when compared with the sham w. SDis group. ^cP < 0.05 when compared with the sham w/o SDis group. One-way ANOVA (see F values, *P < 0.001) followed by Turkey post hoc comparisons.

Expression of GFAP and Localization of Neurocan

The growth-inhibiting gene neurocan has been known to be secreted by reactive astrocytes in response to various CNS injures,²⁴ including ischemic stroke.^25,26 To assess whether the SDis-induced massive increase in neurocan expression was parallel to an increase in reactive astrocytes, we determined the expression of the astrocyte marker GFAP with Taqman assay and the cellular distribution of both GFAP and neurocan with immunoperoxidase staining in the ischemic hemisphere. The GFAP mRNA level increased, although at much smaller scale (2-fold) compared to the increase in the expression of neurocan, in the ischemia without SDis group when compared with sham groups, and there was a slightly (but significantly) further increase in the ischemia/SDis (Figure 7A). The immunoreactivity for both GFAP and neurocan was highly expressed in the peri-infarct area (Figure 7B and C).

Figure 7.

Expression of the GFAP gene (A) and distribution of the immunoreactivity (ir) for GFAP (B) and neurocan (C, D). A: Values are presented as mean ± s.d. One-way ANOVA (F _{3, 8} = 1013. *P < 0.001) followed by Games-Howell comparisons. ^aP < 0.05 when compared with the ischemia w/o SDis group. ^bP < 0.05 when compared with the sham w. SDis group. ^cP < 0.05 when compared with the sham w/o SDis group. B. The GFAP-ir in the peri-infarct area. Note the highly expressed GFAP-ir around the border of the infarct area. Bar, 100 μm. C and D, High-power photomicrograph illustrating expression of the neurocan-ir around parenchyma cells (marked with *). Note the substantially increased expression of the neurocan-ir in the peri-infarct area (C) compared with the weak staining in the contralateral cortex (D). Bars, 20 μm.

Plasma Corticosterone level after SDpv12h and SDis

In the SDpv12h experiment, there was no significant change in the plasma corticosterone level between the ischemia/SDpv12h and its control group. In the SDis experiment, there was a significant decrease in groups subjected to SDis (ischemia and sham) when compared with groups without SDis. There was no significant difference between the ischemia/SDis and sham/SDis (Figure 8).

Figure 8.

Effects of SDpv and SDis on the plasma corticosterone level. Values are presented mean ± s.d. (n = 6 per group) and analyzed by either independent t-test in the SDpv12h experiment or one-way ANOVA followed by Turkey post hoc comparisons in the SDis experiment. ^aP < 0.02 when compared with the ischemia w/o SDis group. ^bP < 0.02 when compared with the sham w/o SDis group.

Changes in Body Weight

Ischemia-operated rats tended to lose about 8-15 g or 5% of their body weight after surgery, compared to 0-4 g of sham-operated rats (Table 4). In neither the ischemia/SDpv12h nor the ischemia/SDis group did the body weight change significantly when compared with their paired ischemia/without SDpv12h or ischemia/without SDis group.

Table 4.

Changes in body weight after SDpv12h and SDis

Experiment	Groups baseline weight (g)	Day 1*	Day 2*	Day 3*
SDpv12h (n = 6 per group)	Ischemia/SDpv 262 ± 31	−14.3 ± 6.1
	Ischemia w/oSDpv 257 ± 23	−14.8 ± 6.1
SDis§ (n = 6 per group)	Ischemia/SDis 272 ± 17	−12.2 ± 2.3a,b	−17.0 ± 6.1a,b	−11.7 ± 3.7b
	Ischemia w/o SDis 324 ± 25	−8.2 ± 5.6 b	−10.5 ± 5.9 b	−10.3 ± 6.4
	Sham/SDis 296 ± 24	−4.0 ± 5.2	−7.8 ± 6.4	−9.8 ± 6.4
	sham w/o SDis 305 ± 14	0.0 ± 5.2	−0.1 ± 5.3	1.3 ± 4.4

^*Values are means of difference from the baseline ± standard deviation.

