Analysis of sleep deprivation-associated Homer1 gene and protein acting on synaptic plasticity by bioinformatics and animal experiments

Homer1, an immediate early gene, is related to sleep deprivation (SD), and its protein products are involved in synaptic plasticity affecting the cognitive process. This study aimed to identify the SD-associated key Homer1 gene in the brain and explore the value of Homer1 proteins acting on synaptic plasticity in SD. GSE9441 was extracted from Gene Expression Omnibus (GEO) database. Differentially expressed genes (DEGs) between SD and Control samples were achieved by R software and were analyzed by the Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway, and gene set enrichment analysis (GSEA). Protein–protein interactions (PPI) network was built by the GeneMANIA databases. In animal experiments, male C57BL/6 J mice (aged 12–13 weeks) were sleep deprived for 6 h, followed by independent behavioral tests and in vitro assays. Morris water maze (MWM) was used to evaluate learning and memory function. The expression of hippocampal Homer1 proteins was detected by Western blot analysis and its distribution in CA1 by immunohistochemistry and immunofluorescence staining. Synaptic plasticity was assessed by Golgi staining and long-term potentiation (LTP) testing in the hippocampal CA1 region. Homer1 was the hub gene most associated with SD, and its protein products specifically acted on the regulation of synaptic plasticity in bioinformatics. SD mice exhibited spatial memory impairment accompanied by increased Homer1a expression in hippocampal tissue and CA1 region. SD did not induce Homer1b/c overexpression of mice in the hippocampus. SD impaired the hippocampal synaptic plasticity of mice by reducing the density of dendritic spines and inhibiting LTP in the hippocampal CA1 region, which may involve the overexpression of Homer1a in the hippocampus. Homer1 gene is a core brain molecule associated with acute SD, and its protein product Homer1a is involved in the changes in cognitive brain function following short-term SD, especially the impact on hippocampal synaptic plasticity.


Introduction
Sleep plays an important role in people's physical and mental health [1]. Good sleep can not only effectively enhance the brain's ability to remove harmful substances, but also help maintain brain function [2], including mental health, supporting learning and memory, and regulating synaptic plasticity, in the process of which genes involved in sleep, especially those expressions with the circadian rhythm in brain, are crucial regulators [3]. Recent research evidence has shown circadian regulation of sleep-dependent gene expression produces short-term or even long-term changes in brain behavior and function [4]. Conversely, sleep deprivation (SD) may damage the brain by disrupting gene expression related to the sleep-wake cycle of synaptic proteins [3,5]. SD can affect gene expression, especially those involved in neuronal plasticity and neuropsychiatric disorders, which are critical for brain health [3,6].
One of the most important effects of sleep deprivation is the regulation of the expression of immediate early genes (IEGs) [7], which may determine the activation or inhibition of cellular signal pathways leading to changes in cellular functions such as synaptic protein synthesis, neurotransmission, synaptic plasticity, and neuronal function [8,9]. An important member of SD-associated IEGs was Homer1, which was encoded by mRNAs and translated into scaffold protein transported to postsynaptic density [10]. Homer1 is widely expressed in the brain, whose protein products consist of two alternative splicing variants, which are short-chain Homer1a and long-chain Homer1b/c, respectively [11]. Homer1a, a key component of the excitatory postsynaptic scaffolding complex, has been shown to affect the process of excitatory synaptic plasticity by regulating the type 1 metabotropic glutamate receptor 1/5 (mGluR1/5) [12], which is strongly linked to neuropsychiatric states such as sleep deprivation, sleep synaptic homeostasis, anti-depression and brain injury [13][14][15][16]. Since it is proteins that carry out the cellular activity and synaptic transmission, Homer1 protein products can act as postsynaptic function actuators, which play important roles in synaptic plasticity [5,14,17]. However, the potential value of the Homer1 gene and protein products in regulating synaptic function in SD is poorly understood, hindering the elucidation of the underlying pathological mechanism of SD-associated neuropsychiatric diseases in the brain.
Studies on immediate early geneexpression of the sleep-wake cycle and SD have shown that Homer1 is up-regulated not only during prolonged wakefulness rather than during sleep [3], but also during sleep rather than during wakefulness [12]. However, Homer1, as a key brain molecule driving sleep and coping with SD [3], lacks a comprehensive network analysis to clarify its bioinformatics characteristics, especially its associated signal pathway with synapse. The Gene Expression Omnibus (GEO) database allows users to freely access gene expression profiles by microarrays for functional enrichment analysis tools, which allow a multi-dimensional analysis of the enrichment of information such as locations, properties, biological functions, and signaling pathways of differentially expressed genes (DEGs) under different sleep/wakefulness states. Interestingly, the integrated bioinformatics analysis approach can effectively identify the candidate hub gene linking to impaired synaptic plasticity in SD and offer a non-biased view on the value of Homer1 in the regulation of synaptic function states after short-term SD. We conducted a comprehensive bioinformatics analysis and integrated the Gene Expression Series (GSE) dataset called GSE9441 from 6-h SD and non-SD bran samples for the analysis of the Homer1 gene, whose gene expressions and protein products are involved in synaptic regulation [3,6,14,18]. In addition, we also applied animal experiments to further explore the role of hippocampal Homer1 proteins and their regulated synaptic plasticity in SD-induced learning and memory impairment. This study will help to understand the role of Homer1 in the impaired synaptic plasticity hypothesis associated with SD.

