Rheumatoid arthritis (RA) represents an autoimmune disorder primarily defined by synovial joint inflammation. Its etiology remains unclear, and its pathogenesis is complex1,2. In its early stages, synovial cells undergo tumor-like hyperplasia, leading to joint swelling and pain. In advanced stages, bone and cartilage destruction occurs, resulting in irreversible structural damage and functional impairment of the affected joints3. This imposes a significant physical and psychological burden on patients. Current primary treatments for RA include traditional disease-modifying drugs and biological agents. Nonetheless, these options are related to multiple side effects: Hepatotoxicity, nephrotoxicity, high costs, and increased risks of infections and cardiovascular diseases. Statistics indicate that only about 70% of RA patients experience satisfactory therapeutic effects4, while 10%−20% do not achieve remission5. This highlights the existence of a“therapeutic ceiling”effect. Additionally, 5%−10% of RA patients exhibit a persistent"difficult-to-treat"pattern of clinical symptoms6.Traditional Chinese Medicine (TCM) possesses a history spanning several thousand years in treating RA and other joint pain disorders. Growing evidence suggests that extracts, monomers, and active components derived from certain traditional herbs exhibit significant therapeutic effects against RA, often with minimal side effects. The Chinese herb Gentiana, commonly known as Qin Jiao, and its related species, such as Gentiana crassicaulis Duthie ex Burk., Gentiana straminea Maxim., and Gentiana dahurica Fisch., are sourced from the dried roots of these plants. Characterized by a bitter taste and neutral properties, Qin Jiao is renowned for its ability to dispel wind dampness and relax tendons and meridians. Qin Jiao can be used in various combinations to treat both acute and chronic rheumatic conditions, establishing its significance as a crucial medication in the clinical management of RA. Gentiopicroside (GEN), a monomeric compound isolated from Qin Jiao, has demonstrated anti-RA properties, including the inhibition of inflammatory responses and osteoclastogenesis7,8. Nevertheless, the precise signaling pathways and molecular targets that contributed to these effects remain elusive. Currently, the analysis of drug targets and mechanisms through network pharmacology (NP) is emerging as a promising avenue for further research. To this end, we employed relevant computational methods to explore the effects of GEN on RA. Additionally, we utilized molecular docking (MOD) alongside in vivo and in vitro experiments to validate the impact of GEN on RA. (Fig. 1)Fig. 1Flow chart of present research.Full size imageMaterials and methodsPotential target predictionSeeking GEN potential target identification (CAS:20831-76−9) for treating RA, we employed an in silico approach. First, we retrieved targets associated with Gentiopicrioside components via the Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform (TCMSP) (https://old.tcmsp-e.com/tcmsp.php), Pubchem (https://pubchem.ncbi.nlm.nih.gov/), Bioinformatics Analysis Tool for Molecular mechANism of Traditional Chinese Medicine (BATMAN-TCM) (http://bionet.ncpsb.org/batman-tcm), Search Tool for Interacting Chemicals (STITCH)(http://stitch.embl.de), as well as Similarity Ensemble Approach Database (SEA) (http://sea.edbc.org). Next, we identified RA-related targets using the Disgenet (https://www.disgent.com) and Genecards (https://www.genecards.org/) databases, ensuring consistent target nomenclature via the UniProt database (https://www.uniprot.org). Eventually, by comparing the identified targets from both sources, we obtained a refined list of potential component-disease targets for GEN in treating RA.NP analysisFollowing the acquisition of component-disease potential target information, they were imported into the STRING (https://string-db.org). Focusing on the “Homo sapiens” species, we utilized a minimum interaction threshold of “medium confidence” (> 0.07) to retrieve the protein-protein interaction (PPI) information and generate the corresponding target association network diagram. The results were then exported for analysis. Using Excel, we calculated the node degree value (number of connections) for each target. Finally, targets exceeding the average node degree value were identified as potentially important targets of GEN for treating RA.Functional enrichment analysisAfter the component-disease potential targets were imported into the Database for Annotation, Visualization, and Integrated Discovery (DAVID) (https://david.ncifcrf.gov/) for Gene Ontology (GO) (http://geneontology.org/) and Kyoto Encyclopedia of Genes and Genomes (KEGG)9 (https://www.kegg.jp/) pathway enrichment analysis, we screened for items exceeding an Enrichment Score of 2, categorized them by analysis type and sorted them by -lgP value. Both analysis results were then visualized using Excel and ImageGP platforms, respectively. Finally, we further analyzed RA-related pathway targets by combining these screening results with relevant prior literature.Molecular dockingTo validate the binding activity of GEN with potential RA-related targets, we performed MOD simulations. The GEN 2D structure (Fig. 2) was retrieved by accessing PubChem (https://pubchem.ncbi.nlm.nih.gov/) and optimized using Chem3D software. AutoDock Vina (version 1.1.2) was employed to prepare the rotatable bonds (Torsion Tree) for docking. Suitable target receptor proteins were identified and downloaded from the Protein Data Bank (PDB) database (https://www.rcsb.org/). Using PyMOL software (version 2.5.0), both solvent and organic molecules were eliminated from the protein structures, duplicates were eliminated, and the structures were then prepared in AutoDock Vina by adding hydrogens, assigning charges, and defining atom types. Grid Box coordinates for the docking simulations were determined based on relevant target receptor literature, and the spatial size of