A total of 48 male Wistar rats (6 weeks, 170–200 g) were bought from Changsha Tianqin Biotechnology Co., Ltd. (Changsha, China). Rats were housed under standard laboratory conditions, with a temperature of 25 $\pm$ 2 °C, humidity of 50 $\pm$ 5%, and a 12-hour light/dark cycle. They were raised in separate cages with 4 in each one, and had free access to food and water for one week of adaptive feeding before experiments. 2.0 mg/mL bovine type II collagen (20022, Chondrex, Woodinville, WA, USA) that dissolved in acetic acid solution was mixed with complete Freund’s adjuvant (7001, Chondrex, Woodinville, WA, USA) in the ice bath at the volume of 1:1, repeated suction by syringe, fully emulsified, to make bovine type II collagen emulsion. The emulsion was injected subcutaneously into the tail (300 µL) of rats. 7 days later, the rats were inoculated with bovine type II collagen emulsion (300 µL) for a second time [27]. Normal rats were injected with normal saline. Groups included Sham, CIA, TWG, Rhodojaponin III Lo, Rhodojaponin III Mi, and Rhodojaponin III Hi, with 8 rats in each group. Rats in the TWG group were given intragastric administration with Tripterygium wilfordii Glycosides (TWG, 50 mg/kg) (Z33020422, Dnd Pharmaceutical Co., Ltd., Shaoxing, China) once a day. Rats in the Rhodojaponin III Lo, Rhodojaponin III Mi, and Rhodojaponin III Hi groups were separately given intragastric administration with Rhodojaponin III (0.06 mg/kg, 0.12 mg/kg, 0.24 mg/kg) (26342-66-5, yuanye Bio-Technology Co., Ltd., Shanghai, China) once a day [23]. Rats in Sham and CIA groups were intragastrically given normal saline. After 4 weeks, all rats were anesthetized to obtain the whole blood by abdominal aorta puncture. The rats were sacrificed by the cervical dislocation method to obtain knee joints and synovium for subsequent experiments. All animal experiments followed the ARRIVE guidelines. This study has been approved by the Experimental Animal Ethics Committee of Hunan University of Chinese Medicine (No. LL2022061401).

The arthritis index (AI) was evaluated according to the swelling degree of joints every 4 days during the experiment [28]. AI was utilized as the criterion to judge the successful construction of the CIA rat model. When AI $\ge$4, the CIA rat model was successfully constructed. AI scoring rule: 0, no swelling; 1, slight swelling of toe joints; 2, swelling of phalangeal joints and toe joints; 3, Paw swelling below the ankle joint; 4, Swelling of all feet, including the ankle joints. Since the incidence of forepaw inflammation was very low, and the joints of the hind feet were more prone to severe swelling, we calculated the AI score of the two hind feet of rats (total scores: 8).

HUVECs (AW-CNH488, Abiowell, Changsha, China) were cultured in Dulbecco’s modified Eagle’s medium (DMEM) with 10% fetal bovine serum (FBS) (10099141, Gibco, Carlsbad, CA, USA) and 1% Penicillin/Streptomycin (SV30010, Beyotime, Shanghai, China) with 5% CO${}_2$ at 37 °C. HUVECs can be used in experiments when the confluence reached 70–80%. The oe-NC plasmids or oe-NIK plasmids (HonorGene, Changsha, China) were transfected separately into HUVECs with Lipofectamine 2000 (11668019, Invitrogen, Carlsbad, CA, USA). The cells used in the experiment were validated and were free from bacteria, fungi, and mycoplasma infections. In addition, the expression of von Willebrand factor (VWF) in HUVECs was evaluated using immunofluorescence (IF) staining, as shown in the Supplementary Figure.

Experiment 1 was divided into the following groups: Control, TNF-$\alpha$, Rhodojaponin III Lo, Rhodojaponin III Mi, and Rhodojaponin III Hi. The cells in the Control group were not treated. Cells in the TNF-$\alpha$ group were induced with 10 ng/mL TNF-$\alpha$ (ab259410, Abcam, Cambridge, UK) for 24 h [24, 29, 30, 31]. Cells in Rhodojaponin III Lo, Rhodojaponin III Mi, and Rhodojaponin III Hi groups were induced with 10 ng/mL TNF-$\alpha$ for 24 h and treated with 9 µg/mL, 18 µg/mL, and 36 µg/mL Rhodojaponin III for 24 h [32]. Experiment 2 was divided into the following groups: TNF-$\alpha$, Rhodojaponin III, Rhodojaponin III+oe-NC, and Rhodojaponin III+oe-NIK. Cells in the TNF-$\alpha$ group were treated as above. Cells in the Rhodojaponin III group were induced with 10 ng/mL TNF-$\alpha$ for 24 h and then treated with 36 µg/mL Rhodojaponin III for 24 h. Cells in the Rhodojaponin III+oe-NC group were treated with oe-NC plasmids, induced with 10 ng/mL TNF-$\alpha$ for 24 h, and then treated with 36 µg/mL Rhodojaponin III for 24 h. Cells in the Rhodojaponin III+oe-NIK group were transfected with oe-NIK plasmids, induced with 10 ng/mL TNF-$\alpha$ for 24 h, and then treated with 36 µg/mL Rhodojaponin III for 24 h.

