Paeoniflorin ameliorates collagen‐induced arthritis via suppressing nuclear factor‐κB signalling pathway in osteoclast differentiation

Summary

Paeoniflorin (PF), extracted from the root of Paeonia lactiflora Pall, exhibits anti‐inflammatory properties in several autoimmune diseases. Osteoclast, the only somatic cell with bone resorbing capacity, was the direct cause of bone destruction in rheumatoid arthritis (RA) and its mouse model, collagen‐induced arthritis (CIA). The objective of this study was to estimate the effect of PF on CIA mice, and explore the mechanism of PF in bone destruction. We demonstrated that PF treatment significantly ameliorated CIA through inflammatory response inhibition and bone destruction suppression. Furthermore, PF treatment markedly decreased osteoclast number through the altered RANKL/RANK/OPG ratio and inflammatory cytokines profile. Consistently, we found that osteoclast differentiation was significantly inhibited by PF through down‐regulation of nuclear factor‐κB activation in vitro. Moreover, we found that PF suppressed nuclear factor‐κB activation by decreasing its translocation to the nucleus in osteoclast precursor cells. Taken together, our new findings provide insights into a novel function of PF in osteoclastogenesis and demonstrate that PF would be a new therapeutic modality as a natural agent for RA treatment and other autoimmune conditions with bone erosion.

Abbreviations

C II

type II collagen

CIA

collagen‐induced arthritis

IL

interleukin

M‐CSF

macrophage colony‐stimulating factor

micro‐CT

micro‐computed tomography

MMP‐9

matrix metalloprotease‐9

NF‐κB

nuclear factor κ‐B

OPG

osteoprotegerin

PBS

phosphate‐buffered saline

PF

Paeoniflorin

RANKL

receptor activator of nuclear factor κ‐B ligand

RANK

receptor activator of nuclear factor κ‐B

RA

rheumatoid arthritis

TGF‐β

transforming growth factor‐β

TRAP

tartrate‐resistant acid phosphatase

Introduction

Physiological bone remodelling is the predominant metabolic process in which there is an exquisite homeostasis between the quantity of bone removed in a resorption cycle and the quantity of bone synthesis to substitute it. Imbalance of bone remodelling can lead to gross perturbations in skeletal structure and function, and probably to morbidity and shortening of lifespan.1, 2, 3 Rheumatoid arthritis (RA) is a systemic autoimmune disease characterized by persistent synovial inflammation and joint destruction. Its bone destruction is caused by dysregulation of bone remodelling, which favours resorption, mainly through high levels of pro‐inflammatory cytokines.4, 5, 6

Osteoclasts are multinucleated cells derived from the fusion of monocyte–macrophage lineage precursor cells and act as the only somatic cells to present bone‐resorbing capacity.7 Macrophage colony‐stimulating factor (M‐CSF) and receptor activator of nuclear factor κ‐B ligand (RANKL) are the two essential cytokines in osteoclast differentiation.8 Lymphocytes and fibroblasts in inflammatory joints secrete RANKL to promote osteoclast differentiation and activation through binding the receptor activator of nuclear factor κ‐B (RANK) expressed on the osteoclast precursors or mature osteoclasts. Excess osteoclasts lead to bone dysregulation and an outcome of bone loss, which explains the bone erosion in RA.9, 10 Focusing on osteoclasts has been a new target for developing therapeutic strategies to block or intervene in the pathological bone erosion in RA.

Collagen‐induced arthritis (CIA) is initiated by autoantibodies that bind to type II collagen (C II) in the cartilage matrix and is a frequently used mouse model to study the effect of novel therapeutics for RA. There are many similarities between CIA and RA, such as symmetrical joint involvement, synovitis and inflammatory cell infiltration.11

In recent years, many advances have been made in the treatment of RA. Treatment strategies currently comprise non‐steroidal anti‐inflammatory drugs, corticosteroids, disease‐modified anti‐rheumatic drugs and biological response modifiers (‘biologicals’).12 However, the currently available treatments remain far from ideal due to cell toxicity, drug resistance and high costs. Compared with synthetic drugs, traditional Chinese medicine has fewer side effects, lower costs and is less prone to causing drug resistance.13

