Deficient Autophagy in Microglia Aggravates Repeated Social Defeat Stress-Induced Social Avoidance

Major depressive disorder (MDD) is associated with repeated exposure to environmental stress. Autophagy is activated under various stress conditions that are associated with several diseases in the brain. This study was aimed at elucidating the autophagy signaling changes in the prefrontal cortex (PFC) under repeated social defeat (RSD) to investigate the involvement of microglial autophagy in RSD-induced behavioral changes. We found that RSD stress, an animal model of MDD, significantly induced initial autophagic signals followed by increased transcription of autophagy-related genes (Atg6, Atg7, and Atg12) in the PFC. Similarly, significantly increased transcripts of ATGs (Atg6, Atg7, Atg12, and Atg5) were confirmed in the postmortem PFC of patients with MDD. The protein levels of the prefrontal cortical LC3B were significantly increased, whereas p62 was significantly decreased in the resilient but not in susceptible mice and patients with MDD. This indicates that enhanced autophagic flux may alleviate stress-induced depression. Furthermore, we identified that FKBP5, an early-stage autophagy regulator, was significantly increased in the PFC of resilient mice at the transcript and protein levels. In addition, the resilient mice exhibited enhanced autophagic flux in the prefrontal cortical microglia, and the autophagic deficiency in microglia aggravated RSD-induced social avoidance, indicating that microglial autophagy involves stress-induced behavioral changes.


Introduction
Major depressive disorder (MDD) is an important social issue that can potentially lead to suicide, and lifetime prevalence estimates are usually high in the general population [1,2]. Repeated environmental stress is w5idely accepted to be involved in the pathogenesis of MDD, promoting its onset or recurrence [3]. Social activity is also avoided by patients with MDD. Social contacts provoke anxiety and depression, making minor stressors overwhelming [4]. Animals exposed to repeated stress have been used to understand the pathophysiology of MDD. For instance, repeated social defeat (RSD) stress causes a robust depression-like phenotype marked by anhedonia, anxiety, and social avoidance behaviors, and these behaviors are helpful for elucidating individual differences [5]. In the central nervous system (CNS), the prefrontal cortex (PFC) mediates the emotional influences on cognitive processes [6]. The PFC circuits are involved in stress responses in mice and patients [7,8].
Recent studies have suggested an association between MDD and autophagy. Autophagy signaling carries its components into the intracellular digestive system and lysosomes and degrades them to promote survival [9]. Previous studies showed increased expression of autophagy-related genes in mononuclear cells in patients with MDD [10]. Attenuation of the mechanistic target of rapamycin (mTOR) signaling in the postmortem brains of depressed patients has been reported [11]. In animal models, chronic mild unpredictable stress in mice has been reported to enhance hippocampal autophagy [12]. Moreover, inhibition of autophagy plays a protective role in reducing depressive-like behavior in rats [13]. Astrocytic autophagic flux involves mitochondrial clearance in a chronic mild stress murine model of depression [14]. These findings suggest that abnormalities in autophagy and subsequent functional changes in the brain are involved in stress-induced depressive behavior.
Although several studies have focused on changes in brain structure and function, we focused on the role of microglia in the current study. Microglia are major immune cells in the central nervous system (CNS) [15]. Their activation has been involved in various psychiatric disorders, including MDD [16]. The involvement of microglia in stress is evident, including changes in microglial density in patients with MDD patients [17] and microglial activation in suicidal and affective disorder patients [18,19]. Animal studies have revealed altered microglial morphology and higher resilience to stress-induced depression-like behavior in microglia-deficient mice [20]. RSD-induced avoidance is caused by microglial activation through toll-like receptors [21]. The microglial inflammatory response has led to an understanding of this pathology. However, it remains poorly understood whether microglial autophagy is also associated with immune response and behavioral changes related to stress and MDD. For instance, microglial autophagy inhibits microglia-derived TNF-α and enhances M1 but reduces M2 markers [22]. Deficient microglial autophagy impairs synaptic pruning and causes autism spectrum disorder-like behavior [23]. Microglial Atg5-deficient mice under chronic unpredictable stress during pregnancy showed decreased behavioral response to the antidepressant fluoxetine at one month postpartum [24]. However, the role of autophagy in microglia-driven behavioral changes in response to chronic stress has not yet been examined.
The current study was aimed at elucidating the autophagy signaling changes in the PFC under RSD to investigate the involvement of microglial autophagy in RSD-induced behavioral changes. We analyzed the transcripts of the autophagy-related gene (Atg) and protein levels of autophagosome markers in mouse PFC microglia under RSD stress. We also investigated the transcripts of ATGs in the postmortem PFC of the microarray database of patients with MDD. Furthermore, the microglial Atg7-knockout mice were used to evaluate the involvement of microglial autophagy in RSD-induced behavioral changes.

