Forty male Sprague-Dawley rats (6–8 weeks old, SPF grade) were purchased from Hunan Slake Jinda Experimental Animals Co., Ltd. (SCXK (Xiang) 2021-0002). Rats were housed in the Animal Experiment Center of Hunan University of Chinese Medicine under controlled conditions (25 °C ± 2 °C, 12 h light/dark cycle), with free access to standard chow and water. All experimental procedures were approved by the Ethics Committee of Hunan University of Chinese Medicine. (ethical approval code: HNUCM21-2312-24).

Medicinal materials for Danggui Shaoyao San (DSS) including A. sinensis (Oliv.) Diels (Danggui) (#CK25111301), Paeonia lactiflora Pall (Shaoyao) (#HH25111401), Atractylodes macrocephala Koidz (Baizhu) (#CK25110302), Poria cocos (Schw.) Wolf (Fuling) (#HH25111101), Alisma orientalis (Sam.) Juzep (Zexie) (#HH25110505), and Ligusticum chuanxiong Hort (Chuanxiong) (#HH25102701) were obtained from the Traditional Chinese Medicine Pharmacy of the First Affiliated Hospital of Hunan University of Chinese Medicine. Other reagents included sodium pentobarbital (Merck KGaA, Germany, #P3761), donepezil hydrochloride (MCE, United States, #HY-B0034), D-galactose (Sigma-Aldrich, Shanghai, China, #V900922), primary antibodies targeting STAT3 and Foxp3 (Proteintech, Wuhan, China, #102532-AP, #22228-1-AP), ROR-γt and p-STAT3-Y705 antibodies (HUABIO, Hangzhou, China, #HA722121, #ET1603-40), JAK2 antibody (Aifang, Hunan, China, #AFRM99330), p-JAK2-Y1007/1008 antibody (ABclonal, Wuhan, China, #AP0531), β-actin antibody (Affinity Biosciences, China, #AF7018), postsynaptic density protein 95 (PSD95) and synaptophysin I antibody (abcam, Cambridge, United Kingdom, #ab18258,#ab254349), secondary antibodies (Elabscience, Wuhan, China, #E-AB-1003), Iba1 antibody (Wako Pure Chemical Industries, Ltd., Japan, #011-27991), primers for β-actin, IL-1β, IL-6, etc., were designed using Primer 6.0 software based on sequences from the NCBI database and synthesized by Sangon Biotech (Shanghai), China, reverse transcription kit (Novoprotein, Suzhou, China, #E047-01B), SYBR Mixture premix and RIPA protein lysis buffer (Comwin Biotech, Jiangsu, China, #CW3008M, #CW2333S), ELISA kits (Jianglai Biotechnology, Shanghai, China, #JL13427, #JL20880), BCA protein concentration assay kit (Elabscience Biotechnology, Wuhan, China, #E-BC-K318-M), TRIzol, SlowFade™ Gold Antifade Mountant with DAPI, and donkey anti-goat Alexa Fluor 647 fluorescent secondary antibody (red) (Thermo Fisher Scientific, United States, #15596-026, #S36942, #A32849), universal two-step detection kit (Zhongshan Golden Bridge, Beijing, China, #PV-9000), Toluidine blue, Triton X-100, and antifade mounting medium with DAPI (Solarbio Science & Technology Co., Ltd., Beijing, China, #G3663, #T8200, #S2110), hematoxylin staining solution and eosin staining solution (Zhongshan Golden Bridge Biotechnology Co., Ltd., Beijing, China, #ZLI-9610, #ZLI-9613), Fixation/permeabilization kits, dyes, and antibodies against CD4, Foxp3, and IL-17A for phenotypic analysis of Th17 and Treg cells were purchased from BD Pharmingen (United States), detailed information is provided in Supplementary Table S1.

The composition of DSS was based on the original ratios from the Synopsis of Prescriptions of the Golden Chamber by Zhang Zhongjing (Danggui 9 g, Baishao 48 g, Baizhu 12 g, Fuling 12 g, Zexie 24 g, and Chuanxiong 24 g), referring to the previous study (Song et al., 2021). The herbal mixture was extracted by decoction with water and concentrated using a rotary evaporator (Hei-VAP ML, Heidolph, China) to yield a final concentration of 5.16 g/mL. Quality control information for the DSS was reported by our group using liquid chromatography-tandem mass spectrometry (LC-MS/MS) (Song et al., 2020).