The knee joint and synovium were firstly fixed with 4% paraformaldehyde (N1012, NCM Biotech, Suzhou, China), then sliced after paraffin embedding. They were dewaxed in xylene followed by dehydration with gradient ethanol (75–100%). Slices were stained with hematoxylin (AWI0001a, Abiowell, Changsha, China) and returned to blue with phosphate-buffered saline (PBS). After being stained with eosin (AWI0029a, Abiowell, Changsha, China), the slices were rinsed with distilled water, followed by dehydration with gradient ethanol (95–100%). The slices were placed in xylene for 10 min for transparency and observed by a microscope after being sealed with neutral gum (AWI0238a, Abiowell, Changsha, China).

Platelet endothelial cell adhesion molecule-1 (CD31) and vascular endothelial cell growth factor (VEGF) expressions in the synovium were detected by IHC staining. The slices were dewaxed in xylene followed by dehydration with gradient ethanol (75–100%). Subsequently, antigen retrieval and inactivation of endogenous enzymes were performed. Slices were incubated with primary antibodies of CD31 (1:300, ab182981, Abcam, Cambridge, UK) and VEGF (1:200, 19003-1-AP, Proteintech, Chicago, IL, USA) at 4 °C overnight followed by 100 µL of horseradish peroxidase (HRP) goat anti-rabbit IgG (1:100, AWS0005, Abiowell, China) at 37 °C for 30 min. Then 100 µL of DAB (ZLI-9017, ZSBG-Bio, Beijing, China) was added for color development. Slices were re-stained with hematoxylin and returned to blue with PBS. The slices were placed in xylene for 10 min for transparency and observed by a microscope after being sealed with neutral gum.

The whole blood, the synovial samples, and cell cultures were pretreated to obtain the supernatant for detection. IL-6 (CSB-E04638h, CUSABIO, Wuhan, China), IL-1$\beta$ (CB-E08053h, CUSABIO, Wuhan, China), and TNF-$\alpha$ (CB-E04740h, CUSABIO, Wuhan, China) ELISA kits were utilized to detect the levels of cytokines (IL-6, IL-1$\beta$, and TNF-$\alpha$).

Logarithmic growth of HUVECs was digested by trypsin (AWC0232, Abiowell, Changsha, China). 1 $\times$ 10${}^4$ cells were inoculated in 96-well plates, 100 µL per well. After cell adhesion, each group was treated with the corresponding drug for 24 h and 48 h, respectively. Then, 10 µL of CCK-8 reagent (NU679, Dojindo, Kumamoto, Japan) was added and incubated at 37 °C for 4 h. Optical density (OD) at 450 nm was analyzed with a multifunctional microplate reader (MB-530, HEALES, Shenzhen, China).

The scratch assay was applied to detect the migration ability of HUVECs. HUVECs from different treatment groups were digested to make cell suspension. 5 $\times$ 10${}^5$ cells were uniformly inoculated in 6-well plates. When the cells had filled the plates, lines were drawn and viewed under a microscope to measure the initial scratch width. After 24 h and 48 h, the scratch widths were measured and photographed.

All reagents and equipment were pre-chilled one day in advance. 100 µL of Matrigel (354262, Corning Inc. Corning, NY, USA) was added to the upper transwell chamber (3428, Corning Inc. Corning, NY, USA) and incubated at 37 °C for 30 min for solidification. HUVECs from different treatment groups were digested to make cell suspension. DMEM without FBS was used to suspend the cells to 2 $\times$ 10${}^6$ cells/mL. 100 µL of cell suspension was added to the upper chamber, and 500 µL of DMEM with 10% FBS was added to the lower chamber followed by incubation at 37 °C for 48 h. The upper chamber was washed with PBS, and the cells not penetrating the membrane were gently wiped off with sterile cotton balls. After fixation with 4% paraformaldehyde and staining with 0.1% crystal violet (G1062, Solarbio, Beijing, China), cells on the outer surface of the upper chamber were observed and counted.