the binding region was adjusted. Finally, AutoDock Vina was deployed to perform the MOD simulations between GEN and the identified RA-related targets. The combination visualized figure between the specific chemical and the target was obtained from PLIP website (https://projects.biotec.tu-dresden.de/plip-web/plip/index)10.Fig. 2Structure of GEN.Full size imageExperimental animals and groupsA total of 30 male SPF Sprague-Dawley (SD) rats (5 weeks, Spearfish Biotechnology Co., Ltd.) were acclimation at 25 ± 1 °C, 45–55% humidity, 12-h light/dark cycle for one week. Ethical approval for the study was granted by the Animal Ethics Committee of Nanjing University of Chinese Medicine (Approval number: 202211 A005) and was performed in conformity to the Guide for the Care and Use of Laboratory Animals. Thereafter, the rats were allocated at random into five groups (n = 6): Control, Model, Methotrexate (MTX), GEN, and High-mobility group box 1 antibody (anti-HMGB1). Except for the Control group, all other groups were exposed to collagen-induced arthritis (CIA) modeling by subcutaneously injecting 0.15 mL of CII emulsion into five sites on each rat: the metatarsal bone of the hindfoot, the neck, the back, and the root of the tail. A booster immunization with the same dosage of CII emulsion was administered 14 days later to establish the CIA rat model. We quantified the CIA rat model using a combined approach of arthritic scoring and joint width measurements. Arthritic scoring was performed based on erythema, swelling, and joint rigidity (0–4 scale per paw)11, while joint width measurements utilized high-precision digital calipers (accuracy ± 0.01 mm)12. Rats with an arthritic score of less than 4 following secondary immunization on Day 14 were excluded from subsequent experiments. Bilateral ankle joint widths (left and right) were recorded daily using digital calipers (SYNTEK, China, JS20 F150) to ensure precision. On the second day post-modeling, rats in each group received daily oral gavage with the following dosages: MTX group (2 mg/kg/day, Meilunbio, China, MB1156), GEN group (20 mg/kg/day, Yongjian Pharmaceutical Technology Co., LTD, China,102342). The Control, Model, and anti-HMGB1 groups administered a daily equivalent volume of distilled water. Additionally, rats in the anti-HMGB1 group received intraperitoneal injection of anti-HMGB1 (100 µg/kg/day, Bioss, Shanghai, China, bs-0064R), while the Control, Model, MTX, and GEN groups received intraperitoneal injections of an equal volume of distilled water. On the 38 th day post-modeling, animals were anesthetized using isoflurane, thereby collecting blood from the abdominal aorta. Subsequently, knee joints were harvested and fixed in a fixative solution (20 times the sample volume) for over 24 h to prepare tissue sections for embedding and paraffin sectioning. Then, we euthanized the rats by intraperitoneal injection of sodium pentobarbital (150 mg/kg).ELISA assaySerum hypoxia-inducible factor-1α (HIF-1α), stromal cell-derived factor-1α (SDF-1α/chemokine C-X-C motif ligand 12(CXCL12)), vascular endothelial growth factor-A (VEGF-A), and angiopoietin-2 (Ang-2) levels were quantified using ELISA kits (Rat HIF-1α ELISA kit, Jianglai Biotechnological Co., Ltd, Shanghai, China, JL20959; Rat SDF-1α ELISA kit, Enzyme Immunity Industry Co., Jiangsu, China, MM-21265R2; Rat VEGF-A ELISA kit, EK-Bioscience, Shanghai, China, EK-R31184; Rat ANG-2 ELISA kit, EK-Bioscience, Shanghai, China, EK-R31191) according to the manufacturer’ s protocol. Absorbance was measured at 450 nm using a microplate reader (SpectraMax, USA, iD3), and sample concentrations were calculated from a standard curve. All assays were repeated three times independently to ensure reproducibility.Histopathological examination (HE)Tissue specimens were fixed in 10% neutral buffered formalin, decalcified, and subsequently processed through standard paraffin embedding and sectioning protocols. HE staining was performed according to established histological procedures. Under a light microscope, pathological conditions were examined and the severity of arthritis was evaluated based on four parameters according to validated grading criteria13: synovial inflammation, bone erosion, proteoglycan loss and cartilage erosion. Each parameter was scored on a scale of 0–3 points.Immunohistochemical (IHC) stainingParaffin-embedded knee tissue sections were subjected to a standard IHC protocol, including dewaxing, antigen retrieval, endogenous peroxidase blockage, serum blockage, primary and secondary antibody incubations, DAB color development, nuclear counterstaining, and dehydration. The primary antibody included CD31 (1:200 dilution, Zhongsui Jinqiao Biotechnological Co., Ltd, China, ZA-0568), and the secondary antibody included Goat Anti-Rabbit IgG Secondary Antibody (HRP, 1:1500 dilution, Sino. BIO, China, SSA004). The IHC-stained slides were examined under a 20× microscope. ImagePro Plus 6 software was employed to quantify the expression of CD31-positive cells. The positive rate was calculated as the ratio of the integrated optical density (IOD) of positive staining to the total area of the section. Three specimens per group were analyzed, with individual positive rates calculated and averaged.Cell culture and groupsThree male Wistar rats (8 weeks, Spearfish Biotechnology Co., Ltd.) were exposed to anesthesia with an intraperitoneal sodium pentobarbital injection. The humerus from the upper limb and the femur and tibia from the lower limbs were harvested. Bone marrow cells were flushed from the cavities of the humerus, femur, and tibia using an EBM-2 medium (3–5 washes per bone). We euthanized the rats by intraperitoneal injection of sodium pentobarbital (150 mg/kg). Subsequently, the cells were centrifuged with Ficoll cell isolation solution to acquire endothelial progenitor mononuclear cells, which were then resuspended in EBM-2 medium (Lonza, USA, CC-3156) that contained 5% fetal bovine serum (Gibco, USA, 10091148) and inoculated