Sixty male DBA/1 mice (weighting 18–22 g) were purchased from Beijing HFK Bioscience Co., Ltd (Beijing, China), and 72 male Wistar rats (weighting 180–220 g) were purchased from Vital River Laboratory Animal Technology Co. Ltd (Beijing, China). Animals were housed in a room on a 12-h light/dark cycle under specific pathogen-free conditions, maintaining the temperature (22–26°C) and relative humidity (55–65%). Standard laboratory chow and water were provided ad libitum. All animal experimental procedures were approved by Experimental Animal Care and Use Committee of the Institute of Materia Medica, Chinese Academy of Medical Sciences & Peking Union Medical College (No. 20140912).

The timeline for the development of CIA and treatment was shown in Figure 2A. Native chicken collagen type II (CII; 2 mg/ml in 0.1 M acetic acid; Sigma, St. Louis, MO, United States) was vortexed at 4°C overnight, and was emulsified in an equal volume of Freund’s complete adjuvant (CFA, Sigma) to a final concentration of 1 mg/ml. Each mouse received two intradermal injections 0.1 ml of CII emulsion at two separated sites tail base on day 1 and day 21, respectively. Mice receiving 0.05 M acetic acid were included as blank control. On day 30, immunized mice were randomly divided into four groups (n = 12) according to its clinical score, and receiving tamarixinin A (12.5 or 50 mg/kg per day through s.c. on the back), methotrexate (MTX, 2 mg/kg per day through gastric gavage) or vehicle, starting from day 30 to day 44 after the first CII injection. Arthritis scoring was performed on alternate days after the initiation of tamarixinin A / methotrexate treatment. On day 44, the mice were sacrificed by euthanasia and blood was collected for serum separation. The hind limbs were dissected and stored in 10% formalin for histopathological assessment.

The timeline for the development of AIA and treatment was shown in Figure 3A. AIA was induced by a single intradermal injection of 0.1 ml CFA (containing 10 mg of heat-killed Mycobacterium tuberculosis in 1 ml of paraffin oil) into the footpad of the right hind paw on day 1. Rats (n = 12) receiving a single dose of 0.1 ml of paraffin oil – water emulsion without Mycobacterium tuberculosis were included as blank control. On day 22, immunized rats were randomly divided into five groups according its clinical score, received tamarixinin A (6.25, 12.5, or 25 mg/kg per day through s.c. on the back), methotrexate (2 mg/kg per day through gastric gavage) or vehicle (equally volumes per day through s.c. on the back), starting from day 22 to day 35. Arthritis scoring, and arthritis diameter (cm) measuring by one tape measure method were performed on alternate days after the initiation of tamarixinin A/methotrexate treatment. On day 35, rats were sacrificed by euthanasia, the hind limbs were dissected, three of them were randomly chosen from each group and stored in 10% formalin for histopathological assessment, another of them were trimmed of skin, cut out ankle joints (sampling sites shown in Figure 4A) and homogenized in 3 ml of RIPA lysis buffer containing protease inhibitor cocktail (Roche Diagnostics, Mannheim, Germany) using a tissue homogenizer. The homogenates were incubated for 2 h on ice to obtain full lysis, and then centrifuged at 15,000 ×g for 15 min at 4°C, the supernatants were transferred in aliquots to new tubes and stored at -70°C before western blotting analysis.

The arthritis scoring were defined by using a semi-quantitative clinical scoring system for each paw, in which 0 = normal, 1 = definite erythema and swelling of the ankle or one digits, 2 = two or more joints involved or mild erythema and swelling of the entire paw, 3 = erythema and swelling extending from the ankle to the metatarsal joints of the entire paw and all digits, and 4 = ankylosing deformity with joint severe erythema and swelling. The arthritis score for each mouse was the sum of all paw scores (totaling a maximum score of 16 per mouse or rat).

The dissected mice or rat left hind paw joints were fixed in 10% formalin solution and decalcified using 12% disodium EDTA. After dehydration, specimens were then paraffin embedded, sectioned (5 μm thickness) for hematoxylin and eosin (H&E) staining, and observed under a light microscope. Histopathological scoring was performed in a blinded manner by experienced pathologists, and classified into four grades for inflammatory activity, synovial hyperplasia, cartilage degradation, and bone erosion, in which 0 = normal, 1 = slight, 2 = moderate, 3 = severe.

Peritoneal macrophages were isolated from C57BL/6 mice 3 days after intraperitoneal injection with 1.2 ml of 4% sterile thioglycolate broth. Then cells were seeded onto 96-well plates at a density of 2 × 10⁵ cells per well in RPMI 1640 medium with 5% fetal bovine serum for 2 h to allow the macrophages to adhere to the plates. The adherent macrophages were cultured in fresh medium containing tamarixinin A (5, 10, and 20 μM, respectively) for 2 h, then incubated with 1 μg/ml of LPS for 24 h. The culture supernatants were collected for further determination of nitric oxide (NO) and cytokines.

