Salidroside ameliorates experimental autoimmune neuritis by dually modulating neuroinflammation and immune homeostasis via PI3K/AKT signaling

Abstract

Guillain-Barré syndrome (GBS), an acute autoimmune peripheral neuritis, remains a clinical challenge due to limited therapeutic efficacy and heterogeneous patient responses. The present study investigated salidroside (SAL), a bioactive phenylethanoid glycoside derived from Rhodiola rosea, for its dual anti-inflammatory and neuroprotective properties using experimental autoimmune neuritis (EAN), a validated GBS model. Integrated network pharmacology and molecular docking analyses identified PI3K/AKT signaling as the principal mechanistic target of SAL. In vivo administration of SAL (100 ​mg/kg/day, intragastric) significantly reduced neurological deficits, alleviated histopathological damage in sciatic nerves, and restored Th1/Th17-Treg immune balance in EAN rats. Mechanistic analysis demonstrated that SAL inhibited NF-κB activation through inhibition of IκBα degradation and p65 nuclear translocation, leading to downregulation of pro-inflammatory cytokines (TNF-α). Concurrently, SAL promoted macrophage polarization from the pro-inflammatory M1 (CD86⁺) to anti-inflammatory M2 (CD206⁺) phenotypes. The findings indicate that SAL mitigates EAN through a dual mechanism involving suppression of NF-κB-mediated neuroinflammation and immunomodulation via macrophage phenotype remodeling. The results further establish the PI3K/AKT pathway as a pharmacologically tractable target in autoimmune neuropathies and provide mechanistic evidence supporting SAL as a multi-target immunomodulatory candidate for GBS therapy.

Graphical abstract

Introduction

Guillain-Barré syndrome (GBS), an acute immune-mediated peripheral neuritis, is pathologically characterized by demyelinating lesions and inflammatory infiltration in nerve roots and peripheral nerves [1,2]. Although intravenous immunoglobulin and plasma exchange have been established as standard therapies, approximately 20% of patients continue to experience severe residual disabilities, with significant heterogeneity in individualized treatment responses [3,4]. The limited therapeutic outcomes reflect the complex immunopathological mechanisms of GBS, including a pro-inflammatory cytokine storm driven by Th1/Th17 hyperactivation, disruption of immune homeostasis due to regulatory T cell (Treg) dysfunction, and persistent neuroinflammation arising from aberrant macrophage polarization [[5], [6], [7], [8]]. Consequently, exploring novel immunomodulators capable of multi-target intervention in neuroinflammatory cascades has emerged as a critical scientific priority for improving GBS prognosis.

In recent years, natural products with immunomodulatory activity have attracted considerable research interest [9]. Salidroside (SAL), a phenylethanoid glycoside derived from Rhodiola rosea, has shown neuroprotective properties, including suppression of microglial activation and attenuation of oxidative stress[[10], [11], [12], [13]]. Preclinical studies suggest that SAL modulates immune cell functions through pathways such as PI3K/AKT and AMPK signaling [14,15]; however, its mechanistic role in peripheral autoimmune neuropathies remains incompletely elucidated. Macrophage phenotypic switching plays a pivotal role in the pathogenesis of GBS. Pro-inflammatory M1 macrophages aggravate neural injury through the release of TNF-α and IL-1β, whereas anti-inflammatory M2 macrophages facilitate tissue repair by secreting IL-10 [[16], [17], [18], [19]]. Dysregulation of this phenotypic transition may contribute to persistent neuroinflammation in GBS.

In the present study, an experimental autoimmune neuritis (EAN) rat model, a validated surrogate for GBS [20], was employed to systematically investigate the therapeutic targets of SAL in reshaping the peripheral nerve immune microenvironment. Network pharmacology prediction combined with molecular docking identified the PI3K/AKT/NF-κB signaling axis as a central regulatory hub [21]. To verify this pathway, ZSTK474, a potent and selective pan-class I PI3K inhibitor targeting the α, β, δ, and γ isoforms, was selected [22]. The choice of ZSTK474 was based on two considerations: its broad and strong inhibitory activity across multiple PI3K isoforms and previous evidence supporting its efficacy in autoimmune and inflammatory models [23,24]. The investigation specifically examined the dual mechanisms of SAL, namely the restoration of Th1/Th17-Treg immune homeostasis and the reprogramming of macrophage polarization, to mitigate EAN pathology. The study seeks to establish a novel paradigm in which natural immunomodulators achieve multi-target regulation of neural-immune crosstalk, thereby providing a theoretical foundation for precision therapies in GBS.

Materials and Methods

Prediction of disease-related targets of SAL

Potential targets of SAL were predicted using Swiss Target Prediction (https://www.swisstargetprediction.ch/), PharmMapper (https://lilab-ecust.cn/pharmmapper/), the Comparative Toxicogenomics Database (CTD; http://ctdbase.org/), HERB (http://herb.ac.cn/), and the Similarity Ensemble Approach (SEA; https://sea.bkslab.org/). GBS-related targets were retrieved from GeneCards (https://www.genecards.org/), DisGeNET (http://www.disgenet.org/), CTD, and Online Mendelian Inheritance in Man (OMIM; https://omim.org/).

