COX-2 inhibition rescues depression-like behaviors via suppressing glial activation, oxidative stress and neuronal apoptosis in rats

Neuropharmacology 160 (2019) 107779

Contents lists available at ScienceDirect

Neuropharmacology journal homepage: www.elsevier.com/locate/neuropharm

COX-2 inhibition rescues depression-like behaviors via suppressing glial activation, oxidative stress and neuronal apoptosis in rats

T

Qiqi Songa, Ya-bo Fengb, Liyan Wangc, Jie Shend, Ye Lia, Cuiqin Fana, Peng Wanga, Shu Yan Yua,e,∗ a

Department of Physiology, Shandong University, School of Basic Medical Sciences, 44 Wenhuaxilu Road, Jinan, Shandong Province, 250012, PR China Department of Neurology, Shandong Provincial Hospital as affiliated with Shandong University, Jingwuweiqi Road 423#, Jinan, Shandong Province, 250012, PR China c Morphological experimental center, Shandong University, School of Basic Medical Sciences, 44 Wenhuaxilu Road, Jinan, Shandong Province, 250012, PR China d Department of Neurosurgery, Qilu Hospital of Shandong University, 107 Wenhuaxilu Road, Jinan, Shandong Province, 250012, PR China e Shandong Provincial Key Laboratory of Mental Disorders, School of Basic Medical Sciences, 44 Wenhuaxilu Road, Jinan, Shandong Province, 250012, PR China b

H I GH L IG H T S

is overexpressed in DG hippocampus of rat depression model. • COX-2 inhibition ameliorated depression-like behaviors in rats. • COX-2 inhibition suppressed oxidative stress and oxidative DNA damage in depressed rats. • COX-2 inhibition suppressed neuroinflammation and neural apoptosis in depressed rats. • COX-2 • Inhibition of oxidative stress attenuated neural and behavioral changes in depressed rats.

A R T I C LE I N FO

A B S T R A C T

Keywords: COX-2 Neuroinflammation Apoptosis Oxidative stress Depression

Depression is considered a neuropsychiatric condition which is associated with neuronal injury within specific brain regions. We previously reported that cyclo-oxygenase (COX)-2, a rate-limiting enzyme for prostaglandin E2 (PGE2) synthesis, significantly enhanced depressive-like disorders induced by chronic stress in rats. However, the underlying molecular mechanisms and identification of potential therapeutic targets for preventing neuronal injury associated with depression remain largely uncharacterized. Here, we show that COX-2 inhibition by celecoxib protects against neuronal injury through suppression of oxidative stress and, in this way, mediates its antidepressant effects. COX-2 is highly expressed in the hippocampal dentate gyrus (DG) of rat depression model and its activity is responsible for depression-like behaviors as demonstrated in two independent rat models of depression. Inhibition of COX-2 exerts neuroprotective actions in DG regions, including suppressing neuroinflammatory response, against oxidative stress and neuronal apoptosis, which are the critical risk factors for neuronal injury and pathophysiology of depression. Moreover, the antioxidant, N-acetylcysteine (NAC), significantly attenuates oxidative stress levels and dendritic spine deficiencies resulting from COX-2 overexpression; and, suppression of oxidative stress by NAC also significantly ameliorates depressive behaviors in rats. These findings suggest that selective inhibition of COX-2 ameliorates depression-like behaviors in rat models of depression. This selective inhibition of COX-2 appears to be protective against oxidative stress and neuronal deterioration resulting from chronic stress. Taken together, these findings have potentially important clinical implications with regard to the development of novel therapeutic approaches in the treatment of neuropsychiatric conditions like depression.

1. Introduction Depression is considered a psychiatric disorder usually resulting

from stress and is associated with alterations in neuronal structures and/or functional injury within specific brain regions (Stockmeier et al., 2004; Oh et al., 2012). Therefore, therapies aimed at preventing

∗ Corresponding author. Department of Physiology, Shandong University, School of Basic Medical Sciences, 44 Wenhuaxilu Road, Jinan, Shandong Province, 250012, PR China. E-mail address: [email protected] (S.Y. Yu).

https://doi.org/10.1016/j.neuropharm.2019.107779 Received 13 August 2019; Received in revised form 11 September 2019; Accepted 14 September 2019 Available online 17 September 2019 0028-3908/ © 2019 Published by Elsevier Ltd.

Neuropharmacology 160 (2019) 107779

Q. Song, et al.

These antidepressant effects of COX-2 inhibition appear to be attributable to decreased levels of oxidative stress, which would normally contribute to neuroinflammation and apoptosis. In support of this conjecture, are the present findings that a pharmacological inhibition of oxidative stress significantly attenuated neuronal injury and depression phenotypes. Given this information, we propose that COX-2 might serve as a new and potentially important therapeutic target for the treatment of depression.

such neuronal injury or promoting neuronal repair at these sites would have major clinical implications for the treatment of this disorder. As the pathophysiological mechanisms underlying neuronal changes involved with this disorder are not fully understood, such therapies remain an enigma. However, it is generally recognized that a critical factor in the production of neuronal injury is oxidative stress, which is suspected to be involved in the development of various of neurodegenerative diseases including Alzheimer's disease (Valko et al., 2007), Parkinson's disease (Hwang, 2013) and depression (Jiménez-Fernández et al., 2015). Recently, results from a number of studies have demonstrated that reactive oxygen species (ROS) play a major role in brain damage (Kennedy et al., 2012). One means of generating ROS is through continuous exposure to harmful environmental sources such as stress, which can then result in DNA damage within specific brain regions (Nissanka and Moraes, 2018). Increased ROS levels within the ventral tegmental area (VTA) induces behavioral, biochemical and anatomical changes associated major depressive disorders in mice (Ibi et al., 2017), while treatment with antioxidants was shown to inhibit these depressive-like behavioral phenotypes in mice (Uchihara et al., 2016). As one of the main sources of ROS, mitochondria, represent the major organelle involved in the regulation of cellular redox status (Quan et al., 2015). In response to feedback, excess ROS production further induces mitochondrial dysfunction and impairs oxidative stress balance (Vivancos et al., 2005). Moreover, excessive generation of ROS will result in various types of DNA damage, including base damage, single-strand breaks (SSBs) and double-strand breaks (DSBs) (Bjelland and Seeberg, 2003; Kryston et al., 2011). Such effects lead to cell senescence or death, which is believed to play a crucial role in the pathogenesis of neurological diseases (Wei et al., 2000). However, whether this association reflects a common, though independent, pathway of stress-driven COX-2 expression and oxidative stress, or whether a cause–effect link exists between these two events in stress-induced animal models of depression is currently unknown. The presence of such a relationship would suggest an important new therapeutic strategy directed toward the design of agents to disrupt oxidative stress as a means for the treatment of depression. Depression has become one of the most prevalent psychiatric disorders of late. Unfortunately, current therapies are usually associated with high relapse rates and side effects (Papakostas and Ionescu, 2015). In particular, the lack of biomarkers and effective medications for treatment in the early stages of depression results in most patients usually progressing to major disorder stages with high suicide rates. Recently, the effects of COX-2 inhibitors have been examined for use in various clinical trials, with results from several reports suggesting that it diminishes cerebral damage and neuronal death (Scali et al., 2003; Govoni et al., 2001). COX-2 enzymatic hyperactivity can prompt production of the pro-inflammatory factor, PGE2, which can facilitate inflammatory responses; and, an upregulation of COX-2 and subsequent enhanced production of the COX-2 derived prostaglandin E2 that occurs with brain injury, is accompanied with an increase in caspase-3 dependent apoptosis (Takadera et al., 2002; Glushakov et al., 2018). Within our laboratory, we have recently shown that a chronic exposure to stress induces a marked overexpression of COX-2 within the hippocampus, while COX-2 inhibition alleviates the display of depressive behaviors in rats (Song et al., 2018). However, the mechanisms underlying the association between COX-2 inhibition and neuronal protection, in particular, whether selective COX-2 inhibition is beneficial against neuronal injury in depression have not been fully elucidated. Therefore, whether and how COX-2 inhibitors may exert their therapeutic effects on depression warrants further investigation. In the present study, we examined the role of COX-2 in the genesis and progression of depression in two animal models of depression. We found that COX-2 was overexpressed within the hippocampal dentate gyrus (DG), one of the key brain regions associated with depression and other psychiatric disorders (Gulyaeva, 2018), while COX-2 inhibition greatly ameliorated the display of depression-like behaviors in rats.

