Peripheral sympathectomy alters neuroin ammatory responses and microglial activity in response to sleep fragmentation in female mice


 BackgroundSleep loss, either induced by obstructive sleep apnea or other forms of sleep dysfunction, induces an inflammatory response, as commonly measured by increased circulating levels of pro-inflammatory cytokines. Increased catecholamines from sympathetic nervous system (SNS) activation regulates this peripheral inflammation. However, the role that catecholamines play in mediating neuroinflammation from sleep perturbations is undescribed. The aims of this study were to determine (i) the effect of peripheral SNS inhibition upon neuroinflammatory responses to sleep fragmentation (SF) and (ii) whether homeostasis can be restored after 1 week of recovery sleep.MethodsWe measured gene expression levels of pro- and anti-inflammatory cytokines and microglial activity in brain (prefrontal cortex, hippocampus and hypothalamus) of female mice that were subjected to acute SF for 24 hours, chronic SF for 8 weeks, or 7 days of recovery after chronic SF. In each experiment, SF and control mice were peripherally sympathectomized with 6-OHDA (6-hydroxydopamine) or injected with vehicle. ResultsSF elevated cytokine mRNA expression in brain and increased microglial density and cell area in some regions. In addition, chronic SF promoted hyper-ramification in resting microglia upon exposure to chronic, but not acute, SF. Effects of chronic SF were more pronounced than acute SF, and 1 week of recovery was not sufficient to alleviate neuroinflammation. Importantly, 6-OHDA treatment significantly alleviated SF-induced inflammation and microglial responses.ConclusionsThis study provides evidence of SNS regulation of neural inflammation from SF, suggesting a potential role for therapeutics that could mitigate neuroinflammatory responses to sleep dysfunction.

In the present study we examined the effects of acute and chronic SF on neural in ammatory responses in female mice. Females were chosen because of their general underrepresentation in neurobiological studies [11]. We also assessed whether 7 days of recovery sleep could su ciently diminish neuroin ammation from chronic SF. To examine the effect of SNS activation on neuroin ammatory responses to SF, we peripherally sympathectomized mice using 6-hydroxydopamine (6-OHDA). We assessed pro-in ammatory cytokine gene expression and microglial morphology in three distinct brain regions (pre-frontal cortex, hippocampus and hypothalamus) that have been explored in previous studies using experimental sleep fragmentation [7,12]. Since peripheral signals of in ammation are proposed to affect neuroin ammation via the neural transmission of vagal afferents reacting to circulating proin ammatory agents or the pro-in ammatory agents themselves crossing the blood-brain barrier to directly affect microglia [13], we hypothesized that peripheral sympathectomy would alleviate the neuroin ammatory responses and diminish microglial activity to acute and chronic SF. Finally, we predicted that a 1-week recovery from chronic SF would reverse neuroin ammatory responses as was previously observed when assessing peripheral tissues [7].

Animals and experimental protocol
Female C57BL/6J mice between 8-12 weeks of age were used. Sleep fragmentation (SF) experiments were performed using automated SF chambers (Lafayette Instrument Company; Lafayette, IN; model 80390) with a thin layer of corn bedding as previously described [7]. These chambers ensure that mice are subjected to sleep fragmentation and not absolute sleep deprivation [12]. Mice were provided ad libitum access to water and food and housed in a 12:12-h light-dark photoperiod (lights on-8am, 21°C ± 1°C) at Western Kentucky University. Subjects were acclimated to the SF chambers for 72 hours prior to the commencement of sleep fragmentation experiments to diminish any carryover effects from the novel cage environment [14]. This study was conducted under the approval of the Institutional Animal Care and Use Committee at Western Kentucky University (#15-11), and procedures followed the National Institutes of Health's "Guide for the Use and Care of Laboratory Animals" and international ethical standards.