Identification of intersection targets and PPI network construction

Overlapping targets between SAL and GBS were identified and visualized with the Venn diagram tool (https://bioinformatics.psb.ugent.be/webtools/Venn/). Protein-protein interaction (PPI) analysis of the shared targets was performed using the STRING database (https://cn.string-db.org/) with Homo sapiens as the selected organism and an interaction score cutoff of 0.9. The PPI network was constructed and analyzed using Cytoscape software (version 3.7.0). Core targets were determined according to degree centrality [25].

Functional enrichment analysis

Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis was carried out using the DAVID database (https://davidbioinformatics.nih.gov/). The results were visualized as bubble charts, in which statistical significance was represented by -log10(P value) and the number of enriched genes was indicated by counts.

Molecular docking of SAL with core targets

Molecular docking was performed using the Molecular Operating Environment (MOE) software. The 2D structure of SAL (PubChem CID: 159278) was imported into MOE and energy-minimized prior to docking. Crystal structures of the core target proteins were retrieved from the Protein Data Bank (PDB; https://www.rcsb.org/). Water molecules were removed, and hydrogen bonding was optimized before conducting full-atom docking. Binding affinities were calculated, and both two-dimensional and three-dimensional interaction diagrams were generated.

Chemicals and reagents

The P2_57–81 peptide was synthesized by Hangzhou Dangang Biotechnology Co., Ltd. (China). Incomplete Freund's adjuvant (IFA) and Mycobacterium tuberculosis H37RA were purchased from Sigma-Aldrich (USA) and BD Difco (USA), respectively. Antibodies and reagents included: GAPDH (Servicebio, Cat# GB 15004-100), p-AKT (Proteintech, Cat# 28731-1-AP), AKT (Proteintech, Cat# 10176-2-AP), p-NF-κB (Proteintech, Cat# 82335-1-RR), NF-κB (Proteintech, Cat# 10745-1-AP), CD86 (Proteintech, Cat# 13395-1-AP), CD206 (Proteintech, Cat# 18704-1-AP), MBP (Proteintech, Cat# 10458-1-AP), p-IκBα (Affinity, Cat# AF2002), IκBα (Affinity, Cat# AF5002), IFN-γ (Absin, Cat# abs119966), IL-17A (Proteintech, Cat# 26163-1-AP), Foxp3 (Affinity, Cat# AF6544), TNF-α (Abmart, Cat# PY19810S), PE-CD3 (eBioscience™, Cat# 12-0030-81), Foxp3-BV421 (eBioscience™, Cat# 404-5773-80), IL-17A-BV786 (eBioscience™, Cat# 417-7177-80), PerCP/Cyanine 5.5-CD4 (BioLegend, Cat# 201519), APC anti-rat CD25 (BioLegend, Cat# 202113), CD45 (BioLegend, Cat# 202218), FITC-IFN-γ (BD Biosciences, Cat# 559498), Transcription Factor Buffer Set (BD Pharmingen Cat# 562574), Fixation/Permeablization Kit (BD Biosciences, Cat# 554714), Leukocyte Activation Cocktail, with BD GolgiPlug (BD Biosciences, Cat# 550583) Fixable Viability Stain 510 (BD Biosciences, Cat# 564406), CD11b (BD Pharmingen, Cat# 562102), CD86 (BD Pharmingen, Cat# 551396), CD206 (Santa Cruz Biotechnology, Cat# sc-58986AF488), and CD32 (BD Biosciences, Cat# 550270). SAL (purity >98 ​%; CAS 10338-51-9) was obtained from Weikeqi Biological Technology Co., Ltd. (Sichuan, China). ZSTK474 (HY-50847) was purchased from MedChemExpress (USA).

Animal experiments

All procedures adhered to the NIH Guide for the Care and Use of Laboratory Animals and were approved by the Ethics Committee of Ningxia Medical University (IACUC-NYLAC-2023-291). Female Lewis rats (8 weeks old, SPF grade) were housed under controlled conditions (12 ​h light/dark cycle, 22 ​± ​1 ​°C). Rats were randomly assigned to six groups (n ​= ​6/group): (1) control; (2) EAN ​+ ​vehicle; (3) EAN ​+ ​salidroside low dose (SAL-L, 50 ​mg/kg); (4) EAN ​+ ​salidroside medium dose (SAL-M, 100 ​mg/kg); (5) EAN ​+ ​salidroside high dose (SAL-H, 200 ​mg/kg); (6) EAN ​+ ​ZSTK474 (ZSTK474, 30 ​mg/kg) [26].