Animals
In this study, 12-13 weeks aged male C57BL/6 J mice were included and housed in the 12-h light/12-h dark cycle with free access to water and food, and were divided into two associated and independent phases, including comprehensive bioinformatics analysis (two groups, n = 3) and animal experiment (three groups, n = 30). Animal research was approved by the Ethics Committee of Tianjin Medical University and conducted following the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals.

SD procedure
For the microarray experiments, mice were sleepdeprived for 6 h (from the light onset, ZT0 to ZT6) and killed with their undisturbed controls. SD was achieved by a gentle-handling method consisting of cage-tapping, the introduction of novel objects in the cage, or approaching a pipette next to the mouse as soon as a sleeping behavior was observed. For the animal experiments, mice were subjected to 6-h SD using a sleepdeprived rod (Model 80,391, Lafayette Instrument, USA). The bar was set to scan once every 30 s to ensure that the animals could not fall asleep. When the bar scanned back and forth, the animals were forced to move or cross the bar to prevent being pushed, thereby achieving completely short-term SD. In order to control the possible movement effect of the rod, we set the control of non-SD (NSD): The feeding conditions, environment, and exposure time are consistent with the SD model. The rod was set to continue scanning for 15 min every one hour with a 7.5 s/scanning cycle for a total of 6 h to ensure that the animals in the NSD experienced the same amount of exercise as the SD group and had sufficient opportunities and time to enter and maintain sleep.

Bioinformatics analysis
(1) Dataset acquisition: GSE9441 was selected as the analysis dataset from the GEO database (https:// www. ncbi. nlm. nih. gov/ geo/). 3 mice receiving sleep-deprived for 6 h were sacrificed along with 3 NC samples to take brain tissues. Microarray datasets of all samples with the Affymetrix Mouse Gene 2.0 ST Array were processed by the Affy package of R language. (2) Identification of DEGs: Data processing was performed by the Robust Multi-array Average algorithm. We used the online analysis tool GEO2R (https:// www. ncbi. nlm. nih. gov/ geo/ geo2r/) to process mRNA expression profiles between the SD and NC samples. In the mRNA profile chip, there were probe ID numbers, P values with differential analysis and log fold change (FC) values, as well as the corresponding gene name. Potential DEGs of the dataset were screened by the Limma package of R language. In this study, DEGs were defined according to the screening criteria of |log FC|> 1. (3) Function Enrichment Analysis of DEGs: We performed Gene Ontology (GO) function enrichment analysis on DEGs using enrichGO package of R language, and pathway enrichment analysis on GEGs in Kyoto Encyclopedia of Genes and Genomes (KEGG) "PATHWAY" using enrichKEGG package of R language. DEGs with adjusted p < 0.05 in the GO and KEGG enrichment analysis were considered statistically significant. The functions and signaling pathways of DEGs were visualized using the enrichplot package of R language. Besides, we used the Gene Set Enrichment Analysis (GSEA) Software for enrichment analysis of DEGs to further study the correlation of functional genes with SD and to explore possible regulatory pathways of DEGs. (4) Protein-Protein Interaction (PPI) Network Construction and Hub Genes Analysis: We used the STRING database to construct a PPI network for selected DEGs with a combined score > 0.4. MCODE was used to screen the most important modules and hub genes in the PPI network of DEGs. In addition, the GeneMANIA database was used to further analyze the protein interaction of hub genes in DEGs. (5) Acquisition of key gene Homer1: In this study, the key gene in the brain of SD was defined according to the following criteria: the absolute value of log FC in DEGs > 2 and adj. p-value < 0.05, and hub gene.
Using the criteria to identify whether Homer1 was the key gene.

