LPS (Escherichia coli O55:B5; purity ≥97%, purified by phenol extraction; specific activity ≥500,000 EU/mg) was obtained from Beijing Solarbio Science & Technology Co., Ltd. (cat. no. L8880), was diluted in sterile phosphate-buffered saline (PBS) to prepare the stock solution of the 10 mg/ml concentration and stored at −20°C.

All animal experiments were performed in strict accordance with the laboratory animal testing guidelines of Zhejiang University and the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health. The primary experimental protocols were reviewed and approved by the Laboratory Animal Welfare and Ethics Review Committee, Zhejiang University (Hangzhou, China; approval no. ZJU20220307). The supplementary in vivo receptor-antagonism control procedures (LPS+SB290157 cohort) were independently authorized and approved by Animal Experimental Ethical Inspection of the Sir Run Run Shaw Hospital, Zhejiang University School of Medicine (Hangzhou, China; approval no. SRRSH2026-0041). The 6-week-old mice (weight, 20–22 g) were provided by Zhejiang Vital River Laboratory Animal Technology Co., Ltd. Alongside a 12/12-h light/dark cycle, the trial maintained a consistent temperature of 22°C. A standard animal care facility supplied food and water for the mice throughout the week. The fewest number of animals possible were employed to minimize their suffering.

A total of 144 male C57BL/6 mice were successfully utilized, evaluated, and euthanized in the present study, with zero premature mortality recorded. Pre-defined humane endpoints were established based on rigorous animal welfare criteria, including a body weight loss exceeding 20%, severe prolonged lethargy, or visible respiratory distress; notably, no animals reached these thresholds during the experimental timelines. The dosage and pre-treatment timeline of SD were selected based on previous in vivo studies (18) and optimized through preliminary pilot experiments. The in vivo phase comprised two distinct procedural blocks: i) The primary therapeutic evaluation cohort, wherein mice received daily intragastric gavage of SD (20 mg/kg) or vehicle for 8 consecutive days starting 3 days prior to a 5-day daily lipopolysaccharide (LPS; 1 mg/kg, i.p.) challenge (19,20), followed by sequential behavioral testing spanning 11 days; and ii) the C3aR inhibition paradigm, where animals were pre-treated with the selective C3aR antagonist SB290157 trifluoroacetate (30 mg/kg, i.p.) every alternate day beginning 2 days prior to any other interventions, followed by an 8-day concurrent SD/LPS regimen and a subsequent 9-day behavioral assessment matrix. At the scheduled experimental endpoints, deep anesthesia was initiated via the inhalation of 4% isoflurane inside a dedicated induction chamber for rapid induction and subsequently maintained at 2% isoflurane during the procedural verification. Anesthetic depth was strictly validated by the complete loss of the pedal withdrawal (toe-pinch) reflex prior to the procedure. Humane euthanasia was sequentially executed via cervical dislocation under this deep anesthetic state. Absolute death was clinically verified by the permanent cessation of cardiac impulses and respiratory movements prior to anatomical tissue harvesting.

To optimize the preventative timing of SD prior to acute LPS challenge, a preliminary pilot study was conducted utilizing smaller animal cohorts (n=3-4 mice per group) using 19 mice. The protective phenotypes were systematically evaluated when mice were administered with SD (20 mg/kg, i.g.) daily for either 1, 2, or 3 consecutive days before the acute systemic challenge, which was subsequently maintained throughout the 5-day repetitive LPS regimen. Pilot behavioral observations indicated that a 3-day pre-treatment regimen offered the most stable, robust, and statistically significant positive effects in safeguarding against LPS-driven behavioral deficits compared with the 1- or 2-day groups. The results of preliminary pilot behavior tests were presented in Fig. S1.

The OFT was conducted following established protocols (21). Prior to the studies, mice were acclimatized to the testing environment and lighting conditions for 30 min. Each mouse was individually placed in a square arena (dimensions: 45×45×45 cm³) and allowed to explore freely for 15 min. After each test, the open field arena was sanitized with 75% ethanol. The locomotor and exploratory behaviors of the mice were recorded using ANY-maze software during the test.

