A Study of Icariin Modulating Hippocampal Neurogenesis in Depression Based on Quantitative CSF Proteomics

Background Hippocampal neurogenesis dysfunction is one of the main pathogenesis of depression. Icariin (ICA) has signicant anti-depression and anti-hippocampal damage effects but could not effectively cross the blood-brain-barrier and accumulate in brain. Based on the proteomics of cerebrospinal uid (CSF), this study aimed to explore the mechanism of ICA against hippocampal neurogenesis dysfunction in depression. Methods In vivo, rats were exposed to 6-week CUMS and treated by ICA to observe the effects of ICA on depressive symptoms, cognitive functions, neurogenesis and number of neurons in DG. Tandem mass tag (TMT) proteomics were used to screen the differentially expressed proteins (DEPs) in CSF co-regulated by CUMS and ICA. Parallel reaction monitoring (PRM) was used to validate 10 DEPs that were related to cell proliferation and survival. In vitro, CSF was conducted on primary hippocampal neural stem cells (NSCs) to observe the proliferation and differentiation under high-corticosterone (CORT) concentration. It was effectively A total 52 DEPs HtrA1 These indicate


Introduction
Depression is one of the most common mental disorders, mainly characterized by despair, anhedonia and even suicide attempts [1]. With the increase of life pressure and social stress, the prevalence and morbidity of depression continue to rise. It has generated huge burdens on individuals, families and society, causing worldwide attention. Depression is a psychologic disease with complex etiology and clinical manifestations, so far, its pathogenesis has not been fully elucidated.
Hippocampal damage is currently recognized as one of the pathogenesis of depression, and hippocampal neurogenesis dysfunction plays an important role in it [2][3][4]. It has been found by clinical autopsy that the hippocampus volume reduced with severe neuronal loss or atrophy in depression patients [5]. Preclinical researches proved that the branches number and the length of hippocampal neurons dendritic were reduced and neurogenesis was impaired in depression animals [6,7]. Therefore, how to repair hippocampal damage has become the focus of current anti-depression treatment [8].
Icariin (ICA) is one of the most dominating bioactive constituents in Herba epimedii. Recently, plenty of evidence con rmed that ICA could effectively alleviate depressive symptoms and protect hippocampus by relieving neuroin ammation and repairing hypothalamic-pituitary-adrenal axis dysfunction [9][10][11]. However, studies have shown that ICA could not effectively cross the blood-brain barrier and always eliminated rapidly and thus it accumulated very little in brain [12]. Except the mild direct effect on the brain parenchyma, whether ICA protects hippocampus in depression by other approach.
Cerebrospinal uid (CSF) is the ultra ltrate from plasma, which is in direct contact with the central nervous system (CNS). CSF is bound up with hippocampus which is adjacent to the lateral ventricle. The substances in CSF may directly affect structure and function of hippocampus, such as regulating neurogenesis [13,14]. CSF plays an important role in molecular exchange and signal transmission in the CNS. Consequently, the alteration of CSF components might result in the occurrence and development of CNS diseases [15][16][17]. As is known that the proteomics of CSF changed in patients with depression and the differential expressed proteins were closely linked to the damage and dysfunction of the CNS [18,19].
Based on the relationship among the alteration of CSF components, depression and hippocampal damage, it was hypothesized that ICA might in uence brain parenchyma through the limbic regions of brain, such as CSF and choroid plexus, and exert anti-depression and against hippocampal damage.
To verify this hypothesis, in vivo, chronic unpredictable mild stress (CUMS) model was applied to observe the effects of ICA on anti-depression and hippocampal protection. The proteomics were used to screen the differentially expressed proteins (DEPs) in CSF co-regulated by CUMS and ICA. In vitro, the stress damage of neural stem cells (NSCs) was simulated under the high-CORT concentration, and the effect of ICA-treated CSF on proliferation and differentiation of NSCs under high-CORT concentration was explored. Through the above experiments, this study aimed to explore the mechanism of ICA against hippocampal damage in depression.

Animals
A total of 45 male Wistar rats weighing 180-220 g and aged 7-8 weeks (License No. SCXK (Yue) 2016-0041) were obtained from the Laboratory Animal Center of Southern Medical University, Guangzhou, China. Rats were housed (5 per cage) for 1 week for acclimation to the environment (23 ±2℃; 48-60% humidity; and on a 12 h light/12 h dark cycle) with diet and water available. All experimental procedures and protocols were approved by the Experimental Animal Ethics Committee of Guangzhou University of Chinese Medicine of China. The experimental procedure followed the United States National Institutes of Health Guide for the Care and Use of Laboratory Animals (NIH Publication No. 85-23, revised 1996).