Materials and Methods
All experimental protocols were performed in accordance with the Guidelines for the Care of Laboratory Animals of Tohoku University Graduate School of Medicine (Sendai, Japan).

Animals.
For all experiments, 8-to 12-week-old male mice were used. C57BL/6J and Slc:ICR (CD-1) mice were purchased from SLC Japan Inc. (Shizuoka, Japan). The mice were individually housed and maintained on a 12 : 12 h light/ dark schedule with ad libitum access to food and water throughout the experimental period. The animals were acclimated for one week in our animal facility. Microglial-specific GFP-expressing CX3CR1 GFP/+ [25] and microglial ATG7deficient CX3CR1-Cre+;Atg7 flox/flox (Cre + ;Atg flox/flox ) mice were used in the current experiments. CX3CR1 GFP/GFP was obtained from the Jackson Laboratory and crossed with C57BL/6J mice. Floxed ATG7 mice obtained from RIKEN (RBRC02759) are generated by Komatsu et al. [26] and crossed with Tg(Cx3cr1-Cre)MW126Gsat mice [27] generated by Heintz (the Rockefeller University, GENSAT); Cx3cr1-Cre mouse [28] lines were generated at Tohoku University for more than ten generations. After weaning on postnatal days (PNDs) 21-28, all mice were housed socially in same-sex groups in a temperature-controlled environment under a 12 : 12 h light/dark cycle (lights on at 09:00 h) with ad libitum access to water and food. Genomic DNA extracted from mouse tails was used for the standard PCR genotyping.

Repeat Social Defeat Stress (RSD).
The RSD procedure was performed as previously reported [5] Corp., Sendai, Japan). Mice were exposed to a different CD1 aggressor mouse for 10 min daily for 10 days by removing the clear perforated Plexiglas divider. After the last exposure session, all the mice were housed individually.

Social Interaction Test (SIT).
On day 11, the SIT [5] was performed to identify subgroups of mice that were susceptible or resilient to social defeat stress. This was accomplished by placing the mice in an open-field test box containing an empty wire mesh cage (10 × 4:5 cm) located at one end. The social interaction of the mice was measured for 2.5 min, followed by 2.5 min in the presence of an unfamiliar aggressor confined in the wire-mesh cage. The "interaction zone" of the test arena encompassed a 14 × 24 cm rectangular area projecting 8 cm around the wiremesh enclosure. The duration spent by the subjects in the "interaction zone" was recorded using a video camera. The interaction ratio was calculated as the time spent in the interaction zone with an aggressor or time spent in the interaction zone without an aggressor. An interaction ratio of 1 was set as the cutoff, whereby mice with 2 Neural Plasticity scores < 1 were defined as "susceptible mice" to social defeat stress and those with scores ≥ 1 were defined as "resilient mice." The number of open and closed arm entries was combined to yield a measure of total entries, which reflected the general exploratory activity during the test.