into 25 cm² culture flasks. The cultures were incubated at 37 °C in a 5% CO₂ for one week. This was followed by harvesting and staining the cells with CD31 (TLD-3 A12) and CD34 (Both from Invitrogen, USA, PA5-78978) antibodies. Flow cytometry was deployed to analyze the positive cell proportion in each group, with samples protected from light during incubation and staining.Endothelial progenitor cells (EPCs) were resuspended in EBM-2 basal medium, seeded into 96-well plates, and assigned into six groups (n = 3): Control, Model, GEN (20 µM), GEN (40 µM), YC-1, and AMD3100. Except for the Control, EPCs in each group were treated with GEN (20 µM, Jiangsu Yongjian Pharmaceutical Technology Co., China, 102342), GEN (40 µM, Jiangsu Yongjian Pharmaceutical Technology Co., China, 102342), YC-1 (50 µM, HIF-1α inhibitor, MCE, USA, HY-14927), and AMD3100 (5 µM, C-X-C chemokine receptor type 4(CXCR4) inhibitor, Selleck, Germany, S8030) for 30 min14,15,16,17, followed by adding HMGB1 (100ng/mL, MCE, USA, HY-P73104) for 24 and 48 h of incubation.Cell proliferation ability assayCell proliferation was determined through the MTT Cell Proliferation and Cytotoxicity Assay Kit (Biyuntian Biotechnology Co., China, C0009S-500 T). Cell viability was calculated as follows:$$\:Cell\:viability\:\left(\%\right)\:=\:\left[\right(A\left(dosing\right)\:-\:A\left(blank\right))\:/\:(A\left(0\:dosing\right)\:-\:A\left(blank\right)\left)\right]\:\times\:\:100$$.where, \(\:A\left(dosing\right)\) reflects the absorbance of wells containing cells, MTT solution, and drug solution; \(\:A\left(blank\right)\) refers to the absorbance of wells containing medium and MTT solution but no cells; and \(\:A\:\left(0\:dosing\right)\) presents the absorbance of wells containing cells, MTT solution, but no drug solution.Transwell migration assayThe EPCs were prepared as a cell suspension in an EBM-2 basal medium (Lonza, USA, CC-3156) and seeded into the upper chamber of 24-well Transwell inserts. The lower chamber was filled with either blank or drug-containing EGM-2 medium, followed by incubating the cells at 37 °C for 12 h. Subsequently, cells were subjected to fixation with 4% paraformaldehyde, air-drying, and staining with 0.1% crystal violet. Non-migrated cells on the membrane’s upper surface were gently wiped with a cotton swab. Five fields of view were chosen at random and imaged under 200× and 400× microscopes to count the number of cells that had migrated through the membrane.Tubular structure formation capability assayA 24-well plate was filled with 300 µL of Matrigel per well and incubated to allow gel formation. Each group was seeded with 1 × 105 cells/well into the 24-well plate matrix gel. Following 0, 24, 48, and 72 h, cells were examined under a 40× microscope, and representative images were captured to document EPC tubular structure formation. The number and total length (µm) of EPC tubular nodes were quantified using the NIH ImageJ software angiogenesis analysis plugin (version 1.53 t, https://imagej.nih.gov/ij/)18.Western blot analysisCells or tissues from different treatment groups were lysed to extract total proteins. Following the separation by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), we transferred the proteins to a polyvinylidene difluoride (PVDF) membrane, blocked them for one hour, added the following primary antibodies against HIF-1α (Cell Signaling Technology, USA, #14179), Ang2 (Abcam, USA, ab155106), VEGF-A (Abcam, USA, ab214424), CXCL12 (Novus Biologicals, USA, NBP2-29480), GAPDH (Tianjin UAB, China, UM4001), β-actin (Tianjin UAB, China, UM4002), and CXCR4 (Abcam, USA, ab124824), and allowed them to incubat overnight at 4 °C. The membrane was rinsed three times with phosphate-buffered saline with Tween-20. Subsequently, horseradish peroxidase (HRP)-conjugated goat anti-mouse IgG (S00001) and anti-rabbit IgG (Both from Affinity, USA, S00002) were added and incubated for one hour. The membrane was developed by an enhanced chemiluminescence (ECL) reagent (Biyuntian Technology Co., China, P0018 FM) and visualized using a chemiluminescence imaging system (Shanghai Tianneng, China, 5200). To quantify protein expression levels, the band’s gray value was measured using GelPro32 software. Three biological replicates per group were subjected to ensure reproducibility.Statistical analysisStatistical analyses were conducted through SPSS 27.0 software, reporting the data as mean ± standard deviation (\(\:\stackrel{-}{x}\pm\:S\)). One-way ANOVA and the chi-square test were employed for comparisons between groups. For chi-square tests, the LSD method was used. For variances that did not meet the assumption of homogeneity, Tamhane’s T2 test and independent samples nonparametric tests were utilized. p < 0.05 indicated statistical significance.ResultsPotential GEN targets on RAHere, we obtained 120 potential targets of GEN from the TCMSP database. Through the Genecards and Disgenet databases, we identified 6,145 disease targets associated with RA. By intersecting the two datasets, we identified 64 shared targets, suggesting these as potential targets for RA treatment with GEN (Fig. 3A).Fig. 3Network pharmacological analyses of GEN on RA. (A) Venn diagram: GEN and RA targets. (B) PPI network (64 nodes and 176 edges) of GEN ingredient targets against RA, which was produced using STRING. (C) The topological parameters of the major target proteins. (D) GO enrichment analysis of the key targets. Different colors indicate different KEGG pathway classifications. (E) KEGG pathway analysis of intersection targets. The numbers on the X-axis indicate the number of target genes in the enrichment analysis, the bubble size and the red depth indicate its significance.Full size imagePPI network analysis64 intersecting targets were imported into the STRING, thereby generating a PPI network graph (Fig. 3B) comprising 64 nodes and 176 edges. By filtering targets with a degree value exceeding twice the average (5.5*2 = 11), 23 significant targets were