Differential gene expression in Brain and Liver tissue of rats after 9-day Rapid Eye Movement Sleep deprivation

Sleep is essential for the survival of most living beings. Numerous researchers have identified a series of genes that are thought to regulate “sleep-state” or the “deprived state”. As sleep has significant effect on body physiology, we believe that lack of REM sleep for a prolonged period would have a profound impact on various body tissues. Therefore, using microarray method, we have tried to determine which genes are up regulated and which are down regulated in the brain and liver of rats after 9-day REM sleep deprivation. Our results suggest that 9-day REM sleep deprivation differentially affects certain genes in the brain and the liver of the same animal. Highlights of the study ➢Gene expression profile of brain and liver tissues of rats was analysed after REM Sleep deprivation for 9 days by using microarray technique. ➢Many of the genes involved in essential physiological processes, such as protein synthesis and neuronal metabolism etc. are affected differently in the brain and liver tissue of rats after 9-day REM sleep deprivation.


Introduction
Many studies have been conducted by using rodents and primates to find out how sleep promotes survival and why prolonged deprivation is mostly fatal. It has also been analysed in non-mammalian species like the fruit fly (Drosophila melanogaster) [1][2][3], the zebrafish (Danio rerio) [4][5][6], the nematode (Caenorhabditis elegans) [7], and bees (Honey bee, Apis mellifera, and Bumble bee, (Bombus terrestris) [8][9][10][11]. Loss of sleep has been shown to have drastic effect on all animals studied thus far [12,13]. The level of physiological changes that sleep loss brings about and the fatality often varies depending upon the nature and duration of sleep deprivation [14,15]. Recent advancement in sleep research has shed light on both, a functional aspect of total sleep in general and rapid eye movement (REM) sleep in particular.
Many theories explain the evolutionary significance and functions of sleep, starting with "null" to "synaptic plasticity" theory [16,17]. Overall, sleep seems to have specific basic standard roles for all species evolved who need it [18]. While the simplest hypothesis advocates that a single basic role cannot be ascribed to sleep, numerous studies links it's loss to the detrimental effects on metabolism, behavior, immunity, cellular functions and hormonal regulations [19][20][21][22]. Sleep is therefore necessary, and a living being cannot be deprived of it for a long time.
There is some behavioral plasticity dependent on social or physiological state, with regulation of sleep [23,24]. In the case of drosophila, where not all sleep is considered necessary for survival [23], questions relating to the critical functions of sleep, plasticity and its overall importance has been studied.
REM sleep is an essential part of total sleep, and is present only in avians and mammals, with few exceptions like reptiles [25]. The functional aspect of REM sleep mainly relates to memory consolidation, brain maturation, muscle re-aeration, special memory acquisition, and maintenance of body physiology [26][27][28][29][30][31][32]. REM sleep is important for hippocampal, pruning, maintenance of new synapses during development and learning and generation of memory [33][34][35]. Some recent studies also suggest that lack of REM sleep can actually cause cell death in tissues and neurons [36][37][38]. Recently, REM sleep deprivation has also been found to be associated with acute phase response of liver, increased synthesis of pro-inflammatory cytokines such as IL1β, IL-6, and IL-12 and increased levels of liver enzymes, alanine transaminase and aspartic transaminase which circulates in the blood [39]. In addition, REM sleep deprivation-induces the production of reactive oxygen species (ROS) and nitric oxide (NO) in hepatocytes, along with an interesting increase in sensitivity to oxidative stress by the hepatocytes [40].
Microarray is a valuable tool for measuring the dynamics of gene expression in a biological system. As a result, microarrays can be used to measure the differences in gene expression profile in different tissues under the same physiologic conditions [18,41,42]. Most sleep studies have been performed using the entire brain, although there are some recent research where other organs has been used. [43][44][45][46][47][48][49].We became interested in comparing the effect on the brain and the liver from the same animal at the same time as our previous studies had indicated that REM sleep may have drastic effect on the liver [37,39,40]. For this purpose, we used the microarray technique to compare gene expression and identify the processes affected in the brain and liver of the same animal after REM sleep deprivation for 9 days. For example, we expected REM sleep loss to affect genes and processes related to circadian rhythm more in the brain and genes related to regeneration and immunity in the liver.
Our findings showed that REM sleep deprivation affected a total of 652 genes in the brain which were different from the 426 genes that were found to be affected in the liver. We found some genes belonging to certain pathways in the brain which were affected by REM sleep deprivation but were not affected in the liver, suggesting that the REM sleep deprivation interestingly affects these two organs differently. We conclude that REM sleep deprivation influences the molecular processes and biological pathways in both brain and liver by affecting different genes, thereby affecting the balance of physiology. Several of these genes identified by us are not recorded in previous studies and therefore taken together, we provide a dataset of genes that have been affected by REM sleep loss in the brain and liver. Our present study could be useful in future work on the detection of candidate genes linked to sleep or sleep loss disorders in general involving a particular organ.