Sucrose Preference Test (SPT).
The SPT employed a twobottle, free-choice sucrose consumption paradigm using previously described methods [29]. The mice were habituated to drink water from two tubes with stoppers fitted with ballpoint sippers (Ancare, Bellmore, NY, USA) for two days. They were then exposed to 1% sucrose or drinking water following habituation for three consecutive days. The weights of the water-or sucrose-containing bottles were measured before and at the end of this period. Sucrose preference was determined using the following equation: 2.6. Quantitative Real-Time PCR. Total RNA was extracted from PFC and used as a template for cDNA synthesis using random primers and the SuperScript VILO cDNA synthesis kit (Invitrogen, Carlsbad, CA, USA). The relative copy number of each transcript in each cDNA sample was measured using specific primers and iQ SYBR Green Supermix (Bio-Rad Inc., Hercules, CA, USA). A standard curve was constructed for each assay to adjust for differences in the amplification efficiency of the primer sets. 18S rRNA was used as an internal control for normalization. The forward and reverse primers for 18S were 5 ′ -GTAACCCGTTGAACCC CATT-3 ′ and 5 ′ -CCATCCA ATCGGTAGTAGCG-3 ′ , respectively. The forward and reverse primers for Atg5 were 5 ′ -GGAGAGAAGAGGAGCCAGGT-3 ′ and 5 ′ -TGTTGC CTCCACTGAACTTG-3′, respectively. The forward and reverse primers for Beclin1 (Atg6) were 5′-GGCCAATAA GATGGGTCTGA-3′ and 5′-GCTGCACACAGTCCAG AAAA-3′, respectively. The forward and reverse primers for Atg7 were 5 ′ -TCCGTTGAAGTCCTCTGCTT-3 ′ and 5 ′ -CCACTGAGGTTCACCATCCT-3 ′ , respectively. The forward and reverse primers for Atg12 were 5 ′ -TCCGTT GAAGTCCTCTGCTT-3 ′ and 5 ′ -CAGCACCGAAATGT CTCTGA-3 ′ , respectively. The forward and reverse primers for Lc3a were 5′-CATGAGCGAGTTGGTCAAGA-3′ and 5′-TTGACTCAGAAGCCGAAGGT-3′, respectively. The forward and reverse primers for Lc3b were 5′-CCCACC AAGATCCCAGTGAT-3′ and 5′-CCAGGAACTTGGTC TTGTCCA-3 ′ , respectively. The forward and reverse primers for Fkbp5 were 5 ′ -GAGTCTGCGAAAGGAC TTGG-3 ′ and 5 ′ -GTGGGTTCTACATCGGCACT-3 ′ , respectively.

Microarray Analyses of Postmortem Human Brains.
The microarray data of postmortem brain tissues (Brodmann area 10: anterior prefrontal cortex) from patients with schizophrenia and healthy controls (SOFT files and CEL files) were downloaded from the Gene Expression Omnibus (GEO) repository (GSE92538) housed at the National Center for Biotechnology Information (NCBI) on their FTP site (ftp://ftp.ncbi.nih.gov/pub/geo/). We used data from the postmortem dorsolateral PFC of patients with MDD. The SOFT and CEL files from the dataset GPL10526 (healthy subjects, n = 56; MDD patients, n = 29) [30], which included 54,120 probe sets, were imported into the BRB-Array Tools v4.6.0 Beta 1 software (https://brb.nci.nih.gov/BRB-ArrayTools/) [31]. Additionally, the dataset GPL17027 (healthy subjects, n = 111; MDD patients, n = 43) with only 12,334 probes were excluded. Signal intensities less than 50 and P values less than 0.05 were rejected. After normalizing the interarray variation among the 85 microarrays using quantile normalization, the significantly differentially expressed genes in each pairwise comparison were identified by a random variance t-test with the Benjamini-Hochberg false discovery (FDR) correction [31].

Statistical Analysis.
A two-tailed unpaired Student's t -test was used to evaluate the differences in the mean values between the two groups. One-or two-way analysis of variance (ANOVA) followed by Tukey's or Sidak post hoc tests was used for comparisons among more than two groups. Statistical analyses were performed using IBM SPSS Statistics for Windows, version 22.0 (IBM Japan, Tokyo, Japan).

Neural Plasticity
compared with those in healthy subjects. Furthermore, the signal intensity of SQRTM1/P62 was increased in MDD patients (P = 0:043) without statistical significance (q value = 0.149; Figure 3(f)). Our findings suggest autophagosome accumulation in the PFC of patients [37]. In addition, in the PFC of MDD patients, Fkbp5 transcripts tended to decrease (P = 0:060) with a statistically significant FDR (q value = 0.18; Figure 3(g)). Although decreased mTOR signal was reported in MDD patients [11], transcripts of mTOR increased in patients with MDD compared with controls in our results (P = 0:017) (FDR q value = 0.089; Figure 3(h)).

