Six-week-old male C57BL/6J mice were obtained from GemPharmatech Co., Ltd. (Jiangsu, China) and housed in a specific pathogen-free facility (temperature: 22°C ± 2°C, humidity: 55 ± 5%) at the Animal Research Center of Nanjing University of Chinese Medicine. The study was reported in accordance with ARRIVE guidelines, and all methods were carried out following relevant guidelines and regulations. The animal husbandry and experimental procedures were approved by the Animal Ethical and Welfare Committee of Nanjing University of Chinese Medicine (Approval No. 202406A042). After acclimating to the environment for 2 weeks, mice were cultured in two batches. The first batch was divided into two groups and treated as follows: exposed to a normal environment (control group, n = 12), while the second group was exposed to a CMS environment (CMS group, n = 12). The second batch was divided into five groups and treated as follows: exposed to a normal environment (control group, n = 12), normal saline treatment (CMS group, n = 12), MgSO₄ gavage at 50 mg/kg (CMS + Mg 50 group, n = 12), MgSO₄ gavage at 100 mg/kg (CMS + Mg 100 group, n = 12), or pioglitazone gavage at 30 mg/kg (CMS + Pi group, n = 12) and exposed to a CMS environment (except for control group). CMS mice were exposed to 2–3 mild stressors daily (restraint, wet/no bedding, loud noise, cage tilting, stroboscopic light, reversed light/dark cycle, food restriction, or tail suspension) after a 2-week acclimation period, continuing for 8 weeks (38). The detailed CMS methodology is shown in Supplementary Table S1. Previous studies have shown that pioglitazone could alleviate depressive-like phenotypes in mouse models induced by a high-fat diet and CMS (39, 40). Therefore, pioglitazone was selected as the positive control group. The administration methods, dosages and duration of MgSO₄ and pioglitazone were based on previous studies (35, 41–44). After behavioral tests, a three-day rest period was provided, and then mice were sacrificed. We anesthetized the mice with 40 mg/kg sodium pentobarbital. After the completion of blood collection from the orbital venous plexus, serum and hippocampal samples were collected and immediately stored at −80°C. Whole brain tissues were preserved for immunofluorescence analysis.

Blood samples from the mice were centrifuged at 1,500 rpm for 15 min, and the serum was collected into clean Eppendorf tubes. Serum Mg²⁺ concentration was measured using Mg²⁺ Detection Kit (ab102506, Abcam, Cambridge, United Kingdom).

The levels of IL-1β, IL-6, and TNF-α in the serum and hippocampus were detected using enzyme-linked immunosorbent assay (ELISA) kits following the manufacturer’s instructions. The kits of IL-1β (MLB00C), IL-6 (M6000B), and TNF-α (MTA00B) were obtained from R&D Systems (Minneapolis, MN, United States).

The mice were sacrificed, and their brains were fixed in 4% paraformaldehyde overnight. The brains were then subjected to sectioning preprocessing using standard protocols at the Analysis Center of Servivebio (Wuhan, China). The brain slices were incubated with primary antibodies, followed by a 2-h incubation with Alexa Fluor-conjugated secondary antibodies at room temperature. Nuclear staining was performed by Hoechst incubation (C1018, Beyotime, Nanjing, China) for another 30 min. The primary antibodies included Iba1 antibody (019-19741, Wako, Japan) and NF-κB p65 antibody (6956s, CST, Framingham, MA, United States). Alexa Fluor-conjugated secondary antibodies included goat anti-rabbit IgG (H + L) cross-adsorbed secondary antibody, Alexa Fluor 488 (A-11008, Invitrogen, CA, United States) as well as Alexa Fluor 555 (A-21428, Invitrogen).

The mice received a tail vein injection of 200 μL dextran-fluorescein isothiocyanate (10 kDa, 20 mg/mL, FD10S, Sigma, St. Louis, United States). Ten minutes later, each mouse was anesthetized, and the brain was harvested. Next steps please refer to the methods of Immunofluorescence staining section for further details above.