Acute Sleep Fragmentation (Acute SF)
Female mice (n = 26) were chemically sympathectomized with a daily subcutaneous injection of 6-OHDA (0.1mg/g body mass/day) for 5 days while the remaining mice (n =26) were injected with the same volume of vehicle (100 µl of 0.9% NaCl/day). Following the nal injection on Day 5, experimental mice (n = 26; n = 13 injected with vehicle and n = 13 injected with 6-OHDA) were subjected to acute SF i.e. a sweeping bar set to move horizontally every 120 seconds for 24 hrs, which simulates the rate of severe sleep apnea among humans ( [15]. Controls (n = 26; n = 13 injected with vehicle and n = 13 injected with 6-OHDA) received no sweeping bar movements. 6-OHDA treatment was effective in suppressing SNS activity because serum epinephrine levels were reduced in SF mice receiving 6-OHDA [7].

Chronic Sleep Fragmentation (Chronic SF)
To induce chronic sleep fragmentation, mice (n = 28) were subjected to a horizontal sweeping bar set to move every 120 seconds (30 swipes/h) during the light phase (i.e. from 8 am to 8 pm) every day for 8 weeks. To account for increased activity from daily chronic SF, control mice (n = 26) were subjected to the same number of bar sweeps as experimental mice for only 3 hours of the light phase (i.e. from 8 am to 11am), albeit at 4 times the rate i.e. 2 swipes/minute, to control for daily activity induced by the bar movement. Mice were subcutaneously injected with either 6-OHDA (0.1mg/g body mass/day; n = 13 of SF mice, n = 13 of control mice) or vehicle (100 µl of 0.9% NaCl/day; n = 15 of SF mice, n = 13 of control mice) between 8-9 am for 5 consecutive days prior to termination of the experiment.

Chronic Sleep Fragmentation + Recovery (Chronic SF + R)
Mice were subjected to the chronic SF protocol above, but post SF, control (n = 16) and SF mice (n = 25) were subjected to a recovery period (no bar movement) for 7 days during which they were injected with either 6-OHDA (0.1mg/g body mass/day; n = 13 of chronic SF + R mice, n = 8 of control mice) or vehicle (100 µl of 0.9% NaCl/day; n = 12 of SF + R mice, n = 8 of control mice) for 5 days.
For immunohistochemical studies, the experimental protocols mentioned above were followed, except that the chronic SF+ R experiment did not have a separate control group. Hence, in immunohistochemistry experiment 2, there were three sleep paradigms: controls, chronic SF and chronic SF+ R, each with vehicle or 6-OHDA treatment (n = 5 per group except vehicle injected chronic SF+ recovery group with n = 4).

Gene expression analysis
In all three experiments, 24 hours following the nal drug administration, brains of mice (n = 8/ group except chronic SF + vehicle group of n = 10 mice) were sampled for RTPCR or immunohistochemistry analysis. For measurement of brain gene expression, brains were dissected from decapitated mice and stored in RNAlater solution (Thermo Fischer Scienti c) at 4°C until RNA extraction.
RNA was extracted from hippocampus, pre-frontal cortex and hypothalamus using a RNeasy mini kit (Qiagen). RNA concentrations were measured using a NanoDrop 2000 Spectrophotometer (ThermoScienti c). Total RNA was reverse transcribed into cDNA using a high-capacity cDNA reverse transcription kit (Life Technologies, Cat number: 4368813). The prepared cDNA was used as template for determining relative cytokine gene expression using an ABI 7300 RTPCR system. Cytokine probes (IL1β, IL-6, TNFα, TGFβ; Applied Biosystems) labelled with uorescent marker 5-FAM at the 5' end and quencher MGB at the 3'end were used for genes of interest along with 18S (primer-limited VIC-labelled probe) as the endogenous control according to the manufacturer's instructions. Samples were run in duplicate and the fold change in mRNA levels was calculated as the relative mRNA expression levels,