All reagents and equipment were pre-chilled one day in advance. 70 µL of Matrigel was evenly spread in the 96-well plates at 37 °C for 30 min for solidification. HUVECs from different treatment groups were digested to make cell suspension. 1 $\times$ 10${}^4$ cells were inoculated in 96-well plates for 24 h. The formation of blood vessels was evaluated by observing and counting closed-loop or pro-angiogenic structures with an inverted microscope.

The three-dimensional structure diagram of NIK was searched from the PDB database (https://www.rcsb.org/), and imported into PyMOL (ver.2.3.1, Schrödinger, LLC., New York, NY, USA) to remove water molecules. Then, it was imported into Autodock tools (ver.1.5.6, The Scripps Research Institute, La Jolla,CA, USA) to add hydrogen atoms and then saved in PDBQT format after calculation of the total charge and set of the atomic type. The Chemical Abstracts Service (CAS) number was searched from the PubChem database (https://pubchem.ncbi.nlm.nih.gov/), and the 3D structure diagram of Rhodojaponin III was downloaded. Its energy was minimized with Chem3D pro and was saved in mol2 format. VINA software (Ver. 1.1.2, The Scripps Research Institute, La Jolla, CA, USA) was used for docking, and the results were visualized with Discovery Studio software (ver. 18.0, BIOVIA, San Diego, CA, USA). Finally, the 3D molecular docking model was output.

Different samples were treated with RIPA lysate (AWB0136, Abiowell, Changsha, China) to extract total proteins. The proteins were transferred to nitrocellulose (NC) membranes after sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). After blocking in 5% bovine serum albumin (BSA) or 5% skimmed milk (AWB0004, Abiowell, Changsha, China), the membranes were incubated with the primary antibodies at 4 °C overnight, including NIK (15 µg/mL, ab203568, Abcam, Cambridge, UK), IKK$\alpha$ (1:1000, ab32041, Abcam, Cambridge, UK), p-IKK$\alpha$ (1:1000, ab38515, Abcam, Cambridge, UK), p52 (1:5000, 10655-1-AP, Proteintech, Chicago, IL, USA), CXCL12 (1:20,000, ab155090, Abcam, Cambridge, UK), and $\beta$-actin (1:5000, 66009-1-Ig, Proteintech, Chicago, IL, USA). The membranes were incubated with HRP-labled secondary antibodies for 1.5 h. Finally, the membranes were incubated with ECL reagent (AWB0005, Abiowell, Changsha, China) and followed by imaging. The protein expressions were analyzed by Quantity One 4.6.6 (Bio-Rad Inc., Hercules, CA, USA) with $\beta$-actin as the reference protein.

All data were analyzed by GraphPad Prism 8.0 (GraphPad Software Inc., San Diego, CA, USA) and presented as mean $\pm$ standard deviation (SD). One-way analysis of variance (ANOVA) and Tukey’s post-hoc test were conducted in comparison between groups. Two-way ANOVA was performed in groups at different time points. The difference was significant as p 0.05. All the experiments followed randomization and blind analysis to avoid experimental bias.

To explore the effect of Rhodojaponin III on RA, rats were induced by bovine type II collagen and then given intragastric administration with different concentrations of Rhodojaponin III with TWG as a positive control drug. High AI score in the CIA group indicated that the CIA rat model was successfully constructed. The AI scores in the TWG group and different concentrations of Rhodojaponin III groups were decreased compared with the CIA group (Fig. 1A). HE staining in knee joints showed that inflammatory infiltration and cartilage damage were significant in the CIA group. However, these lesions were significantly alleviated in the TWG group or different concentrations of Rhodojaponin III groups (Fig. 1B). It was further found by ELISA that IL-6, IL-1$\beta$, and TNF-$\alpha$ levels in serum and synovium of rats in the TWG group or different concentrations of Rhodojaponin III groups were decreased compared with the CIA group (Fig. 1C). Together, the above results proved that Rhodojaponin III alleviated arthritis progression and disease severity in CIA rats.

Fig. 1.

Rhodojaponin III alleviated rheumatoid arthritis (RA) in collagen-induced arthritis (CIA) rats. (A) Arthritis index (AI) score. (B) Hematoxylin-eosin (HE) staining analyzed the pathological changes in the knee joints (n = 8). (C) Enzyme-linked immunosorbent assay (ELISA) detected cytokine levels in serum and synovium (n = 8). * p 0.05 vs. Sham, # p 0.05 vs. CIA. TWG, Tripterygium wilfordii Glycosides.