Tissues or cells were rinsed with ice-cold PBS and lysed in RIPA buffer containing phenylmethylsulfonyl fluoride for 30 min. Then the lysates were centrifuged at 12,000 rpm at 4°C for 15 min. Bicinchoninic acid protein assay Kit (23005, Thermo Fisher Scientific, Wilmington, United States) was used to detect protein concentration. Proteins were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto polyvinylidene fluoride membranes (PVDF, IPVH00010, Millipore, Billerica, United States). The membranes were blocked with 5% milk for 1 h, followed by overnight incubation with primary antibodies at 4°C, and then incubated with peroxidase-conjugated secondary antibodies for 1 h at room temperature. Primary antibodies included rabbit anti-IL-1β (ab254360, Abcam), rabbit anti-IL-6 (ab233706, Abcam, Cambridge, United Kingdom), rabbit anti-TNF-α (ab215188, Abcam), rabbit anti-p-NF-κB p65 (3033, CST), mouse anti-NF-κB p65 (6956s, CST), rabbit anti-p-IKKα/β (2697S, CST), rabbit anti-IKKα/β (R24674, Zenbio, Chengdu, China), mouse anti-p-IκBα (9246S, CST), mouse anti-IκBα (66418-1-1g, Proteintech, Wuhan, China), mouse anti-NLRP3 (68102-1-Ig, Proteintech), rabbit anti-ASC (13833S, CST), rabbit anti-cleaved caspase-1 (89332T, CST), rabbit anti-pro-caspase-1 (24232T, CST), mouse anti-β-actin (4970, CST), mouse anti-lamin B1 (66095-1-Ig, Proteintech) and rabbit anti-GAPDH (10494-1-AP, Proteintech). Secondary antibodies included goat anti-mouse IgG (H + L) HRP (YFSA01, Yfxbio, Nanjing, China), goat anti-rabbit IgG (H + L) HRP (YFSA02, Yfxbio). Relative protein levels were visualized using the Tanon 5200 chemiluminescence imaging system reagent and quantified with ImageJ software (Version 1.50b, National Institutes of Health, Bethesda, United States).

Before tissue collection, the brains of mice were perfused with ice-cold PBS to avoid sampling the circulating blood immune cells. The hippocampus homogenate was filtered through a 70-μm cell strainer and centrifuged again at 1,000 rpm for 3 min at 4°C. The cells were resuspended in cold PBS and centrifuged at 1,000 rpm for 3 min at 4°C. All samples were counted and diluted to a density of 1–2 × 10⁵/mL, and then labeled using a Live/Dead kit for 30 min, centrifuged at 1,000 rpm for 3 min at 4°C. The cells were resuspended in 100 μL PBS buffer and blocked with anti-CD16/32 (553142, BD Biosciences, New Jersey, United States) for 10 min. The cells were incubated with the antibodies for flow cytometry according to the manufacturers’ protocols for 30 min at 4°C. The following antibodies were used in the FACS analysis: CD45-BV510 (561487, BD Biosciences, Franklin Lakes, United States), CD11b-Alexa 488 (557672, BD Biosciences). Cells were detected by a BD Aria III cytometer, and the data were analyzed by FlowJo software.

Total mRNA was isolated from tissues or cells using TRIzol reagent (#15596026, Invitrogen, Carlsbad, CA) according to the manufacturer’s instructions. Total RNA (1 μg) was reverse-transcribed using the HiScript II select qRT supermix (R222-01, Vazyme, Jiangsu, China). Quantitative real-time PCR was performed using gene specific primer sets and SYBR Green (Vazyme) on a quantitative real-time PCR (qRT-PCR) detection system (Bio-Rad, California, United States). All primers were designed by ourselves and synthesized in the Generay Biotech Co., Ltd. (Shanghai, China).