Immunohistochemistry
For immunohistochemistry studies, 24 hours following nal drug administration, mice were deeply anesthetized using iso urane vapors and then transcardially perfused (n = 5/group except recovery + vehicle group of n = 4) with ice-cold saline (0.9%) followed by 4% paraformaldehyde prepared in phosphate buffered saline (PBS). Brains were dissected out from perfused mice, post-xed in the same xative overnight, cryoprotected in 30% sucrose and stored at -80°C until cryosectioned. 30 µm sections from brains of control (n = 5/vehicle or 6-OHDA treatment) and acute SF (n = 5/vehicle or 6-OHDA treatment) mice from the rst study, and control (n = 5/vehicle or 6-OHDA treatment), chronic SF (n = 5/vehicle or 6-OHDA treatment) and chronic SF+ R (n = 4 injected with vehicle, n = 5 injected with 6-OHDA) mice from the second study were processed for immunohistochemistry of IBA-1. Sections were rinsed thrice in 10mM PBS, followed by an incubation in blocking solution (2.5% goat serum, 0.1% triton X) for 2 hours and overnight incubation (at 4°C) in the same blocking solution containing anti-Iba-1 (1:300, Catalogue number 019-19741, Wako Pure Chemicals Industries, China). Sections were then incubated with goat anti-rabbit secondary antibody (Alexa uor 594, catalogue number A11037, Invitrogen, Life Technologies Corporation, Oregon, USA) for 2 hours, followed by washing of sections with 10mM phosphate buffer saline and mounting on slides. After 2 hours of air drying in a dark chamber, slides were mounted and permanently xed with DAPI+antifade (Vectashield, H-1500, Vector Laboratories, Inc., CA 94010, USA).

Image Analysis
To quantify microglia, every sixth section of the brain was selected and immune-labelled with Iba1 antibody. Iba1 immunoreactivity (-ir) was visualized through a Zeiss Axioplan 2 epi uorescence microscope tted with a Leica DFC300 FX digital camera. Omission of primary antibodies in control studies con rmed the absence of nonspeci c immunoreactivity (data not shown). Iba-1-ir cells in captured images were manually counted by two operators blind to the experimental conditions using the cell counting plug-in in Image J (Version 1.45 J; National Institutes of Health, Bethesda, MD, USA). The brightness and contrast of images were adjusted to a rm setting for all sections to aid in visualization of immunostaining. A box of size 80000 µm 2 , 100000 µm 2 and 42000 µm 2 was overlaid on 100x (10x objective, 10x eyepiece) pictures of sections with cingulated cortex (Bregma: +0.74: +0.38), dentate gyrus (Bregma: -1.64: -2.12) and pre-optic area (Bregma: -0.94: -1.28), respectively, and region identi cation was based on the mouse brain atlas (Paxinos and Franklin, 2001; http://www.mbl.org/atlas170/atlas170_frame.html). Cells were counted in a total of 4 sections. Iba1-ir values were averaged from both sides of bilateral nuclei, and the count from 4 sections was summed for each mice. Total numbers of positive cells per animal were multiplied by six to estimate the number of cells per cingulate cortex, dentate gyrus and pre-optic area.
For morphological analysis of microglia, all analysis were performed on maximum intensity projections (Z-project, maximum intensity function in Image J) of the Z-stacks collected with 0.7µm increments on a Olympus Flowview (FV1000) confocal microscope with 10X eyepiece and 100X objective (1000X).
Microglia were classi ed as rami ed (numerous thin processes, radial branching), primed (thickened processes, increased polarity, and proliferation with reduced secondary branching), reactive (thickened stout processes with highly reduced branching), or amoeboid (rounded soma with no branching; Figure   1A-D) based on standard morphological criteria [17]. Sholl analysis was performed over 20 concentric circles, with the inner radius of 6µm and outer radius of 35µm on a total of 15 rami ed microglia (5 from cingulated cortex + 5 from dentate gyrus + 5 from pre-optic area) per brain and 5 brain/group. Analysis was done by 3 operators blind to the experimental conditions. Non-overlapping microglia cells that were in a rami ed state exhibiting intact microglial processes unobscured by either background labelling or other cells were photographed and analysed.