The EPM test was conducted following previously established techniques (21). The maze consisted of two open arms and two closed arms, elevated above the floor. Mice were placed individually at the center of the maze, facing one of the open arms, and their physical activity was recorded for a period of 5 min. Before testing the next mouse, the maze was cleaned with 75% ethanol. The movements of the mice were tracked and measured using ANY-maze software.

The buried marble test relies on documenting the number of marbles buried at the end of the experiment. The test was originally designed to assess anxiety-like behavior. It is conducted in a polypropylene mouse cage measuring 42×24×12 cm, which has a metal grid on the top. The floor of the cage is covered with 5 cm of sawdust, and 20 clean glass marbles (1.5 cm in diameter) are evenly distributed across the surface. During the test, the marbles are arranged with uniform spacing. Mice are placed in this enclosure for 30 min. ~24 h after initial training on the testing day, 15 marbles are distributed evenly within the test cage. The mice are exposed to the marbles for 3 min while being recorded on video. The number of buried marbles is subsequently calculated.

The TST was conducted as outlined in previous research (22). ~1/3 of the mouse's tail was secured and suspended from a base, with the head positioned 20 cm above the ground. Animal activity was recorded from a lateral perspective. The mice were suspended for a total of 6 min, during which an observer, unaware of the treatment the animals received, timed their immobility during the last 4-min assessment. Animals were deemed motionless when they exhibited no body movements and remained passively suspended.

The mice were placed in a glass chamber (35 cm in height, 30 cm in diameter) filled with warm water. The water temperature was maintained between 22 and 24°C, with a depth of 20 centimeters. The water depth was calibrated to prevent the animals from contacting the substrate using their tails or hind limbs. Animal behaviors were recorded via videotape from a lateral perspective. The rodents were allowed to swim freely for 6 min, and an observer blinded to the treatment conditions recorded the duration of the immobility during the final 4-min assessment. Immobility was defined as the time when the animals remained stationary or afloat, displaying only the movements necessary to maintain their balance in the water.

The mice were housed individually and acclimated with two bottles of water for two days, followed by two bottles of 2% sucrose for another two days. After a period of 24 h without water, the mice were provided with one bottle of 2% sucrose and one bottle of water for 2 h during the dark phase. During the 2-h test, the positions of the bottles were swapped after 1 h. The total consumption of each fluid was recorded, and sucrose preference was determined by calculating the average sucrose consumption ratio over the first 2 h. The sucrose intake ratio was computed by dividing total sucrose consumption by the combined intake of water and sucrose.

Total RNA was extracted from the mPFC using the FastPure Cell/Tissue Total RNA Extraction Kit V2 (Vazyme Biotech Co., Ltd.). The quality and quantity of the RNA were assessed with a NanoDrop spectrophotometer (Thermo Fisher Scientific, Inc.). Subsequently, cDNA was synthesized from the RNA using the First Strand cDNA Synthesis Kit (Monad Biotech Co., Ltd.) according to the manufacturer's instructions. qPCR was then performed using SYBR Green I (Foregene Co., Ltd) on an A&B Applied Biosystems device. The thermocycling conditions were as follows: Initial denaturation at 95°C for 30 sec, followed by 40 cycles of denaturation at 95°C for 10 sec and annealing/extension at 60°C for 30 sec. The expression levels of the target genes were normalized to GAPDH using the 2^−ΔΔCq approach (23). The nucleotide sequences of the primers used in RT-qPCR were: TNF-α forward, 5′-CCCTCACACTCAGATCATCTTCT-3′ and reverse, 5′-GCTACGACGTGGGCTACAG-3′; IL-6 forward, 5′-TAGTCCTTCCTACCCCAATTTCC-3′ and reverse, 5′-TTGGTCCTTAGCCACTCCTTC-3′; IL-1β forward, 5′-GCAACTGTTCCTGAACTCAACT-3′ and reverse, 5′-ATCTTTTGGGGTCCGTCAACT-3′; LCN2 forward, 5′-TGGCCCTGAGTGTCATGTG-3′ and reverse, 5′-CTCTTGTAGCTCATAGATGGTGC-3′; C3 forward, 5′-CCAGCTCCCCATTAGCTCTG-3′ and reverse, 5′-GCACTTGCCTCTTTAGGAAGTC-3′; S100a8 forward, 5′-AAATCACCATGCCCTCTACAAG-3′ and reverse, 5′-CCCACTTTTATCACCATCGCAA-3′; matrix metalloproteinase-9 (MMP9) forward, 5′-CTGGACAGCCAGACACTAAAG-3′ and reverse, 5′-CTCGCGGCAAGTCTTCAGAG-3′; Itgal forward, 5′-CCAGACTTTTGCTACTGGGAC-3′ and reverse, 5′-GCTTGTTCGGCAGTGATAGAG-3′; Clec14a forward, 5′-CTTCACCACGCTACCTTCAAG-3′ and reverse, 5′-AAACCCCTTAAAGGCTCTTTCTC-3′; Myo1g forward, 5′-GGCCCTGAGTATGGGAAACC-3′ and reverse, 5′-GATACGAGCACCTCACCAATG-3′; Cybb forward, 5′-TGTGGTTGGGGCTGAATGTC-3′ and reverse, 5′-CTGAGAAAGGAGAGCAGATTTCG-3′; Frg forward, 5′-CGGCTGAAGAACGCTATTACC-3′ and reverse, 5′-GGGCGACGAATATGGTCACTC-3′; Adh1 forward, 5′-GCAAAGCTGCGGTGCTATG-3′ and reverse, 5′-TCACACAAGTCACCCCTTCTC-3′; GAPDH forward, 5′-AGGTCGGTGTGAACGGATTTG-3′ and reverse, 5′-GGGGTCGTTGATGGCAACA-3′.

