Therapeutic STING activation boosts microglia phagocytosis and ameliorates depression-like behaviors during chronic restraint stress

The STING-TBK1-IRF3 signaling pathway involves in modulating host innate immunity, however, the potential role of STING signaling pathway in chronic restraint stress model has not been determined. The aim of this study is to explore the underlying role of STING signaling pathway in regulating neuroinammation, as well as to evaluate the therapeutic potential of STING agonist during chronic restraint stress. C57BL/6 mice were subject to 14-day intermittent restraint stress. Sucrose preference, elevated plus maze and tail suspension tests were measured in chronic restraint stress mice. Expression levels of proinammatory cytokines were tested by QT-PCR and Luminex cytokine assays. The uorescence-labeled latex beads, ow cytometry and CD68 positive cell counts were utilized to evaluate phagocytic abilities of microglia. Then, the ability of intracerebroventricular injection of STING agonist, 2’3-cGAMP, to reverse the depression-like behaviors and inammatory cytokines was examined.


Abstract Background
The STING-TBK1-IRF3 signaling pathway involves in modulating host innate immunity, however, the potential role of STING signaling pathway in chronic restraint stress model has not been determined. The aim of this study is to explore the underlying role of STING signaling pathway in regulating neuroin ammation, as well as to evaluate the therapeutic potential of STING agonist during chronic restraint stress. Methods C57BL/6 mice were subject to 14-day intermittent restraint stress. Sucrose preference, elevated plus maze and tail suspension tests were measured in chronic restraint stress mice. Expression levels of proin ammatory cytokines were tested by QT-PCR and Luminex cytokine assays. The uorescencelabeled latex beads, ow cytometry and CD68 positive cell counts were utilized to evaluate phagocytic abilities of microglia. Then, the ability of intracerebroventricular injection of STING agonist, 2'3-cGAMP, to reverse the depression-like behaviors and in ammatory cytokines was examined.

Results
We found that the expression levels of STING, p-TBK1, and p-IRF3 were remarkably decreased in chronic restraint stress mice, which was associated with decreased IFN-β secretion. Moreover, the STING agonist, 2'3-cGAMP, signi cantly alleviated the neuroin ammation and ameliorated depression-like behavior which depends on the functional STING activation. Furthermore, 2'3-cGAMP promoted microglia phagocytosis through cGAMP-STING-dependent IFN-β release, which was essential for recovery from neuroin ammation during chronic restraint stress.

Conclusions
These ndings demonstrate that STING signaling pathway is a critical mediator in regulating microglia phagocytosis and may serve as a novel therapeutic target for chronic stress-related psychiatric diseases.

Background
Psychosocial stress plays an important role in the pathophysiology of stress-related psychiatric diseases such as depression and anxiety disorders [1], and leads to various health problems including immune de ciencies, digestive disorders and cardiovascular diseases [2,3]. Chronic restraint stress causes activation of the brain's immune responses manifested in microglia activity and leads to depression-like profound microglia alterations characterized by number, morphology and function [4,5]. Several innate immune receptors of microglia have previously been implicated in the chronic stress-induced immune response. For instance, a recent study has suggested crucial roles of TLR2/4 in medial prefrontal cortex for repeated social defeat stress-induced social avoidance [6]. And activation of P2X7 receptor and the NLRP3 in ammasome in hippocampus mediates depression-like behavior induced by chronic stress [7]. A deeper understanding of the roles of intracellular signaling pathways of microglia in chronic stress is useful to optimize preventive and therapeutic strategies to combat the increasing incidence of psychosocial diseases.
Stimulator of interferon gene (STING), as an adaptor protein, predominantly resides in the endoplasmic reticulum and regulates innate immune signaling processes by detecting aberrant cytosolic DNA [8,9].
Previous studies have reported that chronic stress triggers certain biological pathways that ultimately result in the accumulation of DNA damage [10], and a cytosolic DNA sensing pathway (especially cGAS-STING pathway) is the major link between DNA damage and innate immunity [11,12]. STING subsequently initiates further signaling cascades by recruiting and activating TANK-binding kinase 1 (TBK1) and transcription factor interferon regulatory factor 3 (IRF3), ultimately leading to the production of interferon-β (IFN-β) [13,14]. Intriguingly, STING was exclusively expressed by microglia in the brain [15].
STING signaling pathway has been implicated in neuroin ammation during neurodegeneration, but its mechanism of induction and its consequences in chronic restraint stress model remain unclear.
In the current study, we found that chronic restraint stress induces a transient proliferation and activation of microglia, accompanied with the increase of in ammatory cytokines in brain. Moreover, STING is downregulated in chronic restraint stress mice and STING agonist cGAMP induces a type I interferon response in microglia which depends on a functional STING pathway. Treatment with cGAMP enhances microglia phagocytosis by inducing the production of IFN-β, relieves neuroin ammation and ameliorates depression-like behaviors during chronic restraint stress. Collectively, our ndings identify the STING pathway as a regulator of microglia phagocytosis and uncover a potential use of STING agonists for the treatment of stress-related mood disorders.