identified. Figure 3C illustrates the topological parameters of the major target proteins, ranked by degree size.GO and KEGG analysis combined with literature reviewThe GO analysis of GEN’s potential targets in RA revealed involvement in various biological processes: cellular response to lipopolysaccharide, negative regulation of apoptotic process, signal transduction, response to xenobiotic stimulus, inflammatory response, positive regulation of cell population proliferation, positive regulation of MAPK cascade, and immune response. Molecular functions associated with these targets included identical protein binding, enzyme binding, cytokine activity, growth factor activity, and protein kinase activity. While cellular component associated with these targets included cytoplasm, nucleus, protein-containing complex, mitochondrion, chromatin, and endoplasmic reticulum membrane. (Fig. 3D).The KEGG analysis of potential GEN targets in RA identified several significantly enriched pathways, including salmonella infection, AGE-RAGE signaling pathway in diabetic complications, NOD like receptor signaling pathway, and IL-17 signaling pathway (Fig. 3E). Based on ranking by − log10(p-value), RA and the NF-κB signaling pathway were positioned 69 th and 54 th, respectively.On the basis of GO and KEGG analyses, combined with relevant literature, neovascularization emerged as a critical hallmark in the pathological progression of RA. In the “inflammatory hypoxic” environment of RA joints, a plethora of pro-angiogenic factors, including growth factors, matrix remodeling enzymes, and chemokines, activate endothelial cells and induce neovascularization19. This abnormal neovascularization, characterized by its immaturity, irregular morphology, and inadequate perfusion, creates a paradoxical situation where increased vascular density coexists with hypoxia in the RA synovium, perpetuating synovial inflammation20,21. The ability of existing RA therapeutic agents to target this crucial aspect of abnormal neovascularization remains unclear.Further analysis indicates that in RA-affected joints, characterized by an inflammatory hypoxic environment, the production of cytokines can stimulate reactive oxygen species (ROS) production, activating the MAPK and/or NF-κB pathways to promote intracellular signaling and subsequent hypoxia-inducible factor (HIF) production22. Among these factors, HMGB1, associated with the NF-κB pathway, emerges as a critical pro-inflammatory and pro-neovascular mediator, promoting HIF-1 production and thereby inducing neovascularization23,24,25. The elevated production of HIF-1 can further contribute to neovascularization through the CXCL12/CXCR4 axis26,27. Consequently, we hypothesize that there may be a significant association between aberrant neovascularization and the NF-κB-related pathway, as well as the CXCL12/CXCR4 axis, in the pathological progression of RA.Molecular dockingTo validate the molecular targets of GEN within the NF-κB pathway, MOD simulations were performed with CXCL12 and CXCR4. The calculated binding energies for GEN with CXCL12 and CXCR4 were − 5.9 and − 8.0 kcal/mol, respectively, indicating strong binding affinities between GEN and these target proteins (Fig. 4A-B).Fig. 4Molecular docking of GEN with CXCL12 (A) and CXCR4 (B) shown as 3D diagrams. Molecular docking pattern diagram: target protein in blue, compound in red. Binding site schematic: target protein fragments in blue, compounds in orange, labeled with amino acid-protein sequence sites presenting binding sites.Full size imageAnimal experimentCIA rats modelAt 3, 8, 14, and 18 days post-CIA modeling, the rats’ body weight was significantly lower in the Model group than in the Control (p < 0.05). Conversely, the rats’ body weight was significantly higher in the anti-HMGB1 group than in the Model group (p < 0.05). Additionally, the rats’ body weight was significantly higher in the MTX group than in the Model group at 3, 8, and 14 days post-modeling (p < 0.05 or p < 0.01). Similarly, the rats’ body weight was significantly higher in the GEN group than in the Model group at 14 days post-modeling (p < 0.05). However, at 24, 30, and 37 days post-modeling, no significant differences were observed in body weight between any two groups (Fig. 5A).Fig. 5GEN improves physical and biochemical indicators in the rat model of RA. (A) Weight of post-CIA modeling. (B) Swelling degree of the CIA rats’ left ankle. (C) Swelling degree of the CIA rats’ right ankle. (D) The CD31-positive rate via IHC staining. (E) IHC staining (20×). (F) Serum content of angiogenesis-associated factors. (G) Grayscale values of angiogenesis-related factor proteins. (H)The expression of angiogenesis-related factors.Full size imageAs the modeling time progressed, the degree of swelling in both the right and left ankles of rats across all groups exhibited an increasing trend. Unlike the Control, the left and right ankles of rats in the Model, MTX, GEN, and anti-HMGB1 groups displayed significantly increased swelling (p < 0.05 or p < 0.01). At 14 and 18 days post-modeling, the swelling of the right ankle in the MTX, GEN, and anti-HMGB1 groups was significantly diminished in comparison to the Model group (p < 0.05 or p < 0.01, Fig. 5B-C). (Figure S1, S2)HE stainingHistopathological evaluation confirmed the inhibitory effect of GEN on CIA rats (Figure S3, S4). Compared with the Control group, the Model group exhibited marked inflammatory cell infiltration, synovial hyperplasia, pannus formation, and partial cartilage/bone erosion. In contrast, CIA rats treated with MTX, GEN, or anti-HMGB1 demonstrated significant attenuation of synovial inflammatory cell infiltration and hyperplasia compared to the Model group. Histopathological scoring demonstrated that the GEN, MTX, and anti-HMGB1 groups all exhibited significant reductions in scores compared to the Model group (n = 3, p < 0.05). The GEN and MTX groups demonstrated comparable effects on histopathological alterations.IHC stainingThe