Mice were subjected to deep anesthesia and subsequently processed according to a previously established methodology (24) to obtain mPFC tissue. After homogenizing the tissues in ice-cold RIPA Lysis and Extraction buffer (cat. no. 89900; Thermo Fisher Scientific, Inc.) solution containing PMSF, they were centrifuged for 15 min at 4°C at 11,200 × g. The BCA assay (cat. no. P0011; Beyotime Institute of Biotechnology) was utilized to measure concentrations following the manufacturer's instructions. Equal amounts of protein (20 µg per lane) were separated on 4–20% gradient SDS-polyacrylamide gels. SDS-PAGE was performed to separate the proteins, which were then transferred to PVDF membranes. The primary antibodies used included LCN2 (1:1,000; cat. no. A2092), Itgal (1:1,000; cat. no. A23960), Complement C3 (1:1,000; cat. no. A26625PM), MMP9 (1:1,000; cat. no. A0289) and β-actin (1:3,000; cat. no. AC026; all from ABclonal Biotech Co., Ltd.). These antibodies were incubated on membranes with 5% non-fat milk in PBST (0.1% Tween-20) for 1 h at room temperature, followed by overnight incubation at 4°C. The membranes were then allowed to equilibrate at room temperature for 1 h while being treated with the appropriate secondary antibodies (Goat anti-Rabbit IgG (H+L) Secondary Antibody, HRP: 1:5,000; catalog no. 31460; Thermo Fisher Scientific, Inc.; and Goat anti-Mouse IgG (H+L) Secondary Antibody, HRP: 1:5,000; catalog no. 31430; Thermo Fisher Scientific, Inc.). Super ECL Detection Reagent (catalog no. 36208ES76; Shanghai Yeasen Biotechnology Co., Ltd.) was employed for visualization, and images were captured using the ChemiDoc Touch Imaging System (Bio-Rad Laboratories, Inc.). Band intensities were analyzed using ImageJ software (version 1.54p; National Institutes of Health).

In accordance with the methodology detailed in previous studies, serial cryostat sections with a thickness of 25 µm were prepared (15,16). The sections underwent treatment with 5% donkey serum (catalog no. 36136ES60; Shanghai Yeasen Biotechnology Co., Ltd.) and 0.3% Triton X-100 for 1 h at room temperature. This was followed by overnight incubation at 4°C with an antibody diluent containing goat antibodies against Iba-1 (1:200; cat. no. ab178846; Abcam). Subsequently, the sections were washed with PBS at three 10-min intervals. After washing, they were incubated at room temperature for 1 h with Alexa Fluor 594 goat anti-rabbit antibody (1:1,000; catalog no. ab150080; Abcam). Fluorescent images were captured using a Nikon A1 confocal microscope. Image acquisition was conducted using Image-Pro Plus 5.0 software (Media Cybernetics, Inc.), ensuring that all parameters remained consistent throughout the process.