EAN was induced by subcutaneous injection of 200 ​μL of P2 peptide emulsion. The emulsion was prepared by mixing 100 ​μL of incomplete Freund's adjuvant containing 2 ​mg of Mycobacterium tuberculosis with 100 ​μL of PBS containing 0.4 ​mg of P2 peptide, resulting in a final concentration of P2 (2 ​mg/mL). SAL or ZSTK474 was administered by oral gavage once daily from day 10 to day 18 post-immunization. Neurological deficits were evaluated using a standardized clinical scoring system [27]. To ensure consistency of the disease model, several standardized procedures and evaluation criteria were applied. First, all animals were immunized on the same day by the same experimenter, including antigen emulsification and injection, to minimize variability. Second, daily neurological assessments were conducted from day 0 onward using a blinded scoring system. Only animals that developed neurological symptoms (score ≥1) were included in the intervention studies. Rats were euthanized at the disease peak (day 18) for tissue collection.

Neuro-electrophysiological analysis

Compound muscle action potentials (CMAPs) were recorded from the gastrocnemius muscle under anesthesia using an NDI-094 EMG system (Haishen, China). Stimulating electrodes were positioned at the proximal site of the sciatic nerve exit and at the distal sciatic notch. CMAPs amplitude, nerve conduction velocity, and latency were measured in triplicate [28].

Histopathological evaluation

Sciatic nerves were fixed in 4% paraformaldehyde, paraffin-embedded, and sectioned (4 ​μm). Sections were stained with hematoxylin and eosin (H&E) or Luxol fast blue (LFB; Solarbio, China). Sections were stained with hematoxylin and eosin (H&E) or Luxol fast blue (LFB; Solarbio, China) and examined using a digital slide scanner. All analyses were conducted with a standardized and reproducible image analysis pipeline. First, overview images were acquired at low magnification, and random fields of view were selected using the ImageJ Grid tool, excluding vessel areas. Quantification was restricted to these randomly selected fields. For H&E staining, the total neural tissue area was manually outlined in ImageJ, and all inflammatory cells within the region were counted. For LFB staining, a uniform optical density threshold was applied to distinguish blue-stained regions (normal myelin) from pale or unstained regions (demyelination).

Immunofluorescence (IF)

Paraffin sections were dewaxed, rehydrated, and subjected to antigen retrieval in EDTA buffer (pH 8.0). Slides were heated to boiling, maintained at a slight boil for 20 ​min, and then cooled to room temperature (RT). After blocking with 5% goat serum for 1 ​h at RT, primary antibodies MBP (1:200, Proteintech), CD68 (1:100, Servicebio), CD3 (1:100, Servicebio), CD206 (1:200, Proteintech), CD86 (1:200, Proteintech), IL-17A (1:200, Proteintech), Foxp3 (1:200, Affinity), p-AKT (1:200, Proteintech), p-IκBα (1:200, Affinity), p-P65 (1:200, Proteintech) diluted in antibody diluent were applied to fully cover the tissue sections. Sections were incubated overnight at 4 ​°C in a humidified chamber (14–16 ​h). The following day, after equilibration to RT, sections were incubated with fluorescence-conjugated secondary antibodies (Alexa Fluor 488, Alexa Fluor 594) for 1 ​h in a light-protected, humidified chamber. Nuclear counterstaining was performed with DAPI for 5 ​min, followed by mounting with anti-fade medium. Image acquisition was carried out using a digital slide scanner.

Flow cytometry

Spleen cell isolation

Spleens were dissociated through a 70 ​μm strainer (Absin, Cat# abs7306-1). Red blood cells were lysed using BD Pharm Lyse™ Buffer (BD, Cat# 555899). The remaining single-cell suspensions were resuspended in PBS and counted prior to further analysis.

Macrophage phenotyping

Macrophage immunophenotyping was performed using the following protocol: ① Viability staining: Single-cell suspensions (1 ​× ​10⁶ ​cells per sample) were incubated with Fixable Viability Stain 510 in PBS for 15 ​min at 4 ​°C in the dark. ② Fc receptor blocking: Cells were treated with anti-CD32 antibody for 15 ​min at 4 ​°C to minimize nonspecific binding. ③ Surface marker staining: Cells were incubated with the following fluorochrome-conjugated antibodies, including Alexa Fluor® 700-CD45 antibody (Biolegend, 1:50), APC Mouse Anti-Rat CD11b antibody (BD, 1:40), and PE-CD86 antibody (BD, 1:40) for 30 ​min at 4 ​°C in the dark. ④ Intracellular staining: After fixation and permeabilization, intracellular staining was performed with CD206-Alexa Fluor® 488 antibody (Santa Cruz, 1:50) for 30 ​min at 4 ​°C in the dark. Finally, cells were resuspended in 300 ​μL PBS and analyzed using a BD FACSCelesta™ flow cytometer.