Drugs and treatment groups in vivo
After adaption for one week, all rats were solitary-housed for sucrose preference test (SPT; as described in 2.4 Sucrose preference test (SPT)). Unquali ed rats were excluded according to the following SPT results: low sucrose preference (less than 60%), location preference (preferred to drink liquid from a xed location), drinking too little (drinking neither sucrose solution nor pure water) and excessive drinking (total liquid consumption more than twice the average total liquid consumption of all rats).
Rats were randomly divided into three groups based on body weight and sucrose preference: the control group (CON, n=15), the CUMS group (CUMS, n= 15) and the ICA group (ICA, n=15). ICA (Dilger, Nanjing, China; purity≥98%) was dissolved into a suspension (30 mg/mL) with normal saline. Rats in the ICA group were intragastrically administrated with ICA at a dose of 120 mg/kg and the CON group and CUMS group were intragastrically treated with the same volume of normal saline at 16:00 every day during CUMS modeling.

CUMS procedures
Rats in the control group were housed 5 per cage and underwent normal feeding without any stressors. Rats in the CUMS and ICA group were solitary-housed and subjected to CUMS for 6 weeks. According our previous study [4], the CUMS procedures were adopted and modi ed. In brief, rats were randomly exposed to 1-2 of these stimuli once a day for 6 weeks and the same stressors were not scheduled in three consecutive days (as shown in gure 1). Stressors included white noise (85 dB, 5 hours); thermal swimming (45°C, 5 min); stroboscopic illumination (300 ashes/minute, 5 h); soiled cage (10 h); paired with four other stressed animals (10 h); cold swimming (4 ℃, 5 min); tail pinching (3 min); restraint (12 h); water deprivation (12 hours) and food deprivation (12 or 24 hours).
After the 6-week CUMS, behavior tests were carried out, CSF sample and hippocampus slides were prepared (as shown in gure 2).

Sucrose preference test (SPT)
The procedure was performed to assess rats' anhedonia according to our previous method [4]. During the SPT, all rats were kept in single cages. The sucrose preference test was divided into four stages: sucrose training for 48 hours, baseline testing for 36 hours, food and water deprivation for 24 hours, and sucrose preference testing for 12 hours. Two bottles of liquid (1% sucrose solution and pure water) were given to each animal at the same time during the test. In the sucrose preference test stage, after 12 hours of liquid availability (20:30-8:30 the next day), the consumed volume of each solution was recorded to calculate sucrose preference. Sucrose preference (%) = sucrose solution consumption/total liquid intake × 100%.

Open eld test (OFT)
Open-eld test was performed to assess locomotor activity referred to our previous method [4]. The rats were transferred to the behavioral test room (soundproof darkroom) before the OFT for habituation to the environment for 1 hour. Each rat was individually placed into the middle of the open-eld apparatus (100 cm ×100 cm ×48 cm) and then allowed to explore freely for 5 minutes. The total traveling distance was recorded by video-tracking system (Flydy Co.,Ltd, Guangzhou, China).

Forced swimming test (FST)
The immobility time in FST was used to evaluate the desperation. The procedure was referred to our previous method [20]. The rats habituated an hour in the behavioral test room before the test. During FST, the rats were individually placed in a transparent plexiglas cylinder (height: 100 cm, diameter: 30 cm) lled with 35 cm water (25±1 ℃) and forced to swim for 6 min. The immobility time during the last 4 minutes was recorded by three researchers blind to the experimental design. Rats were considered immobile when they ceased struggling and remained oating motionless in the water except the movements necessary to keep their heads above the water. [21] T-maze test was used to evaluate learning and memory ability. The T-maze equipment was constructed of Tshaped elevated maze with a start arm (71cm ×18 cm×30 cm) and two target arms (46cm×18 cm×30cm) made of black and non-re ective panels. Light was avoided throughout the test. The rats were food-deprived and received approximately 75% of common daily intake food during the T-maze test. The T-maze test was divided into the training and testing phases. During the training phase, cheeses were placed at the end of both target arms with both doors open. The rats were placed in the start arm and allowed to freely explore the maze and intake cheese. During the testing phase, cheeses were placed at the end of both target arms. The door of one random target arm namely goal arm was closed. The rat was placed in the start arm and allowed to eat the cheese in the opened target arm. Once the rat entered the opened target arm, the door was closed immediately.