Forty male Sprague-Dawley rats were randomly assigned into five groups (n = 8 per group): control, D-gal model (100 mg/kg), low-dose DSS (L-DSS, 12 g/kg/day, 0.9 mL/day concentrated extract), high-dose DSS (H-DSS, 24 g/kg/day, 1.8 mL/day), and donepezil (0.5 mg/kg/day) (Ayoub et al., 2022) groups. The dosage regimen was determined based on clinical equivalents. The low dose was calculated using the formula: Rat dose = 6.25 × (Clinical dose/Average human body weight of 65 kg). The high-dose group received twice the volume/concentration of the low-dose group (YMe, 2023). Except for the control group, all rats received intraperitoneal injections of D-galactose for 8 weeks to induce an AD-like model (Rehman et al., 2017), which is widely used to induce aging-related oxidative stress and neuroinflammation accompanied by learning and memory deficits. Drug administration started at week 5 and continued for 4 weeks. After behavioral analysis, the rats were sacrificed for subsequent experiments. Control rats received equal volumes of saline via gavage (Figure 1A). During the treatment period, animals were closely monitored for general health status, including body weight, food intake, locomotor activity, and coat condition.

The open field apparatus measured 100 cm × 100 cm × 40 cm, with black-painted inner walls and the floor divided into 16 squares. The central area consisted of 4 squares surrounded by 12 peripheral squares. Each rat was placed gently in the center and allowed to explore freely for 5 min. Between trials, the apparatus was cleaned with alcohol to remove odors. Parameters recorded included total distance traveled, number of entries into the central zone, time spent in central and peripheral zones, and movement trajectory. Data were collected and analyzed using Tracking Master V3.0 (XR-XZR209, Xinruan, Shanghai, China) (Kraeuter et al., 2019).

The water maze pool was divided into four quadrants with distinct visual cues. The hidden platform was submerged 1 cm below the water surface in the target (first) quadrant. Training was conducted over 5 days, with four trials per day from different entry points and a maximum swim time of 60 s. Escape latency (time to reach platform) was recorded. Rats failing to find the platform within 60 s were guided to it for 10 s to facilitate learning. On day 6, the platform was removed for the probe test. Rats were released from the quadrant opposite the original platform location. The number of platform crossings, time in the target quadrant, and swim paths and swimming speed were recorded and analyzed using SuperMaze software (XR-XM101, Xinruan, Shanghai, China) (Vorhees and Williams, 2006).

Following anesthesia with pentobarbital sodium via intraperitoneal injection (40 mg/kg (Reid et al., 2003; Soma, 1983), dissolved in 0.9% NaCl), rats were fixed supine, and a U-shaped abdominal incision was made. Approximately 3 mL blood was collected from the abdominal aorta. The left ventricle was perfused with precooled physiological saline (4 °C), and the right atrium was cut open. Successful perfusion was confirmed by the whitening of visceral organs. All rats were humanely euthanized by rapid decapitation in accordance with approved institutional ethical guidelines. Brains were rapidly extracted; the left hemisphere was fixed in 4% paraformaldehyde for histology, while the hippocampus from the right hemisphere was dissected, snap-frozen in liquid nitrogen, and stored at −80 °C for molecular assays. The spleen was excised, cleared of connective tissue and fat, and placed in cold tissue preservation solution (4 °C) for flow cytometry.

Active components of DSS were screened from the TCMSP (Ru et al., 2014) and PubChem (Kim et al., 2021) databases using the botanical names as keywords, with selection criteria of oral bioavailability ≥30% and drug-likeness ≥0.1. Potential targets of these components were predicted using SwissTargetPrediction. Disease-related targets associated with neuroinflammation were obtained from GeneCards (Stelzer et al., 2011) and OMIM (Amberger et al., 2019). Intersection targets were identified using Venny online tool and imported into STRING (Szklarczyk et al., 2019) for protein-protein interaction network construction. Core targets were screened by topological analysis in Cytoscape with plugins Network Analyzer and cytoNCA. Gene Ontology (GO) and KEGG pathway enrichment analyses were performed via DAVID (Liu and Thomas, 2019; Kanehisa et al., 2017). An active component–core target–pathway network was constructed in Cytoscape 3.7.2 (Kohl et al., 2011). Software details and URLs are listed in Supplementary Table S2.