CD4⁺ T cell analysis

Mononuclear cells were resuspended in complete RPMI 1640 medium at a density of 1 ​× ​10⁶ ​cells/mL and seeded into 24-well plates. Cells were stimulated with Cell Stimulation Cocktail diluted 1:500 in culture medium and incubated in a CO₂ incubator for 5 ​h before collection. Following stimulation, cells were stained with Fixable Viability Stain 510, blocked with anti-CD32 antibody, and incubated with fluorochrome-conjugated antibodies including PE-CD3 (Thermo, 1:100), PerCP/Cyanine 5.5-CD4 (BioLegend, 1:100), and APC-CD25 (BioLegend, 1:50). After fixation and permeabilization according to the manufacturer's protocol, intracellular staining was performed with IL-17A-BV786 (Thermo, 1:100), FITC-IFN-γ (BD, 1:5), and FOXP3-BV421 (Thermo, 1:40). Samples were then analyzed using a BD FACSCelesta™ flow cytometer.

Western blotting

Sciatic nerves from both sides of each rat were pooled, homogenized, and processed for protein extraction to generate one biological replicate. Protein concentrations were determined using the BCA assay, and 20 μg of protein per sample was loaded per lane to ensure equal input. Proteins were separated on 10% SDS-PAGE gels and transferred onto PVDF membranes (Millipore, USA). Membranes were blocked with 5% skim milk for 2 h at RT with gentle shaking. For phosphorylated proteins, membranes were blocked with rapid blocking buffer (Servicebio, Cat# G2052-500 ML) for 20 min at RT. Membranes were then incubated overnight (14–16 h) at 4 °C with primary antibodies, including GAPDH (1:5000), p-AKT (1:2000), AKT (1:2000), p-NF-κB (1:3000), NF-κB (1:3000), CD86 (1:1000), CD206 (1:1000), p-IκBα (1:1000), IκBα (1:1000), IFN-γ (1:1000), IL-17A (1:1000), Foxp3 (1:1000), and TNF-α (1:1000). The following day, membranes were washed three times with TBST for 10 min each, followed by incubation with HRP-conjugated secondary antibodies for 1 h at RT. After additional TBST washes, immunoreactive bands were visualized using the AMERSHAM ImageQuant 800 system and quantified with ImageJ software, with GAPDH serving as the normalization control.

Statistical analysis

Data were analyzed using GraphPad Prism version 9.0. For comparisons among multiple groups, the Kruskal-Wallis and one-way ANOVA followed by Dunnett's multiple comparisons test was applied. Statistical significance was defined at P ​< ​0.05.

Results

Prediction of SAL and GBS targets via Network Pharmacology Analysis

Databases including Swiss Target Prediction, PharmMapper, CTD, HERB, and SEA yielded 168 unique SAL-related target proteins after deduplication, comprising 100 from Swiss Target Prediction, 25 from PharmMapper, 68 from CTD, 50 from HERB, and 50 from SEA. For GBS, 2258 targets were collected from GeneCards, OMIM, DisGeNET, and CTD, refined to 1028 (GeneCards), 329 (OMIM), 171 (DisGeNET), and 1270 (CTD). Intersection analysis identified 93 overlapping targets between SAL and GBS (Fig. 1A). PPI network analysis identified core targets, including AKT1 and NF-κB1 (Fig. 1B–D). KEGG pathway enrichment analysis of these targets highlighted 20 signaling pathways, with the PI3K/AKT pathway being significantly enriched in GBS (Fig. 1E). These results indicate that the PI3K/AKT signaling pathway may serve as a key mechanistic target underlying the therapeutic effects of SAL.

Fig. 1.

Prediction of SAL and GBS targets via Network Pharmacology Analysis (A) Venn diagram of GBS and Salidroside-associated targets. A total of 2258 GBS-related targets (left), 168 salidroside-related targets (right), and 93 overlapping targets (center) were identified. (B) PPI network of salidroside and GBS targets. (C) Cytoscape interactive network showing the highest-scoring MCODE module (40 nodes, 659 edges). (D) The core target gene was analyzed by the CytoHubba Degree of Interaction topology algorithm. (E) KEGG pathway enrichment analysis. Y-axis indicates pathways, X-axis indicates false discovery rate (FDR), and bubble color indicates P-values. Bubble size reflects the number of genes enriched in each pathway.

Molecular docking of SAL with core targets

Molecular docking analysis demonstrated binding energies of -5.8729 and -7.5822 ​kcal/mol for SAL with AKT1 and NF-κB1, respectively (Supplementary Table 1). Binding energies ​< ​-5 ​kcal/mol indicate strong molecular interactions, consistent with favorable docking results. Two- and three-dimensional diagrams confirmed that SAL occupies the active pockets of both AKT1 and NF-κB1 (Fig. 2A and B).

Fig. 2.

Molecular Docking of SAL with core targets (A) Molecular docking results of salidroside with AKT1. (B) Molecular docking results of salidroside with NF-κB1. Arrows indicate the binding orientation of salidroside as a ligand within the active sites of the target proteins. Two- and three-dimensional docking diagrams confirm that salidroside occupies the active binding pockets of AKT1 and NF-κB1.