For western blotting analysis, the macrophages were plated in culture dishes (60 mm in diameter) at a density of 8 × 10⁶ cells, incubated with tamarixinin A (5, 10, and 20 μM, respectively) for 2 h, and then cultured with 1 μg/ml of LPS for 15 min. The cells were washed with PBS three times, and lysed with RIPA lysis buffer containing protease inhibitor cocktail (Roche Diagnostics, Mannheim, Germany) to extract the total cellular protein.

The cytokine levels in the culture supernatants of macrophages and serum of experimental mice or rats were determined by using murine or rat specific ELISA kits (Biolegend, San Diego, CA, United States) for TNF-α, IL-1β and IL-6 according to the manufacturer’s instructions. Serum levels of anti-mouse CII antibody in CIA mice were monitored by ELISA (Senbeijia Biological Technology Co., Ltd., Nanjing, China) as the manufacturer’s recommendation.

Culture supernatants of macrophages were collected and NO levels were determined by using Griess reagent (1% sulfanilamide, 0.1% N-1-naphthylenediamine dihydrochloride, and 2.5% phosphoric acid). The absorbance was measured at 570 nm with a microplate reader.

Analysis of the proteins extracted from macrophages or ankle joints by Western blot was performed using standard methods. Equivalent amounts of protein from each sample were separated in 10% SDS-polyacrylamide gel and blotted onto a PVDF membrane. The membrane was then blocked with 5% milk (BD, Sparks, MD, United States) and incubated with antibodies against p38, p65, p-p38, p-ERK, p-JNK1/2, p-p65, iNOS, and β-actin (Cell Signaling Technology, Danvers, MA, United States) overnight, and then was hybridized with HRP-conjugated secondary antibody for 1 h. The immunoreactive bands were visualized using a ECL system (CLINX, Shanhai, China). The relative intensities of bands were quantified using Image J.

Macrophages were plated in 96-well plates at a density of 5 × 10⁴ cells, treated with tamarixinin A (5, 10, and 20 μM, respectively) for 2 h, and then incubated with 1 μg/ml of LPS for 15 min. The cells were then washed and fixed with 4% paraformaldehyde for 30 min. Cells were incubated sequentially with 0.1% triton X-100 for 15 min and blocked with 5% BSA for 30 min. Next, cells were incubated with primary anti-p-p38 antibody (1:400) overnight at 4°C, and TRITC-conjugated secondary antibody (1:100) at room temperature for 2 h. Nuclei were counter stained with DAPI (Invitrogen, San Francisco, CA, United States) for 10 min. Cells were viewed using high-content screening (IN CELL Analyzer 1000, GE, United States). The quantitative description of cytoplasm to nucleus translocation process of p38 was calculated by comparing the average pixel intensity of the cytoplasmic area around nucleus (RING) with its average pixel intensity inside the nuclear region (CIRC), i.e., “CIRC-RING Avg Inten Diff” calculated by software.

Data were analyzed using SPSS software version 21.0 (SPSS Inc., Chicago, IL, United States) and expressed as mean ± SD. Comparison of multiple groups was performed using one-way analysis of variance (ANOVA), followed by the Bonferroni post hoc test when equal variances assumed, or Dunnett’s T3 post hoc test when equal variances not assumed, or Student’s t-test for comparison of two groups. Statistical significance was set at P < 0.05.

To investigate the pharmacological effects of tamarixinin A on RA development, the CIA model in DBA/1 mice was employed. On day 30, as significant swelling of hind feet developed compared with the blank control, mice received different therapies. As shown in Figure 2B, there was a significant decrease in body weight after immunization, compared to blank controls. After treatment with 50 mg/kg tamarixinin A, this loss was significantly decreased, compared with the CIA model group (P < 0.01).

After immunization, the arthritis symptoms in the CIA mice dramatically progressed and maintained the highest intensity during day 30 to day 44. The representative photographs of the hind paws in each group on day 44 are shown in Figure 2C. There was significant inflammatory swelling and erythema in control paws, and this was obviously relieved after tamarixinin A treatment. From day 35, compared with CIA model group, the mean clinical scores were significantly reduced after MTX, tamarixinin A 12.5 or 50 mg/kg treatment (P < 0.05 or 0.01) (Figure 2D).

As shown in Figures 2E,F, histological examinations showed that the CIA mice exhibited apparent degeneration of joint structure and narrowed joint space in the hind paw ankle joints with massive in soft tissue, synovial lining layer and cartilage destruction, and pannus formation. However, compare with CIA model mice, in tamarixinin A - or MTX-treated mice, the degree of inflammatory cell infiltration, bone erosion and degradation were significantly improved (P < 0.05 or 0.01).

Cytokines, such as TNF-α and IL-1β, are often associated with the progression and severity of arthritis. In this study, the level of TNF-α (Figure 2G) and IL-1β (Figure 2H) in serum were significantly increased compared with blank control group (P < 0.01). After tamarixinin A treatment, levels of TNF-α and IL-1β were significantly decreased, similar to that of MTX treated group.