T-maze test
The rat was taken out after eating the cheese. Thirty seconds later, the rat was placed in the start arm again and allowed to freely explore the maze with both doors open. If the rat entered the goal arm and nished eating the cheese, it was scored as 1 correct time. The testing phase was repeated 10 times a day (each interval was 20 minutes) and lasted for 4 days. Accuracy (%) = correct times/10 × 100%.

CSF sample preparation
CSF sample was prepared 24 hours after the T-maze test. The rats were anesthetized and exposed the foramen magnum. intravenous infusion needle (0.45#) was attached to syringe (1mL). CSF was collected from cisterna magna puncture, centrifuged at 3000 rpm for 15 min at 4 °C. The supernatant of CSF was collected and frozen at -80 °C.

Immuno uorescence of sections
Referred to our previous experimental approach [4], 5-Bromo-2-deoxyUridine (BrdU, Sigma, US) was intraperitoneally injected (three injections, 4 hours apart, 200 mg/kg). One week later, the brain tissues (2.7mm-6.7mm from coronal groove) were dissected and xed in 4% paraformaldehyde for 24 hours at 4°C and then immersed in 30% sucrose until sank to the bottom. The hippocampus was trimmed according to the coronary sulcus (2.7mm-6.7mm from coronal sulcus), and was cut into 40 μm-thick frozen slices.
NeuN was used to label mature neurons and to re ect the number of neurons in DG. The sections were rewarmed, membrane-ruptured with Triton-X, blocked with goat serum and incubated with rabbit anti-NeuN antibody (1:1000, Abcam, UK) overnight at 4 ℃. After washed in TBST, the sections were incubated with AlexaFluor® 488 goat anti-rabbit (1:500, Abcam, UK) at 37°C for 2 hours. After washing and DAPI staining, images were captured by laser confocal microscope (LSM800, ZEISS, Germany). Ratio of NeuN positive cells (%)= NeuN positive cell number/total number of nuclei×100%.

Culture of primary hippocampal NSCs
The embryo was removed from Wistar rats (obtained from the Laboratory Animal Center of Southern Medical University, Guangzhou, China) at pregnant days 16-18 under sterile conditions by etherization and placed in icecold sterile PBS. The bilateral hippocampus was separated, cut into 1 mm × 1 mm × 1 mm pieces, mixed and gently polished with a Pasteur pipette, ltered through a 200-mesh cell sieve to obtain a single cell suspension, and centrifuged at 1200 rpm for 5 minutes. After the supernatant was discarded, the cells were resuspend with neural stem cell medium (DMEM/F12 supplemented with 20 ng/L EGF(Gibco, US), 20 ng/L bFGF(Gibco, US), 2% B27(Gibco, US) and 1% Penicillin /Streptomycin (Gibco, US)). Cells were incubated at a density of 2 × 10 5 cells/mL in 60 mm culture dishes at 37°C in a 5% (v/v) CO2 incubator. The culture medium was half-replenished every 2-3 days.
Cell viability was detected by CCK-8. Referred to our previous method (Wu et.al., 2013), passage1 neutrospheres were digested into single cells by Accutase TM (Gibco, US), resuspended with neural stem cell medium and seeded in 96-well plates at a density of 4 × 10 4 cells/well. Required CSF and CORT were added into each group, the nal volume of culture medium was 100 μL per well. After 72 h, 10 μL of CCK-8 (Dojindo, Japan) was added to each well and cells were cultured for 2 h. NSCs were dispersed into single cells by enzyme, resuspended with neural stem cell culture medium and seeded into 96-well plates at a density of 4 ×10 4 cells per well. The optical density (OD) was measured at 450 nm with a microplate reader (Bio-rad, US). The cell viability was calculated as following: cell viability (%) = [OD (experiment)-OD (blank)] / [OD (control) -OD (blank)] ×100%.