CD31-positive rate displayed significantly higher in the knee joint sections of rats in the Model, MTX, and GEN groups than in the Control (p < 0.05 or p < 0.01). The CD31-positive cells (tan-brown) showed a comparable rate in the anti-HMGB1 group with that of the Control while showing significantly lower rate in the MTX, GEN, and anti-HMGB1 groups than in the Model group (p < 0.05 or p < 0.01, Fig. 5D-E).ELISA analysisELISA analysis revealed significantly overexpressed HIF-1α, CXCL12, VEGF-A, and Ang-2 in the Model, MTX, GEN, and anti-HMGB1 groups compared to the Control (p < 0.01). Unlike the Model group, the MTX, GEN, and anti-HMGB1 groups exhibited significantly decreased HIF-1α, CXCL12, VEGF-A, and Ang-2 levels (for all p < 0.01, Fig. 5F).Western blot detect Angiogenesis-related factor expressionIn comparison to the Control, the Model group exhibited significantly upregulated HIF-1α, CXCL12, VEGF-A, and Ang-2 proteins (p < 0.05 or p < 0.01). Compared to the Model group, the MTX, GEN, and anti-HMGB1 groups displayed significantly reduced gray values for HIF-1α, CXCL12, and VEGF-A proteins (p < 0.05 or p < 0.01). Furthermore, the GEN and anti-HMGB1 groups showed significantly reduced gray values for Ang-2 protein in contrast to the Model group (p < 0.05), with no significant difference between the MTX and Model groups (Fig. 5G-H).Cell experimentEPCs identificationAfter 7 days of cell culture, flow cytometry analysis revealed a positive cell rate of 63.03% and 36.25% for the specific antigens CD31/34, respectively (Fig. 6A).Fig. 6GEN may inhibit the aberrant neovascularization process of EPCs induced by HMGB1, likely through targeting CXCL12 and CXCR4. (A) The EPCs isolated from Wistar rats were confirmed by flow cytometry through the detection of specific antigen CD31/34 expression. (B) The proliferation ability of EPCs at 24 and 48 h. (C) The migratory capacity of EPCs at 24 and 48 h. (D) Migration of EPCs by microscopic observation at 200× and 400× magnification (Red arrows represent EPCs). (E) The number of tubular nodes of EPCs. (F) Total cellular tubular length of EPCs. (G) Observation of the formation of tubular structures in EPCs at 0, 24, 48, and 72 h (Black arrows represent tubular structures of EPCs). (H) The gray value of CXCR4 and CXCL12 in EPCs. (I) Western Blot detect CXCR4 and CXCL12 expression in EPCs. Data are expressed as mean ± SD, n = 3. * P < 0.05 and ** P < 0.01, compared with Control group. # P < 0.05 and ## P < 0.01, compared with Model group.Full size imageCell proliferation assayAt both 24 and 48 h, the EPC proliferation capacity was significantly higher in the Model group than in the Control, with the 48-h difference being statistically significant (p < 0.05). Unlike the Model group, the 24-h YC-1 and AMD3100-treated EPC proliferation ability were significantly lower (p < 0.05), and while the EPC proliferation ability in the GEN (20 µM) and GEN (40 µM) groups was also lower, the difference was not statistically significant. Post-48 h treatment, the EPC proliferation ability in the GEN (20 µM), GEN (40 µM), YC-1, and AMD3100 groups was significantly lower than in the Model group (p < 0.01, Fig. 6B).Transwell migration assayAt 48 h, the migration ability of EPCs in each group was observed with both 200× and 400× microscopic magnifications. Under these observation conditions, the Model group exhibited significantly enhanced EPC migration compared to the Control group (p < 0.01). The number of EPCs migrating across the membrane in the Model group was 1.4- or 1.51-fold higher than that in the Control at 200× and 400× magnification, respectively. In contrast, the EPC migration ability in the GEN (20 µM), GEN (40 µM), YC-1, and AMD3100 groups was significantly lower than in the Model group (p < 0.01). Among these groups, the GEN (40 µM) and YC-1 groups exhibited the most pronounced reduction in EPC migration, with a 0.47/0.25- or 0.53/0.41-fold decrease compared to the Control at 200× and 400× magnification, respectively. The number of EPCs migrating across the membrane was 0.86- or 0.89-fold higher than that of the Control in the GEN (20 µM) group at 200× and 400× magnification, respectively, and 0.67- or 0.71-fold higher in the AMD3100 group at 200× and 400× magnification, respectively. (Fig. 6C-D)Tubular structure formation assayThe formation of tubular structures by EPCs was observed microscopically. The number of tubular nodes of EPCs in the Model group significantly escalated at 0, 48, and 72 h in contrast to the Control (p < 0.05 or p < 0.01), though no intergroup difference reached significance at 24 h. The total tubular length of EPCs in the Model group significantly differed from the Control at 24 and 72 h (p < 0.01), though no intergroup difference reached significance at 0 and 48 h. Unlike the Model group, at 24, 48, and 72 h, the number of cellular tubular nodes and total cellular tubular length of EPCs were significantly lower in the GEN (20 µM), GEN (40 µM), YC-1, and AMD3100 groups (p < 0.01, Fig. 6E-G).Western blot detect CXCR4 and CXCL12 expression in EPCsUnlike the Control, the grayscale values of CXCL12 and CXCR4 protein expression were significantly upregulated in the Model group (p < 0.01 or p < 0.05). Compared to the Model group, the grayscale value of CXCL12 protein expression was significantly decreased in the GEN (40 µM) and YC-1 groups (p < 0.05 and p < 0.01, respectively), with no significant difference between the GEN (20 µM) and AMD3100 compared with Model groups. Consistently, compared to the Model group, the grayscale value of CXCR4 protein expression was significantly diminished in the GEN (20 µM), GEN (40 µM), YC-1, and AMD3100 groups (p < 0.05). (Fig. 6H-I)DiscussionTypically, RA constitutes a chronic and progressive autoimmune disease defined by neovascularization, a crucial factor in the formation of vascular opacities and a hallmark of disease progression. Abnormal neovascularization within RA synovial tissues induces infiltration of numerous immune