In the CIA model, immunization by heterologous chicken CII results in the production of murine CII-specific autoantibodies. Results showed that CIA mice had significantly elevated levels of anti-mouse CII antibodies (P < 0.01). After tamarixinin A treatment, concentrations of anti-mouse CII autoantibodies were significantly reduced (Figure 2I).

This data suggests that tamarixinin A treatment elicits, to a certain extent, a therapeutic effect against arthritis in CIA mice.

Due to the complexity of human RA, using only one animal model cannot fully reflect the effects of the tested compound. Another frequently used animal model, the rat AIA model, was employed to confirm the anti- arthritic effect of tamarixinin A. As shown in Figure 3B, compared to control group, AIA rats exhibited serious swelling and erythema in hind feet since day 22, and maintained serious arthritis status until day 34. Mean clinical scores fluctuated in a narrow range, which were significantly lower after tamarixinin A (12.5 mg/kg and 25 mg/kg) and MTX treatment.

Compared with control rats, both the primary swelling of right hind paws (with adjuvant injected) and the secondary swelling of left hind paws (contralateral hind paw) in AIA rats exhibited significantly increased ankle diameter (P < 0.01) from day 22. Compared with the AIA model group, the primary paw swelling was significantly relieved by tamarixinin A 12.5, 25 mg/kg, and MTX treatment after day 29 (Figure 3C), and the secondary paw swelling was significant relief by tamarixinin A 25 mg/kg and MTX treatment after day 33 (Figure 3D).

Histological results show that the ankle joints in AIA rats exhibited apparent massive inflammatory cell infiltration in soft tissue, synovial lining layer, and cartilage destruction, pannus formation, and narrowed joint space. Tamarixinin A or MTX treatment might restrict the histological development of AIA (P < 0.05 or 0.01), as indicated by semi-quantitative scores in synovium hyperplasia, cartilage degradation, bone erosion, and soft tissue inflammation (Figures 3E,F).

Additionally, the level of TNF-α, IL-6 and IL-1β in joint space flushing fluid were tested by ELISA. Compared with the AIA model group, the levels of IL-6 were significantly decreased with tamarixinin A 12.5 mg/kg and 25 mg/kg treatment(Figure 3G), and the levels of IL-1β were significantly decreased after all doses of tamarixinin A treatment (Figure 3H). However, tamarixinin A treatment had an insignificant effect on the level of TNF-α (data not shown).

To develop RA, p38 and p65 play a critical role in regulating inflammatory signaling pathways. To investigate the potential mechanisms of anti- RA effects of tamarixinin A, the expression of p38, phosphorylation of p38 and p65 were investigated in ankle joints of AIA rats. Compared with the AIA model group, the expression of p65 was significantly decreased with tamarixinin A 12.5 mg/kg and 25 mg/kg treatment (Figure 4B). Compared with the AIA model group, the expression of p38 was significantly decreased with tamarixinin A 12.5 mg/kg and 25 mg/kg treatment (Figure 4C), and the expression of p-p38 was significantly decreased after all doses of tamarixinin A treatment (Figure 4D).

To confirm the anti- inflammatory effects in vitro, tamarixinin A was evaluated for its inhibition of NO and production of pro-inflammatory (TNF-α and IL-6) mediators in murine peritoneal macrophages stimulated by LPS. Compared with the LPS model group, the excessive NO (Figure 5A), TNF-α (Figure 5B) and IL-6 (Figure 5C) production was significantly inhibited (P < 0.05 or 0.01), and showed a dose-dependence.

Excessive NO in inflammatory diseases are produced through the inducible nitric oxide synthase (iNOS) pathway and amplify the inflammatory response. In this study, macrophages stimulated with LPS showed significant induction of iNOS expression, and as expected, the expression was suppressed upon treating the stimulated cells with tamarixinin A (Figure 5D).

To develop RA, MAPK and NF-κB pathways play a critical role. To further investigate whether the anti- RA effects of tamarixinin A are associated with MAPK and NF-κB pathways inhibiton, phosphorylation of ERK, JNK1/2, p38, and p65 was investigated in LPS induced macrophages. As shown in Figure 6, LPS stimulation rapidly induced the phosphorylation of ERK, p38, JNK1/2 and p65 within 15 min in macrophages (P < 0.01), and tamarixinin A (5–20 μM) dose-dependently suppressed the phosphorylation of ERK (Figure 6A), JNK (Figure 6B), p38 (Figure 6C) and p65 (Figure 6D).

Consequent to the above observations of p38 phosphorylation, nuclear translocation of p38 was investigated by high-content immunofluorescence screening. As shown in Figure 7, a significant nuclear translocation of p38 was induced with LPS stimulation which was significantly suppressed in a concentration - dependent manner (5–20 μM) after tamarixinin A treatment.