Immuno uorescence of primary hippocampal NSCs
Previous studies usually selected media containing 10% fetal bovine serum (FBS) as differentiation medium [8].
Referring to our previous method [8], NSCs were dispersed into single cells by Accutase TM and resuspended with the differentiation culture medium (NeuralBasal containing 2% FBS) and then seeded into 15 mm confocal dishes (5000 cells/dish). After 48 h, the CORT and CSF were added as required for each group. The incubation was continued for another 48 h, 10 μM BrdU was added to each dish. After 48 h, 4% paraformaldehyde was xed and BrdU/DCX was labeled by Immuno uorescence, as described in step 2.9.

Tandem mass tag (TMT) analysis of CSF proteomics
Protein was extracted from cerebrospinal uid. After quanti cation and separation, 30 μL protein was taken from each sample for proteolysis. TMT labeling was carried out in accordance with the instructions of the TMT labeling kit (Thermo, US). High pH RP spin column was used for grading. The samples were separated by chromatography (Thermo, US) and analyzed by mass spectrometry (Thermo, US). The original data were identi ed and quantitatively analyzed by Mascot2.2 and Proteome Discoverer1.4. Proteins with a fold change more than 1.2 or less than 0.83 as well as a statistical P-value < 0.05 between two groups were selected as DEPs.

Bioinformatic analysis
Gene Ontology (GO) mapping and annotation of proteins were conducted using the Blast2GO. Kyoto Encyclopedia of Genes and Genomes (KEGG) annotation were was performed using KAAS (KEGG Automatic Annotation Server). Enrichment analysis was performed by Fisher's Exact Test with P-values less than 0.05 and an FDR value less than 0.5.

Parallel reaction monitoring (PRM) quantitative analysis of target protein
The proteins of CSF samples were extracted and hydrolyzed and then separated by HPLC system (Thermo, USA).
The separated peptides were analyzed by PRM mass spectrometry (Thermo, USA). The PRM test was repeated three times. Finally, the software Skyline3.7.0 was used to analyze the data of the original PRM le and to quantify the target protein and target peptide.

Statistical analysis
The data were statistically analyzed using SPSS 22.0 software (IBM, USA). All results were expressed as the mean ± SEM. The data of each group were consistent with a normal distribution (Shapiro-Wilk test). An independent samples t-test was used for comparisons between two groups with homogeneity of variance oneway ANOVA was used for comparisons between three or more groups with homogeneity of variance, and Welch's test was used for heterogeneity of variance. In the pairwise comparisons, the LSD test was used to assess results with homogeneity of variance, and the Games-Howell method was used to assess results with heterogeneity of variance. T-maze accuracy was analyzed by repeated-measures ANOVA. The Greenhouse-Geisser correction was used when the assumption of sphericity was not met. Comparisons of the accuracy between groups at each individual time point were conducted using a multivariate ANOVA. P-values<0.05 were considered statistically signi cant.

ICA could reverse the depression-like behaviors of CUMS rats
3.1.1 The effect of ICA on sucrose preference of CUMS rats As shown in Figure 3A, compared with the CON group, sucrose preference of the CUMS group was signi cantly decreased (n=15, F (2,42) =27.476, p<0.01); compared with the CUMS group, sucrose preference of the ICA group was signi cantly increased (n=15, F (2,42) =27.476, p<0.01), indicating that ICA could alleviate the anhedonia symptom of CUMS rats.
3.1.2 The effect of ICA on total distance in the OFT of CUMS rats As shown in Figure 3B, compared with the CON group, CUMS signi cantly reduced the total traveled distance

Effect of ICA on relatively number of neurons in DG in CUMS rats
As shown in gure 4D and 4E, compared with the CON group, the relatively number of neurons in DG of the CUMS group reduced signi cantly (n=4-5, F (2,12) =27.68, p<0.01). Compared with the CUMS group, ICA signi cantly increased the relatively number of neurons (n=4-5, F (2,12) =27.68, p<0.01). It was suggested that ICA treatment could resist neuronal reduction in DG.

DEPs in the CSF of CUMS rats
It was showed that 1935 DEPs were screened between the CON group and the CUMS group. According to the standard of fold change (CUMS/CON or ICA/CUMS)≥1.2 (or ≤0.83) and P-value≤0.05, 100 signi cantly DEPs were selected (shown in Table 1). Among them, 66 proteins were upregulated compared with the CON group and 34 proteins were downregulated (See Table S1 for details). The results of the GO annotation of these 100 proteins showed that these proteins were involved in binding, catalytic activity, structural molecule activity, molecular function regulator and transcription regulator activity. The biological process they mainly participated in were cellular process, metabolic process, biological regulation, regulation of biological process and response to stimulus. The result of the KEGG pathway annotation indicated that these 100 DEPs were mostly involved in ribosome, uid shear stress and atherosclerosis, transcriptional misregulation in cancer, aminoacyl-tRNA biosynthesis and estrogen signaling pathway (as shown in Figure 6).