The approach (25) was followed to conduct a morphometric evaluation of the microglia. High-resolution images with distinct cellular architecture were obtained using confocal fluorescence microscopy (Nikon Corporation). To analyze the target region layer by layer, an ×40 objective was employed. Additionally, multi-plane virtual-Z mode was used to reconstruct the cellular architecture in three dimensions after the analysis. At least five Iba1⁺ cells were present in each location, and 10 images of individual microglia were captured from each mouse studied. Following the examination of the images, skeletonization was performed using ImageJ's skeleton analysis. The terminal positions of the microglia were counted, and the maximal branch length of the microglia was determined. Sholl analysis was conducted on binary images by utilizing the Sholl analysis tools available in ImageJ. An assessment of the cellular complexity was carried out by counting the number of intersections where concentric circles overlapped.

The BV-2 cell line, obtained from OriCell, was cultured in DMEM basic (1X) media (cat. no. C11995500BT; Gibco; Thermo Fisher Scientific, Inc.) supplemented with 10% fetal bovine serum (FBS; catalog no. A5209402; Gibco; Thermo Fisher Scientific, Inc.) and 1% penicillin/streptomycin (cat. no. 240004004; Beijing Solarbio Science & Technology Co., Ltd.) in a humidity-controlled incubator at 37°C with 5% CO₂.

The BV-2 cell line, maintained under optimal conditions, was inoculated into multi-well plates. The experiment began once the cells reached ~70–80% confluence in medium devoid of FBS. The cells underwent pre-treatment with SD (50 µM) for 2 h. Following pre-treatment, LPS (100 ng/ml) was immediately added to the culture medium for stimulation. Regarding SB290157 trifluoroacetate treatment, either dimethyl sulfoxide (DMSO) or 10 µM SB290157 was introduced to the BV-2 cell 1 h prior to the administration of SD (26). Cells were harvested 6 h after stimulation for RT-qPCR assays. Samples were collected 24 h after treatment for western blot analysis and metabolic extracellular flux assays.

For the concentrate of C3a by ELISA assay BV-2 microglial cells were cultured until achieving ~70–80% confluence, after which the culture medium was replenished with serum-free medium. To evaluate the regulatory effect of SD on complement C3 cleavage, cells were pre-treated with SD (50 µM) for 2 h. A total of 1 h prior to LPS stimulation, recombinant full-length mouse C3 protein (2 µg/ml; cat. no. HY-P78247; MedChemExpress) was introduced into the medium as an exogenous substrate. Subsequently, cells were administrated with LPS (100 ng/ml) to initiate the complement activation cascade. Following a 24-h incubation period, the cell culture supernatants were meticulously collected and centrifuged at 1,000 × g for 10 min at 4°C to remove cellular debris. The concentration of active C3a fragment in the clarified supernatants was quantitatively determined using a commercial Mouse C3a Enzyme-Linked Immunosorbent Assay (ELISA) kit according to the manufacturer's instructions. Concurrently, the adherent cells were washed with ice-cold PBS and lysed using RIPA buffer. The total protein concentration of each well was measured via a BCA protein assay kit.

BV-2 microglial cells were cultured until achieving ~70–80% confluence, after which the culture medium was replenished with serum-free medium. Cells were pre-treated with SD (50 µM) for 2 h. A total of 1 h prior to LPS stimulation, recombinant mouse C3a protein (1 µg/ml; cat. no. HY-P7863; MedChemExpress) was introduced into the medium. Subsequently, cells were administrated with LPS (100 ng/ml) to initiate the inflammation activation cascade. Following a 24-h incubation period, the whole cell harvested and lysed using RIPA buffer. The total protein concentration of each well was measured via a BCA protein assay kit.

