Minocycline Alleviates Abnormal Phagocytosis of Synapses by Microglia in a Mouse Model of Depression

(a mouse model of on microglia, and of in we used a study its effect on behaviors and phagocytosis of in mice. Our results show that the expression levels of PSD-95 in the hippocampus of CSDS-induced depression mice are signicantly reduced, while the microglia are signicantly activated. We co-labeled the synaptic protein PSD-95 with the microglia marker Iba-1, and found that the microglia in the hippocampus of stressed mice contained signicantly more PSD-95 engulfed puncta, which revealed that microglia in stressed mice abnormally phagocytized synapses. Moreover, our results indicated that minocycline treatment dampened microglial activation, reduced synaptic loss, alleviated behavioral impairment, and reduced abnormal phagocytosis of synapses by microglia in stressed mice.


Background
Depression is an affective disorder characterized by low mood and loss of interest. With the continuous development and progress of society, social stress is gradually increasing, and the prevalence of depression is increasing year by year, which has become a serious social public health problem [1].
According to a recent study published in Lancet Psychiatry, the lifetime prevalence of depression in China is 6.8%, of which major depressive disorder (MDD) is 3.4% [2]. For a long time, the monoamine transmitter hypothesis has played a dominant role in the study of the pathogenesis of depression. At present, the mechanism of commonly used antidepressants in the clinic is mainly based on the monoamine transmitter hypothesis. However, this approach has shown multiple problems, such as large individual differences in curative effect, low effective rate, high incidence of side effects, delayed drug effect, and low patient compliance rate, serious withdrawal symptoms [3,4]. In view of the many problems of clinically available antidepressants, it is assumed that in addition to changes in the monoamine transmitter system, there are other dysfunctions in depression. Indeed, the pathogenesis of depression is complex and diverse, and has not been precisely clari ed, thus it is urgent to unravel the pathogenesis of depression anew and nd potential therapeutic targets.
Neuronal synapses are mainly composed of presynaptic structures, synaptic gaps, and postsynaptic structures. It is the structural basis of information transmission between neurons and is a key part of neuroplasticity regulation. Functional brain imaging studies [5] reported that synaptic connections in the hippocampus, prefrontal cortex, and amygdala in patients with depression were signi cantly reduced, suggesting that neuronal synapses may have been lost in depressed patients. In addition, an autopsy report [6] indicated that synaptic structures and function-related genes in the hippocampus of depressed patients were signi cantly reduced, including synaptosomal-associated protein 25 (SNAP25) and glutamate receptors. Another autopsy report [7] stated that post-synaptic density-95 (PSD-95) and γaminobutyric acid receptors in the prefrontal cortex of patients with depression were also signi cantly reduced. Both autopsy reports suggest that there is indeed a loss of synapses in the brains of patients with depression. Researchers have also observed similar experimental results in animal models of depression, where the expressions of PSD-95, Synaptophysin, and growth-associated protein 43 in the hippocampus and prefrontal cortex of depressed rats induced by mother-infant separation were signi cantly reduced[8]. Our group [9] also found that the dendritic spine density in the hippocampus of depressed mice, induced by chronic unpredictable mild stress, (CUMS) was signi cantly reduced. In addition, some drugs, such as Liraglutide [10], Calpeptin [11], and Panaxynol [12] are able to upregulate the expression of synaptophysin, PSD-95, and synapsin I in the hippocampus, or enhance the hippocampal nerve metasynaptic plasticity and long-term potentiation (LTP) to alleviate depressive behaviors of model animals. These results suggest that reversing the loss of synapses can improve depressive behaviors of model animals. In view of this, it is necessary to explore the mechanism of synapse loss, which may provide a new strategy for the treatment of depression.
Shiosaka et al. proposed the concept of "synaptic microenvironment" [13]. Synaptic microenvironment refers to the synapses of neurons and their internal and external environments, including microglia, astrocytes, oligodendrocytes, vascular endothelial cells, extracellular matrix, and cytokines. The synaptic microenvironment has an important in uence on the structure and function of synapses. Therefore, paying attention to the changes of various cells or factors in the synaptic microenvironment can provide new ideas for exploring the mechanism of synaptic loss in depression. Microglia are innate immune cells in the central nervous system, accounting for about 10-12% of adult brain cells, and are one of the important components of the synaptic microenvironment. Under normal circumstances, the main role of microglia is to remove pathogens and phagocyte debris. In addition, microglia also monitor and trim the synapses through phagocytosis to promote the maturation and maintain the function of the synapses [14]. In 2011, Paolicelli RC et al. [15] described in their Science article the nding, that normal mouse microglia contained synaptic fragments. These fragments were synapse-related molecules, including the presynaptic protein SNAP25 and the postsynaptic protein PSD-95, which con rmed the phagocytosis of synapses by glial cells. It is thus rendered obvious, that the loss of neuronal synapses is closely related to the phagocytosis of microglia.
One positron emission tomography (PET) study [16] reported that translocator protein (TSPO), a marker of microglia was signi cantly upregulated in the hippocampus, frontal cortex, and temporal cortex of patients with depression, suggesting that microglial activation may be involved in the pathogenesis of depression. In addition, Du Preez A et al. [17] found that microglia in the hippocampal dentate gyrus of CUMS depression mice were signi cantly activated. Our group performed PET imaging and immunohistochemical studies [18], and found that TSPO in the hippocampus of depressed rats was signi cantly upregulated, and the density of microglia increased signi cantly. Thus, we asked if under pathological conditions, in addition to participating in neuroin ammation, activated microglia can abnormally engulf neuronal synapses.
We speculated that the loss of synapses in the hippocampus of patients with depression may also be caused by the abnormal phagocytosis of activated microglia. In order to verify this conjecture, we conducted two experiments: (1) As shown in Figure 1A, we investigated the effect of chronic social defeat stress (CSDS) on behaviors, synapses, microglia, and phagocytosis of synapses by microglia in mice; (2) As shown in Figure 4A, to con rm the role of microglia in depression, we investigated the effect of minocycline (a microglial activation inhibitor) on behaviors and phagocytosis of synapses in stressed mice. Our work thus aims to explicate the role of microglia phagocytosis of synapses in the pathogenesis of depression for the rst time, in order to provide new insights into the pathogenesis of depression and its treatment.