2. Materials and methods 2.1. Animals and housing conditions Male Wistar rats weighing 160–180 g (5–6 weeks) were obtained from the Shandong University Experimental Animal Centre. All procedures were approved by the Shandong University Animal Care and Use Committee and were performed in accordance with the International Guiding Principles for Animal Research provided by the International Organizations of Medical Sciences Council (CIOMS). Rats were housed in groups of four per cage under controlled temperature (22–24 °C) and unrestricted access to food and water. Rats were acclimatized to these laboratory conditions for at least one week prior to experimental procedures. All efforts were made to minimize pain and numbers of the animals used in the present study. 2.2. Chemicals and antibodies Celecoxib, fluoxetine, lipopolysaccharide (LPS) and dimethyl sulfoxide (DMSO) were purchased from Sigma-Aldrich (St Louis, MO, USA). The N-acetylcysteine (NAC) was purchased from Beyotime (Shanghai, China). The monoclonal rabbit anti-COX2 (CST-12282), polyclonal rabbit anti-p38 MAPK (CST-9212), anti-phospho-p38 MAPK (CST9211), anti-phospho-ERK (CST-9102), anti-NeuN (CST-24307), anticleaved Caspase-3 (CST-9661) and anti-β-actin (CST-4970) were purchased from Cell Signaling (Beverly, MA, USA). Polyclonal goat anti-4hidroxynonenal (4-HNE, ab46545), rabbit anti-CD45 (ab10558), antiCD11b (ab133357), anti-Nrf2 (ab137550), anti–HO–1 (ab13243), antiJNK(ab179461), anti-γ-H2AX (ab26350) and anti-8-OHdG (ab48508) were purchased from Abcam (Cambridge, UK). Polyclonal rabbit antiionized calcium binding adaptor molecule-1 (Iba-1) (019–19741) was purchased from Wako Pure Chemical Inc (Japan). Polyclonal rabbit anti-glial fibrillary acidic protein (GFAP) (60190-1-Ig), anti-GAPDH (10494-1-AP), rhodamine (TRITC)-conjugated goat anti-rabbit IgG and Fluorescein (FITC)-conjugated goat anti-mouse IgG were purchased from the Proteintech Group (Rosemont, IL, USA). Peroxidase-conjugated goat anti-rabbit/mouse IgG was purchased from Zhongshan Golden Bridge Biotechnology. Hoechst 33258 (C0031) was purchased from Solarbio (Beijing, China). The MitoSOX Red (m36008) was purchased from ThermoFisher (MA, USA). 2.3. Depression animal model 2.3.1. CUMS model The CUMS-induced animal model of depression was produced as described previously with minor modifications (Mao et al., 2009). Briefly, rats were housed individually in a separate room and subjected to a daily stress regime for 5 weeks. The unpredictable stressors included physical restraint (2 h), cage shaking (2 h), 5 min cold swimming (at 4 °C), 24 h food deprivation followed by 24 h water deprivation, overnight illumination, wet bedding (24 h) and foot-shock (0.5 mA, 0.5s). One of these stressors was applied daily in a random order to the rats. (Supplemental data Fig. s1). 2.3.2. LPS-induced model LPS (0.5 mg/kg) was injected intraperitoneally (i.p.) daily for one week to induce depression-like behaviors in rats as described previously 2

Neuropharmacology 160 (2019) 107779

Q. Song, et al.

antibodies followed by the appropriate fluorescent-conjugated secondary antibody (Sigma-Aldrich). Sections were then washed three times with PBS and incubated with 4′, 6-diamidino-2-phenylindole dihydrochloride (DAPI) (Thermo Fisher Scientific, USA) at room temperature for 7 min if necessary. Immunofluoresce staining and images were obtained by a LSM780 laser scanning confocal microscope (ZEISS, Germany) system. At least six to eight representative images were taken from each rat for analysis by Image-Pro plus 6.0 software. The statistical results of quantifying fluorescence intensity were expressed as percentage ratios relative to control group.

(Adzic et al., 2015, Cui et al., 2018). (Supplemental data Fig. s1). 2.4. Drug treatments Celecoxib was dissolved in 0.1% dimethyl sulfoxide (DMSO) at a concentration of 10 mg/ml. Fluoxetine was dissolved in sterile-endotoxin-free saline (NaCl, 0.9%) at a concentration of 10 mg/ml. LPS was dissolved in 0.9% saline at a concentration of 10 mg/ml. The antioxidant NAC was dissolved in 0.9% saline at a concentration of 100 mg/ml. In all experiments, celecoxib, fluoxetine or NAC were administered via an i.p. injection at 60 min prior to daily CUMS procedures over the 5-week period of CUMS (Supplemental data Fig. s1). All rats received a daily i.p. injection of saline (10 ml/kg) for three days prior to the experiment for habituation and were then randomly allocated to one of the following groups with N = 18/group: a) control (non-stressed group), b) CUMS, c) LPS (0.5 mg/kg daily), d) CUMS treated with celecoxib (20 mg/kg) (celecoxib + CUMS), e) CUMS treated with fluoxetine (40 mg/kg) (FLX + CUMS), f) LPS treated with celecoxib (celecoxib + LPS), g) LPS treated with fluoxetine (FLX + LPS), h) control (non-stressed group) treated with celecoxib (20 mg/kg) (celecoxib + control), I) CUMS treated with NAC (300 mg/ kg) (NAC + CUMS). The dose and route of celecoxib and fluoxetine administration is based upon previous study that focus on neural injury (Fan et al., 2013; Shan et al., 2016). The dose and intraperitoneal injection regimen of NAC was based on previous results to exerts antioxidative effects (Sen et al., 2014).

2.8. Transmission electron microscopy (TEM) TEM (Philips Tecnai 20 U-Twin, Holland) analysis, as performed by the electron microscopic core lab of Shandong University, was used to assess DG neuronal ultrastructure. Briefly, tissues were fixed with 1% osmium tetroxide for 2 h and then infiltrated with a mixture of one-half propylene oxide overnight. After being embedded in resin, tissues were cut into ultrathin sections (70 nm) and stained with 4% uranyl acetate for 20 min followed by 0.5% lead citrate for 5 min. At least 30 micrographs were randomly taken from each rat for analysis using Image J software (National Institutes of Health, NIH, Bethesda, MD, USA). 2.9. Golgi staining Golgi staining was performed to observe neuronal dendrites and dendritic spines using the FD Rapid GolgiStain™ Kit (FD NeuroTechnologies, MD21041, USA) according to the manufacturers’ instructions. Briefly, brains were immersed in the impregnation solution (A/B = 1:1, total 15 ml) at room temperature in the dark for two weeks, followed by transfer into solution C for three days. Brains were sectioned serially into 100 μm coronal sections and then cleaned in xylene and cover-slipped with Rhamsan gum for light microscopic observation. For each rat, at least 4 to 6 dendritic segments of apical dendrites per neuron were randomly selected, and 5 pyramidal neurons were analyzed per rat. For each group, at least 6 rats were analyzed with use of Image-Pro plus 6.0 software.

2.5. Behavioral tests 2.5.1. Sucrose preference test (SPT) The sucrose preference test was conducted after the 5 weeks of CUMS exposure as described previously with minor modifications (Mao et al., 2009). Briefly, rats were initially placed individually in cages with two bottles of sucrose solution (1%, w/v) for the first 24 h period, and then one bottle of sucrose solution was replaced with tap water for the second 24 h period. After this adaptation phase, rats were deprived of water and food for 24 h and then permitted access to the two bottles for a 3-h test, with one bottle containing 100 ml of 1% sucrose solution and the other containing 100 ml of tap water. The sucrose preference was defined as the sucrose consumption/[water consumption + sucrose consumption] × 100% during the 3-h test period.

2.10. Western blot analysis The dissociated DG tissues were lysed while on ice for 30 min in RIPA buffer with a cocktail of protease/phosphatase inhibitors. The protein concentration was determined with use of the BCA assay kit (Beyotime, China). Protein extracts (30–50 μg) were resolved by SDSPAGE for electrophoretic separation and then transferred onto PVDF membranes. The membrane was blocked overnight with 5% nonfat milk and incubated with appropriate primary antibodies at 4 °C for overnight. Proteins of interest were detected with appropriate horseradish peroxidase conjugated secondary antibodies and developed using an enhanced chemiluminescence kit (ECL; GE Healthcare, Buckinghamshire, UK). Protein band densities were quantified using Image-J software and were normalized to β-actin or GAPDH. The statistical results of different groups were expressed as percentage ratios relative to control group.