The ideal cell density was selected, and the experiments were conducted (27). Quadruplicates or quintuplicates of BV-2 cell lines were cultured on XF-96 plates with LPS and SD. Microglia were washed and assessed in XF Running buffer A (10 mM glucose, 1 mM sodium pyruvate, and 2 mM L-glutamine) for oxygen consumption rate (OCR) or extracellular acidification rate (ECAR) at designated intervals. Measurements were received in real-time before and after pharmacological interventions, including the MitoStress assay with 1 µM oligomycin, 1 µM FCCP and 1 µM rotenone, as well as the glycolysis stress experiment with 10 mM glucose, 1 µM oligomycin, and 50 mM 2-DG. When glucose was supplemented, ECAR measures reflected the basal ECAR, while OCR measures indicated the basal OCR. The difference between basal OCR and the peak OCR following FCCP treatment represented mitochondrial respiration capacity and ATP production. Glycolytic reserve (GR) was calculated as the difference between basal and peak ECAR after oligomycin administration.

The RNA-Seq investigation involved collecting mPFC tissues harvested from the CTRL (n=4), LPS (n=5), SDN (n=5) and LPS+SDN (n=5) groups. The RNA sequencing process encompasses several key steps: Extracting RNA, constructing a library, sequencing, checking quality, mapping reads, analyzing differential expression, and performing functional enrichment. Following the manufacturer's instructions, TRIzol^® Reagent was utilized to extract total RNA from the tissue samples. Subsequently, the 5300 Bioanalyzer (Agilent) was employed to assess the purity of the RNA, while the ND-2000 (NanoDrop Technologies; Thermo Fisher Scientific, Inc.) was used for quantification. For sequencing library creation, only RNA samples meeting strict quality criteria (OD260/280=1.8- 2.2, OD260/230 ≥2.0, RIN ≥6.5, 28S:18S ≥1.0,>1 µg) were selected. RNA purification, reverse transcription, library building, and sequencing were performed by Shanghai Majorbio Bio-pharm Biotechnology Co., Ltd. in accordance with the manufacturer's guidelines (Illumina, Inc.). The Illumina^® Stranded mRNA Prep, Ligation from (Illumina, Inc.) was used to generate the mPFC RNA-seq transcriptome library. A total of 1 µg of total RNA for the experiment. The polyA selection method was employed with oligo(dT) beads to isolate messenger RNA, which was subsequently fragmented using a fragmentation buffer. Following this, the SuperScript double-stranded cDNA synthesis kit (Invitrogen; Thermo Fisher Scientific, Inc.) was used to generate double-stranded cDNA with random hexamer primers (Illumina, Inc.). In accordance with Illumina's library construction protocol, the synthesized cDNA was subjected to end repair, phosphorylation, and the addition of an ‘A’ base. Libraries were then size-selected for cDNA target fragments of 300 bp using 2% low-range ultra-agarose. Next, the fragments were amplified by PCR with Phusion DNA polymerase (New England BioLabs, Inc.) for 15 cycles. After quantification with the Qubit 4.0, the paired-end RNA-seq sequencing library was sequenced on the NovaSeq 6000 sequencer, which produces reads of 2×150 bp. The raw paired-end readings were reduced and fastp was used with the default settings to assess their quality. Subsequently, HISAT2 software (28) was employed to align the cleaned reads to the reference genome in oriented mode. StringTie (29) was utilized to generate mapped reads for each sample in a reference-based manner. To identify differentially expressed genes (DEGs) across two distinct samples, we quantified the expression level of each transcript using the transcripts per million reads methodology. RSEM was employed to measure gene abundances (30). For the differential expression analysis, either DESeq2 (31) or DEGseq (32) were used. DEGs were identified with |log₂FC|>1 and a false discovery rate (FDR) ≤0.05 using DESeq2, or with an FDR <0.001 using DEGseq. Additionally, functional enrichment analysis was performed, including Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways, to identify DEGs that were significantly enriched in GO terms and metabolic pathways, with a Bonferroni-corrected P≤0.05 compared with the whole transcriptome background. GOATOOLS (https://github.com/tanghaibao/goatools) was used for GO functional enrichment analysis, and KOBAS (http://kobas.cbi.pku.edu.cn) was used for KEGG pathway analysis.

