FFXHL pills (SFDA approval number Z11020012) was purchased from Beijing Tongrentang Co.,Ltd.Tongrentang Pharmaceutical Factory. Bovine type II collagen (20022), Incomplete Freund’s Adjuvant (IFA) (7002) were bought from Chondrex, Inc. (United States). Dexamethasone was purchased from Guangdong South China Pharmaceutical Group Co., Ltd. (Guangdong, China). Rheumatoid fibroblast-like synoviocyte transformed with SV40 T antigen (MH7A) was obtained from Jennio Biological Technology (Guangzhou, China). RPMI 1640 culture medium (22400071), trypsin, Pencillin-Streptomycin (P/S), and fetal bovine serum (FBS) were bought from Gibco (United States). Enzyme-linked immunosorbent assay (ELISA) were purchased from Sinouk Bio (Bei Jing, China) and Lianke Bio (Hang Zhou, China). TNF-a protein was obtained from PRPTO TECH (United States). TLR4, Myd88, p65, p-p65, IKKα/β, p-JNK1/2/3, JNK1/2/3, were purchased from Abcam (United Kingdom). p-Erk1/2, Erk1/2, p-p38, p38, β-actin, p-IKKα/β, p-IκBα, IκBα, p-Akt, Akt were purchased from CST (United States).

FFXHL were accurately weighed and ultra-sonicated with 1 mL of 80% methanol for 20 min. The extract was taken out and centrifuged at 13,000 rpm (4°C) and centrifuge for 10 min. Further dilute the solution 100 times with 50% methanol, and place it into a clean sample bottle for LCMS/MS analysis. At the same time, prepare the standard curve with a 50% methanol aqueous solution. The UPLC–MS/MS system consisted of an QTRAP 5500 (AB Sciex, United States) and a Exion LC AD (AB Sciex, United States). Chromatographic separation was performed on Waters XSelect HSS T3 column (2.1 × 100 mm, 2.5 µm) at 40 C with a flow rate of 0.3 mL/min. The mobile phase consisted of acetonitrile (A) and 0 .05% (v/v) aqueous formic acid (B) with gradient elution as follows: 85%–70% B at 0–2 min, 70%–40% B at 2–6 min, 40%–0% B at 6–8 min, 0%–0% B at 8–10 min, 0%–85% B at 10–10.1 min, 85%-85% B at 10.1–13 min. The injection volume was 5 μL. The mass spectrometer was operated in the multiple reaction monitoring (MRM) mode with electrospray positive/negative ionization (ESI). The ionspray voltage was set at 5.5 kV/4.5 kV, and the temperature was maintained at 500°C. The curtain gas was set at 30 psi; collision gas was set at 9; the nebulizer and the heater gas were set to 55 psi. Quantitative parameters are listed in Supplementary Table S3.

After dissolving 50 mg of the FFXHL in 1 mL of ddH₂O water, 4 mL of methanol was added for ultrasonic extraction for 20 min After filtered with 0.22 μm filter, the samples were analysed on I-Class (Waters Corporation, United States) equipped with a ACQUITY UPLC HSS T3 (2.1 × 100 mm, 1.8 μm). The mobile phase consisted of 0.1% formic acid in pure water (A) and acetonitrile (B). A gradient elution program was as follows: 0%–0% B on 0–1.5  min, 0%–100% B on 1.5–70  min, 100%–100% B on 70–72 min. The flow rate and the injection volume were set at 0.25 mL/min and 10 μL respectively. The mass spectrometric data was collected by Q-TOF SYNAPT G2Si system (Waters Corporation, United States) in both positive and negative ion modes. The working parameters of mass spectrometry were as follows: the capillary voltage was 3.0  kV, the desolvation gas flow was 600 L/min, the cone gas flow was 50 L/min, the source temperature was 120°C, and the desolvation temperature was 360°C.

A total of 60 female Sprague-Dawley (SD) rats (6 weeks old; 160 ± 20 g) were provided by Vital River Laboratory Animal Technology Co., Ltd. (Bei Jing, China) [certificate no. SCXK (Jing) 2016-0011].

The rats were fed and housed in a standard pathogen-free environment with a constant temperature of 22°C ± 2°C, 40%–70% humidity, 12-h light/dark cycles and free access to food and water. All animal experiments were approved by the Animal Ethics Committee of Beijing Zhongyan Tongrentang Medicine Research and Development Co., Ltd. (YJY-2021-081801). All experiments were conducted in accordance with the International Guidelines of the Animal Care and Use Committee.