RST protocol
The Experimental Animal Ethics Committee of Xi'an Jiaotong University has approved all the animal experiments involved in this study. Eight-week-old male wild-type C57BL/6 mice weighing between 20-23 g were housed in a speci c pathogen-free animal faculty of the Center for Brain Science, the First A liated Hospital of Xi'an Jiaotong University. According to an established paradigm, the restraint stress experiments were performed after 1 week of habituation. The RST mice were placed in well-ventilated plastic tubes to restrict the movement of the limbs for 6 hours (from 9:30 AM to 15:30 PM) per day lasting 14 days.

Behavioral tests
All behavioral tests were performed during the afternoon in a dedicated sound-proof behavioral facility by researchers blind to treatment. Mice were brought to the procedure room 1 h before the start of each behavioral test and remained in the same room throughout the test. At all times, sound was masked with 55-60 Db of white noise. Mice were allowed to rest for one day after the completion of the 14-day RST protocol.
Sucrose preference tests were performed on day 16. A bottle with tap water and another bottle with 2% sucrose solution were given to mice. The mice had been trained (given a free choice between two bottles) every day to avoid neophobia. To prevent a possible effect of behavior, the left/right location of the bottles was switched every day. Preference was calculated by determining the percentage of sucrose consumed divided by the total uid intake (sucrose intake/total uid intake × 100%).
Tail suspension tests were performed on day 18. As previously described [16],mice were suspended individually by their tails from a metal rod xed 30 cm above the surface of a table. The tip of the tail was xed using adhesive tape (the distance from tip of the tail was 2 cm). The subject mouse was suspended by the tail for 6 min. Mice were acclimated for the rst 1 min, and immobility time was measured for the last 5 min. Immobility was de ned as no movement of the limbs and tails. Mice that climbed up their tails were excluded from the experiment.
An elevated plus maze test was performed on a plus-shaped apparatus with two open and two closed arms. The maze was elevated 50 cm from the oor. Mice were tested in a single 5-min session. The time stayed in open arms and the number of entries into the open arms was scored using a video-tracking system (Smart 3.0). Between each trial, the maze was cleaned with 75% ethanol.

Microglial cell density and morphological analyses
Pictures were taken throughout the entire 12µm thickness of the slice by Z-stack, covering the whole surface of the region (prefrontal cortex and hippocampus). All images from a given slice were acquired with the same exposure time and acquisition parameters. Unbiased stereological counting of Iba1positive cells was done using a modi ed version of the optical fractionator as was previously described [5]. The prefrontal cortex and hippocampus borders were delineated under low magni cation and the stereo-investigator 10 software (MBF Bioscience) was used to determine the surface area and to count cells. The software automatically calculates the number of Iba1-labeled cells within the region of interest.
Microglial morphologic analysis was performed in the blind using ImageJ software (NIH, Bethesda, Maryland, USA). Brie y, confocal images for the selected marker Iba-1 were modi ed as 8-bit projection images. The resulting images were smooth processed, binarized and skeletonized, using the Skeletonize Plugin for ImageJ [17]. For each microglia cell, a polygonal region of interest (ROI) that circumscribed all microglial processes belonging to that cell was set. From each ROI a single cell image was generated, single cell image was subjected to Skeletonize Plugin. To determine the 'number of branches', the 'number of junctions', and 'the maximum branch length' the particle-ltered image was processed by choosing the Analyze Skeleton 2D/3D option in the Skeletonize Plugin, and analyzed using the Fiji plugin 'Analyze Skeleton' to obtained the data from the results table.