Material and Methods
We used male Wistar rats weighing between 220-260 gm for this study. We kept the animals in the institutional animal house facility at 12:12hrs L: D cycle (7:00 am lights on) and provided the rats with food and water ad libitum. We have carried out all the experiments in compliance with the protocol approved by the Institutional Animal Ethics Committee of the University.

REM sleep deprivation procedure
Rats were REM sleep-deprived for nine consecutive days by using the flower pot methods [50,51]. In this method, we keep animals on a small raised platform (6.5 cm in diameter) surrounded by water compared to the control (large platform control, LPC) where the animals are kept on a large platform (12.5 cm in diameter) under identical condition. REM sleepdeprived animals could sit, crouch, and have a NREM-related sleep on this platform. However, due to requirement of atonia of muscle during REM sleep, they are unable to have REM sleep on the small platform. Whenever they try to get REM sleep., they fall into water and next time before the onset of REM sleep, they wake up and are thus deprived of it. Throughout our previous studies, there were no differences between cage control and LPC control group of rats, and thus only the LPC control group as control group was considered in this analysis [40,52]. Further, based on our previous studies we have used 9-day REM sleep deprivation in this study as unlike in the 4 day REM sleep deprivation it significantly affected the acute phase response and caused oxidative stress in the liver. Rats were sacrificed between 10 a.m. and 12 p.m. on day 9 and the total brain and liver were harvested and flash-frozen in liquid nitrogen for further analysis.

RNA extraction and quality analysis
Total RNA was isolated from the entire brain and liver samples using the Qiagen kit. Briefly, rats were anesthetized with isoflurane, and brain and liver samples were immediately dissected and frozen in liquid nitrogen. We isolated the total RNA from the whole brain and liver of each animal using Trizol (Gibco-BRL, Gaithersburg, MD, USA) as directed by the manufacturer.
The concentration of total RNA was measured using Nanodrop and the quality was analyzed using Bioanalyzer and the array was performed using good quality samples.

Microarray: labeling, hybridization, and data analysis
For microarrays, we took an equivalent amount of total RNA from the brain and liver. We outsourced Ocimum Biosolutions (USA) genomics facility for microarray analysis. In short, we used Affymetrix Rat Gene 1.0 ST Arrays containing more than 7000 annotated sequences and 18000 expressed sequence tags (ESTs). The Affymetrix Gene Chip Expression Technical Manual (Affymetrix Inc., Santa Clara, CA, USA) was used for the marking, hybridization and expression study of microarrays, mainly as previously reported [45]. The data analysis was performed using Affymetrix Expression Console and Programming Language-R [53,54].

Gene Ontology analysis
The functional annotation of differentially expressed genes was obtained from the Gene Ontology Consortium database, based on their respective biological process, molecular function and cellular component [55]. The overabundance of a particular term was determined on the basis of the number of significant genes in the experiment, the number of significant genes relevant to the term, the total number of genes for the organism, and the number of genes relevant to the term of the organism. We determined the functional categories enriched within genes that varied between control and REMSD. We used a single-tailed Fisher exact probability test based on the hypergeometric distribution to measure the p values. If the term is significant with say p< 0.05, it has been enriched with genes.