2.5.2. Forced swim test (FST) At one day after the sucrose preference test, rats were subjected to the forced swim test as described previously (Duman et al., 2007). Briefly, rats were placed individually in a cylinder of water (height: 80 cm, diameter: 30 cm, temperature: 25 °C) for 15 min of forced swim training. Then 24 h later, each rat was placed in the cylinder for a 5-min testing phase. In the test session, the immobility (floating with only limited movements to maintain their head above water) and swimming times of each rat were recorded by an observer blinded as to the treatment condition of the rat. 2.6. Brain dissection and tissue preparation

2.11. Reverse transcription PCR and real-time quantitative PCR

Twenty-four hours after behavioral tests, rats were anesthetized with sodium pentobarbital (150 mg/kg, i.p.) and transcardially perfused with 100 ml 0.9% NaCl containing heparin sodium salt followed by fixation with 4% paraformaldehyde (PFA). Brains were then dissected and post-fixed in 4% PFA overnight at 4 °C followed by a graded dehydration. Brain samples were cut into serial coronal frozen sections (30 μm) for immunofluorescence staining.

2.11.1. Reverse transcription PCR (RT-PCR) Total RNA was isolated from samples of DG regions using TRIzol reagent (Invitrogen), and was reverse transcribed into cDNA using reverse transcriptase (TOYOBO). cDNA was subsequently amplified by PCR with specific primers (Supplementary Table 1). PCR products were assessed by electrophoresis on 3% agarose gel and were analyzed using the Gel Image Analysis System (Bio-rad, USA). The band density of each sample were normalized to GAPDH. The statistical results were expressed as percentage ratios relative to control group.

2.7. Immunofluorescence staining and confocal microscopy Frozen slices (30 μm) were incubated with the specific primary 3

Neuropharmacology 160 (2019) 107779

Q. Song, et al.

Fig. 1. COX-2 is overexpressed in hippocampal DG regions in the rat depression model. (A and B) Western blotting and Real-time quantitative PCR analysis of COX-2 expression in hippocampal DG regions of CUMS- and LPS-induced depression models. (C, D, E) Immunofluorescence staining of COX-2 and 4-HNE expression in DG regions of the two models of depression. Scale bar is 50 μm. (F) ELISA analysis of PGE2 in DG of depressed rats after pretreatment of celecoxib (20 mg/kg) for five weeks. Band intensities were quantified by Image J software and normalized to β-actin. Each experiment was repeated at least four times. *P < 0.05, compared to control group; #P < 0.05, compared to CUMS or LPS group. (Cele, Celecoxib).

2.12. Enzyme linked immunosorbent assay (ELISA)

2.11.2. Real-time quantitative PCR Real-time quantitative PCR analysis was performed on a Bio-Rad iCycler system (Bio-Rad, Hercules, CA) using SYBR GREEN mix (TOYOBO). The sequences of specific primers are listed in Supplementary Table 1. GAPDH and β-actin were served as loading control in each sample and mRNA expression levels were evaluated using the 2− (ΔΔCt) method.

Concentrations of PGE2 were measured with use of an ELISA kit (Enzyme-linked Biotechnology Co. Shanghai, China) according to the manufacturers’ instructions. Total protein isolated from DG tissue samples was determined with use of the BCA assay (Thermo Fisher, Waltham, MA). Equal amounts of diluted samples were added to 96well plates. Data were expressed as the amount of PGE2 (pg) per total protein (mg) (mean ± SEM) and assayed in duplicate. The statistical results were expressed as percentage ratios relative to control group. 4

Neuropharmacology 160 (2019) 107779

Q. Song, et al.

Fig. 2. COX-2 inhibition ameliorated depression-like behaviors in stressed rats. (A) Treatment with celecoxib (20 mg/kg) or fluoxetine (40 mg/kg) reversed the decreased consumption of sucrose in CUMS rats. (B) Treatment with celecoxib or fluoxetine reversed the increases in immobility times and decreases in swimming times of CUMS rats in the forced swim test. (C) Treatment with celecoxib or fluoxetine reversed the decreased consumption of sucrose in LPS-induced depression. (D) Treatment with celecoxib or fluoxetine reversed the increases in immobility times and decreases in swimming times in LPS-induced depression. N = 18 per group. *P < 0.05, compared to Control group; #P < 0.05, compared to CUMS or LPS group. (Cele, Celecoxib; FLX, Fluoxetine).

purchased from Jiancheng Inc (Nanjing, China). All assays were performed according to the manufacturers’ guidelines. The statistical results of quantifying enzyme activity were expressed as percentage ratios relative to control group.

2.13. Oxidative stress measures 2.13.1. Measurement of ROS Frozen slices (12 μm) of DG regions were stained with 10 μM dihydroethidium (DHE, Sigma, USA) at 37 °C for 30 min for analysis of ROS production. The mitochondrial ROS level was measured with 10 μM MitoSOX Red fluorescent dye for 15 min at room temperature in the dark. Fluorescence images were captured with use of a laser confocal microscope and quantified using the Image-pro plus image analysis system. The statistical results of quantifying fluorescence intensity were expressed as percentage ratios relative to control group.

2.13.4. Assessment of DNA damage markers Immunofluorescence staining was used to detect DNA damage markers in DG regions. Frozen slices were incubated with anti-phosphoH2AX or anti-8-OHdG antibody, and counterstained with DAPI. The DNA repair enzyme PARP1 andγ-H2AX protein expressions were measured with use of Western blot. All the statistical results were expressed as percentage ratios relative to control group.

2.13.2. Oxidative and nitrosative stress markers The oxidative and nitrosative stress marker, 4-HNE, was measured with use of immunofluorescence staining using specific antibodies (anti-4-HNE, Abcam ab46545). In addition, DG regions were also collected and homogenized for determination of the content of MDA (A003-1) and NO (A013-2) using the assay kits from Jiancheng Inc (Nanjing, China). All samples were assayed in duplicate and results were normalized to total protein content. The statistical results were expressed as percentage ratios relative to control group.

2.14. Statistical analysis Data were analyzed with use of SPSS version 13.0. All data were presented as the mean ± standard error of the mean. The behavioral data were analyzed using a repeated-measures analysis of variance (ANOVA) followed by Bonferroni's post hoc test, and the statistical significance of differences between normal control rats and depression rat model was evaluated with t-tests. Data from all other determinations were analyzed statistically with use of two-way ANOVA followed by the Bonferroni's test for multiple post-hoc comparisons of means. A value of P < 0.05 was required for results to be considered statistically significant.

2.13.3. Antioxidant enzyme activities Activity of antioxidant enzymes in DG tissue homogenates was measured using the superoxide dismutase (SOD) activity assay Kit (A001-3), catalase (CAT) activity assay Kit (A007-1), glutathione peroxidase (GSH-Px) activity assay Kit (A005), total antioxidant capacity (T-AOC) activity assay Kit (A015-2) and the LDH assay Kit (A020-2) all 5

Neuropharmacology 160 (2019) 107779

Q. Song, et al.

Fig. 3. COX-2 inhibition suppressed oxidative stress in hippocampal DG regions of depressed rats. (A) Activity of antioxidant enzymes SOD, CAT and T-AOC. (B) Contents of MDA, NO and LDH were analyzed and levels were normalized to total protein content. (C) Representative images of DHE staining (red) within the DG area from each group of rats for ROS analysis. Nuclei (blue) are stained with DAPI. Scale bar is 50 μm. (D) Representative images of 4-HNE staining (red). Nuclei (blue) are stained with DAPI. Scale bar is 50 μm. (E) Representative images of Mito-SOX staining (red). Nuclei (blue) are stained with DAPI. Scale bar is 50 μm. (F) Western blot analysis of Nrf2, HO-1, phosphorylated p38, JNK, p65 and ERK within DG regions. Band intensities were normalized to βactin. N = 6 per group. *P < 0.05, compared to Control group; #P < 0.05, compared to CUMS group. (Cele, Celecoxib).

