CX3CR1 ^CreER mice were purchased from the Jackson Laboratory (Sacramento, CA, United States). Dlg1 ^flox/flox mice were gifted by Dr. Wanli Liu (Tsinghua University, China) as previously described (Peng et al., 2021). To induce Dlg1 microglia-specific knockout, tamoxifen (S1238, Selleck) was given to mice at the age of 6-week by gavage, and each mouse was administrated a total dose of 18 mg of tamoxifen for three consecutive days. CRS was performed 6 weeks after tamoxifen administration followed by behavioral and pathological analysis. All mice were housed in the Animal Care Facility at the Institute of Basic Medical Sciences with free access to standard rodent chow and clean water. All animal experiments were approved by the Institutional Animal Care and Use Committee of the Beijing Institute of Basic Medical Sciences.

Chronic restraint stress (CRS) was modified from the procedure previously described (Li et al., 2018). Briefly, the mice were placed in a 50 mL conical tube, with 0.5 cm air holes for breathing, for 3 h per day for 14 consecutive days.

Total RNA from brain tissue was extracted using TRIzol™ reagent (# 15596018, Invitrogen). cDNA was synthesized with a One-step First-strand cDNA synthesis kit (AE311-03, TransGen Biotech, Beijing, China). qPCR was performed using 2 × SYBR green master mix to measure the expression of genes. The primer sequences for qPCR are as follows:

GAPDH forward: 5′-AGG​TCG​GTG​TGA​ACG​GAT​TTG-3′;

GAPDH reverse: 5′-TGT​AGA​CCA​TGT​AGT​TGA​GGT​CA-3′;

TNFα forward: 5′-CCC​TCA​CAC​TCA​GAT​CAT​CTT​CT-3′;

TNFα reverse: 5′-GCT​ACG​ACG​TGG​GCT​ACA​G-3′;

Il-6 forward: 5′-CCA​AGA​GGT​GAG​TGC​CTT​CCC-3′;

Il-6 reverse: 5′-CTG​TTG​TTC​AGA​CTC​TCT​CCC​T-3′;

Il-1β forward: 5′-GCA​ACT​GTT​CCT​GAA​CTC​AAC​T-3′;

Il-1β reverse: 5′-ATC​TTT​TGG​GGT​CCG​TCA​ACT-3′.

The hippocampal tissue was lysed on ice using RIPA lysis buffer containing a cocktail of protease and phosphatase inhibitors. The lysates were centrifuged at ×15,000 g for 15 min at 4°C, then supernatants were collected and protein concentration was determined by BCA assays and was adjusted to the same final concentration. After heat denaturation, samples were separated by SDS-PAGE and transferred to Nitrocellulose membranes which were blocked in 5% non-fat milk and subsequently incubated overnight with the primary antibody. Relevant Horseradish Peroxidase (HRP)-conjugated secondary antibodies were incubated, and the immunoreactive proteins were then detected using an enhanced chemiluminescent (ECL) substrate. The protein signal was visualized using an X-Ray Film Processor (Optimax, New York, United States). The following antibodies were used: anti-iNOS/NOS Type II (# 610332, BD Biosciences, San Jose, CA, United States), anti- Iba1 (019-19741, Wako, Richmond, VA, United States), anti-Gapdh (#CW0266A, CWBiotech, Beijing, China).

Mice were anesthetized, perfused and sacrificed. Brains were then removed, fixed in 4% paraformaldehyde for 48 h, and dehydrated with gradient sucrose in phosphate buffered saline (PBS). The brains were embedded in optimal cutting temperature compound (OCT) and sectioned on a freezing microtome (CM3050S, Leica, Wetzlar, Germany). The coronal sections (40 μm in thickness) were incubated overnight with anti-goat Iba1 antibody (1:400, 019-19741, Wako, Richmond, VA, United States) and anti-mouse Gfap antibody (1:400; MAB360, Millipore, Darmstadt, Germany).

For immunohistochemical staining, the slices were incubated with biotinylated immunoglobulin G (IgG), followed by streptavidin-conjugated horseradish peroxidase (VECTASTAIN ABC Kit, Zhongshan Jinqiao, Beijing, China), and finally was visualized by reacting with hydrogen peroxide (DAB Kit, Zhongshan Jinqiao). Images were captured using a Leica SCN400 scanner.

For immunofluorescent staining, the sections were labeled with TRITC AffiniPure goat anti-rabbit IgG (111-025-003, Jackson ImmunoResearch) fluorescent secondary antibody, and the images were acquired by Nikon confocal microscope (Nikon, Melville, NY, United Stats). For activated microglia quantification, we determined the activated microglia by larger soma area, shorter and bolder processes.

For all behavioral tests, mice, in their home cages, were habituated for 2 h in the quiet and dimly lit test room before testing.

During the test, spontaneous activity of mice was monitored for a 5-min period as they moved in the apparatus (50 × 50 × 20 cm), time spent in the central area and total distances traveled were analyzed by ANY-maze software.

Mice were suspended by the tails on the instrument with a clip (50 cm above the floor) using a medical tape placed 1 cm from the tip of the tail. The test was recorded for 6 min and the immobility time was measured in the last 4 min of the test.