Total glucosides of peony, extracted from the root of Paeonia lactiflora Pall, has been broadly used to treat autoimmune diseases like RA,14, 15, 16 multiple sclerosis,17 Sjögren syndrome18and systemic lupus erythematosus.19 Paeoniflorin (PF) is the most abundant bioactive substance of total glucosides of peony and exerts various pharmacological effects including anti‐inflammatory and immune‐regulation activities. It has been reported that PF decreases the inflammatory response by regulating the balance between pro‐inflammatory and anti‐inflammatory cytokines in allergic contact dermatitis.20 Moreover, a number of studies have reported the suppressive effect of PF on immune cells like B cells,21 T helper type 1 and type 17 cells,22 dendritic cells23 and M1 macrophages.24 However, there is no report about the effect of PF on the differentiation of osteoclasts.

Hence, we aimed to explore the potential therapeutic effect of PF on CIA bone destruction and its underlying mechanism on osteoclast differentiation and formation. We hope that this study will identify a novel role of PF in osteoclastogenesis and demonstrate that PF is an effective natural agent for the treatment of RA. It will help to provide a new treatment strategy for other inflammatory diseases with osteoclast‐related bone erosion.

Materials and methods

Mice and reagents

Male DBA/1 mice (8 weeks old) and male C57BL/6 (6 weeks old) were purchased from the Shanghai Laboratory Animal Centre (Chinese Academy of Science, Shanghai, China). All mice were maintained under pathogen‐free conditions at Shanghai JiaoTong University School of Medicine. All experimental procedures were performed according to the Animal Care and Use Committee guidelines. Paeoniflorin was purchased from Yilin Biological Technology Co., Ltd. (Shanghai, China) and had a purity of 98%. It was dissolved in sterile phosphate‐buffered saline (PBS) to provide stock solution.

Induction and treatment of CIA

To induce the arthritis, DBA/1 mice were intradermally injected with C II (150 μg/mouse, Chondrex, Redmond, WA) in Freund's complete adjuvant (Sigma, Ronkonkoma, NY) in the tail, followed by 21 days of injections with C II (75 μg/mouse) in Freund's incomplete adjuvant (Sigma) to boost. After boosting, the mice were divided into two groups randomly and the two groups were given intraperitoneal injections of PF at a dose of 7·5 mg/kg or PBS, respectively. Mice were observed daily and scored for disease severity according to the standards as described previously.25

Histochemical analysis

Mice were killed on day 36, and the knee joints were collected and fixed in 4% paraformaldehyde for 24 hr. Then the joints were decalcified in 10% EDTA decalcification solution for 1 month, embedded in paraffin and sectioned using routine methods. Haematoxylin & eosin, safranin O‐fast green and tartrate‐resistant acid phosphatase (TRAP) staining were performed on the knee sections from different groups.

Micro‐computed tomography

Three‐dimensional micro‐computed tomography (micro‐CT) analysis was performed on the hind legs from naive, PF‐treated or PBS‐treated CIA mice. CT scanning was performed using SkyScan1176. Three‐dimensional microstructural image data were reconstructed using CT VOX software, and structural indices were calculated using CT analyzer software.

RNA extraction and quantitative real‐time PCR

Tissues or cells were lysed in Trizol (Life, Southfield, MI) to extract total RNA. First, cDNA was synthesized using the Prime Script RT Master Mix kit (TaKaRa, Shiga, Japan). Then osteoclastogenesis‐related genes and cytokine genes were measured by real‐time PCR. The reactions were performed in 384‐well plates by an ABI viia 7 real‐time PCR System with Power SYBR Green Master Mix (Life). Actin was used to normalize those genes. Primer sequences used are listed in Table 1.

Table 1.