Microglia isolation and ow cytometry
Brains were dissected into 1-2 mm 3 pieces with small scissors and incubated for 30 min at 37°C in 1 ml HBSS solution containing 2% FBS, 1 mg/ml Collagenase D and 50 µg/ml DNase I. 10 mM EDTA was added to the 1.5 mL tube to stop the enzymatic reaction and homogenates were pipetted for further dissociation. Next the homogenates were ltered through a nylon mesh (70 µm pore size) in a 50 mL tube using the plunger of a 10 mL syringe, washed with cold FACS buffer (2% FBS, 1 mM EDTA in PBS without Ca 2+ or Mg 2+ ) and centrifuged at 300g at room temperature for 5 min. For the enrichment of microglia, the cell pellet was resuspended with 30%-37%-70% Percoll density gradient medium and centrifuged at 800g, no acceleration and braking, at room temperature for 40 min. Next, the 70 − 37% density gradient interphase containing the macrophage subpopulations cells was collected into a clean tube, washed with 5 ml FACS buffer and centrifuged at 500g at 4°C for 5 min, followed by antibody labeling and ow cytometry analysis. Antibodies against PerCP-CD11b (Biolegend, San Diego, CA), PE-CD45 (Biolegend, San Diego, CA), CD16/32 (Biolegend, San Diego, CA), xable viability dye eFluor 520 (ebioscience, San Diego, CA) were used. Analysis was performed on a Calibur (BD Biosciences, BD Diva Software) and analyzed with FlowJo software.

Intracerebroventricular injection of 2'3-cGAMP
Animals were anesthetized with 4% chloral hydrate for a surgical plane of anesthesia throughout the procedure. Animals were placed in a stereotaxic apparatus (RWD Life Science, Shenzhen, China) with the ear bars in the ear canals and the incisors in the tooth bar of the mouse adapter. After making an approximately 1-cm incision in the scalp, the periosteum was scrubbed from the skull with sterile cottontipped applicators to reveal the bregma. Hamilton syringes tted with 22-gauge Huber point removable needles were lled with 10 µl of saline with or without 2'3-cGAMP (diluted from 20 µg/ml in saline, Abcam, Cambridge, UK). The needle was positioned over the bregma and then moved to coordinates 0.22 mm anterior, 1.0 mm right, and 2.5 mm down after the bevel of the needle disappeared through the skull. The needle was injected until the resistance decreased slightly, indicating that the needle had penetrated into the lateral ventricle. 2'3-cGAMP (2.5 µl, 20 µg/ml) was injected gradually by hand over approximately 8 min. After 5 min, the needle was backed out of the skull while applying downward pressure on the animal's skull with a sterile cotton-tipped applicator. The incision was sutured with 1 horizontal mattress stitch using 6 − 0 suture. Animals were allowed to recover from the anesthesia in their home cage. After 5 days, mice were placed in well-ventilated plastic tubes to restrict the movement of the limbs for 6 hours per day lasting 14 days.

Assessment of phagocytosis
Phagocytosis was assessed in BV2 microglia cell line using amino-modi ed polystyrene latex beads (1.0µm mean particle size, Sigma-Aldrich, Boston, MA) as the phagocytosis target. Microglia were plated at the concentration of 1 × 10 4 cells in a confocal dish and treated with 2'3-cGAMP (200 ng/ml in the culture medium) for 24 h. Then 0.5 µL of uorescent latex beads were re-suspended in medium for 2 h at 37°C. After 2 h of incubation, 1 mL of cold PBS was added and washed three times until non-phagocytosed beads were removed. Bead intake was quanti ed by measuring the raw intensity density using the Fiji software and maintaining the same threshold restrictions for all the experimental conditions. We characterized microglia phagocytosis as the amount of internalized bioparticles per cell in actively engul ng cells. In all cases, quanti cations were performed in 50 cells from 5 elds per experimental condition per experiment, with ve independent experiments being performed.
Phagocytosis of uorogenic E. coli particles (pHrodo Green, Invitrogen) was analyzed using BV2 microglial cells. Brie y, cells were plated in 6-well plates at a density of 2 × 10 6 cells and cultured for 24 to 48 hours. pHrodo E. coli bioparticles were dissolved in PBS to a concentration of 1 mg/ml, and a total of 100µg of bioparticles was added per well and incubated for 60 min at 37°C. As a negative control, phagocytosis was inhibited with 10 mM cytochalasin D, which was added 30 min before addition of pHrodo E. coli bioparticles. Cells were harvested by trypsinization, washed two times with FACS sample buffer, and analyzed by ow cytometry on a BD FACS Canto™ II ow cytometer. Data analysis was performed using the FlowJo software.