Data were shown as mean ± SEM. All the data were examined for normality with the Skew test. Two-tailed Student’s t-test was used for comparisons between two groups, and data were analyzed statistically using one-way ANOVA for comparisons of more than two groups with single factor variance. Statistical analyses were performed with GraphPad Prism Software (GraphPad Software, San Diego, United States). Value of p < 0.05 was considered statistically significant (^*p < 0.05, ^**p < 0.01, and ^***p < 0.001).

Mice were suffered with CMS for 8 weeks. At the end of stress, depressive-like behaviors were assessed with OFT, SPT, FST and TST (Figure 1A). The body weight of CMS mice exhibited a significant decrease compared to the control group (Figure 1B). CMS mice demonstrated a decrease in central area activity time, the ratio of the distance of central movement to the total distance in the OFT (Figures 1C,D; Supplementary Figure S1) and a reduced sucrose preference in the SPT (Figure 1E) compared with control group. The CMS procedure led to an increased floating time of mice in the FST (Figure 1F), as well as the immobile time in the TST (Figure 1G). Moreover, it resulted in a decrease in the swimming time and struggling time in the FST and TST, respectively (Figures 1F,H). Taken together, the mice developed depressive-like behaviors after experiencing a protocol of CMS. Notably, we found that CMS mice had a lower Mg²⁺ concentration in the serum when compared to the control group (Figure 1I) and the Mg²⁺ concentration was negatively correlated with the floating time of FST in mice (Figure 1J). These results demonstrated CMS mice displayed depressive-like behaviors accompanied by hypomagnesemia.

The mice subjected to CMS procedure exhibited up-regulation of IL-6, IL-1β, and TNF-α levels in serum compared to the control group (Figures 2A–C). Next, we explored whether CMS developed neuroinflammatory response by counting the number of microglia and observing the morphology of microglia (positive for Iba1). Microglia, a type of glial cells equivalent to macrophages in the brain and spinal cord, are considered as the first and most crucial line of immune defense in the central nervous system. Immunofluorescence results revealed a significant increase in the number of microglia in CMS mice, accompanied by enlarged cell bodies and reduced branches compared with control group (Figures 2D,E). The occurrence of neuroinflammatory response not only displayed activation of microglia, but is often accompanied by damage to the blood-brain barrier (BBB). Haruwaka et al. (48) demonstrated that during sustained inflammation, microglia phagocytosed astrocytic end-feet, impairing the function of BBB. Here, we also found that CMS-induced neuroinflammation and compromised the permeability of the BBB by injecting 10 kDa dextran through the tail vein (Figure 2D). These results suggested that CMS promoted the secretion of inflammatory factor and microglia activation, which may be harm to the BBB.

We have found that depressive-like mice are accompanied by a decrease of Mg²⁺ concentrations in serum. Next, we further explore whether Mg²⁺ supplementation improves the neuroinflammatory response in mice. We observed that oral treatment with MgSO₄ (50 or 100 mg/kg) significantly reduced the cytokine levels of IL-1β, IL-6, and TNF-α in the serum (Figures 3A–C) as well as in the hippocampus (Figures 3D–F) compared with CMS mice. Moreover, MgSO₄ treatment reduced the proportion of Iba1⁺ microglia in the hippocampus of CMS mice (Figures 3G,H). Similarly, pioglitazone significantly decreased the levels of IL-1β, IL-6, and TNF-α, and the number of Iba1⁺ microglia in the hippocampus of CMS mice (Figure 3). These results suggest that treatment with MgSO₄ and pioglitazone effectively inhibits hippocampal neuroinflammation in CMS mice.