Statistical Analysis
Data are presented as mean (±SE). All statistical analyses were done using GraphPad prism (version 6.0), IBM SPSS Statistics version 20 software as appropriate. Two-way ANOVAs assessed the effect of sleep fragmentation (factor 1), 6-OHDA treatment (factor 2) and their interaction on mRNA expression of cytokines and Iba-1 immunohistochemistry in brain. A three-factor analysis, using Univariate General linear model (GLM) tested the effects of sleep fragmentation (factor 1), 6-OHDA treatment (factor 2), and distance from soma (factor 3), and their interactions on morphology of the Iba-1 staining. Bonferroni multiple comparisons were used for posthoc analysis. p < 0.05 was considered statistically signi cant.
In comparison to controls, IL-1β expression under both SF regimes was signi cantly increased in brain regions (except hippocampus of chronic SF exp; Bonferroni posthoc test, p < 0.05, Fig. 2a, e, i, 3a, i). Acute SF caused a signi cant reduction in expression levels of TGFβ in all three brain tissues, and in IL-6 in prefrontal cortex and hypothalamus (Bonferroni posthoc test, p < 0.05, Fig. 2b, d, f, h, j, l), as opposed to chronic SF which lead to a signi cant increase in IL-6 expression in prefrontal cortex, TNFα in hypothalamus and TGFβ in hippocampus (Bonferroni posthoc test, p < 0.05, Fig. 3b, h, k).
A recovery period of 1 week after chronic SF was enough to negate the effects of chronic SF (except IL-6 in prefrontal cortex). In fact, IL-6, TNFα and TGFβ expression in hippocampus and hypothalamus and IL-1β expression in hypothalamus was signi cantly lower in recovery mice compared with controls (Bonferroni posthoc comparison, p < 0.05 , Fig 4f-l). In addition, lack of signi cant differences between 6-OHDA administered controls and experimental groups in all experiments (except hypothalamic TNFα of chronic SF+R experiment), along with signi cant reduction in hypothalamic IL-1β and TNFα in 6-OHDAtreated chronic SF mice, and in TGFβ in hippocampus and cortex of 6-OHDA treated acute and chronic SF mice, indicated that chemical sympathectomy signi cantly attenuated SF-induced neuroin ammatory responses (Bonferroni posthoc test, p < 0.05, g. 2-4).
Hippocampal microglial response was still observed after 1 week of recovery, as rami cation was signi cantly higher in mice subjected to recovery at 12.5 to 14 µm distance from centre of soma (Bonferroni posthoc test, p < 0.05 , Fig 6g, h). Further, 6-OHDA administration diminished the neuroin ammatory response from chronic SF, as evidenced by signi cantly lower microglia rami cation in cortex (at 23 to 33.5 µm distance from centre of soma), hippocampus (at 17 to 23 µm, 26 and 30.5 µm distance from centre of soma) and pre-optic area (at 14, and 20 to 30.5 µm distance from centre of soma) in mice subjected to 6-OHDA treatment following chronic SF, relative to their counterparts injected with vehicle (Bonferroni posthoc test, p < 0.05, Fig 6c, d, g, h, k, l). The pre-optic area of recovery mice injected with 6-OHDA had lower microglia rami cation at 11 to 14 µm distance from centre of soma, compared with recovery mice not injected with 6-OHDA (Bonferroni posthoc test, p < 0.05 , Fig 6k, l).