Microglial Autophagy Associated with Stress-Induced
Depressive-Like Behavior. The microglia are the primary glial cells of the innate immune system of the brain, and autophagy in microglia contributes to neurodegenerative diseases [38]. To examine whether autophagy occurs in microglia under chronic stress, microglial GFP (Cx3cr1-GFP) mice were used to confirm the LC3B-positive puncta that colocalized with p62. After one-way ANOVA with Tukey's multiple comparison analyses, LC3B was significantly increased in the PFC of resilient mice compared with nonstressed controls (P < 0:05) and susceptible mice (P < 0:05), respectively  Figure 2: Repeated social defeat stress induced the expression of autophagic signaling in the prefrontal cortex. (a) Levels of the mRNA encoding the autophagic signaling marker Atg6 relative to those of S18. (b) Levels of the mRNA encoding the autophagic signaling marker Atg7 relative to those of S18. (c) Levels of the mRNA encoding the autophagic signaling marker Atg12 relative to those of S18.

Discussion
Our results determined the increased transcription of the initial autophagy signaling proteins Atgs in the PFC of stressed mice that received RSD. RSD significantly increased the transcript levels of Atgs (Atg6, Atg7, and Atg12) in PFC of both susceptible and resilient mice, which are partially consistent with a previous report that chronic mild unpredictable stress activates hippocampal autophagy in mice [12]. However, enhanced autophagosome formation and autophagosome-lysosome degradation were only observed in resilient mice, whereas susceptible mice (P = 0:065) and MDD patients (P = 0:072) exhibited an increased tendency of failure of autophagosome formation and its accumulation. The ability to cope with stressful events varies across individuals, with resilient ones being able to control stress and susceptible ones being not. The mechanisms underlying these different stress responses have not yet been clarified. A previous study showed that autophagy plays an essential role in synaptic plasticity injury and cognitive decline [39,40]. Stressful event-selective reduction of dendritic spine density in the PFC of susceptible mice [41] could partly explain the different feedback of autophagy activation in resilient but not susceptible mice and patients with MDD. Furthermore, the PFC and the ventral tegmental area (VTA) are key brain regions within the neural circuit of the stress response [42]. RSD-induced mTOR phosphorylation significantly increased in the VTA only in the susceptible mice [43]. Similar to our results, FKBP5 was increased only in resilient mice, suggesting that different autophagy regulators may cause an inefficient and enhanced autophagic flux in susceptible and resilient mice, respectively. However, it is still necessary to define whether variations previously observed after RSD in other Atg expressions are differentially modulated in resilient versus susceptible mice to clarify its association with depressive-like behavior.
Autophagy is a degradative pathway that is essential for tissue homeostasis. Previous studies have shown that autophagy is increased not only by promoting autophagosome formation but also by blocking the disruption of autophagic flux [44]. In the CNS, autophagosome accumulation has been reported in a variety of neurodegenerative disorders, such as Alzheimer's disease, Huntington's disease, and Parkinson's disease [45,46]. After traumatic brain injury, impaired autophagic flux and pathological accumulation of autophagosomes cause neuronal cell death and exacerbate the severity of trauma [47]. However, alterations in autophagic flux in patients with MDD have not been established. Several studies have provided evidence of autophagic molecular changes in patients with MDD. For instance, the peripheral blood mononuclear cells of patients with MDD showed increased expression of autophagy-related genes, such as BECLIN1 (ATG6), ATG12, and LC3 [10]. mTOR signaling was attenuated in the postmortem brains of patients with MDD [11]. These results corroborate with our finding that initial autophagy signaling is activated in patients with MDD. Thus, initial autophagy signaling (Atg5, Atg6, Atg7, and Atg12) was elevated but limited without enhanced autophagic flux, such as increased LC3 and P62, in postmortem brains of MDD patients, similar to the RSD-induced susceptible mice. Together, these findings suggest that the induction of initial autophagy followed by impaired autophagic flux results in the pathological accumulation of autophagosomes and ultimately leads to depression. Interestingly, susceptible mouse PFC showed a significantly reduced autophagosome formation but significantly increased accumulation compared with that of resilient mice. Under RSD exposure, the diverse phases in autophagy flux, from the rate of autophagosome formation to the fusion of autophagosome-lysosome and its degradation, need to be elucidated.