Animals
For all experiments, we used male C57BL/6J mice (7-8 weeks, from Vital River Laboratories, Beijing, China). The animals were provided with food and water (normal diet) available ad libitum and were allowed a one-week habituation period before experimental manipulation, while being housed at 23 ± 2°C on a 12 h light/dark cycle (lights on at 7:00 A.M.). CD-1 retired breeders (male, 8-9 months, from Vital River Laboratories, Beijing, China) were used as the aggressors. This study was carried out in accordance with the National Institute of Health Guide for the Care and Use of Laboratory Animals.

Chronic social defeat stress (CSDS)
The CSDS procedure was performed as previously described [19,20,21]. The C57BL/6J mice were individually introduced to the home cage of an unfamiliar aggressive CD-1 resident mouse for 5-10 min and exposed to physical defeat, after which it was housed together with the CD-1, but separated by a perforated plastic divider to allow for visual, olfactory and auditory contact for the remainder of the 24 h.
The mice were subject to social defeat for 10 consecutive days, each day being exposed to a different CD-1 mouse.

Social interaction test (SIT)
The SIT was performed on day 11, as previously described [19,20,21]. This test was performed in a clean open arena (42 cm × 42 cm). Each time, it consisted of two 2.5 min sessions. In the rst (also called No-Target) session, a C57BL/6J mouse was introduced to the arena with an empty mesh cage (10 cm × 6 cm) at one of its sides. In the second (Target) session, a mesh cage with an unfamiliar CD-1 mouse replaced the empty cage. The area of 26 cm × 14 cm, meaning 8 cm around the mesh cage, was de ned as the interaction zone. The time spent in the interaction zone was measured, and the social interaction rate was de ned as the time spent in the interaction zone with the aggressor present (Target) divided by the time spent in the interaction zone without the aggressor (No-Target).

Open eld test (OFT)
OFT was performed as previously described [21], and was performed in dim light (15 lux). The OFT apparatus consisted of a square open box (50 cm × 50 cm × 40 cm). Mice were individually placed in the center of the arena and allowed to explore for 5 min, while their activity was video-tracked. Total distance traveled, as well as the time spent and distance traveled in the center-zone of the box were recorded. The locomotor activity was tracked by a video recorder and analyzed using specialized software (Shanghai Mobile Datum Information Technology Company, Shanghai, China).

Elevated plus maze test (EPM)
Every EPM test was performed 3 hours after the OFT, and in a way previously described [21]. It took place in dim light (15 lux

Forced swimming test (FST)
The FST [21] was performed in a clear glass cylinder (height 32 cm, diameter 16 cm), lled with 15 cm of water (23-25°C). The test lasted for 5 min, during which the duration of immobility was recorded with a stopwatch by a researcher blind to the grouping of the experimental mice. Immobility is de ned as passive oating or making minimal movements necessary to keep the nostrils above water.