Statistical analysis
All statistical analyses were performed using GraphPad Prism 6.0. In all cases, data are presented as the mean ± SEM. Two-tailed unpaired independent t-test was performed to compare two groups. One-way ANOVA was performed to evaluate the signi cance differences between multiple groups, and Bonferroni post hoc tests were performed following signi cant main effects or interactions in ANOVA where applicable. Two-way ANOVA was performed to evaluate the signi cant differences with conditions (CON vs RST), treatments (vehicle vs cGAMP) or an interaction (stress by cGAMP). At least three independent repeated experiments are performed. *P < 0.05; **P < 0.01; ***P < 0.001.

Chronic restraint stress induces depression-like behaviors and neuroin ammation
During the 14-day experimental period with 6-h restraint stress daily (Fig. 1a), the body weight of RST mice was signi cantly decreased (from 22.5 ± 0.31 g to 20.4 ± 0.46 g) in the rst 5 days compared to CON mice (from 22.4 ± 0.32 g to 23.2 ± 0.45 g, Fig. 1b). The total food intake of RST mice was reduced markedly during the rst three days (from 16.0 g to 12.2 g). However, food intake of RST mice was almost equal with CON mice by day 4, since the food intake of RST mice showed a tendency to increase from the 5th to the 14th day. Next, we observed the behavioral consequences in CON and RST mice after they completed the 14-day paradigm. RST mice displayed anhedonia in a sucrose preference test compared to CON mice (P < 0.01, Fig. 1c). Moreover, the tail suspension test revealed a signi cant increase of the immobility time in RST mice (CON vs. RST, 126.28 ± 10.60 s vs. 160.60 ± 8.68 s, P < 0.05, Fig. 1d). In the elevated plus maze test, RST mice spent relatively less time in the open arms (CON vs. RST, 71.26 ± 10.47 s vs. 40.05 ± 6.72 s, P < 0.05, Fig. 1e). Meanwhile, the total number of open arm entries was also decreased in RST mice (P < 0.05, Fig. 1e). Taken together, these data indicated that chronic restraint stress results in depression-like behaviors.
To further assess whether chronic restraint stress induces neuroin ammation in mice, we measured the mRNA expression of in ammatory cytokines in the prefrontal cortex and hippocampus. We found that the mRNA expression of TNF-α was increased to 2.12 ± 0.12-fold in the prefrontal cortex tissue of RST mice compared to that in CON mice (P < 0.001). Similarly, IL-1β expression was increased to 1.78 ± 0.26fold and IL-6 expression was increased to 1.56 ± 0.21-fold in the RST group (P < 0.05). In addition, CXCL10 and CCL2 expression levels were also signi cantly increased in RST mice (Fig. 1f). Similarly, the mRNA expression of in ammatory cytokines in the hippocampus of RST mice were also increased. These results revealed that chronic restraint stress can enhance the release of pro-in ammatory cytokines in prefrontal cortex and hippocampus.
Microglia are activated towards pro-in ammatory phenotype during chronic restraint stress To consider the effect of chronic restraint stress on microglia density and morphology, we then analyzed changes in microglia cell number by counting Iba1 stained cells and morphology by counting branches and junctions per cell. Our data indicated that the Iba1-positive cell number was signi cantly higher in the prefrontal cortex (CON vs. RST, 7743 ± 159 vs. 9873 ± 141 cells/mm 3 , P < 0.01) and hippocampus (CON vs. RST, 6228 ± 139 vs. 7531 ± 130 cells/mm 3 , P < 0.05) of RST mice (Fig. S1a-c). The analysis of cell morphology in the prefrontal cortex and hippocampus of RST mice revealed a more rami ed phenotype, with a higher number of branches and junctions per cell (P < 0.001, Fig. S1d,e). However, there was no change in the maximum branch length of microglia between RST mice and CON mice (Fig. S1f).