6

Neuropharmacology 160 (2019) 107779

Q. Song, et al.

3. Results

amelioration of depression-like behaviors by celecoxib was due to changes in oxidative stress levels. COX-2 inhibition did, in fact, lead to a significant suppression in oxidative stress in CUMS rats. Two-way ANOVA revealed that the activity of antioxidant enzymes within the DG region was significantly different among the four groups [F (3, 20) = 16.77, P < 0.05]. Post-hoc analysis indicated that the activity in a number of antioxidant enzymes was significantly reduced in CUMS rats (P < 0.05, Fig. 3A). There were no statistically significant differences between the celecoxib pretreated control group and control groups with regard to antioxidant enzymes activity (p > 0.05). Meanwhile, increased levels of the oxidative stress products, malondialdehyde (MDA), nitric oxide (NO) and lactate dehydrogenase (LDH) were observed within DG regions of CUMS rats (P < 0.05, Fig. 3B). In accord with these findings were results from the DHE assay which demonstrated that ROS levels were significantly changed among the four groups [F (3, 20) = 15.89, P < 0.05]. Post-hoc analysis indicated that 5-weeks of CUMS exposure increased the ROS levels in DG regions of CUMS rats as compared with that in unstressed rats (P < 0.05, Fig. 3C). Further evidence for a role of oxidative stress in response to CUMS was indicated by the increased levels of 4-HNE, a key mediator of oxidative stress induced cell death (P < 0.05, Fig. 3D), as well as the increased levels of superoxide in mitochondria, as measured by MitoSox, in CUMS rats (P < 0.05, Fig. 3E). In contrast, levels of the essential antioxidative nuclear transcription factor, Nrf2 and its product HO-1, were decreased, while phosphorylation of p38MAPK, JNK and p65, which usually occurs in response to oxidative stress, was increased in CUMS rats (P < 0.05 respectively, Fig. 3F). Interestingly, all of the above oxidative stress changes were significantly prevented with celecoxib pretreatment (P < 0.05 respectively). These results reveal that an early increase in oxidative damage within the DG is one of the consequences of CUMS exposure, and COX-2 inhibition exerts protective effects on in vivo stress-induced oxidative damage.

3.1. COX-2 is overexpressed in hippocampal DG regions within a rat model of depression COX-2 expression was measured within hippocampal DG regions in two different rat models of depression with use of Western blot. As shown in Fig. 1A, COX-2 levels were significantly increased in both the CUMS-exposure and LPS-treated rats as compared with that in nonstressed control animals (P < 0.05). A similar trend was obtained for COX-2 mRNA (P < 0.05, Fig. 1B). Levels of COX-2 within DG regions were further analyzed using immunofluorescence staining. The intensity and extension of COX-2 staining were significantly increased in stressed, as compared with that in control, rats (P < 0.05, Fig. 1C and D). More interestingly, 4-hydroxynonenal (4-HNE), an important marker of oxidative stress produced by lipid peroxidation in cells, was significantly increased within the DG region (P < 0.05, Fig. 1C, E). Expression of PGE2, a principal component of COX-2 products, was also significantly increased in these two depression models (P < 0.05, Fig. 1F), whereas treatment of celecoxib, an inhibitor of COX-2, was significantly decreased the content of PGE2 in these two depression models (P < 0.05, Fig. 1F). These results indicate that COX-2 is generally overexpressed in hippocampal DG regions in these two rat models of depression. 3.2. COX-2 inhibition ameliorated depression-like behaviors in rat depression model To determine whether COX-2 overexpression contributes to the depression-like behaviors in these two rat models of depression, both CUMS or LPS rats were treated with celecoxib, a specific COX-2 inhibitor, prior to stress exposure. As shown in Fig. 2A, the percent of sucrose consumption was significantly different among the groups [F (4, 86) = 14.87, P < 0.05]. Post-hoc analysis revealed that 5 weeks of CUMS exposure significantly reduced sucrose consumption as compared with that of the non-stressed control group (P < 0.05). Such responses indicated that these CUMS rats were experiencing anhedonia, a core symptom of depression. However, post-hoc analysis showed that pretreatment with celecoxib significantly increased sucrose consumption within CUMS rats (P < 0.05), increased sucrose consumption was also observed in response to the classic antidepressant, fluoxetine, which served as a positive control (P < 0.05). No statistically significant differences were obtained between the celecoxib-treated and fluoxetine-treated CUMS rats with regard to the percent of sucrose consumption (P > 0.05). In the forced swim test, rats showed significantly difference in immobility times [F (4, 86) = 17.36; P < 0.05] and swimming times [F (4, 86) = 15.79; P < 0.05] (Fig. 2B). Post-hoc analysis revealed that CUMS rats showed significantly increased immobility times and decreased swimming times (responses indicative of depression) as compared to the control groups (P < 0.05 respectively), while these CUMS-induced behavioral changes were significantly reversed by a chronic treatment of either celecoxib or fluoxetine (P < 0.05 respectively). There were no significant differences between the celecoxib-treated and fluoxetine-treated CUMS rats with regard to these behaviors (P > 0.05). Moreover, similar antidepressant-like effects of celecoxib and fluoxetine were also observed in response to the LPS-induced stress paradigm (Fig. 2C and D). Taken together, these results provide strong evidence for the involvement of COX-2 in the induction of depression phenotypes in these rat models of depression.

3.4. COX-2 inhibition suppressed oxidative DNA damage in hippocampal DG regions of rat depression model An accumulation of ROS within the brain can result in DNA damage, thus leading to cell death within both neurons and glia. Therefore, we next examined the level of oxidative base damage in CUMS rats with or without celecoxib pretreatment. As shown in Fig. 4A, two-way ANOVA revealed that the expression of 8-oHdG, a marker of oxidative base damage, within the DG region was significantly different among the four groups [F (3, 20) = 19.32, P < 0.01]. Post-hoc analysis revealed that he significant increase in staining intensity of 8-oHdG in depressed rats was effectively reversed by COX-2 inhibition (P < 0.01). There were no statistically significant differences between the celecoxib pretreated control group and control groups with regard to 8-oHdG expression (p > 0.05). Moreover, post-hoc analysis indicated that the increased levels of the double-strand breaks (DSBs) marker, γ-H2AX, in depressed rats were also reversed by celecoxib, as revealed with use of immunofluorescence staining (P < 0.05, Fig. 4B). Western blot results also showed that PARP1, a protein senses in DNA strand breaks which contributes to their repair, was increased in DG regions of CUMS rats, an effect which was reversed by COX-2 inhibition (P < 0.01). Meanwhile, two-way ANOVA followed by post-hoc analysis indicated that the increased p53 levels caused by stress-exposure was also reversed by COX-2 inhibition (P < 0.01, Fig. 4C). As NADPH oxidases (NOX) are a major source of ROS, we next examined whether the expression of NOX family members could be modulated by celecoxib. As shown in Fig. 4D, two-way ANOVA revealed that the increased levels of NOX1 and NOX4 transcripts observed in CUMS rats were significantly attenuated by celecoxib treatment (P < 0.05). These results indicated that COX-2 inhibition plays an important role in cellular responses to oxidative DNA damage.

3.3. COX-2 inhibition suppressed oxidative stress in hippocampal DG regions of rat depression model Depression is usually accompanied by neuronal injury in specific brain regions, along with an excessive production of ROS that may lead to senescence or cell death. Accordingly, we next examined whether the 7

Neuropharmacology 160 (2019) 107779

Q. Song, et al.

Fig. 4. COX-2 inhibition suppressed oxidative DNA damage in hippocampal DG regions of depressed rats. (A) Representative images of 8-OHdG staining (red) within the DG area of each group of rats. Nuclei (blue) are stained with DAPI. Scale bar is 50 μm. (B) Representative images of γ-H2AX staining (red). Nuclei (blue) are stained with DAPI. Scale bar is 50 μm. (C) Western blotting analysis of PARP1 and p53 protein expression of each group. (D) RT-PCR analysis of NOX1and NOX4 mRNA levels of each group. Band intensities were normalized to β-actin or GAPDH. N = 6 per group. *P < 0.05, **P < 0.01, compared to Control group; # P < 0.05, compared to CUMS group. (Cele, Celecoxib).