Morris water maze
Mice were used for the Morris water maze (MWM) test to evaluate the effect of SD on long-term spatial memory. Before the test, animals were put into the pool (without the platform) and swam freely for 120 s to make them familiar with the water maze environment. The positioning navigation test lasted for 5 days, and each rat was trained 4 times per day. The platform was placed in the fourth quadrant, which was recorded as the target quadrant. From any of the four starting points of the pool wall, mice were put into the pool facing the pool wall. The free video recording system tracked the trajectory of the mice, including escape latency, that is, the time to reach the platform. When animals found the hidden platform within 60 s, they were allowed to rest on the platform for 10 s. If the hidden platform is not found within 60 s (the latency was recorded as 60 s), they were manually helped to climb on the platform and rest for 10 s. The average value of the escape latency of the four training tests was taken as the academic performance of mice on that day. Afterward, the animals were treated with 6-h SD and tested for the spatial probe test 24 h after the end of the last training. The hidden platform was removed, and the animals were put into the water from the opposite (the second quadrant) of the target quadrant. The number of entries in the platform zone and time spent in the target quadrant within 60 s were recorded.

Western blot analysis
Mice were euthanized with pentobarbital 100 mg/kg, and hippocampus tissue was decapitated. Hippocampal tissue was prepared with RIPA lysis buffer and PMSF protease inhibitor, lysed for 30 min, centrifuged at 4 ℃ and 12,000 rpm for 10 min, and the supernatant was aspirated. Protein was quantified by BCA assay. Primary antibodies of Homer1a (dilution ratio 1:1000, Santa Cruz, USA), Homer1b/c (dilution ratio 1:1000, Abcam, USA), and GAPDH (dilution ratio 1:1000, Abcam, USA) were added to the membrane and incubated overnight at 4 °C in a shaker. Secondary antibodies (Rabbit anti-mouse or Sheep anti-rabbit, dilution ratio 1:3000, Abcam, USA) were added and continued to incubate at 37 ℃ for 1 h. The optical density values of the target bands were analyzed with Image J software.

Immunohistochemistry staining
Brain tissues from perfused mice were fixed to prepare paraffin sections. Primary antibodies Homer1a (1:100 dilution, Santa Cruz, USA) and Homer1b/c (1:100 dilution, Abcam, USA) were added dropwise to each section, The second antibody was rinsed with PBS three times, and then nuclear stained with DAPI incubation sections. ImageJ software was used to measure the immunofluorescence density of Homer1a and Homer1b/c in the pyramidal cell layer of hippocampal CA1 region of each slice.

Golgi staining
Mice were processed using FD Fast Golgi staining kit to observe the dendritic spines of neurons in hippocampal CA1 region. Brain tissue after Golgi immersion was cut into coronal sections with a thickness of 100 μm. Tissue sections were placed in Golgi staining solution for staining for 10 min and then restained with crystal violet. Gradient alcohol was used for dehydration treatment, xylene was used for transparent treatment, and a neutral resin sealing agent was used for sealing cover glass. Images of dendritic spines in the hippocampal CA1 region were acquired under a 400-fold microscope and analyzed with row dendritic spines density by ImageJ software.