SAL reduces clinical scores and restores the conduction velocity of motor nerves (MNCV) in EAN rats

On days 17 and 18 after immunization, clinical scores in the treatment groups were significantly lower than those in the EAN group (Fig. 3A). At the disease peak (day 18), the median clinical score reached 6.0 in the EAN group, whereas both the SAL-M and ZSTK474 groups showed a median score of 4.0, representing a significant reduction (Kruskal-Wallis with Dunn's multiple comparisons test, P ​= ​0.0155 for EAN vs. SAL-M; P ​= ​0.0251 for EAN vs. ZSTK474; Fig. 3B). Neuro-electrophysiological analysis at the disease peak showed that CMAPs and, particularl, motor nerve conduction velocity (MNCV) were markedly improved in the treatment groups (Fig. 3C). One-way ANOVA with Dunnett's post-hoc test revealed that the SAL-M group exhibited a mean MNCV of 51.94 ​m/s, which was 22.34 ​m/s faster than the EAN group (mean diff. ​= ​-22.34, 95% CI: -43.19 to -1.491, adjusted P ​= ​0.0328). The ZSTK474 group also showed significant improvement (mean MNCV: 52.56 ​m/s; mean diff. ​= ​-22.96, 95% CI: -43.81 to -2.111, adjusted P ​= ​0.0274). The SAL-L and SAL-H groups demonstrated non-significant trends toward increased MNCV compared with the EAN group (mean diff. ​= ​-13.40 ​m/s, adjusted P ​= ​0.3113; mean diff. ​= ​-8.360 ​m/s, adjusted P ​= ​0.7213; Fig. 3D). All in vivo pharmacodynamic evaluations for the groups of interest (EAN, SAL-M) were conducted in two independent and complete animal experiments including behavioral scoring and electrophysiology analysis (Supplemental Fig. 1). Based on these results, SAL-M was selected for subsequent mechanistic studies.

Fig. 3.

SAL reduces clinical scores and restores the conduction velocity of motor nerves (MNCV) in EAN rats (A) Clinical scores of rats in the control (blue), EAN (red), SAL-L (green), SAL-M (purple), SAL-H (orange), and ZSTK474 (black) groups from day 0 to day 18 post-immunization. (B) Clinical scores of the EAN group, SAL-L groups, SAL-M groups, SAL-H groups, and ZSTK474 groups at the peak of disease. (Kruskal-Wallis test with Dunn's comparisons, ∗p ​< ​0.05, n ​= ​6) (C) CMAP recordings from proximal to distal sciatic nerve in control groups, EAN groups, SAL-L groups, SAL-M groups, SAL-H groups, and ZSTK474 groups. (D) Differences in MNCV at the disease peak compared with the EAN group. One-way ANOVA with Dunnett's multiple comparisons test was performed (∗p ​< ​0.05, ∗∗p ​< ​0.01, ∗∗∗p ​< ​0.001, n ​= ​5).

SAL mitigates inflammatory infiltration and demyelination in EAN rats

Representative H&E and LFB staining and quantitative histopathological assessment of sciatic nerves confirmed the therapeutic effects of SAL and ZSTK474 (Fig. 4A–D). HE staining showed that SAL-M and ZSTK474 treatment significantly reduced inflammatory cell infiltration by 26.17% and 29.13%, respectively, compared with the EAN group (EAN vs. SAL-M: mean diff. ​= ​336.0, 95 ​% CI: 31.97 to 640.0, adjusted P ​= ​0.0292; EAN vs. ZSTK474: mean diff. ​= ​374.0, 95% CI: 69.97 to 678.0, adjusted P ​= ​0.015; one-way ANOVA with Dunnett's test). Consistently, LFB staining demonstrated that both treatments largely prevented demyelination, as reflected by preservation of myelinated fiber area (EAN vs. SAL-M: mean diff. ​= ​-10.54, 95% CI: -18.31 to -2.779, adjusted P ​= ​0.0077; EAN vs. ZSTK474: mean diff. ​= ​-11.04, 95% CI: -18.80 to -3.271, adjusted P ​= ​0.0054; one-way ANOVA with Dunnett's test). Immunofluorescence staining further corroborated these findings. SAL-M preserved myelin basic protein (MBP) expression in sciatic nerve tissues (Fig. 4E), while attenuating inflammatory infiltration by T lymphocytes (CD3⁺) and macrophages (CD68⁺) (Fig. 4I, G). (Supplementary Fig. 2 provides single-channel fluorescent staining images of CD68⁺cell and CD3⁺ ​cell). Quantitative analysis confirmed that SAL administration significantly alleviated demyelination (EAN vs. SAL-M: mean diff. ​= ​-35.98, 95% CI: -67.38 to -4.584, adjusted P ​= ​0.0270) (Fig. 4F), reduced T-cell infiltration (EAN vs. SAL-M: mean diff. ​= ​8.670, 95% CI: 2.533 to 14.81, adjusted P ​= ​0.0091) (Fig. 4J), and diminished macrophage infiltration (EAN vs. SAL-M: mean diff. ​= ​7.620, 95% CI: 2.935 to 12.30, adjusted P ​= ​0.0041) (Fig. 4H).