Consistently, ow cytometry analysis revealed Cd11b high CD45 low microglia number was increased in RST mice compared to CON mice (P < 0.05, Fig. S1g). As we known, the M1 phenotype is more likely associated with uncontrolled neuroin ammation, while the M2 phenotype promotes in ammation resolution and tissue repair. Interestingly, we found that the expression levels of M1 microglia markers induced nitric-oxide synthase (iNOS, P < 0.01 in prefrontal cortex, P < 0.05 in hippocampus) and interleukin-6 (IL-6, P < 0.05 in prefrontal cortex, P < 0.01 in hippocampus) were signi cantly increased in RST mice (Fig. 2a-c). In contrast, M2 microglia markers brain-derived neurotrophic factor (BDNF, P < 0.001 in prefrontal cortex, P < 0.001 in hippocampus) and arginase-1 (Arg-1, P < 0.05 in prefrontal cortex, P < 0.05 in hippocampus) expressions were reduced in RST mice (Fig. 2d,e). Above all, these results suggested that microglia are increased in number and more activated towards M1 phenotype during chronic restraint stress.
STING and its downstream effectors p-TBK1 and p-IRF3 protein expressions are inhibited in chronic restraint stress mice To verify whether STING was critically associated with the change of microglia phenotypes, we used lipopolysaccharide and interferon-γ as M1 triggers and interleukin-4 and interleukin-13 as M2 triggers to stimulate BV2 microglia. We observed that STING protein expression was increased by IL-4/IL-13 treatment (1.7 ± 0.2-fold, control vs. IL-4/IL-13) and inhibited by LPS/IFN-γ treatment (0.4 ± 0.1-fold, control vs. LPS/IFN-γ) (Fig. 2f,g). To further demonstrate the role of the STING signaling pathway in vivo, we measured STING, p-TBK1, and p-IRF3 protein expressions in the prefrontal cortex and hippocampus of RST mice. The immuno uorescence staining assay rst revealed that the expression of STING was signi cantly decreased in the prefrontal cortex and hippocampus of RST mice (Fig. S2a,b). This observation was further con rmed by western blotting for STING protein expression after chronic restraint stress. The result showed that the protein expression level of STING was very low in the prefrontal cortex (P < 0.001) and hippocampus (P < 0.01) of RST mice (Fig. S2c,d). Moreover, the phosphoprotein TBK1 (p-TBK1) and the phosphoprotein IRF3 (p-IRF3) protein levels were also decreased in the prefrontal cortex and hippocampus of RST mice Fig. S2e,f). Hence, these results suggested that a dominant proin ammatory microglia phenotype is might characterized by a suppression of the STING signaling pathway during chronic restraint stress.
Treatment with cGAMP ameliorates depression-like behaviors and inhibits production of proin ammatory cytokines through the STING-TBK1-IRF3 axis We next injected the STING agonist 2'3'-cGAMP into the lateral ventricles and assessed the extent of neuroin ammation and depression-like behaviors in mice. As shown in Fig. 3a and 3b, the expression of STING was markedly inhibited in the prefrontal cortex and the hippocampus of RST mice treated with saline, while 2'3'-cGAMP administration partly rescued the expression of STING in RST mice. Consistently, western blotting assays con rmed that the expression of STING was increased in the cGAMP + RST group compared with that in the vehicle + RST group (P < 0.01 in the prefrontal cortex and P < 0.01 in the hippocampus, respectively, Fig. 3c,d), while STING expression showed no difference between the cGAMP group and the cGAMP + RST group. Likewise, cGAMP administration directly activates STING resulted in similar phosphorylation of TBK1 and IRF3 in RST mice (Fig. 3e,f). We also performed RT-PCR to examine the mRNA levels of in ammatory cytokines in the brain including TNF-α, IL-1β, IL-6, IL-10, CCL2 and CXCL10. Pro-in ammatory cytokines TNF-α, IL-1β and IL-6 were higher in vehicle + RST mice, and this effect was nearly abolished in cGAMP + RST mice (Fig. 4a,b). Unexpectedly, analysis of serum in ammatory cytokine levels was consistent with changes in the brain (Fig. S3). Our data showed that STING agonist 2'3'-cGAMP induced less in ammatory cytokines production in RST mice. Importantly, RST mice treated with 2'3'-cGAMP displayed more sucrose consumption in the sucrose preference test, lower immobility in tail suspension test and more times in open arms in the elevated plus maze than vehicle + RST mice (Fig. 4c-e). Collectively, these results suggested that STING activation signi cantly alleviates neuroin ammation and ameliorates restraint stress induced-depression-like behaviors through the STING-TBK1-IRF3 axis.