cells into “inflammatory hypoxic” joints, thereby promoting synovial inflammation and contributing to cartilage and bone damage28. In these “inflammatory hypoxic” joints, a plethora of pro-angiogenic factors, including growth factors, matrix remodeling enzymes, and chemokines, activate endothelial cells, playing a pivotal role in inducing abnormal neovascularization29. The HMGB1, a potent pro-inflammatory and pro-neovascular mediator, promotes aberrant neovascularization by inducing HIF-1α30. The SDF-1, also known as CXCL12, functions to promote endothelial cell migration and tube formation, serving as a critical stromal cell-derived factor required for neovascularization31. The CXCR4 is a specific receptor for CXCL12. Upon binding to CXCR4, CXCL12 activates JAK/STAT, PI3 K/Akt, and ERK1/2 signalings, inducing NF-κB activation and further recruiting numerous pro-inflammatory cytokines, exacerbating “inflammatory hypoxia”. Within this microenvironment, the production of abundant HIF-1α can promote neovascularization through the CXCL12/CXCR4 axis32, further contributing to the pathologic progression of rheumatoid arthritis.The TCM has a history spanning several millennia, and Chinese herbs have been widely used in clinical practice. However, because of the limited reliable scientific evidence, the definitive role of TCM in treating RA remains unclear. The complex nature of TCM, with its numerous components and targets of action, poses challenges in elucidating its mechanism of action. Consequently, the study of active ingredients from Chinese herbs has emerged as a significant area of research. Recently, NP has been increasingly recognized as a rapid and efficacious method for analyzing the effectiveness of Chinese herb components33,34. Gentiana scabra, or Qinjiao, is a commonly used Chinese herb for RA. Its active ingredient, GEN, is believed to possess anti-RA effects, such as inhibiting inflammatory responses and suppressing osteoclastogenesis. Herein, we integrated NP with in vitro and in vivo experiments to investigate the mechanism behind GEN’s therapeutic effect on RA. By analyzing publicly available databases, we predicted the interactions between GEN and its potential targets in RA, as well as the involved signaling pathways and networks. Fifty-one intersecting targets were identified, and 24 potential targets were selected based on the p-value of each enriched pathway and its relevance to RA. The KEGG analysis of the core PPI network revealed that GEN exerts multiple effects on RA treatment by modulating potential targets associated with various pathways, including the NF-κB signaling pathway. Further analysis, combined with a literature review, led us to focus on CXCL12 and CXCR4, which are related to the NF-κB signaling pathway. The MOD simulations indicated that GEN could intervene in RA by regulating CXCL12 and CXCR4. To further investigate whether GEN can inhibit abnormal neovascularization in RA through CXCL12 and CXCR4, we conducted animal and cell experiments.In our study, GEN effects on RA were ascertained using a CIA rat model, manifesting that GEN reduced collagen-induced foot swelling in rats and repressed the angiogenesis-related factor levels, including HIF-1α, CXCL12, VEGF-A, and Ang-2, both in peripheral serum and in the expression of proteins within synovial tissues of CIA rats. It has been well-established that HIF-1α is highly expressed in RA synovial tissues, and its levels correlate with the number of blood vessels, endothelial cell proliferation, synovitis scores, and RA disease activity35. A study on anti-VEGF therapy in tumor cells revealed an “escape pathway” for neovascularization associated with the induction of endothelial cells by CXCL1236. Additionally, CXCL12 has been shown to play a significant synergistic role with classical VEGF-A in the involvement of EPCs in angiogenesis37. Furthermore, CXCL12 promotes the proliferation of fibroblast-like synoviocytes, induces matrix metalloproteinase activity, and increases the expression of osteoclast differentiation factor, leading to bone destruction in RA38. Our study demonstrated that GEN exhibits a favorable inhibitory effect on angiogenesis-related factors, including HIF-1α, CXCL12, VEGF-A, and Ang-2.Neovascularization predominantly occurs in hypoxic environments. Guided by angiogenic factors, neovascular cells migrate toward hypoxic regions, proliferate, and recruit EPCs to establish the neovascular lining18. To further investigate these processes, we isolated and cultured EPCs in vitro and analyzed and compared different drug intervention groups. Our experimental results demonstrated that GEN impeded the HMGB1-provoked EPC proliferation, migration, and tubular structure formation ability. Additionally, GEN decreased the expression of CXCL12 and CXCR4 proteins in EPCs. The CXCR4 is the most abundant chemokine receptor in endothelial cells, and its downstream signaling, including PI3 K/AKT, NF-κB, and JAK/STAT, play crucial roles in processes such as angiogenesis and inflammation39. Notably, CXCR4 is abundantly expressed in RA serum and synovial fluid40. Moreover, T cell entry into the synovium can be regulated by CXCR441, the CXCR4/CXCL12 axis regulates B cell migration into the RA synovium42, and this axis also exhibits angiogenic activity in arthritic symptoms in a CIA model43. Altogether, CXCR4 is significant in RA pathogenesis. Our study further confirmed the inhibitory effect of GEN on the aberrant neovascularization process of EPCs induced by HMGB1, likely through targeting CXCL12 and CXCR4.The present study, combining NP with animal and cellular experiments, confirms GEN’s favorable inhibitory effect on abnormal RA neovascularization. However, our study has certain limitations, including an oversimplified molecular mechanism and a relatively limited scope of experiments. For instance, in our animal experiment, the intervention concentration of GEN was selected based on previous literatures. However, no additional experimental