Then, we studied the effect of Rhodojaponin III on the vascular density in the synovium of inflammatory joints. HE staining showed a large number of synovial hyperplasia and pannus formation in the synovium. However, these were significantly reduced after intervention with TWG or different concentrations of Rhodojaponin III (Fig. 2A). In addition, IHC staining analysis displayed that CD31 and VEGF were highly expressed in the synovium. After intervention with different concentrations of Rhodojaponin III, CD31 and VEGF expressions in the synovium significantly decreased, and the expression level in the Rhodojaponin III Hi group reached in the TWG group (Fig. 2B). These results proved that Rhodojaponin III decreased vascular density in the synovium of inflammatory joints in CIA rats.

Fig. 2.

Rhodojaponin III decreased vascular density in the synovium of inflammatory joints in CIA rats. (A) HE staining in the synovium (n = 8). (B) Platelet endothelial cell adhesion molecule-1 (CD31) and vascular endothelial cell growth factor (VEGF) expressions in the synovium were assessed by immunohistochemistry (IHC) staining (n = 8). * p 0.05 vs. Sham, # p 0.05 vs. CIA.

To determine the specific therapeutic effect of Rhodojaponin III on RA, a cell model of HUVECs was constructed by TNF-$\alpha$ induction. The proliferation ability of HUVECs increased after TNF-$\alpha$ induction and decreased in a concentration-dependent manner after intervention with different concentrations of Rhodojaponin III (Fig. 3A). The migration ability of HUVECs enhanced after TNF-$\alpha$ induction. However, after intervention with different concentrations of Rhodojaponin III, the migration ability of HUVECs markedly decreased (Fig. 3B). The number of HUVECs passing through the membrane increased after TNF-$\alpha$ induction. After intervention with different concentrations of Rhodojaponin III, the number of cells passing through the membrane considerably decreased (Fig. 3C). In addition, TNF-$\alpha$ induced the formation of numerous tubes from HUVECs. After intervention with different concentrations of Rhodojaponin III, the number of tubes decreased significantly (Fig. 3D). Next, ELISA illustrated that IL-6, IL-1$\beta$, and TNF-$\alpha$ levels in the supernatant of HUVECs increased after TNF-$\alpha$ induction, while they were significantly reduced after intervention with different concentrations of Rhodojaponin III (Fig. 3E). These results showed that Rhodojaponin III inhibited migration, invasion, angiogenesis, and inflammation of TNF-$\alpha$-induced HUVECs.

Fig. 3.

Rhodojaponin III inhibited migration, invasion, angiogenesis, and inflammation of tumor necrosis factor-alpha (TNF-$\alpha$)-induced human umbilical vein endothelial cells (HUVECs).(A) Cell proliferation was assessed by CCK-8 (n = 3). (B) Cell migration was detected by scratch assay (n = 3). (C) Cell invasion was detected by Transwell invasion assay (n = 3). (D) Tube formation assay (n = 3). (E) Cytokine levels in the cell supernatant were determined by ELISA (n = 3). * p 0.05 vs. Control, # p 0.05 vs. TNF-$\alpha$.

The binding of Rhodojaponin III and NIK was simulated by molecular docking. In Fig. 4, the binding energy of Rhodojaponin III to NIK was –7.2 kcal/mol, indicating that Rhodojaponin III could spontaneously bind to NIK. The optimal docking conformation between Rhodojaponin III and NIK was shown on the left. Rhodojaponin III bound to the hydrophobic cavity of NIK and interacted with surrounding amino acid residues. Amino acid residues around Rhodojaponin III were shown in the truncated part on the right, including 13 amino acid residues: ASP A:519, SER A410, LYS A:517, ARG A:408, GLY A:409, ARG A:405, GLY A:407, GLU B:396, GLU B:395, GLU B:461, VAL B:397, ARG B:394, and GLU B:375. Rhodojaponin III was bound to ASP A:519, ARG A:408, ARG A:405, and GLU B:375 by conventional hydrogen bonds. Rhodojaponin III was bound to SER A410, LYS A:517, GLY A:409, GLY A:407, GLU B:396, GLU B:395, GLU B:461, VAL B:397, and ARG B:394 by van der Waals force. These results proved that Rhodojaponin III interacted with NIK.

Fig. 4.

Rhodojaponin III interacted with NF-κB-inducing kinase (NIK). Molecular docking model of Rhodojaponin III and NIK (Binding energy: –7.2 kcal/mol).