The top 10 key active components of DSS were selected for docking with four target proteins (JAK2, STAT3, IL-10, IL-17) based on KEGG pathway analysis. Crystal structures were retrieved from the RCSB PDB database and preprocessed via PyMOL by removing ligands and water molecules, followed by structural optimization. Ligand structures were optimized using Chem3D. Docking was performed using CB-Dock2 online platform to obtain binding affinity scores (Vina Scores) and optimal conformations. Binding affinity scores less than −5.0 kcal/mol indicated strong binding. Scores were visualized and statistically analyzed on MicroBioinformatics platform. PyMOL 3.0 was used to visualize interactions with high binding affinity (Burley et al., 2021; Khanal et al., 2022).

Untargeted serum metabolomics was conducted following established protocols (He et al., 2023). Serum samples were thawed at 4 °C, and 100 µL aliquots were mixed with 400 µL pre-cooled methanol-acetonitrile (1:1, v/v) for protein precipitation. After vortexing, samples were incubated at −20 °C for 1 h, centrifuged at 14,000 × g for 20 min at 4 °C, and supernatants collected for LC-MS analysis. Chromatographic separation used a Waters ACQUITY UPLC HSS T3 column (2.1 × 100 mm, 1.8 µm) at 40 °C with mobile phases A (water +0.1% formic acid) and B (acetonitrile +0.1% formic acid), flow rate 0.3 mL/min, injection volume 2 µL. Mass spectrometry was performed on Thermo Q Exactive HF-X with full scans in positive and negative ion modes (m/z 70–1050). Data processing included peak extraction, alignment, and normalization via Progenesis QI. PCA and OPLS-DA were employed for group discrimination. Differential metabolites met criteria of VIP >1.0, p < 0.05 (t-test), and fold change >1.5 or <0.67. Metabolite annotation and pathway enrichment were conducted using HMDB and KEGG databases.

After fixation and paraffin embedding, for each hemisphere, a series of 30 consecutive coronal sections (5 µm thick) encompassing the entire dorsal hippocampus were collected for analysis. HE staining: sections were baked, dewaxed, stained with hematoxylin (5 min), differentiated in hydrochloric acid-ethanol, blued in distilled water, dehydrated through graded alcohols, cleared in xylene, and mounted with neutral balsam. Nissl staining: sections were dewaxed, stained with 1% toluidine blue at 37 °C for 25 min, differentiated with 95% ethanol (30 s), dehydrated, and mounted. Images were captured using a Motic BA410E microscope (Motic, China) to evaluate hippocampal histology and neuronal Nissl body distribution.

Total RNA was extracted from right hippocampal tissue using TRIzol reagent. RNA concentration and purity were measured before reverse transcription to cDNA using a commercial kit. qPCR was performed on a Bio-Rad CFX96 system with SYBR Green dye under the following conditions: initial denaturation at 95 °C for 5 min, followed by 40 cycles of 95 °C for 30 s and 58 °C for 30 s. Reactions were run in triplicate. β-actin served as the internal reference gene, and relative expression levels were calculated by the 2^−ΔΔCt method. Primer sequences are listed in Table 1.

Hippocampal tissues from rats were collected, and total protein was extracted using RIPA lysis buffer. Protein concentration was quantified by BCA assay. Equal amounts of protein (30 μg) were separated by SDS-PAGE and transferred onto membranes. Membranes were blocked with 5% skim milk (or 5% BSA for phosphorylated antibodies) at room temperature for 1 h, followed by overnight incubation at 4 °C with primary antibodies: PSD95 (1:2000), synaptophysin I (1:1500), RORγt (1:1000), Foxp3 (1:1000), p-STAT3 (1:5000), STAT3 (1:10,000), p-JAK2 (1:500), JAK2 (1:5000), and β-actin (1:5000). After washing, membranes were incubated with HRP-conjugated secondary antibodies (1:10,000) for 1 h at room temperature. Chemiluminescent signals were detected using a Bio-Rad ChemiDoc XRS + imaging system, and band intensities were quantified with ImageJ software.