weeks of CUMS exposure significantly increased Iba1 positive cells in DG regions (P < 0.01, Fig. 5A), as well as an overexpression in CD11b and CD45 proteins, two important microglial markers within the brain (P < 0.01 respectively, Fig. 5B). These results suggested that CUMS exposure induced microglial activation within DG regions. Astroglial activation also plays a crucial role in neuroinflammatory responses. Consistently, immunofluorescence staining and Western blot assay showed that the number of GFAP positive astroglia and GFAP protein

3.5. COX-2 inhibition suppressed neuroinflammatory responses in hippocampal DG regions of rat depression model We next examined some of the possible mechanisms through which ROS, in turn, produces depression-like behaviors in rats. Results from our immunofluorescence assay showed that he number of Iba-1+ microglia within the DG region was significantly different among the four groups [F (3, 20) = 18.06, P < 0.01]. Post-hoc analysis indicated that 58

Neuropharmacology 160 (2019) 107779

Q. Song, et al.

Fig. 5. COX-2 inhibition suppressed neuroinflammatory responses in hippocampal DG regions of stressed rats. (A) Immunofluorescence staining of Iba1 positive microglial cells within the DG region. Scale bar is 50 μm. (B) Western blotting analysis of CD45 and CD11b protein expressions of each group. (C) Western blotting analysis of GFAP protein expression of each group. (D) Immunofluorescence staining of GFAP positive astrocytes within the DG region. Scale bar is 50 μm. (E) RT-PCR assays of mRNA expression levels of IL1β, TNF-α and IFN-γ within the DG regions of rats. Band intensities were normalized to β-actin or GAPDH. N = 6 per group. *P < 0.05, **P < 0.01, compared to Control group; #P < 0.05, ##P < 0.01 compared to CUMS group. (Cele, Celecoxib).

9

Neuropharmacology 160 (2019) 107779

Q. Song, et al.

As confirmed in the present study, CUMS-exposure induced a marked overexpression of COX-2. If oxidative stress mediates the antidepressant-like effects of COX-2 inhibition in CUMS rats, NAC pretreatment would be expected to alleviate the neural inflammation and apoptosis resulting from COX-2 overexpression. Indeed, two-way ANOVA followed by post-hoc analysis revealed that increased mRNA levels of the pro-inflammatory cytokines IL-1β, IFN-γ and TNF-α (P < 0.05 respectively, Fig. 7C) as well as mRNA levels of pro-apoptotic factors (P < 0.05 respectively, Fig. 7D) induced by CUMS-exposure were significantly decreased by NAC pretreatment. In addition, as shown in Fig. 7E and F, dendritic spine and synapse deficiencies in DG neurons resulting from CUMS exposure were significantly rescued by NAC pretreatment (P < 0.05 respectively). These results indicate that oxidative stress can promote processes involved with neuroinflammation and apoptosis, two important pathogenic mechanisms in neuronal injury. Finally, NAC pretreatment also produced a pronounced amelioration in depression-like behaviors (P < 0.05, Fig. 7G). Taken together, these results suggest that suppression of oxidative stress may serve as one of the important mechanistic bases for the antidepressantlike effects of COX-2 inhibition.

levels within the DG regions were significantly increased in CUMS rats (P < 0.05 respectively, Fig. 5C and D). As activation of glia cells usually triggers the secretion of cytokines, we next evaluated the expressions of several critical pro-inflammatory cytokines. Two-way ANOVA showed that there were significant differences among the four groups with regard to the mRNA levels of interleukin-1β (IL-1β) [F (3, 20) = 16.31, P < 0.05], interferon gamma (IFN-γ) [F (3, 20) = 17.97, P < 0.01] and tumor necrosis factor-a (TNF-a) [F (3, 20) = 18.24, P < 0.01]. Post-hoc analysis indicated that CUMS exposure significantly increased mRNA expression levels of the pro-inflammatory cytokinesIL-1β, IFN-γ and TNF-a within the DG as compared to that observed in the non-stressed control group (Fig. 5E). More important, this glial activation was significantly prevented by celecoxib treatment (P < 0.01), which indicated that COX-2 inhibition effectively suppressed the substantial increases in inflammatory responses produced by CUMS exposure in rats. 3.6. COX-2 inhibition decreased neural apoptosis in hippocampal DG of rat depression model Excessive production of ROS may cause cell death or senescence. In order to further investigate whether COX-2 inhibition could regulate neural apoptosis in CUMS rats, results obtained from immunofluorescence staining revealed that cleaved Caspase 3, a terminal effector of the apoptotic cascade, was co-expressed with the neuronal marker NeuN in DG regions. Two-way ANOVA showed that there were significant differences among the four groups with regard to the number of positive double labeled cells within the DG region [F (3, 20) = 18.15, P < 0.01]. Post-hoc analysis indicated that these positive double labeled cells were significantly increased within the DG regions of CUMS rats (P < 0.01, Fig. 6A). Moreover, Hoechst-33258 staining and transmission electron microscopy were used to observe morphological changes of nuclei within DG cells. While no significant ultrastructural changes were observed within the non-stressed control group, the CUMS-exposure groups exhibited characteristics of apoptosis, including nuclear chromatin margination, aggregation and condensation (Fig. 6B). In this way, two-way ANOVA followed by post-hoc analysis indicated that the significant increases in transcriptional levels of the apoptosis-related proteins Bax (P < 0.05), capase 3 (P < 0.05) and caspase-9 (P < 0.01) were accompanied with decreased expression levels of Bcl-2 (P < 0.05) within the DG of the CUMS rats (Fig. 6C). In response to celecoxib pretreatment, these CUMS exposureinduced apoptotic morphological changes and expressions of proapoptotic factors within the DG region were suppressed. These results provide further evidence that COX-2 inhibition contributes to the suppression of neuronal apoptosis which may then be responsible for the amelioration of depression-like behaviors.

4. Discussion Despite recent advances in the treatment of depression in patients (Dean and Keshavan, 2017), much work remains for the understanding and management of this neuropsychiatric disorder. For example, identification of the key and universal molecules involved in this heterogeneous disease may provide novel insights for the development of therapeutic strategies for such patients. In this report we observed that COX-2 is overexpressed in both CUMS-induced and LPS-induced animal models of depression. This COX-2 overexpression results in the generation of ROS in the mitochondria and enhancement of DNA oxidative damage, eventually leading to neuroinflammation and apoptosis in DG regions of rat depression moedel. Conversely, COX-2 inhibition significantly suppressed oxidative stress as well as depressive phenotypes in these rats. Moreover, treatment with the antioxidant, NAC, attenuated this neuronal injury and thus the depressive behaviors. These results suggest that strategies involved with COX-2 inhibition may be useful for the design potential therapies in the treatment of depression disorders. The findings presented in this study indicate that down-regulation of COX-2 expression significantly rescued depression-like behaviors, effects similar to that in response to treatment with the classic antidepressant, fluoxetine. Interestingly, inhibition of COX-2 enzymatic activity with celecoxib decreases its product PGE2, a potential pro-inflammatory factor. Celecoxib is a nonsteroidal anti-inflammatory drug (NSAID) that specifically inhibits COX-2. In clinical practice, celecoxib reduces hormonal levels, for example the PGE2 responsible for inflammation and pain in the body (Lin et al., 2014). Here, we found that increased PGE2 expression resulting from CUMS exposure is prevented by celecoxib. In addition, the significant enhancement in the co-expression of COX-2 along with 4-HNE, a sensitive marker of oxidative damage and lipid peroxidation, within DG regions of depressed rats suggest that increased levels of COX-2 synthesis may be associated with the expression of a state involving increased levels of oxidative stress. In our current study, we demonstrate that CUMS exposure elevates oxidative stress and induces oxidative DNA damage in hippocampal DG regions of rat depression model. Specifically, in response to chronic stress DHE and 4-HNE were significantly upregulated in DG regions; effects which were accompanied by DNA damage resulting from excessive ROS as indicated by elevated levels two markers of oxidative DNA damage, 8-OHdG and γ-H2AX. Interestingly, celecoxib effectively attenuated ROS production and mitigated this oxidative DNA damage. Such findings provide strong evidence indicating that the overexpression of COX-2 represents an important factor in the generation of depression-like behaviors, likely by promoting oxidative stress. It has