groups with varying concentrations of GEN were established for further observation and analysis. The observation time points in cell experiments could be further extended for more comprehensive analysis. Additionally, due to current experimental constraints, further knockout or overexpression experiments targeting CXCL12/CXCR4 were not performed to elucidate the underlying mechanisms of GEN. Future research is necessary to further support our outcomes, offer additional experimental evidence, and expand the investigation into the detailed mechanism of action behind GEN on RA.ConclusionsIn this study, we analyzed the targets of action and signaling pathways of GEN in RA using network pharmacology and validated them through molecular docking and experimental approaches. Our findings indicate that GEN demonstrates a significant effect in inhibiting abnormal neovascularization and can regulate CXCL12 and CXCR4 to intervene in the pathological progression of RA. This study provides a novel approach for the treatment of RA and a theoretical foundation for further research.Data availabilityThe datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.ReferencesWu, D. et al. Systemic complications of rheumatoid arthritis: focus on pathogenesis and treatment. Front. Immunol. 13, 1051082 (2022).Article CAS PubMed PubMed Central Google Scholar Giannini, D. et al. One year in review 2020: pathogenesis of rheumatoid arthritis. Clin. Exp. Rheumatol. 38(3), 387–397 (2020).PubMed Google Scholar Aureal, M., Machuca-Gayet, I. & Coury, F. Rheumatoid arthritis in the view of osteoimmunology. Biomolecules 11(1) (2020).Zhou, S., Zou, H., Chen, G. & Huang, G. Synthesis and biological activities of chemical drugs for the treatment of rheumatoid arthritis. Top. Curr. Chem. (Cham). 377(5), 28 (2019).Article PubMed Google Scholar Melville, A. R., Kearsley-Fleet, L., Buch, M. H. & Hyrich, K. L. Understanding refractory rheumatoid arthritis: implications for a therapeutic approach. Drugs 80(9), 849–857 (2020).Article PubMed Google Scholar de Hair, M., Jacobs, J., Schoneveld, J. & van Laar, J. M. Difficult-to-treat rheumatoid arthritis: an area of unmet clinical need. Rheumatol. (Oxford). 57(7), 1135–1144 (2018).Google Scholar Xie, X. et al. Therapeutic effects of Gentiopicroside on adjuvant-induced arthritis by inhibiting inflammation and oxidative stress in rats. Int. Immunopharmacol. 76, 105840 (2019).Article CAS PubMed Google Scholar Chen, F. et al. Gentiopicroside inhibits RANKL-induced osteoclastogenesis by regulating NF-kappa B and JNK signaling pathways. Biomed. Pharmacother. 100, 142–146 (2018).Article CAS PubMed Google Scholar Minoru, K., Miho, F., Yoko, S., Yuriko, M. & Mari, I. W. KEGG: biological systems database as a model of the real world. Nucleic Acids Res. 53(D1), D672–D677 (2025).Article Google Scholar Melissa, F. A. et al. PLIP 2021: expanding the scope of the protein-ligand interaction profiler to DNA and RNA. Nucleic Acids Res. 49(W1), W530–W534 (2021).Article Google Scholar Khachigian, L. M. Collagen antibody-induced arthritis. Nat. Protoc. 1(5), 2512–2516 (2006).Article CAS PubMed Google Scholar Pietrosimone, K. M., Jin, M., Poston, B. & Liu, P. Collagen-Induced arthritis: A model for murine autoimmune arthritis. Bio Protoc 5(20) (2015).Hayer, S. et al. SMASH’ recommendations for standardised microscopic arthritis scoring of histological sections from inflammatory arthritis animal models. Ann. Rheum. Dis. 80(6), 714–726 (2021).Article CAS PubMed Google Scholar Zhang, N. et al. Gentiopicrin exerts anti-rheumatic effect in human fibroblast-like synoviocytes via Inhibition of p38MAPK/NF-κB pathway. Cell. Mol. Biol. 65(6), 85–90 (2019).Article PubMed Google Scholar Chen, F. Q. et al. Gentiopicroside inhibits RANKL-induced osteoclastogenesis by regulating NF-κB and JNK signaling pathways. Biomed. Pharmacother. 100, 142–146 (2018).Article CAS PubMed Google Scholar Yang, L. et al. Regulatory effect of Astragaloside IV on SDF-1α/CXCR4 in EPCs. Chin. J. Pathophysiology. 35(9), 1587–1593 (2019).Google Scholar Liu, Y. et al. Study of Astragaloside IV promoting migration of high glucose (HGS)-impaired human endothelial progenitor cells via SDF-1α/CXCR4 axis. CJTCMP 37(8), 4714–4719 (2022).CAS Google Scholar Caroline, A. S., Wayne, S. R. & Kevin, W. E. NIH image to imageJ: 25 years of image analysis. Nat. Methods. 9, 671–675 (2012).Article Google Scholar Elshabrawy, H. A. et al. The pathogenic role of angiogenesis in rheumatoid arthritis. Angiogenesis 18(4), 433–448 (2015).Article CAS PubMed PubMed Central Google Scholar Quinonez-Flores, C. M., Gonzalez-Chavez, S. A. & Pacheco-Tena, C. Hypoxia and its implications in rheumatoid arthritis. J. Biomed. Sci. 23(1), 62 (2016).Article PubMed PubMed Central Google Scholar Fu, H., Hu, D., Zhang, L. & Tang, P. Role of extracellular vesicles in rheumatoid arthritis. Mol. Immunol. 93, 125–132 (2018).Article CAS PubMed Google Scholar Deng, W., Feng, X., Li, X., Wang, D. & Sun, L. Hypoxia-inducible factor 1 in autoimmune diseases. Cell. Immunol. 303, 7–15 (2016).Article CAS PubMed Google Scholar Kong, L. et al. Inhibition of hypoxia inducible factor 1 by YC-1 attenuates tissue plasminogen activator induced hemorrhagic transformation by suppressing HMGB1/TLR4/NF-kappa B mediated neutrophil infiltration in thromboembolic stroke rats. Int. Immunopharmacol. 94, 107507 (2021).Article CAS PubMed Google Scholar Peng, T., Du, S. Y., Son, M. & Diamond, B. HIF-1alpha is a negative regulator of interferon regulatory factors: implications for interferon production by hypoxic monocytes. Proc Natl. Acad. Sci. U S A 118(26) (2021).Park, S. Y. et al. HMGB1 induces angiogenesis in rheumatoid arthritis via HIF-1alpha activation. Eur. J. Immunol. 