Microglia phagocytosis is enhanced through cGAMP-STING-dependent IFN-β release during chronic restraint stress
Previous study demonstrated that cGAMP administration enhanced the expressions of type I IFN and IFNrelated genes [18]. IFN-β produced by microglia activates phagocytic activity and increases the clearance of myelin debris, which is suggested to cause downregulation of pro-in ammatory cytokines. To determine whether STING signaling pathway was involved in microglia phagocytosis and whether STING activation could affect the phagocytic activity by inducing the release of IFN-β, we assessed microglia phagocytosis and measured the production of IFN-β in vitro and in vivo. In vitro, STING was successfully knocked down by using siRNA transfection method (Fig. 5a). Phagocytosis of uorescent beads analyses indicated that LPS-stimulated microglia treated with cGAMP displayed an increase red-bead occupancy compared with LPS-stimulated microglia (LPS vs. LPS + cGAMP 19.8 ± 6.3% vs. 32 ± 7.9%, P < 0.01), while a decrease red-bead occupancy was observed in LPS + cGAMP + siRNA group (LPS + cGAMP vs. LPS + cGAMP + siRNA, 19.8 ± 6.3% vs. 13.7 ± 3.04%, P < 0.001, Fig. 5b,c). The cellular IFN-β production of externally added cGAMP are expected to be similarly affected. We observed that the levels of IFN-β were signi cantly increased in LPS + cGAMP group compared with that in LPS group (P < 0.05, Fig. 5d). In vivo, we then assessed the number of CD68-positive phagocytic microglia and measured the IFN-β expression in the brain. Our results indicated that CD68-positive phagocytic microglia number were signi cantly increased in the cGAMP + RST group, although the CD68-positive microglia numbers were not changed in the cGAMP group compared with the vehicle group (Fig. 5e-h). The IFN-β mRNA expression in the prefrontal cortex was signi cantly increased in cGAMP-treated RST mice compared with vehicle + RST mice (P < 0.001, Fig. 5i).
To further clarify whether the phagocytic machinery was dependent on the cGAMP-STING-IFN-β signaling, microglia were treated with recombinant IFNβ (r-IFNβ) or STING agonist cGAMP and incubated with pHrodo-Green labeled E. coli bioparticles for measurement of phagocytosis e ciency by ow cytometry. We found that LPS-stimulated microglia phagocytosed E. coli bioparticles more e ciently following treatment with r-IFNβ or STING agonist cGAMP (P < 0.05, Fig. 5j). In contrast, treatment with cGAMP, the IFNβ-knockdown microglia exhibited a decrease in phagocytotic rates.
In order to better understand the interaction between microglia phagocytosis and depression-like behavior when STING pathway was activated, a Pearson correlation analysis was conducted to examine whether there is relationship between CD68 positive cell numbers, TNF-α or IL-6 level and times in open or closed in the elevated plus maze test (Fig. 6). In the cGAMP-treated restraint stress mice, we observed signi cant negative correlation between phagocytic microglia cell counts and decreased plasma TNF-α level (r = − 0.599, P = 0.011, Fig. 6a), decreased plasma IL-6 level (r = − 0.597, P = 0.015, Fig. 6b). Moreover, a signi cant strong positive correlation was observed between phagocytic microglia cell counts and times in open arms of depressive-like behavior (r = 0.534, P = 0.027, Fig. 6c). However, there was no signi cant correlation between phagocytic microglia cell counts and times in closed arms (r = 0.116, P = 0.656, Fig. 6d). Taken together, these data indicated that cGAMP-STING-dependent IFN-β signaling is required to stimulate microglia phagocytosis, and STING activation might alleviate chronic restraint stress-induced neuroin ammation and depression-like behavior through boosting microglia phagocytosis.