^§Repeated measures ANOVA (F_{3, 20} = 5.56, P = 0.006) followed by independent t-test.

^aP < 0.05 when compared with sham/SDis.

^bP < 0.05 when compared with the sham w/o SDis.

DISCUSSION

The findings that both sleep deprivation and repeated sleep disruption over 3 days aggravate acute brain damage (Figure 3 and 4) provide the first direct evidence for a detrimental effect of sleep disruption on stroke. The underlying mechanism(s) remains, however, to be specified. It has been known that high levels of plasma glucocorticoids after ischemia increase neuronal vulnerability,²⁷ whereas low levels reduce brain damage.²⁸ In this study, SDpv12h did not significantly alter the corticosterone level (Figure 8), while SDpv12h for consecutive 3 days (the SDis experiment) decreased the level (Figure 8). Thus, the plasma corticosterone level does not appear to be related to the worsened brain damage after SDpv/SDis. Ischemic stroke, due to a sudden reduction of blood flow, triggers a series of events, including energy failure, excitotoxicity, free radical generation, and inflammation—a cascade that determines the fate of infarction.^4,5 Short-term SDpv has been shown to elevate the brain temperature²⁹ and brain energy consumption, evidenced by increasing 2-deoxyglucose uptake⁸ and decreasing the level of brain glycogen, the principal energy store in the brain.⁶ Short-term SDpv also increases the extracellular glutamate concentration,⁹ glutamate receptor levels and neuronal activities,³⁰ which would worsen ischemia-induced excitotoxicity. Furthermore, prolonged SDpv has been shown to decrease levels of some antioxidative stress markers, such as glutathione, glutathione peroxidase, and superoxide dismutase,^31–33 which in turn could intensify cell damage caused by free oxygen radical in the ischemia site. Prolonged SDpv or sleep restriction also increases levels of certain proinflammatory cytokines such as IL-1ß, IL-6, and TNFα^10,11 that play an important role in the inflammatory phase of cerebral ischemia.³⁴ Taken together, substantial experimental evidence suggests that SDpv/SDis could aggravate stroke through potentiation of mechanisms implicated in its pathophysiology.^4,5 Further studies are needed to identify essential contributors.

SDis altered the expression of several genes that mediate initiation of post-stroke axonal sprouting at the early phase after stroke (Figure 6). The substantial changes occurred in the ischemic hemisphere where SDis induced a massive increase in expression of the growth-inhibiting gene neurocan (Figure 6). Neurocan is one of major chondroitin sulfate proteoglycans (CSPGs) and secreted by reactive astrocytes in response to various CNS injures.^24–26 Its inhibitory property to axonal growth has been demonstrated both in vitro^35,36 and in vivo,^26,37 possibly by activation of the Rho/ROCK pathway.³⁸ Around the border of the infarct area, densely packed reactive astrocytes (Figure 7B) together with growth-inhibitory molecules including neurocan (Figure 7C) form a physical and biochemical barrier that hinders the neuronal reconnection. Interestingly, the present data that SDis increases neurocan expression after ischemic stroke is in line with our previous observations that the sleep stimulant gamma-hydroxybutyrate has an opposite (decrease) effect on the ischemia-induced neurocan expression,¹⁶ suggesting that neurocan be an important molecule implicated in the sleep-modulated post-stroke brain plasticity. In this context, large scale screening of neuroplasticity-related genes and their protein products is needed to understand molecule mechanisms involved in sleep-modulated neuroplasticity after ischemic stroke. In addition, it remains to demonstrate the influence of sleep disruption on post-stroke axonal sprouting and finally on functional recovery.

In summary, this study demonstrate that at the early phase of stroke, sleep disruption aggravates brain damage and increases expression of genes that inhibit post-stroke axonal sprouting. Thus, prevention of sleep disturbances and improvement of sleep quality may become a new therapeutic window to improve outcome.

DISCLOSURE STATEMENT

This was not an industry supported study. The authors have indicated no financial conflicts of interest.