Discussion
Sleep disturbances can promote neuroin ammation and blood-brain barrier disruption mediated by in ammatory molecules produced from astrocytes and microglia. For instance, sleep restriction for 72 h elevated IL-1β, IL-6 and TNFα in hippocampus and basal forebrain of rats [18] and chronic sleep restriction for 21 days elevated IL-6 and TNFα in hippocampus and TNFα in prefrontal cortex of rats [19]. Similarly, in the present study, sleep fragmentation (SF) promoted neuroin ammatory responses albeit in a brain region-speci c and SF-duration dependent manner. In addition to having proin ammatory roles, elevated IL-1β in brain following acute and chronic SF (except in hippocampus), and TNFα in hypothalamus following chronic SF may indicate an accumulation of SRS (sleep regulatory substances) following a period of prolonged wakefulness [20]. The increase in neural TNFα response after chronic but not acute SF is in corroboration with previous reports [12,21,22]. An induction of an anti-in ammatory environment in the brain after long-term SF has been reported before [21,23]. Increased hippocampal TGFβ expression following chronic SF perhaps provided evidence for neuroprotection from in ammation. Further, while primarily pro-in ammatory, IL-6 has been described to play anti-in ammatory roles in some scenarios, causing stimulation of IL-10 synthesis [24] and inhibition of IL-1 and TNFα production [25].
While the precise functioning of IL-6 is not deciphered in the present study, opposite trends in its mRNA expression under acute and chronic SF seems to suggest an effort to closely regulate neuroin ammatory response to SF.
Microglial activation is considered a main source of neuroin ammation among other factors [9]. Sleep loss has been shown to affect microglial morphology [26]. In mature CNS, microglia are in a 'resting state' with highly rami ed and motile processes, continuously surveying the surrounding environment. In response to injury, microglia shift from resting to a primed phagocytic 'activated' state [26]. Sleep loss, in the absence of injury, has been reported to increase phagocytic activity of microglia [8]. On the other hand, psychological and mechanical stress has also been shown to increase the microglial secondary branching in the cortex of rats [27,28], thus suggesting that microglial activation in response to homeostatic perturbations can range from mild non-in ammatory hyper-rami ed phenotype to proin ammatory, apoptotic and phagocytic phenotypes [26]. Unlike chronic OSA and SF, neither OSA nor sleep deprivation/restriction of 1 night has been previously reported to increase the phagocytic activity of microglia [8,29]. Therefore, we surmise that perhaps acute SF of 1 night might affect the mild 'nonin ammatory' instead of the 'activated' phagocytic morphology of microglia. Absence of microglial hyperrami cation in acute SF mice suggested that SF-induced neuroin ammatory changes may not necessarily be associated with changes in microglial morphology [26]. On the other hand, increased rami cation after chronic SF suggests that chronic sleep perturbations induced microglial morphological changes that may include an increase in 'activated' phenotype involved in enhanced phagocytosis of synaptic elements [8] and an increased rami cation of the 'resting' phenotype possibly involved in increased surveillance and scanning of microenvironment [27]. Given the reversibility and complexity of microglia morphology, especially across brain regions [30], we analysed only the rami ed morphology of microglia. SF-induced alterations in the phagocytic phenotype need to be tested in future studies, and thus presently cannot be ruled out.
Usually, in ammation from sleep loss return to basal levels after sleep recovery. However, depending on duration of sleep loss and recovery, some immune components may remain altered after sleep recovery [31]. In this study, while seven days of recovery completely alleviates in ammatory responses in peripheral tissues from a previous study [7], it was not su cient to restore homeostasis in brain. Elevated Iba-ir in all three brain regions and increased IL-6 expression in cortex of recovery mice indicated that neuroin ammatory responses persisted at least 1 week following chronic SF. Consistent with our nding, a recent study reported that even after three weeks of recovery from sleep loss, neuronal apoptosis, microglial activation and IL-6 response still occurred in hippocampus of mice [32]. Furthermore, similar to our results, anti-in ammatory responses lower than baseline levels have been reported in mice subjected to 20 days of recovery post SD of 192 hours [21]. In addition, we documented a signi cant reduction in cytokine gene expression in brain after 7 days of recovery sleep. Perhaps, these changes may be a compensatory response that limits neurological and cardiovascular dysfunction related to SF-induced hyperactivation of in ammation over prolonged periods of SF. Irrespective of direction of deviation from baseline, it is plausible that one possible mediator of delayed return of neuroin ammatory to baseline homeostatic state was SNS and not the hypothalamic-pituitary-adrenal axis, since high levels of NE were found after 1 week of recovery sleep [7].
Our initial hypothesis that the SNS contributes to neuroin ammatory responses to sleep loss has been supported, as chemical sympathectomy signi cantly altered the in ammatory responses in all three SF paradigms examined. These results reiterate the complexities of potential NEI (neuro-endocrine-immune) interactions by which SF can affect immune functioning [31,33]. Peripheral sympathectomy also successfully alleviated SF-induced neuroin ammatory responses. Perhaps one or more of the following established mechanisms contributed to these results: (1) a passive diffusion of peripheral cytokines to brain via circumventricular organs or an active transport with carrier proteins across blood brain barrier (BBB) affected the microglial activity and neuroin ammatory processes (2) peripheral signals stimulated the endothelial cells of BBB to secrete molecules affecting activity of neurons and glia and/or (3) the autonomic nervous system itself via vagal afferents activated the neuroin ammatory responses [10].
Interestingly, 6-OHDA treatment also altered baseline cytokine levels. More speci cally, TGFβ in hippocampus and IL-6 in hypothalamus was signi cantly lowered, while IL-1β in hippocampus was signi cantly increased in control mice after chemical sympathectomy in acute SF experiment. IL-6, TNFα and TGFβ in hypothalamus and hippocampus (except TNFα) of control mice were signi cantly attenuated after 6-OHDA treatment in the recovery experiment. The effects of chemical sympathectomy on baseline immune function in chronic SF experiment were less profound with only a signi cant reduction in hypothalamic TNFα expression in controls treated with 6-OHDA. Such differences between the three experiments could be attributed to the differences in experimental paradigms and duration of treatments involved (cf. Figure 1). Lastly, our data suggests that baseline neuroin ammatory responses of hypothalamus and hippocampus were more susceptible to peripheral 6-OHDA treatment than the cortex. Further studies are warranted to understand how peripheral signals activate neuroin ammation across different brain regions.
Historically, the bias towards use of male rodents in research has resulted in data gaps to effectively discern the effects of sleep abnormalities upon female physiology. Given the sexual differences in susceptibility to diseases including neurological and immune diseases [34,35] , this study provides critical insight regarding effects of sleep disruption on female peripheral and neural immune responses. Sex steroids regulate the development and maturation of immune cells and responses [35]. As result, in ammatory pro les [34,36] and microglial activity [36] differ between the sexes. Therefore, caution is warranted before extrapolating these results to responses in males. Paradoxical SD affects estrous cyclicity in rats [37]. Since we the reproductive status was not tested in mice of our study, the involvement of reproductive hormones in regulating SF-induced in ammation and microglial responses cannot be ruled out. Restraint stress also affects phasing of the estrous cycle in female rats, although there is large individual variation [38]. Because sample sizes in our study were small, future studies should evalute the effect of estrous cycling upon SF-induced in ammation using a greater sample size.