Discussion
Chronic restraint stress produced signi cant physiological and behavioral changes, notably reduced weight gain and depression-like behaviors. These changes were also accompanied by signs of microglia activation in the brain. This is consistent with past evidence that exposure to stress can trigger a neuroin ammatory response [3]. To our knowledge, neuroin ammation heavily relies on innate immune responses that are primarily mediated by microglia. As a signaling hub in innate immunity, STING is a key adaptor molecule in orchestrating the body's response to in ammation, cell death and autophagy [19,20]. Although STING is best known for its role in immune responses to cytoplasmic DNA sensed by cGAS [21], and microglia utilize the cGAS-STING pathway to orchestrate the antiviral response in the brain [22], the molecular mechanisms underlying cGAS-STING pathway contributions to depression-like behaviors remains unknown. Here, an intriguing suggestion that arises from the present study is that activation of the STING-TBK1-IRF3 pathway might be served as a potential mechanism to boost microglia phagocytosis through promoting the production of IFN-β in chronic restraint stress mice, consequently alleviating neuroin ammation and ameliorating depression-like behavior.
Of particular interest is that human microglia constitutively express cGAS and its critical downstream adaptor STING [23]. In a Parkinson's disease mouse model, STING-mediated in ammation has been implicated in promoting microglia activation and neuronal death [19]. Microglia are activated via the cGAS-STING pathway to induce the production of interferons to launch antiviral defense pathways [24]. Moreover, in the multiple sclerosis model [15], the antiviral drug Ganciclovir inhibits in ammation and leads to protection through activation of the STING pathway. It is interesting to note that excessive engagement of the cGAS-STING pathway in the brain can lead to neuroin ammation and neurodegeneration. It is clearly indicated that the number of microglia in certain stress-sensitive brain regions is increased during chronic stress [4]. This result was also con rmed in our study, indicating that microglia are increased in number and more activated towards M1 phenotypes in prefrontal cortex and hippocampus of RST mice. The expressions of M1-polarized microglia markers as well as the production of pro-in ammatory cytokines were increased in RST mice. Interestingly, in BV2 microglia, we observed that M1-polarized microglia have the lower level of STING expression. Thus, we wondered that whether activation of the STING pathway could ameliorate chronic restraint stress-induced depressive symptoms and relieve neuroin ammation. CDG, CDA, 3'3-cGAMP, and 2'3-cGAMP are the native agonists of STING [25,26]. 2'3-cGAMP, as a higha nity ligand of STING, is a non-classical cyclic dinucleotide that activates STING and triggers potent immune responses [14,27]. The cGAS-STING pathway is a dual-edged sword, which may be activated or inhibited to arrive at the desired outcome. The activation of STING plays a pivotal role in mediating innate immune responses and in removing multiple pathogens including both viruses and bacteria [28,29]. In contrast, in chronic neurodegenerative states such as the prion disease, activation of the cGAS-STING signaling pathway ameliorated in ammation and disease progression [30]. Supporting these data, other studies showed that STING activation worsens acute pancreatitis severity in experimental models [31], but plays a protective role in chronic pancreatitis models [32]. Although the detrimental and protective roles of STING in regulating immune response have been studied, the precise role of STING signaling pathway in the chronic stress model has not been clari ed. We rst observed that STING activation reduced levels of brain pro-in ammatory cytokines or chemokines and ameliorated depression-like behaviors in RST mice.
In the brain, STING, which elicits the interferon response, is expressed predominantly in microglia [33]. Microglia plays a major role in the non-autonomous clearance of protein aggregates in the central nervous system [34]. Previous study reported macrophage proteins that are affected by STING and demonstrated the relationship between STING and phagocytosis [35]. In agreement with this nding, we rst found that microglia phagocytosis is affected by STING activation in RST mice. A question arising from our work relates to how cGAMP improved microglia phagocytosis in the procession of chronic stress. One possibility is that higher levels of IFN-β caused by activation of STING signaling pathway may have contributed to the enhanced phagocytosis and anti-depressive effect observed in the present study.
It should be noted that IFN-β is also central mediator of CNS in ammation during autoimmunity, where depending on the disease, they can be either protective or detrimental [36]. Previously, Toll/interleukin-1 receptor domain-containing adapter inducing IFN-β (TRIF) de cient microglia exhibited an increased threshold for activation of interferon-regulated genes, suggesting that IFN-β may enhance phagocytic activity [37]. In experimental autoimmune encephalomyelitis, IFN-β producing microglia mediated an enhanced removal of myelin debris compared to IFN-β non-producing microglia [38]. Indeed, we observed that STING-activated microglia exhibit a superior phagocytic capacity, and IFN-β might guide positioning of microglia in the in amed central nervous system during chronic stress. In vitro, we observed the similar results that cGAMP-treated microglia displayed an increase in phagocytotic rates with LPS stimulation.
However, without LPS stimulation, STING activation does not affect microglia phagocytosis, although the release of IFN-β is signi cantly increased. In support of this, other study con rmed that treatment of macrophages with exogenous IFN-β alone is not su cient to induce the production of pro-in ammatory cytokines, but retreatment with IFN-β could lead to an increase in the anti-in ammatory cytokine and a decrease in pro-in ammatory cytokine expression in response to LPS stimulation [39]. On the other hand, a slight decrease was observed in open eld activity in the IFN-β-treated mice, and this amelioration of sickness was associated with trivial IL-6 production [40]. Hence, we concluded that microglia phagocytosis was enhanced via cGAMP-STING-dependent IFN-β signaling in chronic restraint stress model, which drives behavioral changes. Besides microglia phagocytosis, neurogenesis impairments [41], dysregulation of tryptophan-kynurenine metabolism [40] and dysfunction of the HPA axis [42] were also contributed to the depression-like behaviors. More detail of mechanism should be given in future study.
In addition, STING pathway activation is linked to multiple inhibitory feedback loops, such as in ammasome activation, autophagy induction, and autocrine IFN signaling. A previous study con rmed that chronic activation of the STING pathway is associated with reduced mTORC1 signaling in metabolically relevant tissues [43]. The STING pathway is also involved in interacting with the autophagy or mitophagy machinery in innate immune responses [44][45][46]. Moreover, another study revealed that STING knockout could attenuate cardiac injury by inhibiting NLRP3 in ammasome mediated in ammation, apoptosis and pyroptosis [13]. Further investigations will be needed to elucidate the molecular details, as well as the consequences of the interaction between STING and other signaling pathways in different diseases.