Animal and in vitro studies have shown the potential roles of autophagy in the mechanism of antidepressant action. Microglial autophagy deficiency inhibits the behavioral effects of fluoxetine treatment on chronic unpredictable stress [24]. Desipramine elevated the autophagic protein levels of Beclin1 and LC3 in C6 glioma cells [48]. Imipramine stimulated autophagy progression in human U-87MG glioma cells [49], while ketamine promotes neural 9 Neural Plasticity differentiation of mouse embryonic stem cells via mTOR activation [50,51]. Thus, antidepressants have diverse effects on autophagy modulation. It is well known that mTOR activation is a crucial regulator of autophagy induction in the nervous system [52]. A previous study showed decreased mTOR protein levels in the postmortem PFC of patients with MDD [11]. However, our analyses of postmortem PFC in MDD patients 5showed different alterations, and mTOR transcripts were undetectable in the PFC of stressed and control mice. Recently, FK506 binding protein 51 (FKBP5) has been linked with autophagy regulators independent of mTOR signaling. Notably, FKBP5 can enhance autophagy and synergize with antidepressant action [53]. Under restraint stress, Fkbp5 mRNA levels were increased in the hippocampus, amygdala, and paraventricular nucleus of the hypothalamus [36]. In our study, the PFC showed significantly increased FKPB5, indicating that variable expression of FKBP5 in a specific brain region may affect different stress response patterns, and it is necessary to focus on the role of FKPB5 in astrocytes or neurons in the future. Moreover, the antidepressant ketamine inhibits mTOR sig-naling, although its anesthetic and hallucinogenic effects limit its clinical use in most countries. The prefrontal cortical FKBP5 induction may be used as an mTOR-independent antidepressant to prevent depression.
The microglial activation function in pruning and removing dead cells and releasing humoral factors for immune responses may be involved in the pathogenesis of MDD [15,54,55]. Recently, various studies have suggested the essential role of autophagy in microglia in the pathophysiology in the CNS. For instance, microglial Agt5 knockdown was sufficient to trigger M1 microglial polarization, while upregulation of autophagy promoted microglial polarization toward the M2 phenotype [22]. Microglial Atg7 deficiency was associated with reduced microglia-mediated neurotoxicity resulting in impaired microglial proinflammatory response [56]. In the animal studies, microglial autophagy is important for refining synapses during development, and defects cause autism spectrum disorder-like behavior [23]. Previous studies have shown that microglial autophagy dysfunction does not exhibit anxiety and depressive-like behavior [24], which is consistent with the lack of effects . One-way ANOVA followed by Turkey's post hoc test was applied to all comparisons. Data are presented as the mean ± SEM. * P < 0:05, * * * P < 0:001 vs. Cre-negative or Cre + ; repeated social defeat, Atg+/+; Cre-negative mice, Atg-/-; and Cre + ;Atg flox/flox mice. Naive: nonstressed; RSD: repeated social defeat. 10 Neural Plasticity on EPM and SPT in our results ( Figure 5). Furthermore, deficient autophagy in microglia impaired synaptic pruning [23] which may potentially explain the increased susceptibility and aggravated social avoidance in the microglial autophagy deficiency mice by RSD in our results. Additionally, in the CNS, activation of Toll-like receptors (TLRs) in microglia leads to impaired microglial autophagy [57]. Microglial TLR2/4 deficiency also abolishes RSD-induced social avoidance [21], suggesting that microglial autophagic regulation via TLR activation may affect stress-induced avoidance changes. In addition, enhanced autophagy in the PFC may occur in cells other than microglia in resilient mice. Thus, the behavioral roles of autophagy in astrocytes and neurons in anxiety-depressive-like behaviors remain to be studied [58].

Conclusion
Repeated social stress induced the initial activation of autophagy in the PFC of stressed mice and patients with MDD. The enhanced autophagic flux was only determined in the prefrontal cortical microglia of resilient mice, revealing the relationship between autophagy and stress-induced depressive behavior. Furthermore, microglia autophagy deficiency impaired stress-induced avoidance behavior, but not anxiety and depressive-like behaviors. These findings help to better understand microglial autophagic functions for stress and depression and might lead to the autophagybased development of antidepressants.

Data Availability
The data that support the findings of this study are available on request from the corresponding author.

Additional Points
Main Points. (i) Repeated social stress (RSD) induced initial autophagy signals and enhanced autophagic flux in the stress resilience. (ii) Patients with depression exhibited enhanced initial autophagy signaling. (iii) Autophagy deficiency in microglia aggravates RSD-induced avoidance.