Conclusion
This study systematically evaluated the SF-associated neuroin ammation and microglial responses of female mice. Changes in in ammatory responses seemed to be re ective of stress-axes activation and were related to the duration of SF [7]. Unlike previous reports, our experimental approach of exposing mice to 8 weeks of SF can more accurately be expected to mimic sleep abnormalities, such as OSA and the diseases associated with it, in the modern world. Moreover, it is important to emphasize that a recovery sleep of seven nights following such chronic sleep fragmentation was clearly insu cient for a full return to homeostasis. Importantly, to our knowledge this is the rst study showing chronic SFinduced subtle changes in the morphology of microglia primarily characterised as 'resting'. Finally, this study provides evidence of a critical contribution of SNS in development of an in ammatory state. These ndings could lead to novel therapeutic interventions that target the SNS for treating in ammationdependent disorders, such as cardiovascular diseases.

Availability of data and materials
The data generated for this study are available upon request to the corresponding author. weeks (experiment 2: Chronic SF), or 8 weeks of chronic SF followed by undisturbed sleep of 1 week (experiment 3: Chronic SF + recovery). A horizontally moving automated sweeping bar set to swipe across the chamber every 2 minutes ensured experimental group mice were sleep fragmented. Controls in chronic SF and recovery experiment were also subjected to a sweeping bar, although at a higher speed, and for lesser duration (2 swipes/minute for 3 hours aligned with lights on) to control for the increased activity induced effects between groups. During the recovery phase, mice of both-control and experimental groups were not subjected to sweeping bar swipes. On last 5 days of the experiments, mice were injected subcutaneously with 0.9% saline or 6-OHDA (6-hydroxydopamine). After 24 hours of nal injection, mice were either decapitated for tissue gene expression study or transcardially perfused for immunohistochemical study. All mice were > 8 weeks of age and were subjected to 12h light: 12h dark cycles with lights on at 8am and Lights off at 8pm, and provided food and water ad libitum.