List of primer sequences

Name	5′–3′	Sequence
Tartrate‐resistant acid phosphatase	Forward	CCAATGCCAAAGAGATCGCC
	Reverse	TCTGTGCAGAGACGTTGCCAAG
Receptor activator of nuclear factor κ‐B	Forward	CCAGGAGAGGCATTATGAGCA
	Reverse	ACTGTCGGAGGTAGGAGTGC
Matrix metalloproteinase‐9	Forward	CTGGACAGCCAGACACTAAAG
	Reverse	CTCGCGGCAAGTCTTCAGAG
Cathepsin	Forward	GACGCAGCGATGCTAACTAA
	Reverse	CCAGCACAGAGTCCACAACT
Receptor activator of nuclear factor κ‐B ligand	Forward	GCAGATTTGCAGGACTCGACT
	Reverse	CCCCACAATGTGTTGCAGTT
Osteoprotegerin	Forward	CCTTGCCCTGACCACTCTTAT
	Reverse	CACACACTCGGTTGTGGGT
Matrix metalloproteinase‐3	Forward	ACATGGAGACTTTGTCCCTTTTG
	Reverse	TTGGCTGAGTGGTAGAGTCCC
Interleukin‐1β	Forward	TTTTTGTTGTTCATCTCGGAGCCTGTAG
	Reverse	GAGCACCTTCTTTTCCTTCATCTTTG
Interleukin‐6	Forward	TAGTCCTTCCTACCCCAATTTCC
	Reverse	TTGGTCCTTAGCCACTCCTTC
Tumour necrosis factor‐α	Forward	CCCTCACACTCAGATCATCTTCT
	Reverse	GCTACGACGTGGGCTACAG
Interleukin‐10	Forward	CTTACTGACTGGCATGAGGATCA
	Reverse	GCAGCTCTAGGAGCATGTGG
Transforming growth factor‐β	Forward	CTCCCGTGGCTTCTAGTGC
	Reverse	GCCTTAGTTTGGACAGGATCTG
β‐actin	Forward	TGTCCACCTTCCAGCAGATGT
	Reverse	AGCTCAGTAACAGTCCGCCTAG

Splenocyte culture and cytokine detection

Spleens were removed from the dead mice on day 36 and homogenized to single‐cell suspensions. After red blood cells were lysed, cells were suspended in RPMI‐1640 complete medium containing 10% fetal bovine serum, and seeded in round‐bottom 96‐well plates at a density of 10⁶ cells per well. The splenocytes were stimulated with C II (40 µg/ml) for 72 hr and its supernatants were collected and stored at −80°. The concentrations of the cytokines like interleukin‐1β (IL‐1β), IL‐6, tumour necrosis factor‐α, transforming growth factor‐β (TGF‐β) and IL‐10 were detected by ELISA Kit (R&D Systems, Minneapolis, MN) according to the manufacturers' instructions.

Osteoclast differentiation and TRAP staining

Bone marrow cells were flushed from tibiae and femora of 6‐week‐old male C57BL/6 mice, seeded on a 24‐well plate at a density of 0·5 million cells per well. Then bone marrow cells were induced to osteoclast precursors with M‐CSF (50 ng/ml, PeproTech, Rocky Hill, NJ) for 4 days in α‐MEM medium containing 10% fetal bovine serum. Then osteoclast precursors were cultured to osteoclasts for another 4 days with RANKL (50 ng/ml, R&D Systems) and M‐CSF (50 ng/ml). Meanwhile, medium was replaced every 2 days. Paeoniflorin was administered during culture from day 3 at different concentrations (0, 2, 10, 50 μm) to explore the effect of PF on osteoclast differentiation. After osteoclasts had formed, cells were washed with PBS and fixed with 4% paraformaldehyde. TRAP staining using a kit (Sigma) was employed for the cultured osteoclasts following the manufacturer's instructions. The cells were observed under a light microscope. TRAP‐positive cells containing more than three nuclei were regarded as osteoclasts.

Western blotting analysis

Osteoclast precursors were stimulated by RANKL for 0, 5 and 15 min. Then cells were lysed with RIPA buffer containing phenylmethylsulfonyl fluoride and protease inhibitor cocktail. After protein quantification, the samples were separated by 10% SDS–PAGE and transferred onto nitrocellulose membranes. The protein bands were visualized by enhanced chemiluminescence (Thermo, Waltham, WA) and the densitometry was quantified using image J software (NIH, Waltham, WA).