3.7. Antioxidant treatment rescues the neural inflammation, apoptosis and depression-like behaviors resulting from COX-2 overexpression The main product of COX-2, PGE2, is involved in many biological processes. Accordingly, it is possible that the antidepressant-like effects exerted by COX-2 inhibition may due to processes other than that involved with suppressing oxidative stress. As an approach to determine whether oxidative stress plays a critical role in the pathogenesis of depression, we examined the potential for an antioxidant to rescue the depression phenotypes resulting from CUMS exposure. As shown in Fig. 7A, post-hoc analysis indicated that levels of MDA induced by CUMS-exposure were significantly reduced by N-acetylcysteine (NAC) pretreatment (P < 0.05), effects which were accompanied with elevated activity of the antioxidant stress kinase, CAT (P < 0.05) and GSH-Px (P < 0.05). Results from our Western blot assay showed that the expression levels of HO-1 and phosphorylation of ERK were increased by NAC pretreatment, while phosphorylation levels of p38MAPK and JNK were significantly attenuated (P < 0.05, Fig. 7B). 10

Neuropharmacology 160 (2019) 107779

Q. Song, et al.

Fig. 6. COX-2 inhibition decreased neural apoptosis in hippocampal DG of stressed rats. (A) Representative images of NeuN/cleaved Caspase 3 double labeled positive cells within the DG region. Scale bar is 50 μm. (B) Representative images of Hoechst-33258 staining (Scale bar is 5 μm) and TEM to observe morphological changes of nuclei (Scale bar is 2 μm) and ultrastructure (Scale bar is 0.5 μm) of DG neurons. Arrow indicates swelling mitochondria and vacuolar endoplasmic reticulum. (C) RT-PCR assays of mRNA expression levels of Bcl-2, Bax, cleaved caspase-3 and caspase-9 within the DG region. Band intensities were normalized to GAPDH. N = 6 per group. *P < 0.05, **P < 0.01, compared to Control group; #P < 0.05, ##P < 0.01 compared to CUMS group. (Cele, Celecoxib).

depression model. PARP1, a protein that senses DNA strand breaks, plays an important role in response to oxidative DNA damage (Luo and Kraus, 2012). Cleavage of PARP-1 by caspase-3 is believed to represent a universal hallmark of apoptotic cell death via inhibiting DNA repair (D'Amours et al., 2001). Hence, the findings that PARP levels were decreased by COX-2 inhibitors in depression model suggest that the decreased persistence of oxidative DNA damage when COX-2 is inhibited may be due to processes involved with repairing this damage. In addition, we found that NOX1 and NOX4 were upregulated by CUMS exposure and are involved in these effects of COX-2 inhibition.

been reported previously that mitochondrial dysfunction in various brain regions is associated with the pathogenesis of depression (Bansal and Kuhad, 2016). Mitochondria, which are major sites of ROS generation, are sensitive to oxidative stress and exhibit limited function to eliminate oxidants (Moldovan and Moldovan, 2004). In addition to enhancing activity of antioxidative enzymes, COX-2 inhibition also suppresses mitochondrial stress, upregulates NRF2 antioxidant function and inhibits the downstream mitogen-activated protein kinase (MAPK) c-Jun N-terminal kinase (JNK) and p38 activation. All of these processes contribute to the maintenance of redox homeostasis in rat 11

Neuropharmacology 160 (2019) 107779

Q. Song, et al.

(caption on next page)

12

Neuropharmacology 160 (2019) 107779

Q. Song, et al.

Fig. 7. Antioxidant treatment rescues neural inflammation and apoptosis resulting from COX-2 overexpression in depressed rats. (A) Contents of MDA and NO and activity of GSH-Px were analyzed. (B) Western blotting analysis of HO-1, phosphorylated p38, JNK and ERK protein expression of each group. (C) RT-PCR assays of mRNA expression levels of IL1β, TNF-α and IFN-γ within DG regions. (D) RT-PCR assays of mRNA expression levels of Bcl-2, Bax, cleaved caspase-3 and caspase-9 within DG regions. (E) Representative electron micrograph of vmPFC neurons in rats from each group. Arrows indicate spine synapses. Scale bar is 0.2 μm. (F) Representative images of the spine densities of dendrites within DG areas as assessed with Golgi staining. Scale bar is 10 μm. (G) Depression-like behaviors were verified with use of sucrose preference and forced swim tests. Band intensities were normalized to β-actin or GAPDH. N = 6 per group. *P < 0.05, **P < 0.01, compared to Control group; #P < 0.05, ##P < 0.01, compared to CUMS group. (Cele, Celecoxib).

pathway (Borsini et al., 2018), while antidepressants reversed IL-1βinduced reduction of human neurogenesis (Borsini et al., 2017). Related to these findings, the apoptosis observed in our rat model could likely be due to the enhanced inflammation resulting from COX-2 overexpression, which eventually leading to depression. There is also evidence that the activation of apoptotic cell death is dependent on elevated levels of ROS (Morris and Berk, 2015; McManus et al., 2014), that subsequently provoke the phosphorylation of MAPK JNK, an effector molecule of the apoptotic cascade (Katagiri et al., 2010). Accordingly, the oxidative stress induced by COX-2 can either promote apoptosis directly, or indirectly, through COX-2-induced neuroinflammation. As a result, the series of neuronal changes present within stressed rats, ranging from oxidative stress, inflammation, DNA damage to apoptosis, can all, at least partially, be traced back to COX-2 overexpression. To further verify the protective action of antioxidant defenses in the pathology of depression, in the present study we show that the oxidative stress inhibitor, NAC, ameliorates neuronal injury and depressive behaviors in this rat model by inhibiting inflammatory responses and apoptosis. These observations supported that a reciprocal relationship exists between the oxidative stress and neuroinflammatory responses which may then, in part, jointly contribute to the behavioral alterations underlying depression. NAC supplementation in CUMS rats inhibited the hyperactivity of oxidative stress by reducing levels of lipid peroxidation, NO, and increasing the activity of GSH-Px. In addition, we also evaluated the effects of pretreatment with NAC on neuronal dendritic spine morphology of synaptic structures in depression model. Previous evidence has demonstrated that impairments in neuroplasticity are mediated through various cascading mechanisms including the inflammatory immune system and apoptosis, as observed under pathophysiological conditions associated with depression (Anisman, 2009; Kubera et al., 2010). Here, we show that chronic stress resulted in defective terminal dendrite spine densities as accompanied by a loss of spines and synapses, which would likely result in synaptic transmission deficiencies and thus behavioral changes. Findings from recent reports have indicated that high ROS concentrations were associated with deficits in synaptic plasticity and a decline in cognitive functions in some neurodegenerative disorders and age-dependent decays in neuroplasticity (Hu et al., 2006; González-Fraguela et al., 2018). We found that NAC treatment significantly ameliorated these decreases in synaptic spine densities and cytoplasmic organelle damage, effects that were associated with improving behavioral deficits resulting from chronic stress. These findings suggest a potential link between COX-2 expression and enhanced neuronal deterioration in depression rat model. Specifically, the presence of COX-2 in DG regions appears to support an active role for COX-2 in oxidative stress which may then induce depression-like behaviors. Besides COX-2 inhibition, the profound antioxidant, anti-inflammatory and anti-apoptosis capacities of NAC would also suggest that it may prove an effective therapeutic approach in the treatment of neurological and neuropsychiatric disease. In conclusion, our studies provide the first evidence that COX-2 inhibition protects against neuronal injury and depressive behaviors. This capacity appears to, at least in part, be exerted through suppression of the oxidative stress pathway. These observations suggest that targeting the COX-2 inhibition to develop new therapeutic strategies for the treatment of depression might provide a novel and valuable route for future research in this area.