Figure 2
Acute Sleep fragmentation altered cytokine mRNA expression levels in brain tissues. Mean (± SE) gene expression levels of IL-1, IL-6, TNFα, and TGFβ in Prefrontal cortex (a-d), hippocampus (e-h), and hypothalamus (i-l) of vehicle-injected or chemically sympathectomized mice subjected to control or acute SF (n = 8/group). Asterisk (*) and pound (#) indicate a signi cant effect of SF and chemical sympathectomy, respectively, as determined by Bonferroni posthoc test followed by 2-way ANOVA. For statistical signi cance, alpha (α) was set at 0.05.

Figure 3
Chronic Sleep fragmentation altered cytokine mRNA expression levels in brain tissues. Mean (± SE) gene expression levels of IL-1, IL-6, TNFα, and TGFβ in Prefrontal cortex (a-d), hippocampus (e-h), and hypothalamus (i-l) of vehicle-injected or chemically sympathectomized mice subjected to control or chronic SF (n = 8/group except for chronic SF + vehicle group of n = 10 mice). Asterisk (*) and pound (#) indicate a signi cant effect of SF and chemical sympathectomy, respectively, as determined by Bonferroni posthoc test followed by 2-way ANOVA. For statistical signi cance, alpha (α) was set at 0.05.

Figure 4
One week of recovery altered cytokine mRNA expression levels in brain tissues. Mean (± SE) gene expression levels of IL-1, IL-6, TNFα, and TGFβ in Prefrontal cortex (a-d), hippocampus (e-h), and hypothalamus (i-l) of vehicle-injected or chemically sympathectomized mice subjected to control or one week recovery following chronic SF (n = 8/group). Asterisk (*) and pound (#) indicate a signi cant effect of SF and chemical sympathectomy, respectively, as determined by Bonferroni posthoc test followed by 2-way ANOVA. For statistical signi cance, alpha (α) was set at 0.05. Acute sleep fragmentation altered microglia activation. Mean (± SE) Iba-1-ir cell number (a, e, i), cell area (b, f, j), and rami cation as determined by sholl analysis (c-d, g-h, k-l) in cortex (cingulated cortex; upper panel), hippocampus (dentate gyrus; middle panel) and hypothalamus (pre-optic area; lower panel) of vehicle-injected or chemically sympathectomized mice subjected to control or acute SF (n = 5/group).
The magni ed view in each micrograph represents rami ed microglia (magni cation: 1000x) subjected to Chronic Sleep fragmentation and one week of recovery altered microglia activation. Mean (± SE) Iba-1-ir cell number (a, e, i), cell area (b, f, j), and rami cation as determined by sholl analysis (c-d, g-h, k-l) in cortex (cingulated cortex; upper panel), hippocampus (dentate gyrus; middle panel) and hypothalamus (pre-optic area; lower panel) of vehicle-injected or chemically sympathectomized mice subjected to control, chronic SF, or a recovery of 1 week following chronic SF (n = 5/group except vehicle treated Page 24/24 chronic SF + recovery group of n = 4 mice). The magni ed view in each micrograph represents the rami ed microglia (magni cation: 1000x) subjected to sholl analysis. Asterisk (*) and pound (#) indicate a signi cant effect of SF and chemical sympathectomy, respectively, on Iba-1-ir cell number and area, as determined by Bonferroni posthoc test followed by 2-way ANOVA. General linear model followed by Bonferroni posthoc test tested the effects on microglia rami cation. α and β indicate a signi cant difference of acute SF from control and recovery mice, respectively, while γ indicates a signi cant difference between recovery and control mice. Drug-induced effects are indicated by # and &, presenting a signi cant difference between vehicle and 6-OHDA treated mice of chronic SF and recovery, respectively. For statistical signi cance, alpha (α) was set at 0.05.