Intraperitoneal drug administration
Minocycline (Cat: M22254) was obtained from ABCONE (Shanghai, P.R. China). Along with the social defeat stress procedure, control mice and defeated mice were daily injected with minocycline (50 mg/kg) or its vehicle (0.9% NaCl) intraperitoneally.

Statistical analysis
The data are presented as the mean ± standard error (SEM). All statistical analyses were performed using GraphPad Prism 6.01 (GraphPad Software Inc., San Diego, CA). The statistical signi cance of differences between groups was analyzed using a T-test or two-way analysis of variance (ANOVA). The statistical signi cance level was set to p < 0.05. The graphs were plotted using GraphPad Prism 6.01.
2.11. Experimental design 2.11.1 Experiment 1: Effect of CSDS on behaviors, synapses, microglia, and phagocytosis of synapses by microglia in mice To investigate the effect of CSDS on behaviors, synapses, microglia, and phagocytosis of synapses by microglia in mice ( Figure 1A), the rst experiment was performed. Mice were allowed a one-week habituation period before any experimental manipulation. Then, the mice were divided into two groups: control group (No CSDS) and CSDS group. CSDS mice were exposed to ten days of CSDS. On day 11, the SIT, OFT and EPM were conducted. On day 12, the FST was performed. On day 13, the mice were sacri ced. Then, the synapses, microglia, and phagocytosis of synapses by microglia in hippocampi were examined.

Experiment 2: Effect of minocycline (a microglial activation inhibitor) on behaviors and phagocytosis of synapses by microglia in stressed mice
To illustrate the effect of minocycline on behaviors and phagocytosis of synapses by microglia, we administered the control and stressed mice with minocycline or its vehicle by intraperitoneal injection ( Figure 4A). Mice were allowed a one-week habituation period before any experimental manipulation.
Then, mice were divided into four groups: No CSDS + Saline group, No CSDS + minocycline group, CSDS + Saline group, and CSDS + minocycline group. CSDS mice were exposed to ten days of CSDS. During that time minocycline or its vehicle were administered. On day 11, the SIT, OFT and EPM were performed. On day 12, the FST was conducted. On day 13, the mice were sacri ced and the phagocytosis of synapses by microglia in their hippocampi was detected.

CSDS induced depressive-like behaviors in mice
The depression state of mice was successfully achieved through the CSDS model ( Figure 1A). As shown in Figure 1, compared with the control mice, the social interaction rate of CSDS mice in the SIT was signi cantly reduced ( Figure 1B), the immobility time in the FST was markedly increased (Figure 1C), and the time percentage spent in the open arms in the EPM was signi cantly reduced ( Figure 1D). In the OFT, although the total distance of the model mice was not different from the control mice, the time spent and distance traveled in the center-zone were both signi cantly reduced ( Figure 1E-G). These results demonstrate that CSDS induced signi cant depressive-like behaviors in mice.

CSDS induced microglial activation and synaptic loss in mice
As shown in Figure 2A-C, the relative intensity and the soma size of microglia in the hippocampus of the stressed mice were signi cantly increased, compared with control mice. Western blotting results also showed that the protein levels of Iba1 were signi cantly increased in the stressed mice ( Figure 2D, E). These results indicate that CSDS induced microglial activation in stressed mice. Besides, the expression levels of PSD-95 ( Figure 2F, G) and synaptophysin ( Figure 2H, I) in the hippocampus of stressed mice were signi cantly reduced. The above results indicate that CSDS caused both synaptic loss as well as microglial activation in model mice. Since microglia can phagocytize synapses, we put forward the following question: Is the synaptic loss caused by abnormal activation of microglia in stressed mice?

CSDS induced abnormal phagocytosis of synapses by microglia in mice
To clarify, whether the synaptic loss was caused by the abnormal activation of microglia in stressed mice, we co-labeled the synaptic PSD-95 or synaptophysin with the microglia marker Iba-1 respectively. Compared with the control mice, the microglia in the hippocampus of stressed mice contained signi cantly more PSD-95 engulfed puncta ( Figure 3A, B), which revealed that microglia in stressed mice abnormally phagocytized synapses. However, we couldn't nd any synaptophysin in microglia of control or stressed mice. The reason for that could be that synaptophysin is unstable and easily degraded in the process of phagocytosis by microglia. To further con rm the role of microglia in depression, we used a microglial activation inhibitor to study its effect on behaviors and phagocytosis of synapses in stressed mice.