Therefore, we speculate that the antidepressant effects of COX-2 inhibition may attributable to decreased oxidative stress, which, in part, is partially mediated by the upregulation of the NADPH oxidases NOX1 and NOX4. Taken together, it appears that this orchestrated cascade of events resulting in oxidative stress and oxidative DNA damage, appear to be critical components underlying depression, and that a COX-2 inhibition-based therapy may serve as potential new approach for the treatment of this disorder. In recent years, neuroinflammation has emerged as a critical factor in the etiology of depression (Miller et al., 2009). Normalization of inflammatory markers is often associated with remission of depressive symptoms in some patients (Schmidt et al., 2016). ROS are typically characterized as toxic molecules involved with oxidizing membrane lipids, producing conformational changes in proteins, damaging nucleic acids and instigating neuronal inflammatory responses (Forrester et al., 2018). In addition, overexpression of PGE2 was reported to stimulate localized production of pro-inflammatory cytokines by resident glia cells (Johansson et al., 2013). As a result of these two pathways, activation of the neuroinflammatory system might then serve as an important underlying mechanism in the development of depression-like behaviors initiated by COX-2 overexpression. Consistent with this suggestion, our data show that COX-2 inhibition significantly attenuated the activation of microglia and astrocytes within the DG hippocampus. In addition, we provide further evidence that links the activation of the COX-2 and subsequent production of PGE2 with the release of the pro-inflammatory cytokines IL-1β, TNF-a and IFN-γ in this depression model. Further support for a role of inflammation has been provided by results obtained following anti-inflammatory suppression of COX-2 by celecoxib. This treatment down-regulated the production of these cytokines within the DG region, which may then ameliorate depressive phenotypes induced by inflammation. Here, our experimental results verified that selective inhibition of COX-2 ameliorated depression-like behaviors in a CUMS rat model of depression, which appears, in part, to be attributable to protection against neuronal inflammation and thus a lowering in the degree of neuronal injury. In addition to the established pleiotropic neuroprotective effects of COX-2 inhibition, such as anti-oxidation and anti-inflammation, we also found that COX-2 inhibition can be protective against neuronal death in DG regions. Similar to previous findings, our present results showed that the upregulation of COX-2 was accompanied with an associated increase in caspase-3 dependent apoptosis. This apoptosis is usually indicated by nuclear chromatic agglutination and margination. Here, we demonstrate that the nuclear apoptotic damage resulting from CUMS exposure was prevented by the selective inhibition of COX-2 and this selective COX-2 inhibition also appears to decrease the double labelling of neurons with NeuN and cleaved caspase 3 in this rat model of depression. Moreover, selective COX-2 inhibition leads to downregulation of the pro-apoptotic proteins, Bak, Caspase-3 and Caspase-9 and an upregulation of the anti-apoptotic protein, Bcl-2, within the DG of rats. These results imply that COX-2 inhibition may lower neuronal apoptotic rates and thus diminish the display of depression-like behaviors. Results from previous studies have shown that increased levels of pro-inflammatory cytokines, such as IL-1β, are present in depressed patients (Zou et al., 2018; Su et al., 2017), which was proposed to be the key contributor to neuronal deterioration in depression (Kaufmann et al., 2017). Moreover, recent studies reported that IFN-α could alter adult neurogenesis and induces cell death via activation of STAT1