Mice were allowed to swim in water for 2 min the day before the test for habituation. During the test, mice were placed in a clear glass cylinder (diameter 25 cm), which was filled with water (21°C–23°C) up to a height of 25 cm from the bottom. The test was recorded for 6 min and the immobility time was measured in the last 4 min of the test.

The sucrose preference test is performed to evaluate anhedonia in mice. Briefly, mice were given two identical standard drinking bottles and habituated to two bottles filled with water (1 day) and with 1% sucrose water (1 day) for 2 days. After being deprived of food and water for 12 h, mice had free access to two drinking bottles, one containing 1% sucrose and the other with normal drinking water for 5 days. The positions of these two bottles were switched every 12 h to avoid side preferences. The overall sucrose and water intakes were measured, and the sucrose preference rate was calculated as 100% × (sucrose intake/sucrose intake + water intake).

Skeleton analysis of microglia was performed as described in our previous study (Cheng et al., 2021). Briefly, microglial images were obtained on Nikon A1 Confocal Microscope with ×20 objective at ×5 zoom. Skeletonize [two-dimensional (2D)/3D] plugin and Analyze Skeleton plugin of the software ImageJ were used for the skeleton analysis.

Data are expressed as mean ± SEM of at least two independent experiments. Statistical differences between the two groups were analyzed using Student’s t-test. Statistical differences for multiple comparisons were analyzed by application of ordinary two-way ANOVA (Tukey’s multiple comparison test). Statistical significance was determined as follows: *p < 0.05, **p < 0.01, and ***p < 0.001. GraphPad Prism software (version 9.0, GraphPad, San Diego, United States) was used for statistical analysis.

To evaluate the role of microglial Dlg1 on depression, we established chronic restraint stress (CRS) model, which has higher etiological similarities to depression. Specifically, Dlg1 cKO mice (Dlg1 ^f/f; CX3CR1creER) and their control littermates (Dlg1 ^f/f, term as Dlg1 f/f) were subjected to tamoxifen at 6 weeks of age to induce microglia-specific knockout of Dlg1. These mice were then randomly assigned to either the control or CRS group. 6 weeks after tamoxifen administration, mice in the CRS group underwent chronic restraint, whereas mice in the control group were handled in a separate home cage at an equivalent length of time. After 2 weeks, behavioral tests, namely sucrose preference test (SPT), open field test (OFT), tail suspension test (TST), and forced swimming test (FST) were carried out (Figure 1A). We found that CRS mice exhibited no overt anxiety-like behavior (Figure 1B), but they displayed more prominent depression-like behaviors than non-stressed controls (Figures 1C–F). Stressed Dlg1 f/f mice showed reduced locomotion in OFT (Figure 1C), less sucrose consumption in SPT (Figure 1D), and more immobility time in TST and FST (Figures 1E, F), indicating that the CRS model was established successfully. However, these depression-like behaviors induced by CRS were significantly alleviated in microglial Dlg1 deleted mice. These results suggest that ablation of Dlg1 in microglia ameliorates CRS-induced depression-like behavior.

After characterizing that specific knockout of Dlg1 in microglia alleviates depression-like behavior, we next sought to investigate the effects of deleting Dlg1 on microglia. To address this, we performed immunohistochemical staining of brain slices from Dlg1 cKO and their control littermates in both control and CRS group (Figure 2A). The number of microglia in different brain regions, including the cortex (CTX), dentate gyrus (DG), and CA1 of the hippocampus, was nearly unchanged in Dlg1 f/f mice subjected to CRS compared to Dlg1 f/f mice in the control group. Deletion of Dlg1 also showed no effect on microglia density in either the control or CRS groups (Figure 2B). Considering that astrocytes are known to communicate with microglia in a variety of disease states, in addition, we performed immunostaining for Gfap, a commonly used marker of astrocytes, and found that the number of astrocytes were not affected by CRS or deletion of Dlg1 from microglia (Figures 2C, D).

All four groups had compatible microglial densities, but some microglia in the CRS group exhibited alterations in morphology. In light of this, we performed immunofluorescence staining for Iba1, a microglia-specific marker, and found that the number of activated microglia, which possessed a larger volume of soma as well as shorter and bolder processes, were markedly increased in the Dlg1 f/f mice subjected to CRS (Figures 3A–C). It is interesting to note that this increase was significantly reversed by microglia-specific knockout of Dlg1 (Figure 3C). In light of the fact that microglia morphology and function are closely linked, we further analyzed the detail changes in microglia morphology (Figure 3D). We found that the soma area of microglia was notably increased while branch number and total branch length were markedly reduced in stressed Dlg1 f/f mice, As expected, these activated morphology changes were significantly absent in Dlg1 deleted microglia (Figures 3E–H). These results demonstrate that CRS provokes microglial activation, and knockout of Dlg1 prominently reverses this effect.

We next determined microglial status by examining the inflammatory cytokines. The protein level of inducible nitric oxide synthase (iNOS), termed as a marker of inflammation, was significantly upregulated in Dlg1 f/f mice of CRS group, and Dlg1 knockout reversed this upregulation (Figures 4A, B). In contrast, other cytokines like tumor necrosis factor α (TNFα), interleukin 6 (IL-6) and interleukin 1β (IL-1β) were found to have similar levels among four groups (Figures 3D–F). These results indicate that microglia display a mild activated state in CRS model, and deletion of Dlg1 in microglia recruits its normal state.