Minocycline treatment dampened microglial activation in stressed mice
Minocycline, as a microglial activation inhibitor, was used to assess the role of microglial activation in CSDS-induced abnormal behaviors and pathology. To illustrate the role of microglial activation in stressed mice, we administered the control and stressed mice with minocycline or vehicle by intraperitoneal injection ( Figure 4A). As expected, minocycline treatment signi cantly dampened the increased intensity and soma size of microglia in the stressed mice ( Figure 4B-D). Western blotting results also showed that the protein levels of Iba1 were signi cantly reduced in stressed mice treated with minocycline ( Figure 4E, F). These results indicate that minocycline treatment dampened microglial activation in stressed mice.

Minocycline treatment alleviated synaptic loss and behavioral impairments in stressed mice
As shown in Figure 5, chronic treatment with minocycline signi cantly reversed the reduced levels of PSD-95 ( Figure 5A

Minocycline treatment reduced abnormal phagocytosis of synapses by microglia in stressed mice
The immunohistochemical co-labelling experiment revealed that minocycline treatment signi cantly reduced the number of PSD-95 engulfed puncta contained in microglia in the hippocampus of stressed mice ( Figure 7A, B), which hints that minocycline treatment reduced abnormal phagocytosis of synapses by microglia in stressed mice.

Discussion
As both functional brain imaging studies [5] and autopsy reports[6] [7] indicated that there is indeed a loss of synapses in the brains of patients with depression, it is necessary to explore the mechanism of synapse loss, which may provide a new strategy for the treatment of depression. In this study, the results showed that the expression levels of PSD-95 and synaptophysin in the hippocampus of CSDS-induced depression mice are signi cantly reduced, while the microglia are signi cantly activated. Microglia are the major immune cells in the central nervous system and they control the in ammation in brain through releasing in ammatory cytokines, clearing apoptotic cells, regulating synaptic plasticity and synapse pruning [22]. As microglia, the brain's specialized phagocytes, can eliminate synapses [23], we proposed that the synaptic loss may be caused by the abnormal activation of microglia in stressed mice. In order to verify this conjecture, we co-labeled the synaptic protein PSD-95 with the microglia marker Iba-1, and found that compared with the control mice, the microglia in the hippocampus of stressed mice contained signi cantly more PSD-95 engulfed puncta, which revealed that microglia in stressed mice abnormally phagocytized synapses. To con rm the role of microglia in depression, we used minocycline, a microglial activation inhibitor, to study its effect on behaviors and phagocytosis of synapses in stressed mice. The results indicated that minocycline treatment dampened microglial activation, reduced synaptic loss and alleviated behavioral impairments in stressed mice. In addition, minocycline treatment reduced abnormal phagocytosis of synapses by microglia in stressed mice. Thus, under depressive pathological conditions, the activated microglia may abnormally engulf neuronal synapses causing synaptic loss.
This study proposed for the rst time the role of microglia's phagocytosis of synapses in the pathogenesis of depression, which is innovative. However, the mechanism of microglial synaptic phagocytosis in depression is still not clear. There are several possible mechanisms: (1) One study indicated the absence of triggering receptor expression on myeloid cells 2 (TREM2) in microglia resulted in impaired synapse elimination in Trem2(-/-) mice, thus TREM2 might be essential for microgliamediated synaptic re nement [24]; (2) Another study revealed that transforming growth factor (TGF)-beta may regulate neuronal C1q expression to initiate C1q/C3/CR3 (CD11b) complement-and microgliamediated synaptic pruning. Besides, mice lacking TGF-beta receptor II (TGFbetaRII) in retinal neurons were found to have reduced C1q expression and reduced synaptic localization of complement [25]; (3) Yet another study showed that REV-ERBalpha deletion increased synapse loss and phagocytosis of synapses by microglia in the CA3 region of the hippocampus. It was found that REV-ERBalpha mediated complement expression and diurnal regulation of phagocytosis of synapses by microglia [26]. Thus, the mechanism of dysfunctional molecular interaction between microglia and synapses in depression needs to be studied further.

Conclusions
Taken together, we conclude, that under depressive pathological conditions, the activated microglia may abnormally engulf neuronal synapses causing synaptic loss. However, the mechanism of dysfunctional molecular interaction between microglia and synapses in depression has not been found yet. In spite of this limitation, our study adds to the accumulating evidence, that microglial activation is involved in the pathogenesis of depression. One of the major strengths of this study is that it provides insight into the role of abnormal phagocytosis of synapses by microglia in depression. Undeniably, our ndings are important for the discovery of novel drugs for the treatment of depression.