Microglial activation is characterized not only by an increase in cell number, but also by morphological changes, often involving the M1-to-M2 shift phenotype (15, 49). Compared to the control group, CMS significantly upregulates the mRNA levels of M1 markers in the hippocampus of mice, including CD86, TNF-α, IL-1β, IL-6, COX-2, and iNOS, while treatment with MgSO₄ and pioglitazone reverses the mRNA expression of M1 markers (Figures 4A–F). In addition, CMS treatment significantly decreases the mRNA levels of M2 markers, including CD206, IL-4, G-CSF, GM-CSF, TGF-β1, and IGF-1. MgSO₄ and pioglitazone treatment increased the expression of M2 markers (Figures 4G–L). In vitro, MgSO₄ also significantly reversed the M1-to-M2 transition induced by lipopolysaccharide (LPS) in BV2 microglia. Compared to the control group, LPS significantly upregulated the mRNA levels of M1 markers and downregulated the M2 markers in BV2 cells, while treatment with MgSO₄ and pioglitazone reverses these changes (Supplementary Figures S2A–J). Therefore, we determined that MgSO₄ has a specific effect on microglia. Moreover, immunofluorescence results of microglia showed that after CMS induction, the size of cell bodies increased and the ramification of distal branches decreased, and MgSO₄ and pioglitazone reversed these morphological transformations of microglia (Figures 4M–O). Notably, MgSO₄ and pioglitazone also alleviated CMS-induced dextran leakage of BBB (Figure 4M). Therefore, MgSO₄ treatment could inhibit microglia activation and BBB damage induced by CMS.

NF-κB signaling and NLRP3 inflammasome are important transduction signals of inflammatory reaction in mammals (50, 51). The CMS-induced mice mentioned above exhibited higher levels of IL-1β, IL-6, and TNF-α in both serum and hippocampus compared to the control mice (Figures 3A–E). Here, we observed increased protein levels of p-NF-κB p65, p-IKKα/β, p-IκBα, nuclear NF-κB, as well as NLRP3, ASC, and pro-caspase-1 in the hippocampus of CMS mice. Similarly, nuclear proteins were extracted, revealing that CMS facilitated the translocation of NF-κB into the nucleus. This observation indicates that CMS promotes the activation of NF-κB signaling and the NLRP3 inflammasome in the hippocampus (Figures 5A–F). MgSO₄ and pioglitazone reduced p-NF-κB p65, p-IκBα and p-IKKα/β expression, and inhibited NF-κB entry into the nucleus (Figures 5A–D). Similarly, in vitro, NF-κB (green fluorescence) entered the nucleus in LPS-induced BV2 cells compared to the control group, while MgSO₄ significantly reversed the phenomenon of NF-κB entering the nucleus (Supplementary Figure S2K). Moreover, MgSO₄ and pioglitazone inhibited the activation of the NLRP3 inflammasome (Figures 5E,F) in the hippocampus of CMS mice, which was consistent with the alleviation of hippocampal neuroinflammation and BBB damage mentioned above. These results suggested that MgSO₄ and pioglitazone may inhibit the activation of microglia by inhibiting NF-κB signaling and NLRP3 inflammasome to alleviate hippocampal neuroinflammation and repair BBB damage.

To further investigate the effects of Mg²⁺ supplementation on CMS-induced depressive-like behaviors in mice, we performed OFT, SPT, FST and TST experiments. MgSO₄ were unable to restore the body weight of CMS mice (Figure 6A), but increased the central area activity time, distance and the ratio of the distance of central movement to the total distance in the OFT (Figures 6B,C; Supplementary Figure S3) and sucrose preference in the SPT (Figure 6D) in CMS mice. Moreover, Both MgSO₄ and pioglitazone reduced the floating time of CMS mice in the FST (Figure 6E), as well as the immobile time in the TST (Figure 6F), while increasing swimming time and struggling time in the FST and TST, respectively (Figures 6E,G). Notably, MgSO₄ significantly alleviated serum Mg²⁺ levels in CMS-induced mice (Figure 6H).

Therefore, the depressive-like behaviors of mice experienced CMS could be alleviated through treatment with MgSO₄ and pioglitazone. The mechanism by which MgSO₄ improves depressive-like behavior in CMS mice is illustrated in Figure 7.