DEPs regulated by both CUMS and ICA
52 DEPs regulated by CUMS and ICA were identi ed Figure 7A and Table 1 . Among them, 44 DEPs were upregulated by CUMS and down-regulated by ICA; 8 DEPs were down-regulated by CUMS and up-regulated by ICA.
The results of the GO annotation of these 52 DEPs showed that they were involved in the formation of cellular components such as cytoplasm, organelles, and cell membrane. They mainly participate in molecular functions such as nucleic acid binding, protein binding, structural composition of ribosomes, and biological processes such as stress response, biological metabolism and gene expression (Table S2). The result of the KEGG pathway annotation indicated that these 100 DEPs were mostly involved in the ribosome pathway, PI3K-Akt signaling pathway and IL-17 signaling pathway ( Figure 7B).

Quanti cation of target proteins by PRM
Based on the results of GO annotation and KEGG enrichment analysis, 10 cell proliferation relating DEPs (Rps3, Rps12, Rps4x, Rps14, Rps19, Hsp90b1, Hsp90aa1, Calm1, Cpd and HtrA1) were selected to be analyzed by PRM.
The results showed that Rps4x, Rps12, Rps14, Rps19, Hsp90b1 and Hsp90aa1 were up-regulated by CUMS and down-regulated by ICA; HtrA1 was down-regulated by CUMS and up-regulated by ICA ( Table 2). The PRM results of such 7 proteins were consistent with TMT results, indicating the TMT results were reliable. In addition, Rps3, Calm1 and Cpd were not validated by PRM in this study.