LTP testing
Brain samples were placed in the ice water mixed artificial cerebrospinal fluid pretreated with 95% O 2 + 5% CO 2 mixed oxygen saturation for continuous saturation oxidation. A stimulating electrode tip (MCE-100/200) was placed in the hippocampal Schaffer collateral, and a recording glass microelectrode was placed at the apical dendrite of the hippocampal CA1 cone neuron strip in the range of 200-250 μm from the stimulating electrode tip. A micro manipulator system (Scientifica) was activated to accurately control the position of the recording electrode. Classical field Excitatory Postsynaptic Potentials (fEPSPs) were recorded at 15-20 min. LTP was induced by highfrequency stimulation (3 continuous stimuli, 5 s apart, 100 stimuli per 100 Hz), and the signal at 60 min after induction was recorded. The fEPSPs electrophysiological signals of the hippocampal brain slices were collected, recorded, and analyzed by using a computer brought in with the patch-clamp system, pClamp, and Clampfit software.

Statistical analysis
In this study, R language, SPSS 22.0, and GraphPad Prism 9.3 software were used for statistical analysis. Measurement data were described as mean ± standard deviation (SD). One-way ANOVA was used for comparison among multiple groups, and the Tukey test was used for further pairwise comparison. The value of p < 0.05 was considered statistically significant.

Key gene homer1 analysis
According to the screening criteria, A total of 10 DEGs associated with SD in 8 relevant samples of the GSE9441 dataset were selected: 7 up-regulated genes and 3 down-regulated genes (Supplementary Table 1). The heatmap and volcano plot of DEGs were drawn, respectively ( Fig. 1), showing that Homer1 was the most significant up-regulated expression pattern (marked in red) in the SD group and that DEGs such as Hspa1b, Hspb1, and BDNF were also significantly highly expressed in the SD group. In functional enrichment analysis, the GO enrichment analysis of the bioinformatic function of DEGs presented by the bar and circle plots (Fig. 2) showed that 27 terms in biological process (BP) (Fig. 2A), and the function of Homer1 gene was mainly involved in regulating protein localization to synapses, protein localization to cell junctions, G protein-coupled glutamate receptor binding, synaptic structural components, and postsynaptic structural components (Fig. 2B). In addition, the GSEA showed that highly expressed BDNF, Hspb1, and Hspa1b were mainly enriched in the MAPK signaling pathway (Supplementary Fig. 1). According to the filter threshold, 5 Hub genes, including Homer1, BDNF, Hspa5, Hspa1b and Hspb1, were obtained, and constructed the PPI network based on the GeneMANIA database, indicating that Homer1 was located in the most significant module (Fig. 3).

SD-induced spatial memory impairment
Mice in the three groups underwent the same 5-day spatial learning training in MWM, while SD mice showed less number of entries in the platform (p < 0.05) and shortened time spent in the target quadrant (p < 0.05) in the probe test, indicating impaired spatial memory. Mice in the NSD group that received only exercise control without SD had complete spatial memory function as the Controls (Fig. 4).

The changes of Homer1 proteins in the hippocampus and CA1 region following SD
Western blot analysis of hippocampal protein extracts showed that SD induced Homer1a protein upregulation (Fig. 5). Compared with the Control and NSD groups, the expression levels of Homer1a protein in the hippocampus of mice in the SD group were significantly increased (p < 0.05), while Homer1b/c protein had no similar changes. No significant difference in Hippocampal Homer1a protein expression was shown between the Control and NSD groups. In immunohistochemistry staining in the hippocampal CA1 region, Homer1 proteins were stained into yellowish brown or brownish yellow granules, mainly distributed in the cytoplasm and a small part in the nucleus (Fig. 6). The AOD value of Homer1a protein in the hippocampal CA1 region of mice in the SD group was significantly higher than that in the Control and NSD groups (p < 0.05), while the AOD value of Homer1a in the hippocampal CA1 region was similar between the Control and NSD groups. There was no difference in Homer1b/c immunoreactivity in hippocampal CA1 region of mice in the three groups.
Immunofluorescence staining of the hippocampal CA1 region showed that the fluorescence density of Homer1a protein in the pyramidal cell layer of the hippocampal CA1 region in SD group was significantly higher than that in control group and NSD group (p < 0.05), while the fluorescence density of Homer1a protein in the hippocampal CA1 region in control group and NSD group was similar (Fig. 7). There was no difference in Homer1b/c immunofluorescence density of pyramidal cell layer in the hippocampal CA1 region among the three groups.