Paraffin-embedded dorsal hippocampal sections were dewaxed, hydrated, and subjected to antigen retrieval. Sections were permeabilized with 0.3% Triton X-100 and blocked with 5% BSA for 30 min at room temperature. Primary antibodies against Iba-1, IL-17A, FOXP3, and p-JAK2 (each 1:50) were applied overnight at 4 °C. After washing with PBS, appropriate fluorescent secondary antibodies (1:500) were incubated for 1 h at room temperature in the dark. Nuclei were counterstained with DAPI, and sections were mounted with antifade medium. Images were acquired using the TissueFAXS Plus system, and positive cells were quantified.

Brain sections (5 μm) were prepared using a paraffin microtome. After dewaxing, hydration, and antigen retrieval, endogenous peroxidase activity was quenched with 3% H₂O₂. Sections were blocked with 5% BSA and incubated overnight with anti-p-STAT3 antibody (1:500) at 4 °C. After washing, secondary antibody (1:500) was applied for 30 min at room temperature. DAB was used for color development, followed by hematoxylin counterstaining, dehydration, clearing, and mounting. Images were captured under a light microscope, and positive staining was quantified.

Collected arterial blood was allowed to clot at room temperature for 2 h, then centrifuged at 3,000 rpm for 10 min. Serum levels of IL-17A and IL-10 were measured according to the manufacturer’s instructions using ELISA kits. Absorbance was read on a microplate reader, and cytokine concentrations were calculated based on standard curves.

Spleens were harvested aseptically, minced, and filtered to obtain single-cell suspensions. After red blood cell lysis, cells were washed and resuspended in PBS. Surface staining with anti-CD4 antibody was performed at 4 °C for 30 min in the dark. Following permeabilization, intracellular staining for Foxp3 and IL-17 was conducted for 1 h at 4 °C. Cells were washed, resuspended in staining buffer, and analyzed on a BD LSRFortessa flow cytometer to quantify CD4⁺Foxp3⁺ Treg and CD4⁺IL-17⁺ Th17 cell populations.

Statistical analyses were conducted using SPSS 27.0 and GraphPad Prism 9.5.0. Data are presented as mean ± standard deviation (SD). For in vivo experiments, the number of biological replicates (n) refers to the number of individual animals used in each group. Behavioral assessments were conducted using independent cohorts of animals (n = 5 per group), while biochemical, histological, flow cytometry, and molecular analyses were performed using tissue samples randomly collected from the five animals (n = 3 per group). For normally distributed data with homogeneity of variance, one-way ANOVA was used; otherwise, the Kruskal–Wallis H test was applied. MWM acquisition data were analyzed using two-way repeated-measures ANOVA, with group as the between-subject factor and training day as the within-subject factor. Statistical significance was set at p < 0.05.

To investigate the effects of DSS on neuroinflammation and cognitive function in AD rat model, we established D-galactose-induced AD rats, with the experimental timeline depicted in Figure 1A (By Fig draw). During the treatment period, animals were closely monitored for general health status, including body weight, food intake, locomotor activity, and coat condition. No signs of overt toxicity or abnormal behavior were observed in DSS-treated groups. Body weight changes over time have now been included in the revised manuscript (Figure 1B). Our team has performed comprehensive preclinical toxicological assessments, which indicate that DSS is a safe pharmaceutical formulation (Luo et al., 2024; Wang et al., 2023). MWM testing revealed that, compared with controls, model rats exhibited disorganized swimming trajectories during the navigation phase, reflecting impaired spatial exploration (Figure 1C). No significant differences in swimming speed were observed among groups on day 6, indicating that motor ability and general motivation during the task were comparable (Figure 1D). The model group demonstrated prolonged escape latency throughout training, indicative of deficits in spatial learning and memory (Figure 1E). Conversely, rats treated with low- and high-dose DSS (DSS-L, DSS-H) as well as donepezil showed markedly reduced escape latencies. In the spatial probe trial, the model group manifested a significant decline in both platform crossings (p < 0.05) and time spent in the target quadrant (p < 0.01), effects that were significantly reversed by DSS-H treatment (p < 0.01; Figures 1F,G). Open Field Test (OFT) results demonstrated that the model group exhibited reduced total movement distance, fewer entries into the central area, and decreased time spent therein (p < 0.01), indicative of diminished spontaneous locomotion. The treatment groups (DSS-H, donepezil) significantly ameliorated these behavioral deficits compared to the model group (p < 0.01; Figures 1H–K), suggesting that DSS improves behavioral performance in AD rats.