Discussion
Depression is highly prevalent in the general population and is associated with grave consequences, including excessive mortality, disability, secondary morbidity and high socioeconomic costs. However, the e cacy of current antidepressant is not optimistic --almost 40% of patients do not recover following an antidepressant trial [4]. Scholars have been devoting themselves to exploring the pathogenesis of depression in an attempt to nd more effective and safer antidepressants. Currently, there is an increasing interest in anti-depression effects of natural compounds due to their lower toxicity and diverse biological properties. In this study, it has been observed that ICA showed signi cant anti-depressant e cacy in depressive-like rats by alleviating typical depressive symptoms such as anhedonia, decreased locomotor activity and despair.
The hippocampus is vulnerable to be damaged from a variety of psychological stressors [22]. Clinically, depressed individuals exhibit cognitive impairment such as memory and learning de cits, implicating hippocampal dysfunction in the pathophysiology of depression. Hippocampal neurogenesis dysfunction is one of the main pathogenesis of depression. Therefore, how to alleviate hippocampal damage is currently a concern of anti-depression research. In this study, ICA showed signi cant effect in repairing hippocampal damage, learning-memory de cit, hippocampal neurogenesis dysfunction and neuron deletion in DG of depression rats.
ICA could not effectively cross through the blood-brain barriers and accumulate in the brain [12]. Therefore, how ICA exerts anti-depression and anti-hippocampal damage effects has aroused our interest. CSF is the pivot that connects CNS with the periphery and other brain regions, in which substances might directly regulate hippocampal neurogenesis. It was speculated that ICA probably protected hippocampal neurogenesis and promoted neuronal survival by changing CSF components.
To prove this hypothesis, in vitro, the chronic stress-induced damage of NSCs was simulated under the high-CORT concentration. It was observed that high-CORT concentration could signi cantly inhibit the proliferation and differentiation of NSCs. The CSF of CUMS rats could not repair the CORT-induced injuries, while the CSF of ICA rats could effectively resist the damage. The results suggested that ICA might adjust some components in CSF to regulate the proliferation and differentiation of NSCs.
Proteins are the speci c executor of life activities and biological functions. Changes in CSF proteomics may be involved in the regulation of ICA on the proliferation and differentiation of NSCs. A total of 52 signi cantly differential expressed proteins in CSF co-regulated by CUMS and ICA were screened in CSF proteomics in this study. Among these proteins, 44 were up-regulated in CUMS exposure and down-regulated after ICA intervention, while 8 were the down-regulated in CUMS exposure and up-regulated after ICA intervention.
The GO annotation and KEGG pathway enrichment of common DEPs indicated that three possible pathways were involved in the e cacies of ICA in anti-depression, against dysfunctional neurogenesis and neuron deletion, namely ribosome pathway, PI3K-Akt signal pathway and IL-17 signal pathway. To support the TMT analysis results, PRM quantitative analysis was conducted on 10 DEPs that were closely connected with cell proliferation and survival. It had been veri ed that the regulatory trends of 7 DEPs (Rps4x, Rps12, Rps14, Rps19, Hsp90b1, Hsp90aa1 and Htra1) were consistent with the TMT results.
(1) Proteins related to ribosome pathway Among the 52 signi cantly DEPs co-regulated by CUMS and ICA, 11 of them were enriched into the ribosome pathway, which was the most enriched KEGG pathway. All these 11 proteins in CSF were up-regulated by CUMS.
A large number of studies have con rmed that abnormal ribosomal proteins expression or transcription exists in patients with depression and animal models. Most of such studies have observed that the expression of ribosomal proteins is up-regulated or increased transcription in blood, liver, hippocampus and other tissues of depressed individuals [23][24][25]. It has been also observed that the expression of ribosomal proteins in the hippocampus of rats was signi cantly down-regulated under 3-week exposure to CUMS [26]. According to our previous study [4], although rats showed depression-like behaviors after 3-week exposure to CUMS, rats were still in the stress compensation stage, which did not cause hippocampal damage. Different brain regions may not respond to mental stress simultaneously and may react differently towards various stressors. Nevertheless, abnormal expression or transcription of ribosomal proteins is found in both central and peripheral regions of depressed individuals.
It is worth noting that in PRM analysis, the expressed differences of ribosomal proteins showed higher signi cance than in high-throughput TMT results. It was suggested that these ribosomal proteins might play an important role in the development and remission of depression.
Ribosomes are consisting of ribosomal RNAs and ribosomal proteins and are the main sites for intracellular RNA translation controlling protein synthesis. During the stress process, to cope against stress-induced damages, the assembly of ribosomes and synthesis of proteins speed up as the demand for proteins increases [27][28][29]. In this study, more up-expressed DEPs rather than down-expressed ones were observed in the CSF of CUMS rats.
However, ribosome assembly requires extremely high rates, coordinated synthesis and assembly of macromolecules across cellular compartments. Ribosomal synthesis dysfunction may easily appear under stimulates during the assembly process, resulting in rapid accumulation of ribosomal proteins [30][31][32]. In addition, numerous studies have shown that multiple cellular stresses could directly act on the nucleus and trigger the overexpression of ribosomal proteins. Therefore, the overexpression of ribosomal proteins may indicate ribosomal assembly dysfunction under stress [33,34].