Conclusion
In summary, our study is the rst to elucidate the mechanism by which cGAMP treatment boosts microglia phagocytosis through activating IFN-β response in a STING-dependent manner. Furthermore, this current study demonstrated that the STING-TBK1-IRF3 pathway as a key player in mediating chronic stress-induced low-grade neuroin ammation has indicated an exciting new direction to elucidate the mechanism underlying psychosocial diseases.

Availability of data and materials
All data generated or analyzed during this study are included in this published article.

Competing interests
The authors declared that they had no con icts of interest with respect to their research, authorship or the publication of this article. and chemokines were increased in RST mice. mRNA expression levels of TNF-α, IL-1β, IL-6, IL-10, CXCL10 and CCL2 in brain tissues were examined using real-time PCR. Data are presented as the mean ± SEM. *P < 0.05, **P < 0.01 and ***P < 0.001 between CON vs. RST in t-tests (N = 7-10 mice per group).

Figure 2
In RST mice, microglia were activated towards to M1 phenotype in the prefrontal cortex and hippocampus. a Photomicrographs show protein levels of induced nitric-oxide synthase (iNOS), interleukin-6 (IL-6), arginase-1 (Arg-1), and brain-derived neurotrophic factor (BDNF). β-tubulin was used as a loading control. b Protein expression of iNOS in CON and RST mice. c Protein expression of IL-6 in CON and RST mice. d Protein expression of Arg-1 in CON and RST mice. e Protein expression of BDNF in CON and RST mice. f Protein expression of STING in resting microglia, IL-4/IL-13-induced M2 phenotype microglia and LPS/IFN-γ-induced M1 phenotype microglia, which are represented by green uorescence in g. Data are presented as the mean ± SEM. *P < 0.05,**P < 0.01 and ***P < 0.001(N =6 mice per group) Figure 3 Suppression of the STING-TBK1-IRF3 signaling pathway in RST mice was reversed after delivery of 2'3-cGAMP into the brain. a Immuno uorescent staining for Iba-1 (red) and STING (green) in the prefrontal cortex of vehicle, vehicle+RST, cGAMP, and cGAMP+RST mice. b Immuno uorescent staining for Iba-1 (red) and STING (green) in the hippocampus of vehicle, vehicle+RST, cGAMP, cGAMP+RST mice. The scale bar denotes 50 μm. c Representative western blot analyses of protein expression levels of STING, TBK1, and IRF3, which were measured in the prefrontal cortex and hippocampus of vehicle, vehicle+RST, cGAMP, and cGAMP+RST mice. d Protein level of STING was signi cantly increased in cGAMP mice, whereas there was no change in the cGAMP+RST group compared with the vehicle group. e Increased phosphorylation of TBK1 normalized to their total protein expression was observed in cGAMP mice. There was no signi cant difference between the vehicle and cGAMP+RST groups. f Meanwhile, increased phosphorylation of IRF3 normalized to their total protein expression was observed in cGAMP mice. βtubulin was used to normalize the total expression of proteins. Data are presented as the mean ± SEM. *P < 0.05, **P < 0.01, #P < 0.01 and ###P < 0.001 (N = 6-8 mice per group).

Figure 4
Effects of cGAMP on the brain in ammatory pro les and depression-like behaviors in RST mice. a mRNA expression levels of TNF-α, IL-1β, IL-6 in the brains of vehicle, vehicle+RST, cGAMP, and cGAMP+RST mice. b mRNA expression of IL-10, CXCL10 and CCL2 in vehicle, vehicle+RST, cGAMP, and cGAMP+RST Page 22/24 mice. c The sucrose preference ratio of vehicle, vehicle+RST, cGAMP, and cGAMP+RST mice. d The immobility time of restraint stress mice treated with cGAMP was signi cantly reduced in the tail suspension test. e The immobility time in open arms was increased in cGAMP+RST group compared with that in vehicle+RST group, and the total number of open arm entries in cGAMP+RST mice was also increased. Data are presented as the mean ± SEM. *P < 0.05 and **P < 0.01 (N =10-12 mice per group).