To systematically assess the neuropathological impact of DSS in D-galactose-induced AD rats, histological analyses of the hippocampal CA2 and CA3 regions were performed. Hematoxylin and eosin (H&E) staining (Figure 2A) showed well-organized neurons with clear nuclear morphology in controls, whereas the model group exhibited disordered neuronal arrangement with signs of injury such as pyknosis and nuclear condensation. Treatment with DSS (both doses) and donepezil restored neuronal order and reduced nuclear condensation and neuronal loss. Nissl staining (Figure 2B) revealed reduced neuronal count, irregular arrangement, and diminished Nissl body content in the model group; DSS intervention counteracted these changes, increasing neuronal density and Nissl body restoration (p < 0.01, Figure 2C), indicating effective mitigation of D-galactose-induced structural neuronal damage. To explore DSS’s regulatory effect on neuroinflammation, we evaluated microglial activation and inflammatory cytokine expression. Immunofluorescence staining (Figure 2D) showed a significant increase of Iba1⁺ activated microglia in the model group (p < 0.01), which was significantly suppressed by DSS and donepezil treatments (p < 0.01, Figure 2E). Western blot analysis was performed to detect the expression of AD-related pathological markers synaptic proteins PSD95 and Synapsin I in the hippocampus of each group. Compared with the control group, the model group exhibited significantly decreased expression levels of both PSD95 (p < 0.01) and Synapsin I (p < 0.01). Conversely, treatment with a high dose of DSS restored the expression of these proteins compared to the model group (p < 0.05) (Figures 2F–H). qPCR analysis revealed elevated mRNA levels of pro-inflammatory cytokines IL-6, IL-1β, and IL-17A and decreased expression of anti-inflammatory cytokines TGF-β, IL-4, and IL-10 in the hippocampus of model rats. DSS treatment significantly reversed these expression patterns (p < 0.01, Figures 2I–N), with the high-dose DSS group demonstrating more pronounced effects.

To elucidate the potential mechanisms by which DSS counteracts neuroinflammation, network pharmacology analysis was conducted. Thirty-two active compounds were identified from DSS (Supplementary Table S3), corresponding to 296 potential drug targets, while 1206 neuroinflammation-related targets were retrieved from disease databases. Venn diagram analysis (Venny 2.1) revealed 132 overlapping targets (Figure 3A; Supplementary Table S4), highlighting potential DSS intervention sites. An “Active Component-Target-Disease” network was constructed (Figure 3B; Supplementary Tables S5, S6), demonstrating the characteristic multi-component, multi-target synergy of DSS. Protein-protein interaction (PPI) network analysis identified core hub targets including AKT1, TNF-α, STAT3, and JAK2 (Figure 3C; Supplementary Tables S7, S8), suggesting pivotal roles in DSS-mediated neuroinflammation regulation. GO enrichment analysis showed significant involvement in biological processes (BP) such as inflammatory response, JNK cascade activation, and T cell costimulation; cellular components (CC) related primarily to dendrites and plasma membranes; and molecular functions (MF) centered on protein kinase activity (Figure 3D; Supplementary Table S9). KEGG pathway analysis indicated significant enrichment in immune-related pathways, including JAK-STAT, IL-17, T cell receptor signaling, and differentiation of Th1/Th2 and Th17 cells (Figure 3E; Supplementary Table S10). Molecular docking revealed that ten key DSS compounds exhibited strong binding affinities (binding energy < −5.0 kcal/mol) toward JAK2, STAT3, IL17, and IL10 (Figure 3F; Supplementary Figure S1), indicating that multiple active constituents of DSS may target these kinases. Collectively, these data suggest that DSS ameliorates AD-associated neuroinflammation primarily via modulation of T cell immunity and related signaling pathways.