Fig. 4.

SAL mitigates inflammatory infiltration and demyelination in EAN rats (A, B) Representative H&E and LFB staining in normal, EAN, SAL, and ZSTK474 groups (magnification: ​× ​200, scale bar: 50 ​μm; magnification: ​× ​400, scale bar: 20 ​μm). (C, D) Quantitative analysis of H&E and LFB staining. (E, F) Expression and quantitative analysis of MBP in sciatic nerve tissue. (G–J) Infiltration and quantitative analysis of CD68⁺ macrophages and CD3⁺ T cells in sciatic nerve tissue. Statistical analysis was performed using one-way ANOVA with Dunnett's multiple comparisons test (∗p ​< ​0.05, ∗∗p ​< ​0.01, ∗∗∗p ​< ​0.001).

SAL modulates macrophage polarization and T-cell responses in EAN rats

Results of flow cytometric analysis for macrophages and helper T cells (Fig. 5A–C). Flow cytometry analysis demonstrated that SAL-M treatment significantly reduced the proportion of CD86⁺ M1 macrophages (EAN vs. SAL-M: mean diff. ​= ​5.917, 95% CI: 1.204 to 10.63, P ​= ​0.0172) while increasing CD206⁺ M2 macrophages (EAN vs. SAL-M: mean diff. ​= ​-13.82, 95% CI: -27.38 to -0.2596, P ​= ​0.0461). Additionally, SAL promoted Treg cell (CD25⁺Foxp3⁺) expansion (EAN vs. SAL-M: mean diff. ​= ​-2.21, 95% CI: -3.071 to -1.349, P ​= ​0.0002) while suppressing Th1 (IFN-γ⁺) and Th17 (IL-17A⁺) cell populations (EAN vs. SAL-M: mean diff. ​= ​6.867, 95% CI: 2.554 to 11.18, P ​= ​0.0046; mean diff. ​= ​0.6333, 95% CI: 0.1551 to 1.112, P ​= ​0.0013), respectively (Fig. 5D–H). The gating strategy for macrophages and helper T cells is presented (Fig. 5I–K).

Fig. 5.

Salidroside regulates macrophage polarization and T cell responses in the spleen of EAN. (A–C) Percentage of M1 (CD45⁺CD11b⁺CD86⁺), M2 (CD45⁺CD11b⁺CD206⁺), Th1 (CD3⁺CD4⁺IFN-γ ⁺), Th17 (CD3⁺CD4⁺IL-17A⁺) and Treg (CD3⁺CD4⁺CD25⁺Foxp3⁺) cells. (D–H) Statistical analysis of M1, M2, Th1, Th17, and Treg subsets in spleens. One-way ANOVA was used to compare the EAN group with treatment groups (∗p ​< ​0.05, ∗∗p ​< ​0.01, ∗∗∗p ​< ​0.001, ∗∗∗∗p ​< ​0.0001, n ​= ​3). Gating strategies for flow cytometric analysis of: (I) macrophage subsets (M1 and M2); (J) Th1 and Th17 ​cell subsets; (K) Treg cells.

IF of sciatic nerve sections was conducted to characterize macrophage subtypes, including CD86⁺ M1 and CD206⁺ M2 macrophages (Fig. 6A, C), as well as T-cell subsets, represented by IL-17A⁺ Th17 (pro-inflammatory) and Foxp3⁺ Treg (anti-inflammatory) cells (Fig. 6E, G). The staining results revealed distinct infiltration patterns of these immune cell populations within the nerve tissue. Quantitative analysis showed that SAL treatment markedly reduced infiltration of pro-inflammatory CD86⁺ M1 macrophages and IL-17A⁺ Th17 ​cells (EAN vs. SAL-M: mean diff. ​= ​5.226, 95% CI: 0.3629 to 10.09, P ​= ​0.0364; mean diff. ​= ​4.040, 95% CI: 0.05281 to 8.028, P ​= ​0.0473) (Fig. 6B, F) while enhancing the presence of anti-inflammatory CD206⁺ M2 macrophages and Foxp3⁺ Treg (EAN vs. SAL-M: mean diff. ​= ​-5.299, 95% CI: -10.35 to -0.4439, P ​= ​0.0342; mean diff. ​= ​-9.494, 95% CI: -16.70 to -2.287, P ​= ​0.0134) (Fig. 6D, H). (Supplementary Fig. 3 provides single-channel fluorescent staining images of CD86⁺ cells, CD206⁺cells, IL-17A⁺cells, and Foxp3⁺cells).

Fig. 6.