Pathway analysis
A pathway analysis of microarray data was performed using the Kyoto Encyclopedia of Genes and Genomes (KEGG) software. A number of biochemical pathways are identified by physiological processes documented in the KEGG databank. Since rat species-specific functional gene annotations are still few for many biological processes.These considerations make it difficult for the pathways to have a functional analysis of the results of the microarray.
We therefore tried to identify general pathways and species-specific pathways. When data from the rat genome were not available, we looked at other species databases and compared the reference pathways used in the analysis. We used the KEGG map pathway to illustrate the maximum visual impact of REM sleep loss on highly up-regulated genes involved in protein translation processes.

Validation of array expression with Real-Time quantitative qPCR
Following analysis of microarray results, a group of genes was selected for validation by qPCR based on their degree of change in expression. We conducted a correlation study between the effects of the microarray and the qPCR, and statistical significance was calculated (Fig. 1). For the microarray, the data input to the correlation study was the Log2 ratio value of the weighted average for each gene on the composite array representing all replicate animals. For qPCR, we used the mean Log2 ratio value stated by qPCR for all replicate animals. We selected six transcripts for validation of microarray analysis using RT-PCR (Table ST-1). Controls were established, along with experimental conditions, to rule out any possibilities for external variables to affect our experimental setup. We tested the respective mRNA levels with RT-PCR in real-time. Briefly, samples obtained from total liver and total brain tissue were frozen and stored separately at −80 0 C before mRNA was quantified. Total RNA was isolated using Trizol methods and re-transcribed using the ABI reverse transcription kit (Applied Biosystems, Catalog number: 4368814). TaqMan gene expression Master Mix (Applied Biosystems, Catalog Number: 4369016) and probes (Applied Biosystems, Supplementary Table ST1) used for quantitative analysisof mRNA. Each cDNA sample was analyzed in triplicate. The RT-PCR reactions for all focal genes and Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) were measured from the same cDNA sample and loaded onto the same 96-well analysis plate. We quantified the gene levels using 2 − ΔΔCt methods and GAPDH as a reference control gene for expression level normalization.

Statistics
The results of qRT-PCR are presented as a mean of ± SE. We used the Kolmogorov -Smirnov normality test to estimate the data normality. The non-parametric Mann-Whitney U test was used to compare the pairwise expression of the microarray and RT-PCR expression for the respective genes used for liver and brain validation. Expression validation experiments were performed on the basis of five rats per group. The array experiments were analyzed, maintaining a p<0.5 significance level. The KEGG bioinformatics map and diagrams used to illustrate findings and discussions. All statistical analyzes considered p< 0.05 to be significant and were performed and plotted using Graph Pad 5.1 and Origin 6.0 software.

General results
In the current analysis, we used the Affymetrix Rat Gene 1.0 ST Array. Primary data analysis was performed using Affymetrix Expression Console and R. A total of 311 up-regulated genes (Table ST- (Table ST-2, and ST-3D) in the liver (Fig. 2). Out of this pool, we found a set of genes that were commonly affected, either in the same or opposite direction, between the brain and the liver. For example, 4 of the 11 genes identified (Table ST- 1) were up-regulated in the brain and down-regulated in the liver (Fig. 2).In addition, we identified 3 genes (Table, ST-4C; namely Histocompatibility 2, class II DR alpha (RT1-Da, NM 001008847), Zinc Finger And BTB Domain Containing 6 (Zbtb6, NM 001108953), and Transmembrane protein 106B (Tmem 106a, NM 001004267) out of a total of 4 genes that were up-regulated in the liver and down-regulated in the brain (Fig. 2). These findings underpinned our hypothesis that REM sleep loss will affect different tissues differently. In order to further explore and refine this, we continued with the study of gene ontology and KEGG pathway analysis.