The concentration of active C3a fragment in the clarified supernatants was quantitatively determined using a commercial Mouse C3a ELISA kit (cat. no. NBP2-70037; Novus Biologicals, LLC) according to the manufacturer's instructions. To ensure experimental rigor and eliminate variations in cell density, the absolute concentrations of C3a obtained from the ELISA were normalized against the total protein content of the respective cell lysates and expressed as pg/mg total protein.

All data were collected and analyzed blindly. Data are presented as the mean ± SEM. Single comparisons were determined using a two-tailed Student's t-test. Two-way ANOVA with Tukey's multiple comparisons test was applied for multiple comparisons. Sample sizes were based on our previous study and relevant literature to ensure reliable measurements. Statistical analyses were conducted using GraphPad Prism 8.0 (GraphPad Software; Dotmatics). P<0.05 was considered to indicate a statistically significant difference. All statistical analyses are detailed in the figure legends and Table SI.

Adult wide-type C57BL/6 mice received with LPS (1 mg/kg) daily for 5 consecutive days. Mice were pretreated with SD (20 mg/kg, i.g.) for 3 days before LPS administration and continued to receive SD for a total of 8 consecutive days (Fig. 1A). LPS treatment reduced body weight compared with the sham group, whereas SD significantly increased body weight in LPS-treated mice (Fig. 1B). Neither LPS nor SD affected locomotor activity, as indicated by the absence of significant differences in the total distance traveled during the OFT (Fig. 1C). Similarly, no significant difference was observed in the center time of the OFT test or in the time spent in the open arms of the EPM test among the experimental groups (Fig. 1C-E). The number of buried marbles also remained unchanged across all four groups (Fig. 1F). These findings indicate that neither LPS nor SD induced anxiety-like behavior. In the SPT, LPS-treated mice exhibited a significant reduction in sucrose preference, indicating the development of anhedonia-like behavior. SD treatment markedly increased the sucrose preference in the LPS-treated mice (Fig. 1G). In addition, LPS administration significantly increased immobility time in both the TST and FST compared with the control groups, indicating the induction of depressive-like behaviors. SD treatment effectively reversed these behavioral deficits (Fig. 1H and I). Collectively, these results demonstrate that SD alleviates LPS-induced depressive-like behaviors.

To evaluate the effects of SD on LPS-induced neuroinflammation, the mRNA expression levels of the pro-inflammatory cytokines, tumor necrosis factor α (TNF-α), interleukin-6 (IL-6) and interleukin-1β (IL-1β) were measured in the mPFC. RT-qPCR analysis showed that the mRNA levels of TNF-α, IL-6 and L-1β were significantly increased in the LPS-treated mice compared with sham mice controls (Fig. 2A). SD treatment led to a substantial reduction of these inflammatory cytokines in LPS-treated mice (Fig. 2A). To further examine microglial activation, morphometric analyses of microglia were performed in the mPFC. The representative images of immunofluorescence of Iba1 were shown in Fig. 2B. LPS-treated mice exhibited increased soma size and greater maximum branch length, morphological features indicative of microglial activation. SD treatment significantly attenuated these changes (Fig. 2C and D). Together, these results demonstrate that SD suppresses LPS-induced neuroinflammatory responses and microglial activation in the mPFC.