Ribosomes control the translation of all proteins in cells, the normal synthesis of ribosome is essential for cell survival, growth and proliferation. In the process of ribosome synthesis, either over-expression or underexpression of ribosomal protein could disrupt ribosome synthesis, lead to ribosomal assembly dysfunctions and cause cell cycle arrest, senescence or apoptosis [35]. In addition, the accumulation of ribosomal proteins caused by the ribosome synthesis dysfunction could lead to protein folding homeostasis collapse, inhibiting cell growth [31,32]. It may also cause cell cycle arrest or apoptosis through extraribosomal functions of ribosomal proteins [36]. For example, Rps14 activates p53 pathway to inhibit cell proliferation and induce apoptosis [37]. In the CNS, the normal synthesis of ribosome also plays an important role in the development of neurons. Some forms of synaptic plasticity require rapid, local activation of protein synthesis in ribosome [38]. Dysfunction in ribosomal gene expression may be associated with a decrease of proliferation in DG [39].
In this study, multiple increased ribosomal proteins (including Rps4x, Rps12, Rps14, Rps19) in CSF of depression rats suggested ribosomal synthesis dysfunction in depression rats, which may be related to neurogenesis dysfunction and neuronal loss in DG. ICA could signi cantly down-regulate the expression of such ribosomal proteins in CSF of depression rats, promote the normal synthesis of ribosome, provide more proteins needed for the repair of neurons, proliferation and differentiation of NSCs.
In summary, ribosomal synthesis dysfunction may affect cell proliferation and survival by abnormal expression of ribosomal proteins. ICA may resist neurogenesis dysfunction and neuronal loss in rats with depression by repairing ribosomal synthesis dysfunction.
(2) Proteins related to PI3K-Akt pathway and IL-17 pathway PI3K-Akt pathway is one of the classical pathways regulating cell cycle, which is of great signi cance in promoting the survival of neurons and neurogenesis [40]. It is also one of the effective pathways of many antidepressants [41,42]. Hsp90b1 and Hsp90aa1 are important proteins and targets in the PI3K-Akt pathway.
Abnormal expression of Hsp90b1 and Hsp90aa1 could both inhibit PI3K-Akt pathway activity [43,44]. Chronic stress could cause overexpression of HSP90 family proteins in the brain of depression animals [45], and overexpressed Hsp90b1 and Hsp90aa1 would combine with HIF1α to promote NDRG2 expression and play a key negative regulatory role in PI3K-Akt signaling [43].
Hsp90b1 and Hsp90aa1 are also important members of the IL-17 pathway. HSP90 functions as a chaperone that facilitates the folding and assembly of its client proteins. Loss of HSP90 chaperone function results in the degradation of its client proteins-Act1. As Act1 is required for IL-17 signaling and is a client protein of heat shock protein 90 (Hsp90) family proteins, so that Hsp90 activity is required for IL-17 signaling [46]. Epithelial cells, endothelial cells and glial cells are all target cells of IL-17. IL-17 is activated after binding to the receptors of target cells and ultimately promotes the release of large amounts of in ammatory cytokines from target cells, which is one of many causes of neuronal death in individuals with depression [47]. In addition, some pathological phenomena induced by IL-17 can be reversed by inhibiting the activity of Hsp90 family proteins [48].
In this study, PI3K-Akt pathway and IL-17 pathway were enriched, Hsp90b1 and Hsp90aa1 were overexpressed in CSF, neurogenesis dysfunction and hippocampal neuronal loss were observed in depression rats. It was suggested that the overexpressed Hsp90b1 and Hsp90aa1 might inhibit PI3K-Akt pathway and activate IL-17 pathway in hippocampus, causing neurogenesis dysfunction and neuronal loss in depression rats. The e cacy of ICA in repairing neurogenesis dysfunction and neuronal loss may be related to the reduction of Hsp90b1 and Hsp90aa1 in CSF.
In addition, though not enriched in KEGG pathway, some DEPs co-regulated by CUMS and ICA may be involved in repairment of dysfunctional neurogenesis and neural reduction. For example, HtrA serine peptidase 1 (HtrA1) is abundantly expressed in astrocytes. Down-expressed HtrA1 is related to a variety of neurological diseases. HtrA1 could mediate transforming growth factor-β hydrolysis and bone morphogenetic protein inhibition to prevent neural stem cell proliferation and differentiation inhibition [49]. In this study, the treatment of ICA reversed CUMSinduced Htra1 reduction in CSF, neurogenesis dysfunction and neuronal loss in DG, suggesting that the antihippocampal damage effect of ICA may be related to the regulation of Htra1 content in CSF.
In conclusion, 6-week ICA intervention could signi cantly alleviate the hippocampal neurogenesis dysfunction, neuronal loss in DG and memory and learning de cits in depression rats. Such e cacies of ICA may be related to its regulation of Rps14, Hsp90b1, Htra1 and other proteins in CSF. The pathways involved mainly include ribosome pathway, PI3K-Akt pathway and IL-17 pathway. It was suggested that ICA might protect hippocampus by changing CSF proteomics (as shown in Figure 8).

Conclusions
This study showed prospect of ICA as a potential antidepressant, but there remain some problems. The pathways and targets of ICA in anti-hippocampal damage screened in this study need to be further veri ed. As we known, brain metabolites could directly transfer into the CSF. It cannot be ruled out that a small amount of ICA crossing the blood-brain barrier might affect brain regions' metabolism and lead to the changing of CSF proteomics.   Figure 1 CUMS procedure. One or two arbitrary mild stressors were randomly applied for 6weeks and the same stressors were not scheduled in three consecutive days.    Data were expressed as the mean ± SEM. *p 0.05, **p 0.01.

Figure 7
Differentially expressed proteins regulated by both CUMS and ICA, A) VEEN map. B) KEGG pathway enrichment.

Figure 8
The possible mechanism of ICA against hippocampal neurogenesis dysfunction via CSF in depression.