Gene Ontology analysis
Functional categories within genes that vary between regulation and REM sleep deprivation in the brain and liver have been established. We also grouped all processes and components into three main groups, namely biological processes, molecular functions, and cellular components ( Fig. 3-5). In addition, we classified each group into two subcategories based on their direction of regulation (upregulation and downregulation), e.g. biological processes (Fig.3, A-D), molecular functions (Fig. 4, A-D) and cellular components (Fig. 5, A-D).In our analysis, we used only significantly (p<0.05) affected processes. They were plotted on 3 dimensional graphs considering their p-values, the affected genes (node count) and the total genes in the database (database count, log-transformed). This gave us a better understanding of the processes most profoundly affected and the nature of regulation.
The top five brain-regulated processes / functions included translational elongation, translation, anatomical rRNA processing and erythrocyte differentiation (Fig. 3A, Biological Processes), structural components of ribosomes, protein binding, rRNA binding, translation regulator activity, and mRNA binding (Fig. 4A, molecular function) and ribosomes, cytosols, intracellular, small ribosomal subunits, and nucleolus (Fig. 5A, cellular component).While REM sleep loss negatively affected processes in brain related to behavioral fear response, locomotive behavior, dopamine receptor signaling pathway, dopamine synaptic transmission, and brain visual learning (Fig. 3B, biological processes), protein binding, serotonin receptor activity, serotonin binding, drug binding and G-protein coupled receptor activity (Fig . 4B, molecular function), plasma membrane, axon, membrane, dendrite, and extracellular space ( Fig. 5B, cellular component). Similarly, the liver was positively affected by Sarcoma, Gluconeogenesis, triglyceride metabolic process, negative regulation of fatty acid biosynthetic process and response to an organic cyclic substance (Fig. 3C, biological processes), pyridoxal phosphate binding, structural constituents of the cytoskeleton, identical protein binding, protein binding, and dimethyl arginase activity (Fig. 4C, molecular function), and endoplasmic reticulum, plasma membrane, membrane, lysosome, and cytosol (Fig. 5C, cellular component).

Pathway analysis
We have used the KEGG analysis to evaluate the pathways affected by REM sleep loss. We also categorized positive and negatively affected pathways in the brain and liver. They were plotted against their significance level (p<0.05), database count, and multiple genes affected by each pathway (node count).We illustrated up-regulated and down-regulated pathways in the brain (Fig. 6) and up-regulated pathways in the liver (Fig. 7A) and down-regulated pathways ( Fig. 7B). Based on these analyzes, we selected some of the most significantly affected pathways in each tissue. We further described them by illustrating the affected genes on the KEGG pathway map ( Fig. SF 1-4), which allowed us to visualize the components, proteins, and genes most affected by the process and further helped us in our discussion. We used subunits of ribosomes most strongly negatively regulated (Fig. SF1) and neuroactive legend-receptor interaction pathways (Fig. SF2) that were positively regulated in the brain. We also identified the cancer pathway (Fig. SF3) and glycerophospholipid metabolism pathway ( Fig.   SF4) that were negatively and positively regulated due to REM sleep loss. We further expanded and examined our results in the light of the available literature and tried to conclude our findings.