To further explore the metabolic mechanisms underlying DSS-mediated cognitive improvement, untargeted metabolomic profiling of rat serum was performed using UHPLC-QE-MS. Quality control confirmed data reliability. Thirteen metabolite categories were identified, with lipids and lipid-like molecules comprising the largest proportion (17.916%) (Figure 4A; Supplementary Table S11). Orthogonal partial least squares-discriminant analysis (OPLS-DA) revealed clear separations among Control, Model, and DSS groups, indicating significant metabolic disturbances in AD rats that were partly reversed by DSS (Figures 4B,C). Volcano plot analysis identified 123 upregulated and 160 downregulated metabolites in the Model vs. Control comparison, while DSS treatment led to 64 upregulated and 169 downregulated metabolites relative to the Model group (Figures 4D,E). Hierarchical clustering confirmed that DSS reversed dysregulation of key metabolites, including polyunsaturated fatty acids (docosahexaenoic acid, α-linolenic acid, γ-linolenic acid) and the second messenger cAMP, implying a multi-pathway coordination role (Figures 4F,G; Supplementary Tables S12, S13) KEGG pathway enrichment showed significant involvement of fatty acid and amino acid metabolism pathways—particularly alanine, aspartate and glutamate metabolism, and nicotinate and nicotinamide metabolism (Figures 4H–J; Supplementary Tables S14–S16). Cross-validation of network pharmacology outputs and metabolomics datasets revealed that core targets JAK2 and STAT3 were linked to four key metabolites (methylimidazoleacetic acid, pi-methylimidazoleacetic acid (hydrochloride), α-linolenic acid, γ-linolenic acid), which participate in histidine metabolism, unsaturated fatty acid biosynthesis, and fatty acid metabolism (Figures 4K,L; Supplementary Table S4). Together with network pharmacology’s designation of JAK2 as a hub target, these observations imply that DSS could regulate the JAK2/STAT3 signaling axis through targeted modulation of these metabolic cascades.

To determine whether DSS influences the Th17/Treg balance in AD rats, flow cytometric analysis of splenic CD4⁺ T cell subsets was conducted. Compared with controls, model rats showed a significant increase in Th17 cells (CD4⁺IL-17A⁺) and a marked decrease in Treg cells (CD4⁺Foxp3⁺) (p < 0.01). DSS treatment significantly ameliorated this imbalance (p < 0.05, Figures 5A–E). ELISA further demonstrated that DSS lowered serum IL-17A levels (p < 0.01) while elevating IL-10 concentrations (p < 0.05) in model rats (Figures 5F,G). These results indicate that DSS effectively restores Th17/Treg cell ratios and associated cytokine profiles in both peripheral immune organs and systemic circulation.

The regulatory effect of DSS on central nervous system Th17/Treg -related immune dysregulation was assessed via qRT-PCR and Western blot of hippocampal tissue. The model group exhibited significantly upregulated mRNA and protein levels of RORγt (p < 0.01), the Th17-specific transcription factor, accompanied by downregulation of Foxp3 (p < 0.01), the key Treg transcription factor. DSS-H treatment reversed these changes, significantly decreasing RORγt and increasing Foxp3 expression (p < 0.01) (Figures 5H–J, (Figures 6C,D). Immunofluorescence staining further confirmed increased IL-17A and decreased Foxp3 signals in the hippocampal CA2 region of the model group (p < 0.01), both of which were significantly normalized by DSS (Figures 6A,B,E,F). These findings demonstrate that DSS not only corrects Th17/Treg -related immune dysregulation in peripheral immune organs and circulation but also directly modulates key immune molecules in the brain, thereby restoring systemic and central immune dysregulation and alleviating AD-associated neuroinflammation.

Given the implication of the JAK/STAT pathway in DSS’s anti-neuroinflammatory effects from network pharmacology and metabolomics analyses, we experimentally validated this mechanism. Western blot data showed significantly elevated phosphorylated JAK2 (p-JAK2) and STAT3 (p-STAT3) protein levels in hippocampal tissue of model rats compared to controls (p < 0.01), while total JAK2 and STAT3 levels remained unchanged. Treatment with DSS markedly reduced p-JAK2 and p-STAT3 levels (p < 0.01), indicating selective inhibition of phosphorylation-dependent activation of the pathway (Figures 7A–C). Immunofluorescence and immunohistochemistry analyses corroborated these findings, with increased p-JAK2 and p-STAT3 signals observed in the model group hippocampus, which were significantly diminished following DSS treatment (p < 0.01, Figures 7D–G). Collectively, these results indicate that DSS alleviates neuroinflammation in AD rats may via suppression of the JAK2/STAT3 signaling pathway.