Immunofluorescence analysis

Bone marrow cells were cultured on slices and put into 6‐well‐plates to generate osteoclast precursors as described previously with or without PF treatment. Then cells were stimulated with 50 ng/ml RANKL for 15 min. Non‐treated cells were regarded as a negative control. Cells were fixed in paraformaldehyde for 15 min and then washed with PBS for 5 min. Triton‐X‐100 was used to improve the permeability of the membrane. After that, slices were washed with PBS for 5 min and blocked in 1% bovine serum albumin for 2 hr. Next, they were incubated with p65 antibody (CST, Danvers, MA) at 4° overnight. Secondary antibody labelled with Alexa Fluor 488 (Thermo) and DAPI were used to stain the proteins and nuclei, respectively. The nuclear factor‐κB (NF‐κB)/p65 translocation was observed under confocal microscope.

Luciferase reporter gene assay

RAW 264.7 cells were transfected with pGL4.32 (luc2P/NF‐κB‐RE/Hygro) plasmid (Promega, Madison, WI). The cells were plated in 24‐well plates in triplicate and pre‐incubated with PF for 24 hr. Then cells were lysed in lysis buffer (Promega) and the luciferase was detected in a multi‐well plate luminometer by a Luciferase Assay System (Promega) per the manufacturer's instructions.

Statistical analysis

The data are presented as the mean ± standard error of the mean (SEM) and analysed using prism for windows (version 5·0, GraphPad Software, Inc., San Diego, CA). Statistical analysis was performed by a two‐tailed Student's t‐test and P‐values < 0·05 were considered to be statistically significant.

Results

PF attenuates CIA severity

First, we verified that PF could ameliorate CIA, a mouse model induced by collagen with adjuvant to mimic the pathology of RA. We initiated PF treatment on day 21. Our data showed that PF diminished disease severity and joint swelling in CIA (Fig. 1a,b). In addition, PF alleviated joint inflammation by decreasing inflammatory cell infiltration in CIA (Fig. 1c). These results indicated that PF could alleviate the clinical symptoms and systematic inflammation of CIA.

Figure 1.

Paeoniflorin (PF) alleviated the clinical symptoms of collagen‐induced arthritis (CIA) in mice. CIA was induced in DBA/1 mice by intradermal injection with collagen type II (C II) in the tail on day 0 and day 21. Then mice were injected intraperitoneally with phosphate‐buffered saline (PBS) or PF every day from day 21. The DBA/1 mice in the naive group were without any treatment, as a blank control. Mice were killed on day 36. (a) Clinical scores of arthritis; n = 4 per group. Data are shown as mean ± SEM. *P < 0·05. (b) Representative images of swollen paws. (c) Representative haematoxylin & eosin staining of knee joint sections.

PF ameliorates joint damage by inhibiting osteoclast formation

Inflammation in RA is accompanied by cartilage and bone destruction. We explored the effect of PF on cartilage damage by safranin O‐fast green staining in the joints. The results showed that naive mice had thick cartilage with a smooth surface and without any structural destruction, whereas CIA mice had seriously destroyed structure and rough surface, which revealed that PF could rescue CIA mice from severe cartilage destruction (Fig. 2a).

Figure 2.

Paeoniflorin (PF) ameliorated articular cartilage and bone destruction in collagen‐induced arthritis (CIA) mice. Knee joint sections and hind legs were obtained from naive, phosphate‐buffered saline (PBS) ‐treated and PF‐treated CIA mice on day 36. (a) Representative safranin O‐fast green staining visualized the degree of cartilage destruction of the knee joint sections. (b) Representative three‐dimensional renditions of hind paws and knee joints, scanned by micro‐computed tomography. Arrows indicate bone erosion areas. (c) Representative radiographs of knees and three‐dimensional renditions of tibia. (d) Representative tartrate‐resistant acid phosphatase staining of knee joints. Arrows indicate TRAP⁺ cells.