Expression of major markers of macrophages (M1, M2) and helper T cells (Th1, Th17, Treg) in sciatic nerve tissue (A–D) IF of Macrophage Subtype (CD86⁺-red, CD206⁺-green) and quantitative analysis of sciatic nerve (∗p ​< ​0.05, ∗∗p ​< ​0.01 n ​= ​3); (E–H) IF and quantitative analysis of helper T-cell subsets (IL-17A⁺, red; Foxp3⁺, green). (I) Representative Western blot images showing CD206⁺ (M2), CD86⁺ (M1), Foxp3⁺ (Treg), IFN-γ⁺ (Th1) and IL-17A⁺ (Th17) expression. (J–N) Densitometric analysis of protein expression normalized to GAPDH (∗p ​< ​0.05, ∗∗p ​< ​0.01, ∗∗∗p ​< ​0.001, n ​= ​3).

Western blotting was performed to examine the expression of macrophage and T-cell markers, including CD86⁺ (M1), CD206⁺ (M2), IFN-γ⁺ (Th1), IL-17A⁺ (Th17), and Foxp3⁺ (Treg), in sciatic nerve tissue (Fig. 6I). Protein levels were quantified by densitometric analysis and normalized to GAPDH. The results demonstrated a significant reduction in CD86⁺ M1 macrophages, IFN-γ⁺ Th1 cells, and IL-17A⁺ Th17 ​cells following SAL treatment (EAN vs. SAL-M: mean diff. ​= ​1.961, 95% CI: 1.640 to 2.858, P ​= ​0.0006; mean diff. ​= ​1.507, 95% CI: 0.2504 to 2.763, P ​= ​0.0216; mean diff. ​= ​1.233, 95% CI: 0.4526 to 2.013, P ​= ​0.0048) (Fig. 6K-M). In contrast, CD206⁺ M2 macrophages and Foxp3⁺ Tregs were significantly increased in the SAL-M group (EAN vs. SAL-M: mean diff. ​= ​-0.3779, 95% CI: -0.6704 to -0.08533, P ​= ​0.0148; mean diff. ​= ​-0.7065, 95% CI: -1.090 to -0.3234, P ​= ​0.0019) (Fig. 6J, N). One-way ANOVA with Dunnett's multiple comparisons test confirmed these differences. We have prepared uncropped original blot images and have uploaded them as Supplementary Material (Supplemental Fig. 5). Collectively, these findings provide direct evidence that SAL reprograms the inflammatory microenvironment.

SAL inhibits activation of AKT/NF-κB signaling axis in sciatic nerve tissue of EAN rats

To explore the mechanism underlying the protective effect of SAL-M in EAN, the activation status of the AKT/NF-κB pathway was examined, as this signaling cascade is critical in regulating inflammatory responses. IF and Western blotting of sciatic nerve tissues collected at the peak of disease (day 18 post-immunization) demonstrated marked changes in pathway activation (Fig. 7).

Fig. 7.

SAL inhibits activation of AKT/NF-κB signaling axis in sciatic nerve tissue of EAN rats (A–F) IF of p-AKT, p-IκBα, p-P65 and quantitative of sciatic nerve (∗p ​< ​0.05, ∗∗p ​< ​0.01, n ​= ​4); (G) Representative Western blot images showing p-AKT, AKT, p-IκBα, IκBα, P65, p-P65 and TNF-α expression. (H–J) Quantification of phosphorylated protein levels relative to total protein. (K) Quantification of TNF-α levels relative to GAPDH (∗p ​< ​0.05, ∗∗p ​< ​0.01, ∗∗∗p ​< ​0.001, n ​= ​3).

Densitometric analysis revealed that phosphorylated AKT (p-AKT), phosphorylated IκBα (p-IκBα), phosphorylated NF-κB p65 (p-P65), and the downstream cytokine TNF-α were significantly elevated in the EAN group compared with controls. SAL-M treatment markedly reduced the expression of these activated proteins. Quantitative analysis confirmed the inhibitory effect of SAL-M on pathway activation. Expression of p-AKT increased in the EAN group but was significantly reduced after SAL-M treatment (EAN vs. SAL-M: mean diff. ​= ​0.2802, 95% CI: 0.01594 to 0.5444, P ​= ​0.0387) (Fig. 7H). A similar reduction was observed for p-IκBα (EAN vs. SAL-M: mean diff. ​= ​0.8294, 95% CI: 0.03916 to 1.620, P ​= ​0.0405) (Fig. 7I) and p-P65 (EAN vs. SAL-M: mean diff. ​= ​1.2, 95% CI: 0.1092 to 2.229, P ​= ​0.0327) (Fig. 7J). The elevated production of TNF-α in the EAN group was also significantly suppressed by SAL-M (EAN vs. SAL-M: mean diff. ​= ​1.002, 95% CI: 0.2027 to 1.802, P ​= ​0.0173) (Fig. 7K). And consistent results were obtained from IF of p-AKT, p-IκBα, p-P65 and quantitative of sciatic nerve (Fig. 7A–F). (Supplementary Fig. 4 provides single-channel fluorescent staining images of p-AKT, p-IκBα, p-P65 positive cells). Collectively, these findings indicate that the therapeutic benefit of SAL in EAN involves inhibition of AKT/NF-κB signaling pathway activation.