Discussion
The 9-day REM sleep loss study showed that many biological, cellular, and molecular processes were differentially affected in the brain and liver. These findings are consistent with our hypothesis and prediction that different body tissues will respond differently to REM sleep loss. Out of the hundreds of genes found to be significantly affected in the brain and liver, we identified few genes that were mainly shared between the tissues examined and the direction of regulation. Of the 11 genes widely regulated in the brain and liver, only four were identified as Wee1, slc2a12, Hrk, and Fam110b ( Fig. 2 and Table, ST-4A). The GO term analysis of molecular / biological functions illustrates the involvement of protein tyrosine kinase activity, carbohydrate transmembrane transport activity, and processes such as apoptosis (Table, ST-4A). Recent preprints and other studies support this concept of REM sleep deprivation that effectively results in the apoptotic death of neuronal and hepatocytic cells [36][37][38]. We could not identify any common genes between brain and liver which are positively affected (out of 2 probes).In addition, a total of 3 genes (out of 6 differentially expressed probes) that were upregulated in the brain and down-regulated in the liver were found to be ribosomal proteins (Hba-a1, Hba-a2 & Mup5) responsible for drug transport, oxidoreductase activity, heme binding, fatty acid biosynthesis processes and catalytic activity (  [56,57]. We assume that REM sleep loss is causing a cerebral hypoxia-like condition in the brain, and the Hba-a2 gene has been induced to respond to it as found in our array studies. The explanation why it has been regulated in the liver is difficult to explain and requires a more detailed study. Sleep restriction study showing an increase in free fatty acids in healthy men provided us to speculate that REM sleep deprivation can affect genes such as Mup5 which is involved in fatty acid biosynthetic processes differently under REM sleep deprivation condition. [58]. Genes that were found to be up-regulated in the liver and at the same time down-regulated in the brain (RT1Da, Zbtb6, and Tmem 106b) were associated with GO term function linked to stimulus response (Table, ST-4C). Zbtb6 is a homologous gene of zinc finger protein containing the BTB domain in mammals with unknown functions. Tmem 106b is found to be associated with lesser known functions except protein binding, dendrite morphogenesis, and lysosomal transport [58,59]. A recent research shows the dementia-related functions of Tmem 106b, regulated by microRNAs [60].
In addition, our analysis of biological, molecular and cellular processes in response to REM sleep loss showed that many of the important processes that are associated with maintenance functions linked to protein translations required for damage control are differentially effected in the brain and liver ( Fig. 2-4). Overall, it shows that brain processes are mainly up-regulated in response to REM sleep loss. At the same time, the liver becomes more actively involved in energy conservation. Therefore, negative responses indicate that REM sleep loss influences processes linked to the brain's fear response and locomotive activity related to the peripheral circadian clock, hemoglobin level, and transport of oxygen throughout the liver. Now, we have identified pathways that have been significantly influenced either positively or negatively using the threshold values (p<0.05) and have illustrated it (Fig. 6, 7A & 7B).
We further defined some of the pathways most affected in either direction due to REM sleep loss in the brain and liver to visually explain the effects using the KEGG pathway map ( Fig.   SF 1-4). We found that many of the ribosomal proteins involved in protein synthesis processes were positively affected in response to REM sleep loss in brain (Fig. SF1). Similarly, longterm sleep loss has been found to control several genes in the brain linked to DNA binding / regulation of transcription, immunoglobulin synthesis , and stress response [61,62].
Irrespective of the notion that Homer-1a is a key brain molecule in response to sleep loss in the mouse model, no substantial modulation of Homer's gene expression was found in our studies involving REM sleep loss in rats [48].
Additionally, REM sleep loss negatively affected several genes linked to neuroactive ligandreceptor interaction pathways in brain, primarily related to gamma-Aminobutyric acid, Human Thrombin receptor, and associated receptor signaling dopamine (Fig. SF2). A recent research reviewed sleep and protein-dependent synaptic plasticity, indicating that sleep deprivation impairs many of the related biological and physiological processes [63]. We have found that many of the pathways in the liver which have been upregulated are linked to metabolism, immunity, and depression (Fig. 7A). On the other hand, only a few downregulated pathways in liver have been established which includes nitrogen metabolism and circadian rhythm (Fig.   7B). These findings further indicate that REM sleep deprivation may affect the different set of genes as measured in our study compared to the total sleep loss studies in mice. It is also important to understand the species-specific and sleep phase-specific response of the loss of sleep function in order to understand its consequences.