We further conducted micro‐CT and three‐dimensional reconstruction on the hind limbs to investigate further the effect of PF on bone damage. We found that both paws and knee joints from the CIA group exhibited severe bone destruction but there was a trend to decrease the degree of bone erosion after PF treatment (Fig. 2b). In addition, the radiographs of knees revealed that the bone erosion in CIA mice was more severe than in naive mice and PF could rescue the bone loss in CIA mice (Fig. 2c). This demonstrated that PF could protect CIA mice from severe bone destruction.

Osteoclast is the only direct bone resorbing cell characterized by TRAP expression. Through TRAP staining, we found that osteoclast number was decreased in the PF‐treated group compared with the CIA group (Fig. 2d). Collectively, these results suggested that PF suppressed bone destruction by decreasing osteoclast formation in CIA mice.

PF decreases the mRNA expression level of osteoclastogenesis‐related genes in pathogenic sites

TRAP, matrix metalloproteinase‐9 (MMP‐9) and cathepsin are characteristic genes during osteoclastogenesis. MMP‐3 can activate various pro‐MMPs and cleave extracellular components. The RANKL/RANK/osteoprotegerin (OPG) regulatory axis plays a crucial role in osteoclast differentiation and function. To validate PF inhibitory effect on osteoclasts, we examined mRNA expression levels of osteoclastogenesis‐related genes in pathogenetic sites. We found that PF not only significantly decreased the expression of bone‐destruction‐related genes, including TRAP, MMP‐9, MMP‐3 and RANKL, but also increased the expression of the bone protective gene OPG in CIA mice. In addition, the ratio of RANKL/OPG was also decreased by PF treatment (Fig. 3a).

Figure 3.

Paeoniflorin (PF) altered osteoclastogenesis‐related mediator and cytokine profile of collagen‐induced arthritis (CIA) mice. The mRNA levels of the osteoclastogenesis‐related markers (a), and cytokines (b) in the paws were detected by real‐time PCR. (c) The concentrations of cytokines in the splenocyte culture supernatants were detected by ELISA. Data are shown as mean ± SEM. *P < 0·05, **P < 0·01, ***P < 0·001.

We measured the mRNA levels of a panel of cytokines that were critically involved in osteoclast differentiation and found a concomitant down‐regulation of pro‐inflammatory cytokine genes (such as IL‐1β) and up‐regulation of anti‐inflammatory genes (such as IL‐10 and TGF‐β) in the paws of PF‐treated mice, compared with CIA mice (Fig. 3b). Furthermore, we detected a panel of cytokines in splenocyte culture supernatants. We found a decreased production of IL‐1β and IL‐6, and an increased production of TGF‐β in the PF‐treated group compared with the CIA group (Fig. 3c). Interleukin‐10 in the culture system was also observed, but the concentration was too low to be detected. Collectively, these results showed that PF altered the osteoclastogenesis mediators and cytokine profile, which reveals that PF may decrease osteoclast formation by changing cytokine composition in CIA.

PF inhibits RANKL‐induced osteoclastogenesis in vitro

Given the PF inhibitory effect on osteoclast formation, we established an RANKL‐induced osteoclast differentiation system in vitro. Bone marrow cells were differentiated into osteoclast precursors followed by further RANKL stimulation to osteoclasts. We found that the osteoclast precursor viability was not affected by PF at concentrations up to 250 µm in vitro (data not shown) and PF significantly suppressed osteoclast formation in a dose‐dependent manner (Fig. 4a,b). Moreover, the related genes of osteoclast formation such as TRAP, cathepsin and MMP‐9 were up‐regulated in the RANKL‐elicited group and reversed after PF administration (Fig. 4c). These results provided compelling evidence that PF treatment suppressed osteoclast differentiation.

Figure 4.