13

Neuropharmacology 160 (2019) 107779

Q. Song, et al.

Conflicts of interest statement

modification of NMDA receptor 1. J. Neurosci. 37 (15), 4200–4212. Jiménez-Fernández, S., Gurpegui, M., Díaz-Atienza, F., Pérez-Costillas, L., Gerstenberg, M., Correll, C.U., 2015. Oxidative stress and antioxidant parameters in patients with major depressive disorder compared to healthy controls before and after antidepressant treatment: results from a meta-analysis. J. Clin. Psychiatr. 76 (12), 1658–1667. Johansson, J.U., Pradhan, S., Lokteva, L.A., Woodling, N.S., Ko, N., Brown, H.D., Wang, Q., Loh, C., Cekanaviciute, E., Buckwalter, M., Manning-Bog, A.B., Andreasson, K.I., 2013. Suppression of inflammation with conditional deletion of the prostaglandin E2 EP2 receptor in macrophages and brain microglia. J. Neurosci. 33 (40), 16016–16032. Katagiri, K., Matsuzawa, A., Ichijo, H., 2010. Regulation of apoptosis signal-regulating kinase 1 in redox signaling. Methods Enzymol 474, 277–288. Kaufmann, F.N., Costa, A.P., Ghisleni, G., Diaz, A.P., Rodrigues, A.L.S., Peluffo, H., Kaster, M.P., 2017. NLRP3 inflammasome-driven pathways. in depression: clinical and preclinical findings. Brain Behav. Immun. 64, 367–383. Kennedy, K.A., Sandiford, S.D., Skerjanc, I.S., Li, S.S., 2012. Reactive oxygen species and the neuronal fate. Cell Mol. Life Sci. 69 (2), 215–221. Kryston, T.B., Georgiev, A.B., Pissis, P., Georgakilas, A.G., 2011. Role of oxidative stress and DNA damage in human carcinogenesis. Mutat. Res. 711 (1–2), 193–201. Kubera, M., Obuchowicz, E., Goehler, L., Brzeszcz, J., Maes, M., 2010. In animal models, psychosocial stress-induced (neuro) inflammation, apoptosis and reduced neurogenesis are associated to the onset of depression. Prog. Neuropsychopharmacol. Biol. Psychiatry. 35 (3), 744–759. Lin, T.Y., Lu, C.W., Wang, C.C., Huang, S.K., Wang, S.J., 2014. Cyclooxygenase 2 inhibitor celecoxib inhibits glutamate release by attenuating the PGE2/EP2 pathway in rat cerebral cortex endings. J. Pharmacol. Exp. Ther. 351 (1), 134–145. Luo, X., Kraus, W.L., 2012. On PAR with PARP: cellular stress signaling through poly (ADPribose) and PARP-1. Genes Dev 26 (5), 417–432. Mao, Q.Q., Ip, S.P., Ko, K.M., Tsai, S.H., Che, C.T., 2009. Peony glycosides produce antidepressant-like action in mice exposed to chronic unpredictable mild stress: effects on hypothalamic-pituitary-adrenal function and brain-derived neurotrophic factor. Prog. Neuropsychopharmacol. Biol. Psychiatr. 33, 1211–1216. McManus, M.J., Murphy, M.P., Franklin, J.L., 2014. Mitochondria-derived reactive oxygen species mediate caspase-dependent and -independent neuronal deaths. Mol. Cell Neurosci. 63, 13–23. Miller, A.H., Maletic, V., Raison, C.L., 2009. Inflammation and its discontents: the role of cytokines in the pathophysiology of major depression. Biol. Psychiatry 65, 732–741. Moldovan, L., Moldovan, N.I., 2004. Oxygen free radicals and redox biology of organelles. Histochem. Cell Biol. 122, 395–412. Morris, G., Berk, M., 2015. The many roads to mitochondrial dysfunction in neuroimmune and neuropsychiatric disorders. BMC Med 13 (68). Nissanka, N., Moraes, C.T., 2018. Mitochondrial DNA damage and reactive oxygen species in neurodegenerative disease. FEBS Lett 592 (5), 728–742. Oh, D.H., Son, H., Hwang, S., Kim, S.H., 2012. Neuropathological abnormalities of astrocytes, GABAergic neurons, and pyramidal neurons in the dorsolateral prefrontal cortices of patients with major depressive disorder. Eur. Neuropsychopharmacol. 22, 330–338. Papakostas, G.I., Ionescu, D.F., 2015. Towards new mechanisms: an update on therapeutics for treatment-resistant major depressive disorder. Mol. Psychiatry. 20 (10), 1142–1150. Quan, C., Cho, M.K., Perry, D., Quan, T., 2015. Age-associated reduction of cell spreading induces mitochondrial DNA common deletion by oxidative stress in human skin dermal fibroblasts: implication for human skin connective tissue aging. J. Biomed. Sci. 22, 62. Scali, C., Giovannini, M.G., Prosperi, C., Bellucci, A., Pepeu, G., Casamenti, F., 2003. The selective cyclooxygenase-2 inhibitor rofecoxib suppresses brain inflammation and protects cholinergic neurons from excitotoxic degeneration in vivo. Neuroscience 117 (4), 909–919. Schmidt, F.M., Schröder, T., Kirkby, K.C., Sander, C., Suslow, T., Holdt, L.M., Teupser, D., Hegerl, U., Himmerich, H., 2016. Pro- and anti-inflammatory cytokines, but not CRP, are inversely correlated with severity and symptoms of major depression. Psychiatry Research 239, 85–91. Sen, H., Deniz, S., Yedekci, A.E., Inangil, G., Muftuoglu, T., Haholu, A., Ozkan, S., 2014. Effects of dexpanthenol and N-acetylcysteine pretreatment in rats before renal ischemia/reperfusion injury. Ren. Fail. 36 (10), 1570–1574. Shan, H., Bian, Y., Shu, Z., Zhang, L., Zhu, J., Ding, J., Lu, M., Xiao, M., Hu, G., 2016. Fluoxetine protects against IL-1β-induced neuronal apoptosis via downregulation of p53. Neuropharmacology 107, 68–78. Song, Q., Fan, C., Wang, P., Li, Y., Yang, M., Yu, S.Y., 2018. Hippocampal CA1 βCaMKII mediates neuroinflammatory responses via COX-2/PGE2 signaling pathways in depression. J. Neuroinflamm. 15 (3), 8. Stockmeier, C.A., Mahajan, G.L., Konick, L.C., Overholser, J.C., Jurjus, G.J., Meltzer, H.Y., Uylings, H.B., Friedman, L., Rajkowska, G., 2004. Cellular changes in the postmortem hippocampus in major depression. Biol. Psychiatry 56, 640–650. Su, W.J., Zhang, Y., Chen, Y., Gong, H., Lian, Y.J., Peng, W., Liu, Y.Z., Wang, Y.X., You, Z.L., Feng, S.J., Zong, Y., Lu, G.C., Jiang, C.L., 2017. NLRP3 gene knockout blocks NFκB and MAPK signaling pathway in CUMS-induced depression mouse model. Behav. Brain Res. 322, 1–8. Takadera, T., Yumoto, H., Tozuka, Y., Ohyashiki, T., 2002. Prostaglandin E (2) induces caspase-dependent apoptosis in rat cortical cells. Neurosci. Lett. 317, 61–64. Uchihara, Y., Tanaka, K., Asano, T., Tamura, F., Mizushima, T., 2016. Superoxide dismutase overexpression protects against glucocorticoid-induced depressive-like behavioral phenotypes in mice. Biophys. Res. Commun. 469 (4), 873–877. Valko, M., Leibfritz, D., Moncol, J., Cronin, M.T., Mazur, M., Telser, J., 2007. Free radicals and antioxidants in normal physiological functions and human disease. Int. J.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. Acknowledgments This study was supported by grants to Shu Yan Yu from the National Natural Science Foundation of China (NSFC81873796; NSFC81471371), Shandong Provincial Key Research and Development Plan (2017CXGC1504), the Fundamental Research Funds of Shandong University (2018JC008) and the Key Research and Development Foundation of Shandong Province (2018GSF118050). Appendix A. Supplementary data Supplementary data to this article can be found online at https:// doi.org/10.1016/j.neuropharm.2019.107779. References Adzic, M., Djordjevic, J., Mitic, M., Brkic, Z., Lukic, I., Radojcic, M., 2015. The contribution of hypothalamic neuroendocrine, neuroplastic and neuroinflammatory processes to lipopolysaccharide-induced depressivelike behaviour in female and male rats: involvement of glucocorticoid receptor and C/EBP-β. Behav. Brain Res. 291, 130–139. Anisman, H., 2009. Cascading effects of stressors and inflammatory immune system activation: implications for major depressive disorder. J. Psychiatry Neurosci. 34 (1), 4–20. Bansal, Y., Kuhad, A., 2016. Mitochondrial dysfunction in depression. Curr. Neuropharmacol. 14 (6), 610–618. Bjelland, S., Seeberg, E., 2003. Mutagenicity, toxicity and repair of DNA base damage induced by oxidation. Mutat. Res. 531 (1–2), 37–80. Borsini, A., Cattaneo, A., Malpighi, C., Thuret, S., Harrison, N.A., MRC ImmunoPsychiatry Consortium, Zunszain, P.A., Pariante, C.M., 2018. Interferon-alpha reduces human hippocampal neurogenesis and increases apoptosis via activation of distinct STAT1dependent mechanisms. Int. J. Neuropsychopharmacol. 21 (2), 187–200. Borsini, A., Alboni, S., Horowitz, M.A., Tojo, L.M., Cannazza, G., Su, K.P., Pariante, C.M., Zunszain, P.A., 2017. Rescue of IL-1β-induced reduction of human neurogenesis by omega-3 fatty acids and antidepressants. Brain Behav. Immun. 65, 230–238. Cui, Y., Yang, Y., Ni, Z., Dong, Y., Cai, G., Foncelle, A., Ma, S., Sang, K., Tang, S., Li, Y., Shen, Y., Berry, H., Wu, S., Hu, H., 2018. Astroglial Kir4.1 in the lateral habenula drives neuronal bursts in depression. Nature 554 (7692), 323–327. D'Amours, D., Sallmann, F.R., Dixit, V.M., Poirie, r G.G., 2001. Gain-of-function of poly (ADP-ribose) polymerase-1 upon cleavage by apoptotic proteases: implications for apoptosis. J. Cell Sci. 114 (Pt 20), 3771–3778. Dean, J., Keshavan, M., 2017. The neurobiology of depression: an integrated view. Asian J. Psychiatr. 27, 101–111. Duman, C.H., Schlesinger, L., Kodama, M., Russell, D.S., Duman, R.S., 2007. A role for MAP kinase signaling in behavioral models of depression and antidepressant treatment. Biol. Psychiatr. 61, 661–670. Fan, L.W., Kaizaki, A., Tien, L.T., Pang, Y., Tanaka, S., Numazawa, S., Bhatt, A.J., Cai, Z., 2013. Celecoxib attenuates systemic lipopolysaccharide‐induced brain inflammation and white matter injury in the neonatal rats. Neuroscience 240, 27–38. Forrester, S.J., Kikuchi, D.S., Hernandes, M.S., Xu, Q., Griendling, K.K., 2018. Reactive oxygen species in metabolic and inflammatory signaling. Circ. Res. 122 (6), 877–902. Glushakov, A.O., Glushakova, O.Y., Korol, T.Y., Acosta, S.A., Borlongan, C.V., Valadka, A.B., Hayes, R.L., Glushakov, A.V., 2018. Chronic upregulation of cleaved-caspase-3 associated with chronic myelin pathology and microvascular reorganization in the thalamus after traumatic brain injury in rats. Int. J. Mol. Sci. 19 (10). González-Fraguela, M.E., Blanco-Lezcano, L., Fernandez-Verdecia, C.I., Serrano Sanchez, T., Robinson Agramonte, M.L.A., Cardellá Rosales, L.L., 2018. Cellular redox imbalance and neurochemical effect in cognitive-deficient old rats. Behav. Sci. (Basel). 8 (10). Govoni, S., Masoero, E., Favalli, L., Rozza, A., Scelsi, R., Viappiani, S., Buccellati, C., Sala, A., Folco, G., 2001. The cycloxygenase-2 inhibitor SC58236 is neuroprotective in an in vivo model of focal ischemia in the rat. Neurosci. Lett. 303, 91–94. Gulyaeva, N.V., 2018. Functional neurochemistry of the ventral and dorsal Hippocampus: stress, depression, dementia and remote hippocampal damage. Neurochem. Res. 44 (6), 1306–1322. Hwang, O., 2013. Role of oxidative stress in Parkinson's disease. Exp. Neurobiol. 22 (1), 11–17. Hu, D., Serrano, F., Oury, T.D., Klann, E., 2006. Aging-dependent alterations in synaptic plasticity and memory in mice that overexpress extracellular superoxide dismutase. J. Neurosci. 26 (15), 3933–3941. Ibi, M., Liu, J., Arakawa, N., Kitaoka, S., Kawaji, A., Matsuda, K.I., Iwata, K., Matsumoto, M., Katsuyama, M., Zhu, K., Teramukai, S., Furuyashiki, T., Yabe-Nishimura, C., 2017. Depressive-like behaviors are regulated by NOX1/NADPH oxidase by redox

14

Neuropharmacology 160 (2019) 107779

Q. Song, et al.

apoptosis in neuronal cells. Biochim. Biophys. Acta. 1498, 72–79. Zou, W., Feng, R., Yang, Y., 2018. Changes in the serum levels of inflammatory cytokines in antidepressant drug-naïve patients with major depression. PLoS One 13 (6), e0197267.

Biochem. Cell Biol. 39 (1), 44–84. Vivancos, A.P., Castillo, E.A., Biteau, B., Nicot, C., Ayté, J., Toledano, M.B., Hidalgo, E., 2005. A cysteine-sulfinic acid in peroxiredoxin regulates H2O2-sensing by the antioxidant Pap1 pathway. Proc. Natl. Acad. Sci. 2012, 8875–8880. Wei, T., Chen, C., Hou, J., Xin, J., Mori, A., 2000. Nitric oxide induces oxidative stress and

15