Bovine type II collagen solution was emulsified with an equal volume of IFA to create a stable emulsion of 1 mg/mL in an ice-cold water bath. Excluding the normal group (control, n = 7), 0.2 mL of the emulsion was injected on the tail, as previously described with minor modification (Song et al., 2015). On day 7, 0.1 mL of the emulsion was administered as a booster to establish CIA models, avoiding the original injection sites. The normal group received the same amount of saline. The dosage of FFXHL for humans is 0.2 g/kg/day. Therefore, the dosage was 1.2 g/kg per day (medium dose), 2.4 g/kg per day (high dose) and 0.6 g/kg per day (low dose) for rats. After excluding rats where the model was not successfully established, the remaining rats were randomly divided into five groups (n = 7 per group): model group (CIA rats), FFXHL high-dose group (CIA rats with 2.4 g/kg of FFXHL, FH), FFXHL medium-dose group (CIA rats with 1.2 g/kg of FFXHL, FM), FFXHL low-dose group (CIA rats with 0.6 g/kg of FFXHL, FL), and positive drug group (CIA rats with 0.3 mg/kg of dexamethasone, designated as DXM). From day 16, the rats in FFXHL and DXM groups were orally administered FFXHL and DXM once a day, while the normal and CIA group rats received the same volume of drinking water using the same method. After 20 days of treatment, these rats were fasted for overnight (with free access to water) and then sacrificed on day 36 (Supplementary Figure S1).

On the day 0, the rats underwent measurements for body weight and hind paws volumes to establish baseline values. The hind paws volumes were determined by plethysmometry, and these measurements were continued until the day before the rats were sacrificed, along with assessments of body weight, hot-plate latent pain response test, and arthritis score.

The baseline for the hot-plate latent pain response test and arthritis score was established on the day before administering the drugs. In the hot-plate latent pain response test, all rats were placed on a 55°C ± 0.5°C hot-plate to observe pain responses, including paw licking and jumping. The reaction time was defined as the interval between the moment the animal was placed on the plate and the time it initiated licking its claws or raising its paws.

For the arthritis score, in accordance with the criteria from previous literature (Zhao et al., 2021), a scoring system with four levels was used: 0 for no swelling or redness, 1 for slight erythema or swelling of the toes, 2 for swelling and redness of the paws, 3 for severe redness and swelling of the ankles, and 4 for serious redness and swelling of the ankle, foot, and digits. The total score for each rat was the combined score of both hind paws, with a maximum value capped at 8.

MH7A cells were maintained in 1640 medium, supplemented with 10% FBS and 1%P/S in a CO₂ incubator at 37°C.

The ankle joints subsequently harvested from the rats were fixed in 4% paraformaldehyde. Samples were decalcified in 10% ethylenediaminetetraacetic acid (EDTA) at room temperature (25°C–30°C) for approximately 4 weeks. After decalcification, the tissues were dehydrated by ethanol gradient, cleared in xylene. Finally, the samples were paraffin-embedded, cut into 4 mm slices by using a microtome and stained with hematoxylin-eosin (H&E) for routine histological evaluations with a light microscope (Nikon Eclipse E100, Japan).

The paraffin sections were deparaffinised and subsequently treated with a 3% H₂O₂ solution at room temperature in the dark for 25 min. Afterward, the sections underwent three washes in PBS (pH 7.4) and were then incubated with 3% BSA at room temperature for 30 min. Following this, the sections were exposed to the primary anti-NF-κB p65 antibody (diluted at 1:200) overnight at 4°C, and subsequently, they were treated with HRP goat anti-rabbit secondary antibody (Servicebio, Wuhan, China) at room temperature for 50 min. The sections were developed using 3,3 diaminobenzidine (DAB) and counterstained with hematoxylin at room temperature. Positive area quantification was performed by calculating the ratio of the positive area to the total tissue area using Image-Pro 6.0 analysis software.

The blood samples were gathered from abdominal aorta and the samples were centrifuged at 4°C for 20 min (3,000 rpm). The levels of serum TNF-α, IL-1β, IL-10, and IL-6 were detected by kits from Sinouk Bio (Beijing, China). MH7A cells were seeded in 6-well plates, after treatment, culture medium was collected, perform ELISA detection according to the requirements of the kits (Lianke Bio, Zhe Jiang, China).