Paeoniflorin (PF) suppressed osteoclast differentiation in vitro. Bone marrow (BM) cells from naive C57BL/6 mice were induced to osteoclasts with receptor activator of nuclear factor κ‐B ligand (RANKL) and macrophage colony‐stimulating factor (M‐CSF). PF concentrations ranged from 0 to 50 μm as indicated above. (a) Representative tartrate‐resistant acid phosphatase (TRAP) staining in the in vitro cultured system. (b) Quantification of TRAP‐positive multinuclear cells (nuclei > 3) in TRAP staining. (c) The mRNA levels of osteoclast‐specific markers were detected by real‐time PCR. Data are shown as mean ± SEM of three independent experiments. *P < 0·05, **P < 0·01, ***P < 0·001.

PF suppresses RANKL‐induced NF‐κB signalling pathway

During osteoclastogenesis, the NF‐κB signalling pathway is an essential molecular event induced by RANKL. Phosphorylation of p65 (p‐p65) plays an important role in NF‐κB pathway. Therefore, we explored whether PF had any inhibitory effect on the NF‐κB signalling pathway in osteoclast differentiation through immunoblotting. The result indicated that the p‐p65 expression was suppressed by PF at a 50 µm concentration (Fig. 5a,b), suggesting that PF might suppress osteoclast formation through negative regulation of the NF‐κB pathway.

Figure 5.

Paeoniflorin (PF) inhibited p‐p65 expression in the process of osteoclast differentiation. Cell lysates were analyzed by western blot and probed with anti‐p‐p65, anti‐p65 and anti‐actin. (a‐b) Bone marrow (BM) cells were induced to osteoclast precursors for 4 days in macrophage colony‐stimulating factor (M‐CSF) and then induced to osteoclasts in receptor activator of nuclear factor κ‐B ligand (RANKL) and M‐CSF for another 4 days. PF was administrated from day 3 at different concentrations in vitro. (a) Representative immunoblotting images of p‐p65, p65 and actin from osteoclasts. (b) Grey level of p‐p65 in osteoclasts. (c‐d) BM cells were induced to osteoclast precursors for 4 days in M‐CSF and then stimulated with RANKL for different time as indicated. PF was administrated on day 3. (c) Representative immunoblotting images of p‐p65, p65 and actin from osteoclast precursors. (d) Grey level of p‐p65 in osteoclast precursors. Data are shown as mean ± SEM of three independent experiments. *P < 0·05, **P < 0·01.

Furthermore, we assessed the p65 and p‐p65 expression levels in osteoclast precursors stimulated by RANKL for different times. After RANKL stimulation, the p‐p65 expression was significantly down‐regulated by PF, which verified that PF suppressed NF‐κB activation in osteoclast culture (Fig. 5c,d). Immunofluorescence staining demonstrated that PF decreased p65 translocation from the cytoplasm into the nucleus (Fig. 6a). Paeoniflorin consistently inhibited RANKL‐induced NF‐κB activation in RANKL‐stimulated RAW264.7 cells, using the luciferase activity assay (Fig. 6b). RAW264.7 cells have been widely used as a model for luciferase reporter gene assay as a substitute for bone marrow cells in exploring the osteoclast signalling pathway.26, 27 Hence we used RANKL to stimulate RAW264.7 cells for the luciferase activity assay and found that PF could inhibit RANKL‐induced NF‐κB activation in the system (Fig. 6b). In conclusion, these data showed that PF inhibited osteoclast formation mainly through NF‐κB pathway inhibition.

Figure 6.

The effect of Paeoniflorin (PF) on nuclear factor‐κB (NF‐κB) signalling pathway. Bone marrow (BM) cells were induced to osteoclast precursors for 4 days and pre‐incubated with PF (50 μm) from day 3. Then cells were stimulated with receptor activator of nuclear factor κ‐B ligand (RANKL) for 15 min. (a) Representative images of p65 immunofluorescence staining. Arrows indicate NF‐κB protein in the nucleus. (b) NF‐κB luciferase activity of RAW264.7 cells. Cells were transfected with pGL4.32 plasmid and pre‐incubated with PF for 24 hr. Then cells were stimulated by RANKL for 15 min. Data are shown as mean ± SEM of three independent experiments. **P < 0·01.