Discussion

The present study systematically elucidates the multidimensional mechanisms through which SAL ameliorates EAN via modulation of the PI3K/AKT axis. In contrast to previous investigations, the current work establishes a comprehensive chain of evidence for SAL in the treatment of peripheral autoimmune neuropathies by integrating dose-response evaluation, molecular mechanism exploration, and immunoregulatory assessment. Administration of SAL at 100 ​mg/kg markedly attenuated P2-induced autoimmune peripheral neuropathy, achieving therapeutic efficacy comparable to ZSTK474, a selective PI3K/AKT inhibitor, while avoiding the potential toxic side effects commonly associated with chemical inhibitors. These findings provide important insights for the development of natural therapeutics with low toxicity and high efficiency.

Prior studies have shown that aberrant activation of the PI3K/AKT pathway drives NF-κB signaling cascades by promoting IκBα degradation and p65 nuclear translocation, thereby amplifying the release of pro-inflammatory cytokines such as TNF-α [29,30]. Experimental data confirmed that SAL-M markedly suppressed AKT phosphorylation and inhibited activation of downstream NF-κB components, including TNF-α. These results indicate that SAL-M reduces inflammatory mediators in EAN rats by targeting the AKT/NF-κB signaling axis. The interpretation is further supported by previous reports[[31], [32], [33], [34]], which consistently showed that inhibition of this pathway, reflected by reduced phosphorylation levels, correlates with the alleviation of inflammation.

Molecular docking analysis revealed that SAL established stable hydrogen bond networks within the active pockets of AKT1 (binding energy, -5.87 ​kcal/mol) and NF-κB1 (-7.58 ​kcal/mol), indicating a potential conformational locking effect that interferes with ATP binding at kinase catalytic domains [35,36]. SAL treatment demonstrated robust efficacy in ameliorating MNCV and histopathological damage, resulting in functional and structural outcomes that are comparable to, or potentially exceed, those achieved by other therapies in EAN rats [37]. This suggests that SAL warrants further investigation as a promising therapeutic candidate for GBS.

At the molecular level, SAL exerts dual regulatory effects. First, it suppresses pathological PI3K/AKT activation in a dose-dependent manner, thereby attenuating downstream NF-κB signaling, as reflected by reduced p-P65 levels and subsequent downregulation of TNF-α expression. Second, SAL dynamically modulates macrophage polarization, which emerges as a key immunoregulatory mechanism. Flow cytometry analysis demonstrated a significant reduction in the proportion of M1 macrophages (CD86⁺), accompanied by an increase in M2 macrophages, a phenotypic shift associated with decreased pro-inflammatory cytokine production. In addition, IF of sciatic nerve tissue revealed that SAL induces M2 macrophage polarization more potently than a PI3K inhibitor, indicating its enhanced capability to modulate innate immune responses in EAN rats. Within the adaptive immune compartment, SAL reduced Th1/Th17 ​cell frequencies while promoting Treg expansion, potentially through PI3K/AKT-mediated regulation of transcription factors governing T-cell differentiation. Dose-response experiments further indicated that 100 ​mg/kg achieved optimal therapeutic efficacy, whereas efficacy declined at 200 ​mg/kg, suggesting receptor saturation or negative feedback regulation. Such an “inverted U-shaped” dose-response pattern is consistent with observations reported for natural compounds by published study [38].

However, the study has several limitations. Although published evidence supports the computational predictions regarding the regulation of AKT, NF-κB p65, and the PI3K/AKT/NF-κB pathway by SAL [30,39], computational data should be supported by proteomic analyses. In addition, the lack of conditional knockout animal models limits definitive verification of key signaling nodes. Future work will incorporate proteomic approaches to directly quantify protein expression and phosphorylation within the PI3K/AKT pathway. Moreover, the use of AKT1 conditional knockout models combined with single-cell RNA sequencing will facilitate precise mapping of SAL-mediated immunoregulatory networks and promote clinical translation.

In conclusion, the study elucidates the PI3K/AKT-dependent therapeutic mechanisms of SAL in experimental autoimmune neuritis, highlighting its dual role in suppressing NF-κB-mediated neuroinflammation and restoring immune homeostasis through macrophage polarization and Th1/Th17-Treg balance modulation. These findings provide both a theoretical framework and experimental foundation for advancing the development of natural product–derived therapeutics for neuroimmune disorders.

Author Contributions

Tao Gu: Writing – original draft, Methodology, Investigation. Li Wang: Methodology, Investigation. Xue Kang: Methodology, Investigation, Conceptualization. Chunguang Yu: Investigation. Kaichun Shen: Methodology, Conceptualization. Man Song: Methodology. Yingying Gao: Methodology. Guowei Wang: Writing – review & editing. Zhenhai Wang: Supervision, Resources, Project administration, Funding acquisition.

Funding

This work was supported by the Science and Technology Key Research Program of Ningxia (Grant No. 2023BEG02073), the National Natural Science Foundation of China (Grant No. 82371358) and the National Science and Technology Major Project of China (Grant No. 2025ZD0546002).

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following are the Supplementary data to this article.