The effect of Homer1 on synaptic plasticity
The KEGG pathway enrichment analysis was presented by the bubble and circle plots (Fig. 8) and showed that 13 signaling pathways were involved in DEGs, among which Homer1 specifically acted on the glutamatergic synaptic signaling pathway. In addition, the KEGG pathway analysis found Homer1, as an endogenous ligand of mGluR5, regulated excitatory neuronal plasticity ( Supplementary  Fig. 2).
Our animal experiments showed that the overexpression of Homer1a after SD was accompanied by the inhibition of synaptic structure and function in hippocampal CA1 region. Golgi staining of neurons in hippocampal CA1 region (Fig. 9) exhibited that the density of dendritic spines of mice in the SD group was significantly lower than that in the Control and NSD groups (p < 0.01). There was no significant difference in dendritic spine density between the Control and NSD groups. In the LTP testing, the baseline fEPSP slope levels of mice in the three groups were consistent, while they had significant changes within 60 min after HFS (Fig. 10A). Mice treated with SD had a significant decrease in the fEPSP slope of CA3-CA1 region of hippocampus (p < 0.01), while mice in the Control and NSD groups had no such changes (Fig. 10B).

Discussion
One of the most important effects of SD was the changes in gene expression [3], which reflected activation or inhibition of cellular processes that led to altered neuronal function and synaptic connections [19]. Therefore, analysis of the changes in gene expression would provide a direction for understanding the effects of SD on brain function. This study showed that there were significant differences in 10 transcripts, among which Homer1,  One-way ANOVA and Tukey test were used for the analyze the data with mean ± SD. *p < 0.05, **p < 0.01, ns, no significance Hspa1b, Hspb1, P4ha1, BDNF, Dio2, and Hspa5 were the up-regulated DEGs, while Cirbp, Tagtp2///Tatp1, and Oasl2 were the down-regulated DEGs. The DEGs, including HOMER1, BDNF, HSPA5, HSPA1B, and HSPB1, were proven to be hub genes in the PPI network. SD-associated Homer1 was found to be the key gene in the submodule and specifically enriched in glutamatergic synaptic plasticity. In addition, Homer1a protein, not Homer1b/c  One-way ANOVA and Tukey test were used for the analyze the data with mean ± SD. *p < 0.05, ns, no significance protein, was up-regulated in the hippocampus in the SD-induced memory impairment animal model and was accompanied by the inhibition of synaptic plasticity. Current studies aimed at analyzing gene expression of SD are often difficult to compare due to the use of different sleep-deprivation methods, durations, and animal models. In addition, the methods used to quantify gene expression levels vary widely [20]. For example, the decision to measure protein or mRNA may be biased towards different outcomes depending on the timescale of the experiment. Therefore, when using bioinformatics analysis to screen SD-related gene expression in the GEO database, it is also necessary to unify experimental standards and conditions and try to avoid potential confounding factors and heterogeneity. In this study, we unified the baseline levels of the filtered microarray dataset provided by GSE9441, including the rearing environment, species, age, and mRNA microarray type. Notably, gene differential expression varied greatly with the duration of SD [3]. Our data showed that after 6 h of SD, only a few genes were significantly altered in expression, among which the Homer1 gene was the most closely related to SD. Homer1 has two subtypes, namely the short variant Homer1a and the long variant Homer1b/c [10]. Homer1a, as an important member of IEGs [21], was widely up-regulated in the brain of sleep-deprived rats [3,22]. In a study consistent with the SD duration of our study, the results showed that Homer1 was up-regulated in the basal forebrain and cerebral cortex of SD mice [7]. Another study of Homer1 gene expression after 3, 6, 9, and 12 h of SD found that the transcription of the cerebral cortex was upregulated in all these periods [23]. A meta-analysis of SD-related gene expression found that Homer1a levels were upregulated in the cerebral cortex, hippocampus, Fig. 7 immunofluorometric assay of Homer1 proteins in the hippocampal CA1 region of three groups of mice (n = 6). A Immunofluorescence staining of Homer1a and Homer1b/c in pyramidal cell layer of the hippocampal CA1 region under 100 × high power microscope. B Comparison of fluorescence density of Homer1a immunoreactive substance. C Comparison of fluorescence density of Homer1b/c immunoreactive substance. One-way ANOVA and Tukey test were used for the analyze the data with mean ± SD. ****p < 0.0001, ns, no significance olfactory bulb and striatum in mice with 6 h of SD [24]. Although some studies on SD-related gene expression did not