The prefrontal cortex, particularly the mPFC, plays a critical role in social cognition and socio-emotional processing (3). To examine the transcriptional changes associated with the effects of SD on LPS-induced depressive-like behaviors, high-throughput RNA sequencing of the mPFC was performed as an unbiased transcriptomic screening approach to identify DEGs. The DEG analysis identified 468 genes that were significantly altered between the LPS + SDN mice and LPS groups using a cutoff value of a fold change >2.0 and P<0.05. Among these genes, 22 were upregulated and 438 were downregulated. Compared with the CTRL group, LPS treatment increased the expression of 686 genes and decreased the expression of 29 genes in the mPFC. By contrast, comparison of the SDN and CTRL groups identified 107 altered genes, including 21 upregulated and 86 downregulated genes (Fig. 3A). The heatmap analysis revealed marked alterations in the mPFC transcriptomic profile following LPS treatment, whereas SD largely reversed the LPS-induced upregulation of gene expression (Fig. 3B). Consistently, the Venn diagram showed an overlap between genes upregulated in the LPS versus CTRL comparison and genes downregulated in the LPS + SDN versus LPS comparison (Fig. 3C). KEGG pathway enrichment analysis of the 120 overlapping genes provided preliminary insight into the molecular pathways underlying the effects of SD on LPS-induced depressive-like behaviors. Enriched pathways included ‘Neutrophil extracellular trap formation’, ‘Cell adhesion molecules’, ‘Neuractive ligand-receptor interaction’ and ‘Coronavirus disease’ (Fig. 3D). GO analysis further indicated enrichment in biological processes and cellular components related to ‘Leukocyte migration’, ‘Positive regulation of response to external stimulus’, ‘Leukocyte cell’, ‘Collagen’ and ‘Regulation of cell’ (Fig. 3E). These findings suggest that immune- and inflammation-related pathways contribute to the protective effects of SD against LPS-induced depressive-like behaviors.

To further validate the transcriptomic findings, the 10 most significantly altered DEGs among the 120 overlapping genes identified in the mPFC were examined using RT-qPCR. LPS treatment significantly increased the mRNA level of LCN2, C3, S100a8, MMP9, Itgal, Myo1g, Cybb and Fgr. SD treatment significantly reduced the mRNA levels of LCN2, C3, Itgal and MMP9 in LPS-treated mice (Fig. 4A). The protein expression of LCN2, C3, Itgal and MMP9 was next assessed by western blot analysis. Among these candidates, C3 and MMP9 protein levels were significantly elevated following LPS administration. SD treatment effectively suppressed the LPS-induced upregulation of both C3 and MMP9 (Fig. 4B and C). These unbiased transcriptomic analyses identified C3 and MMP9 as key candidate targets associated with the effects of SD.

The consistent changes observed at both the mRNA and protein levels further support the involvement of C3 and MMP9 in the protective actions of SD against LPS-induced pathological alterations.

To further validate the RNA-sequencing results, an in vitro BV-2 microglial cell model was established and the effects of LPS and SD treatment were examined. RT-qPCR analysis indicated that SD significantly attenuated the LPS-induced upregulation of pro-inflammatory cytokines in the BV-2 cells (Fig. 5A). Consistent with the transcriptomic findings, SD also reversed the LPS-induced increases in C3 and MMP9 expression (Fig. 5B and C). Glycolysis provides a rapid source of energy to support microglial proliferation, migration, chemotaxis, and other cellular functions during inflammatory responses (33). To determine whether SD affects microglial metabolic reprogramming, the ECAR and OCR in BV-2 cells were measured following LPS and SD treatment. LPS stimulation significantly increased basal glycolysis and GR, as reflected by elevated ECAR values (Fig. 5D and E). By contrast, LPS significantly reduced basal respiration, maximal respiration and ATP production, as indicated by OCR analysis, compared with untreated control cells (Fig. 5F and G). These findings demonstrate that LPS promotes a metabolic shift toward aerobic glycolysis in microglia, which is closely associated with inflammatory activation. Notably, SD significantly suppressed the LPS-induced increases in basal glycolysis and GR. Moreover, SD restored basal and maximal OCR and improved ATP production in LPS-treated BV-2 cells (Fig. 5D-G). Together, these results indicate that SD attenuates LPS-induced inflammatory activation and metabolic reprogramming in microglia.

To determine whether MMP9 expression is regulated by upstream complement activation, an in vitro rescue experiment was performed in BV-2 cells using exogenous recombinant mouse C3a protein. Western blot analysis showed that SD significantly suppressed the LPS-induced increase in MMP9 expression. However, treatment with exogenous C3a effectively reversed this inhibitory effect and restored MMP9 protein expression despite the presence of SD (Fig. 5H and I). These findings provide direct evidence that C3a signaling acts upstream of MMP9 and regulates its expression.

Collectively, the results suggest that the protective effects of SD on neuroinflammation and microglial metabolic reprogramming are closely associated with modulation of the C3a/C3aR signaling pathway and its downstream target, MMP9.