The MH7A cells were plated in 96-well plates. FFXHL was dissolved in DMSO (final concentration is 0.5%) and subjected to ultrasonic processing. FFXHL concentrations ranging from 0.05 to 1 mg/mL (0.05, 0.1, 0.2, 0.4, 0.8, and 1 mg/mL) were applied to the cells for 24 h before treatment with or without TNF-α (20 ng/mL) for an additional 24 h. Subsequently, 10 μL of cell counting kit-8 (CCK-8) solution (Applygen, Beijing, China) was directly introduced to each well, and after a 2-h incubation, the microplate was read at OD480 using a plate reader (BioTek, VT, United States).

MH7A cells were cultured in black 96-well plates and stimulated with FFXHL (0.05, 0.1, 0.2 mg/mL) for 24 h, followed by stimulation with TNF-α (20 ng/mL) for an additional 24 h. The experimental procedures were conducted in accordance with the guidelines of CellROX™ Deep Red (Thermo Fisher Scientific, United States). NucBlue™ Live ReadyProbes™ reagent was added at a concentration of 1 drop/mL (Thermo Fisher Scientific, United States), and the cells were then incubated at 37°C for 30 min. After PBS washing, the results were assessed using Operetta CLS High-Content Screening (Revvity, Shanghai, China).

MH7A cells cultured in black 96-well plates and subjected to the aforementioned treatment. After treatment, cells were fixed with 4% paraformaldehyde for 15 min at room temperature, permeabilized with 0.1% Triton-X100 for 15 min, and then blocked with 10% goat serum for 1 h at room temperature. Subsequently, the cells were incubated overnight with anti- NF-κB p65 antibody in 1:200 (Abcam, United Kingdom) and then incubated with Alexa Fluor^® 488 Goat Anti-Rabbit IgG secondary antibodies (Abcam, United Kingdom) at room temperature for 1 h. After washing in PBS, nuclei were stained by NucBlue™ Live ReadyProbes™ reagent at 1 drop/mL for 15 min. The stained cells were analysed with by Operetta CLS High-Content Screening.

MH7A cells were plated in 6-well dishes and subjected to the aforementioned treatment. Total RNA extraction was carried out using trizol, followed by reverse transcription of RNA into cDNA using the Revert Aid First Strand cDNA Synthesis Kit (Thermo Fisher Scientific, United States). qPCR reactions were conducted with Fast Start Universal SYBR Green Master (Roche, Switzerland). All reactions were performed using the SLAN-96P fluorescent quantitative PCR instrument (Hongshitech, Shanghai, China). Quantification of results was done using the 2^−ΔΔCT method, with GAPDH mRNA serving as the internal control. The primer details are provided in Supplementary Table S4.

MH7A cells were seeded in 10 cm culture dishes (430167, Corning, United States) and subjected to the aforementioned treatment. All samples were lysed in RIPA cell lysis buffer (Beyotime, Shanghai, China), containing both phosphatase and protease inhibitors (Roche, Switzerland). Protein concentrations were determined using the BCA Protein Assay Kit (Beyotime, Shanghai, China). Equivalent amounts (70 ug) of each sample was separated by electrophoresis on a 4%–20% SDS-PAGE gel and transferred to PVDF membranes. After blocked with 5% bovine serum albumin for 1 h at room temperature, the membranes were incubation in primary antibody (1:300-1:1000, specific dosage in Supplementary Table S5) overnight at 4°C. The secondary HRP-linked antibody was diluted in 1:5000 (CST) or IRDye secondary antibody was diluted in 1:10000 (LI-COR, United States) and membranes were incubated at room temperature for 1 h. HRP-linked antibody binding was visualized by enhanced chemiluminescence (ECL Prime; Cytiva, China). Blots were imaged using the Odyssey^® XF (LICOR, United States) and analysed with ImageJ.

All data were presented as mean ± standard deviation (mean ± SEM). Statistical analysis was carried out with SPSS 27.0 and prism 9.0. One-way analysis of variance (ANOVA) with Bonferroni post hoc test. The Wilcoxon rank-sum test is used for arthritis score and histopathologic score. p < 0.05 was considered as statistically significant difference.