Proto-Oncogene (K-Ras) and Fos Proto-Oncogene (Fos)
which are marked in cancer pathways ( Fig. SF 3). A number of recent studies have shown that sleep dysfunction / loss and cancer processes are very related [64][65][66][67][68][69][70]. However, some emerging evidence also suggests that sleep loss / insomnia prior to the onset of cancer is independently associated with cancer risk [65,69,71,72]. Ptgs2, an enzyme, plays a key role in various pathological processes by catalyzing conversion of arachidonic acid to prostaglandins [73]. Studies have shown that overexpression of Ptgs2 is associated with angiogenesis, metastases, and immunosuppression [75,76]. Pgst2 is also found to be associated with the chemoresistance of some malignant tumors, including liver, pancreatic, lung, esophageal, and gastric cancers [77][78][79]. Inhibition of Ptgs2 effectively increased the sensitivity of tumors to drugs [74]. Similarly, Bcl-2, Kit, K-Ras, and Fos genes have been found associated with cancer [75][76][77][78]. These genes play an important role in the sleep-wake cycle regulation and are shown to be correlated with sleep [36,37,[79][80][81]. At the same time, glycerophospholipid metabolism pathway was found significantly upregulated in liver ( Fig. SF 4). This includes the genes Phospholipase, PLa2g
The chronic sleep deprivation in rats affected the protein profile of Gpd2 in hypothalamic Astrocytes [89] The functional aspect of other genes affected (e.g., Pla2g12a, Cds2 and Plpp2) is lacking and needs further exploration.
Studies are very inconsistent with REM sleep deprivation and the behavior of locomotives in rodents. Several studies have shown that REM sleep loss induces locomotor activity [51,90,91] while others have shown decreased locomotor activity [92]. We consider these ambiguities to be related to procedural changes or the degree of REM sleep loss. To illustrate this, few studies used multiple pots compared to our classic single flower pot method for deprivation, and other studies have used less time for deprivation (72-96 hrs.) compared to ours, which was ~216 hrs.

Recent research supporting that REM sleep deprivation can affect locomotor activity in rats in
an inverted-U manner has been published [93,94].The widely accepted view in the scientific community is that sleep deprivation decreases the pain tolerance and increases the transmission of pain in multiple chronic pain conditions [95][96][97][98][99][100]. We also found conflicting reports on the sensory perception of pain that were negatively affected in our studies relative to others [101,102]. There are few studies indicating that it is only total sleep deprivation that raises the intensity to pain rather than REM sleep deprivation [103,104]. Nonetheless, selective REM sleep deprivation is correlated with enhanced placebo analgesia effects [105].
Furthermore, there are some report describing individual variations in REM sleep loss and its association with the perception of pain [106]. We assume that total and short-term sleep deprivation could lower the pain threshold, while long-term sleep deprivation increases it.
REM sleep deprivation and pain is also found to be significantly correlated with environmental conditions (e.g. dry or wet conditions). Whereas pain sensitivity was found to be enhanced in dry test conditions and no difference in wet conditions. [107]. This suggests that further work is needed to understand deeply the relationship between experience of pain and lack of sleep.
Reports suggest that chronic REM sleep deprivation is associated with increased metabolic rate in brown adipose tissue [108]. Several studies are suggesting that there is a strong correlation between sleep loss and an increased risk of obesity and diabetes [109][110][111].
Our approach to GO term and KEGG pathway analysis is quite relevant in the current era of genomics and sequencing, but it also involves discrepancies in gene function across organisms, distributed biases and biases linked to positive and negative annotations (more information in [56][57][58]. Like GO term analysis, KEGG analysis also has its limitations, apart from reducing the complexity of the data and helping to increase the explanatory power. One of the key disadvantages of KEGG is the independent consideration of pathways, even though crosses and overlaps occur in the natural system [115][116][117][118]. Therefore, the findings of our current study indicate that REM sleep deprivation affects the physiology of different tissues and organs differently. We therefore need further research into sleep deprivation to determine stage and tissue-specific effects in order to explain particular effects and to evaluate the influence of sleep deprivation in related disorders.

Conclusion
Our microarray analysis found that many of the physiological processes and the genes involved in the pathways are regulated differently in the brain and liver tissues in response to REM sleep loss. This supports the general concept of the body reacting to the stress induced by REM sleep loss, where the liver helps to combat the body's stress response. In general, brain cells have been shown to be more receptive and influence processes such as apoptosis, learning and memory, oxidative stress, and circadian rhythms in response to REM sleep loss. On the other hand, hepatocytes were found to be more affected by processes such as protein synthesis, stress balance and detoxification. Our study provides an excellent platform for visualizing the effects of REM sleep loss in the body and can be used as the underlying basis for further studies on sleep deprivation.

Acknowledgments
This research was carried out and funded by a laboratory grant from the SKK Laboratory at the