specify which Homer1 subtypes were measured, it was likely Homer1a, as most other studies clearly indicated that Homer1a was up-regulated by SD [3,12,13,15,25], while Homer1b/c was much less affected by SD [25]. In addition, SD upregulated Homer1a mRNA levels in the morning (sleep period) but not at night (Awakening period) [25,26], while also indicating that Homer1a, an important regulatory molecule in driving sleep [12], was dramatically affected by SD. A large amount of evidence has shown that SD impairs learning and memory function [6,19], in which impaired synaptic plasticity may act as a key mechanism [27]. Homer1 is an interesting brain gene associated with SD [3], whose protein products play an important role in the regulation of excitatory synaptic structure and function and intracellular signal transduction [14]. In this study, Homer1a has been identified from genomics and animal experiments as a specific brain marker molecule for SD-induced spatial memory impairment. Homer1a is widely upregulated in the brain of short-term (6 h-48 h) sleep-deprived animals, including the hippocampus, a key brain area that regulates cognitive function [3]. Our 6-h SD animal models also obviously induced the overexpression of Homer1a protein in the hippocampus, and its high distribution in the hippocampal CA1 area that determines the storage of spatial memory, which may be an important mechanism molecule for acute SD to damage hippocampal synaptic plasticity. A Science study showed that Homer1a is also a key driver of sleep synaptic homeostasis [12]. Acute SD for 4 h significantly induced the overexpression of Homer1a in mouse hippocampal synaptosomes, which was 125% higher than the sleep level, so that synapses that should have been "elongated" in the awake period were still in a "shrinking" state. The imbalance of sleep synaptic homeostasis is an important mechanism for SD-induced cognitive dysfunction. Since the overexpression of Homer1a is induced by SD, Homer1a protein with splice variants regulates synaptic structure and function in an activity-dependent manner in a negative feedback loop [10], which may disrupt sleep synaptic homeostasis and become an important mechanism of SD-induced cognitive dysfunction. Dendritic spines are the main sites of synaptic transmission in the central nervous system [28]. Together with LTP, an important form of learning and memory, they participate in the regulation of hippocampal synaptic plasticity [29]. Rodent studies have shown that SD reduces dendritic spine density and weakens synaptic transmission function in the hippocampus, including inhibition of LTP [14,30,31], which is also validated in our animal experiments. SD affects synaptic plasticity in different ways, among which the hot signal molecules are Homer1 gene transcription and protein translation that can change and regulate synaptic strength and control synaptic plasticity during sleep and SD [3,12]. Prior studies have shown that Homer1a reduces the density and size of dendritic spines in the hippocampus and dominates the "shrinking" process of synapses [12,30]. A cytological study has also confirmed that Homer1a not only destroys the morphology of dendritic spines of cultured hippocampal neurons, but also reduces the excitatory synaptic response of CA1 pyramidal neurons [32]. In addition, in the neuronal cells transfected with Homer1a by gene knockout, decreased amplitude of alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor-mediated excitatory postsynaptic current was detected, which is consistent with the decreased expression of AMPA receptor on the synaptic surface detected by immunohistochemistry [33]. Our animal experiments confirmed histologically that SD induced high expression of Homer1a in the hippocampus and CA1 region, accompanied by decreased dendritic spine density and weakened synaptic plasticity LTP. Integrating research on the association between SD and synapse and our bioinformatics analysis and animal experiments, it can be seen that Homer1a, a protein product of the Homer1 gene, may be involved in synaptic plasticity damage associated with SD.
This study still has some limitations. First, the sample size of this study is small, and there may be heterogeneity in the interpretation of the results. Additionally, we only explore the genetic changes after 6 h of SD, which may not be fully representative. It still needs a comprehensive and detailed analysis based on a larger sample size to further verify our findings. Although this study also found that SD-related synaptic plasticity inhibition was accompanied by hippocampal Homer1a protein overexpression, no method of inhibiting Homer1a was used to verify the causal mechanism. In addition, this study lacks the downregulation of Homer1 gene to observe and verify whether SD induced cognitive impairment and synaptic plasticity damage can be improved.