The identification and quantification of chemicals are essential for TCM studies to ensure repeatable and consistent results. According to the Chinese Pharmacopoeia - Traditional Chinese Medicine Formulations, benzoylhypacoitine, benzoylmesaconine, benzoylaconitine and paeoniflorin were representative metabolites for quality control of FFXHL. The UPLC-MS/MS was used for the qualitative and quantitative analysis of metabolites in FFXHL. The results of qualitative analysis of FFXHL were shown in Figures 1A, B and Supplementary Table S6. There were total 161 metabolites in FFXHL, including terpenoid, lactone, alkaloid, volatile oils, flavone and amino acid. The contents of benzoylhypacoitine, benzoylmesaconine, benzoylaconitine and paeoniflorin were 0.135 ± 0.002, 0.411 ± 0.004, 0.0781 ± 0.001 and 1.418 ± 0.01 mg/pill, respectively in FFXHL (3 g per pill) (Figures 1D–F).

To investigate the therapeutic effect of FFXHL on RA, a CIA model was induced in rats. Throughout the entire experiment, paw swelling, body weights, and arthritic indexes were assessed. In comparison to the control group, a significant decrease in body weight was observed in the model group and DXM group at days 27 and 34 (Figure 2A). However, rats treated with FFXHL did not show a significant weight increase compared to the CIA model rats throughout the test, with an increasing trend observed in the FM group, though not statistically significant (Figure 2A).

After two injections, all groups, except the control group, displayed significant swelling of the feet (Figures 2B,E). Following treatment with different concentrations of FFXHL, the paw swelling in FH, FM, and FL groups markedly diminished by the final day (Figures 2B,E). In comparison to the FFXHL treatment groups, the DXM treatment group exhibited a more favourable outcome, as DXM is the first-line medication for RA, and its therapeutic effect was evident on day 27, earlier than FFXHL (Figures 2B,E). Consistent with the paw swelling results, arthritis scores also demonstrated a decreasing trend after rats received FFXHL treatment (Figure 2C).

Joint pain is another symptom of RA, and therefore, the pain-relieving effect of FFXHL during RA occurrence was tested. Figure 2D showed that, compared to the model groups, the latency time in FM and FL groups was significantly prolonged in both hind paws. Although the FH group increased the latency time, the results were not significant.

To evaluate the impact of FFXHL on joint damage in CIA rats, histopathological changes were examined, including synovial cell proliferation, inflammation, and cartilage damage. As shown in Figure 3A, the articular cartilage surface in the control group was smooth, the morphology and structure of the chondrocytes are normal, without noticeable synovial hyperplasia or inflammation in the synovial tissue. In contrast, the model groups (Figure 3B) exhibited large amount of bone tissue necrosis and dissolution visible, replaced by extensively proliferated connective tissue (black arrow). There is severe proliferation of the synovium on both sides (yellow arrow) and severe inflammatory cells infiltration (red arrows) and some of them have infiltrated into the joint cavity (blue arrow). Treatment with FFXHL (Figures 3C,D,E) and DXM (Figure 3F)markedly reduced these histological abnormalities (Figure 3G). In FH, FM and FL groups, there was a noticeable reduction in bone necrosis and dissolution. Additionally, these groups showed minimal synovial thickening and only a modest presence of infiltrative inflammatory cells.

The expression levels of TNF-α, IL-1β, IL-6, and IL-10 in serum from different groups were subsequently examined. Compared with the control group, the serum levels of TNF-α, IL-1β, and IL-6 were significantly increased in the model group (Figures 4A–C). All FFXHL-treated rats exhibited effectively reduced levels of TNF-α, IL-1β, and IL-6, indicating an anti-inflammatory role of FFXLH in CIA model rats. IL-10, as an anti-inflammatory cytokine, showed a sharp reduction trend in model groups; however, after FFXHL treatment, the serum level of IL-10 increased significantly (Figure 4D).

The effects of FFXHL on NF-κB p65 expression in the joints of CIA rats was immunohistochemically examined. As shown in Figures 5A,B, NF-κB p65 was strongly expressed in synovium of the CIA rats, while in control rats, there were barely positive signals can be observed. After treatment with FFXHL and DXM, the positive signals of NF-κB p65 in the synovium of CIA rats were limited and weak (Figures 5C–G).