45(4), 1216–1227 (2015).Article CAS PubMed Google Scholar Zhang, Y. et al. GDF11 promotes wound healing in diabetic mice via stimulating HIF-1a-VEGF/SDF-1a-mediated endothelial progenitor cell mobilization and neovascularization. Acta Pharmacol. Sin. 44(5), 999–1013 (2023).Article CAS PubMed Google Scholar Korbecki, J. et al. The effect of hypoxia on the expression of CXC chemokines and CXC chemokine Receptors-A review of literature. Int J. Mol. Sci 22(2) (2023).Villalvilla, A., Gomez, R., Roman-Blas, J. A., Largo, R. & Herrero-Beaumont, G. SDF-1 signaling: a promising target in rheumatic diseases. Expert Opin. Ther. Targets. 18(9), 1077–1087 (2014).Article CAS PubMed Google Scholar Noort, A. R. et al. NF-kappaB-inducing kinase is a key regulator of inflammation-induced and tumour-associated angiogenesis. J. Pathol. 234(3), 375–385 (2014).Article CAS PubMed PubMed Central Google Scholar Liu, Y. et al. Knockdown of HIF-1alpha impairs post-ischemic vascular reconstruction in the brain via deficient homing and sprouting BmEPCs. Brain Pathol. 28(6), 860–874 (2018).Article CAS PubMed PubMed Central Google Scholar Asquith, D. L., Bryce, S. A. & Nibbs, R. J. Targeting cell migration in rheumatoid arthritis. Curr. Opin. Rheumatol. 27(2), 204–211 (2015).Article CAS PubMed Google Scholar Devignes, C. S. et al. HIF signaling in osteoblast-lineage cells promotes systemic breast cancer growth and metastasis in mice. Proc. Natl. Acad. Sci. U S A. 115(5), E992–E1001 (2018).Article CAS PubMed PubMed Central Google Scholar Zhou, C. et al. A systems Biology-Based approach to Uncovering molecular mechanisms underlying effects of traditional Chinese medicine Qingdai in chronic myelogenous leukemia, involving integration of network Pharmacology and molecular Docking technology. Med. Sci. Monit. 24, 4305–4316 (2018).Article CAS PubMed PubMed Central Google Scholar Zhang, Y. F., Huang, Y., Ni, Y. H. & Xu, Z. M. Systematic Elucidation of the mechanism of geraniol via network Pharmacology. Drug Des. Devel Ther. 13, 1069–1075 (2019).Article CAS PubMed PubMed Central Google Scholar Paradowska-Gorycka, A. et al. HIF-1A gene polymorphisms and its protein level in patients with rheumatoid arthritis: a case-control study. Inflamm. Res. 67(5), 423–433 (2018).Article CAS PubMed Google Scholar Ziegler, M. E., Hatch, M. M., Wu, N., Muawad, S. A. & Hughes, C. C. mTORC2 mediates CXCL12-induced angiogenesis. Angiogenesis 19(3), 359–371 (2016).Article CAS PubMed PubMed Central Google Scholar Odent, G. G. et al. Synergic effects of VEGF-A and SDF-1 on the angiogenic properties of endothelial progenitor cells. J. Tissue Eng. Regen Med. 11(11), 3241–3252 (2017).Article CAS Google Scholar Kim, H. R. et al. Reciprocal activation of CD4 + T cells and synovial fibroblasts by stromal cell-derived factor 1 promotes RANKL expression and osteoclastogenesis in rheumatoid arthritis. Arthritis Rheumatol. 66(3), 538–548 (2014).Article CAS PubMed Google Scholar Pozzobon, T., Goldoni, G., Viola, A. & Molon, B. CXCR4 signaling in health and disease. Immunol. Lett. 177, 6–15 (2016).Article CAS PubMed Google Scholar Peng, L. et al. Expression levels of CXCR4 and CXCL12 in patients with rheumatoid arthritis and its correlation with disease activity. Exp. Ther. Med. 20(3), 1925–1934 (2020).MathSciNet CAS PubMed PubMed Central Google Scholar Nanki, T. et al. Stromal cell-derived factor-1-CXC chemokine receptor 4 interactions play a central role in CD4 + T cell accumulation in rheumatoid arthritis synovium. J. Immunol. 165(11), 6590–6598 (2000).Article CAS PubMed Google Scholar Nanki, T. et al. Chemokine receptor expression and functional effects of chemokines on B cells: implication in the pathogenesis of rheumatoid arthritis. Arthritis Res. Ther. 11(5), R149 (2009).Article PubMed PubMed Central Google Scholar Chung, S. H. et al. CXC chemokine receptor 4 expressed in T cells plays an important role in the development of collagen-induced arthritis. Arthritis Res. Ther. 12(5), R188 (2010).Article PubMed PubMed Central Google Scholar Download referencesFundingThis work was supported by the Natural Science Foundation of Nanjing University of Chinese Medicine (Grant No. XZR2022047) and the Project supported by Research Fund of Nanjing Hospital of Chinese Medicine (Grant No. YJLC202305).Author informationAuthors and AffiliationsDepartment of Rheumatology, Nanjing Hospital of Chinese Medicine Affiliated to Nanjing University of Chinese Medicine, Nanjing, 210022, ChinaRongyue Jing, Yueyue Chen, Meimei Xu & Suling WuGraduate School, Nanjing University of Chinese Medicine, Nanjing, 210023, ChinaXudan ZouAuthorsRongyue JingView author publicationsYou can also search for this author inPubMed Google ScholarYueyue ChenView author publicationsYou can also search for this author inPubMed Google ScholarMeimei XuView author publicationsYou can also search for this author inPubMed Google ScholarXudan ZouView author publicationsYou can also search for this author inPubMed Google ScholarSuling WuView author publicationsYou can also search for this author inPubMed Google ScholarContributionsThe authors confirm contribution to the paper as follows: Conceptualization: RJ and SW; Data curation, Formal analysis, Investigation and Methodology: RJ, YC and MX; Funding acquisition: RJ. Project administration, Resources and Supervision: RJ and SW; Software: XZ and RJ; Validation: RJ and YC; Visualization and Writing-original draft: RJ; Writing-review & editing: SW. All authors reviewed the results and approved the final version of the manuscript.Corresponding authorsCorrespondence to Rongyue Jing or Suling Wu.Ethics declarationsCompeting interestsThe authors declare no competing interests.Ethics statementThe animal study protocol was approved by the Experimental Animal Ethics Committee of Nanjing University of Chinese Medicine (Approval number: 202211 A05). All methods are reported in accordance with ARRIVE guidelines.Additional informationPublisher’s noteSpringer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.Electronic supplementary materialBelow is the link to the electronic supplementary material.Supplementary Material 1Supplementary Material 2Rights and permissionsOpen Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.Reprints and permissionsAbout this article