To elucidate the specific molecular mechanism of Rhodojaponin III on RA, Western blot was applied to detect the expressions of NIK pathway-associated proteins in CIA rats and TNF-$\alpha$-induced HUVECs. Compared with the Sham group, NIK, p52, and CXCL12 expressions in the synovium of rats in the CIA group were up-regulated, and the phosphorylation level of IKK$\alpha$ was increased. Compared with the CIA group, NIK, p52, and CXCL12 expressions in the synovium of rats in the TWG group or different concentrations of Rhodojaponin III groups were significantly down-regulated, and the phosphorylation level of IKK$\alpha$ was decreased (Fig. 5A). NIK, p52, and CXCL12 expressions were up-regulated, and the phosphorylation level of IKK$\alpha$ was increased in HUVECs after TNF-$\alpha$ induction. Compared with the TNF-$\alpha$ group, NIK, p52, and CXCL12 expressions in HUVECs intervened with different concentrations of Rhodojaponin III was significantly down-regulated, and IKK$\alpha$ phosphorylation was decreased (Fig. 5B). These results displayed that Rhodojaponin III inhibited the NIK pathway activation in CIA rats and TNF-$\alpha$-induced HUVECs.

Fig. 5.

Rhodojaponin III inhibited the NIK pathway activation in CIA rats and TNF-$\alpha$-induced HUVECs. (A,B) NIK, IκB kinase-alpha (IKK$\alpha$), p-IKK$\alpha$, p52, and C-X-C motif chemokine ligand 12 (CXCL12) expressions in the synovium of rats (n = 8) and HUVECs (n = 3) were detected by Western blot. * p 0.05 vs. Sham, # p 0.05 vs. CIA, & p 0.05 vs. Control, @ p 0.05 vs. TNF-$\alpha$.

To verify the role of the NIK/NF-$\kappa$B pathway in RA, HUVECs were transfected with oe-NIK plasmids, induced by TNF-$\alpha$, and then intervened with Rhodojaponin III. Compared with the TNF-$\alpha$ group, NIK, p52, and CXCL12 expressions in HUVECs in the Rhodojaponin III group were down-regulated, and the phosphorylation level of IKK$\alpha$ was decreased. Compared with the Rhodojaponin III+oe-NC group, NIK, p52, and CXCL12 expressions in HUVECs in the Rhodojaponin III+oe-NIK group were significantly up-regulated, and the phosphorylation level of IKK$\alpha$ was increased (Fig. 6A). The migration ability of HUVECs in the Rhodojaponin III group was decreased compared with the TNF-$\alpha$ group. The migration ability of HUVECs in the Rhodojaponin III+oe-NIK group was increased compared with the Rhodojaponin III+oe-NC group (Fig. 6B). The proliferation ability of HUVECs in the Rhodojaponin III group decreased compared with the TNF-$\alpha$ group. The proliferation ability of HUVECs in the Rhodojaponin III+oe-NIK group was increased compared with the Rhodojaponin III+oe-NC group (Fig. 6C). The number of HUVECs passing through the membrane in the Rhodojaponin III group decreased compared with the TNF-$\alpha$ group. The number of HUVECs passing through the membrane in the Rhodojaponin III+oe-NIK group was increased compared with the Rhodojaponin III+oe-NC group (Fig. 6D). The number of tubes in HUVECs in the Rhodojaponin III group decreased compared with the TNF-$\alpha$ group. The number of tubes in HUVECs in the Rhodojaponin III+oe-NIK group increased compared with the Rhodojaponin III+oe-NC group (Fig. 6E). It was found by ELISA that compared with the TNF-$\alpha$ group, IL-6, IL-1$\beta$and, TNF-$\alpha$ levels in the supernatant of HUVECs in the Rhodojaponin III group were decreased. Compared with the Rhodojaponin III+oe-NC group, IL-6, IL-1$\beta$, and TNF-$\alpha$ levels in the supernatant of HUVECs in the Rhodojaponin III+oe-NIK group were increased (Fig. 6F). These results indicated that Rhodojaponin III inhibited the migration, invasion, angiogenesis, and inflammation of HUVECs by inhibiting the NIK/NF-$\kappa$B pathway.

Fig. 6.

Rhodojaponin III inhibited the migration, invasion, angiogenesis, and inflammation of HUVECs by inhibiting the NIK/nuclear factor kappa B (NF-$\kappa$B) pathway. (A) NIK, IKK$\alpha$, p-IKK$\alpha$, p52, and CXCL12 expressions in HUVECs were detected by Western blot (n = 3). (B) Cell proliferation was detected by CCK-8. (C) Cell migration was detected by scratch assay (n = 3). (D) Cell invasion was detected by Transwell invasion assay (n = 3). (E) Tube formation assay (n = 3). (F) Cytokine levels in the cell supernatant were determined by ELISA (n = 3). * p 0.05 vs. TNF-$\alpha$, # p 0.05 vs. Rhodojaponin III+oe-NC.