To further understand the possible underlying mechanism of FFXHL on treatment RA, an in vitro models of cell was employed. The MH7A synovial fibroblast cell line is derived from the synovium of a female patient with RA and has been chemically modified using the SV40 T antigen. This cell line is capable of stable propagation across 10 to 15 passages. Its use has significantly facilitated the screening of drugs for RA and the investigation into the mechanisms of action of RA treatments in recent years (Dai et al., 2022). Firstly, the potential cytotoxicity of FFXHL on MH7A cells was evaluated. Different concentrations (0.05, 0.1, 0.2, 0.4, 0.8, and 1 mg/mL) of FFXHL were added to MH7A cells for 24 h, and cell viability was tested using CCK8 (Figure 6A). The results demonstrated that treatment with FFXHL at concentrations of 0.05, 0.1, and 0.2 mg/mL had no effect on cell viability, so these three concentrations would be used in the next step. Next, to investigate the effects of FFXHL on TNF-α-stimulated cell proliferation, 0.05, 0.1, and 0.2 mg/mL FFXHL were administered to cells for 24 h before 20 ng/mL TNF-α stimulation. TNF-α is a major inflammatory mediator in the pathogenesis of RA as it can promote RA symptoms like secretion of inflammatory cytokines and fibroblast proliferation. Therefore, TNF-α is frequently utilized in vitro to treat fibroblasts in order to stimulate RA (Dai et al., 2022). The results showed that after 24 h of treatment with TNF-α, MH7A cells proliferated significantly, and FFXHL significantly inhibited this cell proliferation (Figure 6B).

In order to create an in vitro system for evaluating anti-inflammatory activity, MH7A cells were stimulated with 20 ng/mL TNF-α. After 24 h, there was a significant increase in the expression of IL-1 and IL-6 in the culture supernatant. However, 24 h of FFXHL treatment reduced the release of IL-1β and IL-6 in the supernatant, as well as the mRNA expression in TNF-α-stimulated MH7A cells (Figures 6C–F).

Reactive oxygen species (ROS) contribute to the destruction of collagen tissues, and a high level of ROS is associated with damage leading to cartilage degradation (Ahsan et al., 2022). To further elucidate the role of ROS in TNF-α-stimulated inflammation of MH7A cells, cells were pre-treated with FFXHL (0.05, 0.1, 0.2 mg/mL) for 24 h and then stimulated with 20 ng/mL TNF-α. The results showed that TNF-α significantly increased the intracellular ROS level (Figures 7A,B); 0.1 and 0.2 mg/mL FFXHL effectively reduced the accumulation of ROS in MH7A cells.

To further investigate the translocation events that govern NF-κB p65 function by FFXHL, the immunofluorescence technique was used. The cells were treated with different concentrations of FFXHL before TNF-α induction and then incubated with NF-κB p65 antibody. As shown in Figures 8A,B, TNF-α challenge induced NF-κB p65 nuclear translocation, and the process of NF-κB from the cytoplasm into the nucleus was significantly suppressed in MH7A cells by FFXHL.

MMPs play a crucial role in the pathogenesis of RA by mediating joint cartilage and bone destruction. In this study, the mRNA levels of MMP-3, MMP-9, and MMP-13 were examined. After stimulation with 20 ng/mL TNF-α, the mRNA levels of MMP-3, MMP-9, and MMP-13 (Figures 9A–C) significantly increased in MH7A cells compared with the control group. However, in groups pretreated with FFXHL, the mRNA levels of these MMPs were significantly inhibited.

To elucidate the mechanism of the anti-inflammatory effect of FHXHL in MH7A cells, the effects of FHXHL on inflammation-related signalling cascades, including TLR4/MyD88/NF-κB, Akt, and MAPKs pathways, were analysed. As shown in Figures 9D–I and Figures 10A–E, exposure of MH7A cells to TNF-α elevated the protein levels of p-Akt, p-NF-κB p65, TLR4, and MyD88, without affecting the total protein levels of Akt and NF-κB p65. Treatment with FFXHL attenuated this increase in MH7A cells (Figures 9E, F, I, 10E).

Furthermore, TNF-α significantly increased the phosphorylation levels of IKKα/β and the degradation of IκBα, but treatment with FFXHL strongly inhibited these phosphorylation and degradation events (Figures 9G, H). In Figures 10A–D, the results showed that after TNF-α stimulation, the MAPK pathway was activated, and the phosphorylation levels of p38, ERK, and JNK increased significantly, while treatment with FFXHL significantly inhibited their phosphorylation levels.