To determine whether SD regulates C3 expression or its downstream activation through proteolytic cleavage, additional in vitro and in vivo ELISA assays were performed to quantify C3a levels.

For the in vitro assay, BV-2 cells were pretreated with SD (50 µM) for 2 h. Recombinant full-length mouse C3 protein (2 µg/ml) was added to the culture medium 1 h before LPS stimulation to provide an exogenous C3 substrate. Cells were then exposed to LPS (100 ng/ml) to activate the complement cascade. After 24 h, C3a levels in the culture supernatants were measured by ELISA. LPS stimulation significantly increased C3a production, indicating enhanced cleavage of C3 into its active fragment. SD significantly reduced C3a generation in response to LPS. Importantly, this inhibitory effect persisted even in the presence of excess exogenous full-length C3 protein, demonstrating that SD continued to suppress C3a production despite substrate replenishment (Fig. 6A). These findings suggest that SD interferes with the activation process of the complement cascade by inhibiting C3 cleavage rather than simply reducing endogenous C3 availability.

To validate these findings in vivo, C3a levels in mPFC tissues from the Ctrl, LPS, SD, and LPS + SD groups were quantified. ELISA analysis showed that 5 consecutive days of LPS administration significantly increased C3a levels in the mPFC, whereas SD treatment significantly attenuated this elevation (Fig. 6B). When considered together with the total C3 and MMP9 expression data presented in Fig. 4, these ELISA results indicate that SD exerts its protective effects by suppressing both C3 expression and the subsequent activation of the C3a signaling pathway.

Complement activation is centered on complement component C3 (C3), which undergoes proteolytic processing to generate several biologically active fragments, including C3a. As an active complement peptide, C3a exerts antimicrobial effects and regulates inflammatory responses .SB290157 trifluoroacetate is a selective antagonist of the C3a receptor (C3aR). Supplementary data showed that SB290157 alone did not significantly affect sucrose preference in LPS-treated mice. Similarly, no significant differences in immobility time during the TST or FST were observed between the LPS and LPS + SB290157 groups (Fig. S2A-C). According to the experimental design shown in Fig. 7A, mice received SB290157 trifluoroacetate (30 mg/kg, i.p.) for two days before SD and LPS administration. SD partially reversed the body weight loss induced by LPS. However, this protective effect was abolished following SB290157 treatment (Fig. 7B). Likewise, no significant difference in sucrose preference was observed between the LPS and LPS + SD + SB290157 groups (Fig. 7C). In addition, SD failed to reduce the LPS-induced increase in immobility time in both the FST and TST when SB290157 administration occurred (Fig. 7D and E). Western blot assays showed that SD mitigated the LPS-induced upregulation of C3 and MMP9. However, this effect was no longer observed in mice treated with SB290157 (Fig. 7F and G).

Together, these findings indicate that the antidepressant-like and anti-inflammatory effects of SD depend on an intact C3a/C3aR signaling pathway, which acts upstream of MMP9 regulation.

The effects of SD on inflammation and aerobic glycolysis were further examined following C3aR inhibition using the BV-2 cell model. SD significantly reduced the LPS-induced increase in pro-inflammatory cytokine expression. However, after treatment with SB290157 trifluoroacetate, SD no longer restored cytokine levels in LPS-stimulated BV-2 cells (Fig. 8A). Similarly, the inhibitory effects of SD on the LPS-induced upregulation of C3 and MMP9 expression were abolished following SB290157 administration (Fig. 8B and C). Metabolic analysis further showed that SD failed to suppress the LPS-induced increases in basal glycolysis and GR, as measured by ECAR, in the presence of SB290157 (Fig. 8D and E). Likewise, in OCR assays, SD lost its ability to restore basal respiration and ATP consumption after C3aR inhibition (Fig. 8F and G). Notably, although SD partially preserved its effect on maximal oxygen consumption, it largely failed to reverse LPS-induced inflammatory responses and metabolic reprogramming following SB290157 treatment.

These findings further demonstrate that the anti-inflammatory and anti-glycolytic effects of SD under LPS stimulation are predominantly mediated through the C3a/C3aR signaling pathway.