Discussion

Total glucosides of peony exhibit various pharmacological effects, including diminishing pain and joint swelling and alleviating disease severity in experimental arthritis.28 However, the mechanism of how PF treats CIA is little known. This study first provides the proof that PF inhibited osteoclast differentiation through the NF‐κB signalling pathway, subsequently suppressing bone destruction, resulting in CIA alleviation.

First, our study verified that PF diminished the symptoms of arthritis mainly through reduction of joint inflammation and reduction of destruction of cartilage and bone. Bone undergoes the process of remodelling, which involves resorption and formation and is the predominant metabolic process regulating bone structure and function with the key participants being the osteoclasts. As osteoclasts act as the only somatic cells to produce bone‐resorbing capacity, we went on to detect their number in the joints. The results demonstrated that PF alleviated joint destruction by decreasing the number of osteoclasts, leading to disease amelioration in CIA. TRAP,29 cathepsin30 and MMP‐931 are typical genes of osteoclast lineage. The expression of osteoclastogenesis‐related genes like TRAP and MMP‐9 was down‐regulated in the pathogenic sites of PF‐treated CIA mice, which is in accordance with osteoclast reduction.

RANKL is a member of the tumour necrosis factor superfamily and a potent osteoclast induction mediator.32, 33 RANK, expressed on the osteoclast precursors and mature osteoclasts, can bind to RANKL. OPG is a decoy receptor of RANKL that negatively controls osteoclast formation.34 Hence, the RANKL/RANK/OPG regulatory axis plays a crucial role in the differentiation and function of osteoclasts. Our data showed that PF down‐regulated RANKL gene expression and up‐regulated OPG gene expression in arthritic mice, which revealed a potential therapeutic mechanism of PF by altering RANKL/RANK/OPG ratio, which accounts for the osteoclast number reduction.

Joint inflammation is mediated by increased pro‐inflammatory cytokines as well as decreased anti‐inflammatory cytokines. Inflammatory cytokines have a particular role in the pathogenesis of focal bone erosion in inflammatory arthritis. Both IL‐1β and IL‐6, predominant pro‐inflammatory cytokines in potentiating osteoclastogenesis, are produced in substantial quantities by inflamed rheumatoid synovial tissues.35, 36 Interleukin‐10 and TGF‐β are anti‐inflammatory cytokines that can suppress bone resorption.37 Our results demonstrated that PF can weaken IL‐1β and IL‐6 expression and collaboratively enhance TGF‐β and IL‐10 expression in CIA, which indicated a protective mechanism of PF on joint inflammation and bone destruction. Those cytokine changes were accompanied by a reduction of inflammatory cell infiltration and alleviation of joint synovium inflammation. Taken together, PF decreased osteoclast number and so alleviated bone destruction through regulating RANKL/RANK/OPG proportion and cytokine profile in the arthritis model.

Our study has provided a detailed account of the novel role of PF in osteoclast differentiation using multiple in vivo and in vitro experimental systems. Given the proof that PF can treat CIA in vivo, we explored the direct effect of PF on osteoclast differentiation in vitro and found that PF pre‐treatment of osteoclast precursors can inhibit osteoclast differentiation. Moreover, it remarkably decreased the expression of osteoclast‐specific genes, such as TRAP, cathepsin and MMP‐9. This was in accordance with the reduction of osteoclast number in vivo, which provided a novel feature of the mechanism of PF in bone erosion.

RANKL binds to RANK to activate distinct signalling cascades. Among them, the NF‐κB pathway is one of the root downstream signalling pathways to initiate osteoclast differentiation.38, 39, 40 p65, a subunit of NF‐κB, can translocate into the cell nucleus from cytoplasm and be phosphorylated to work. It has been reported that NF‐κB can induce expression of TRAP, cathepsin and other genes associated with osteoclast resorptive function.41 As PF can decrease the typical gene expression of osteoclasts, we assessed the expression of NF‐κB signalling pathway related proteins in osteoclast differentiation and found that PF suppressed p65 activation. In addition to detection of the NF‐κB signalling pathway in osteoclast precursors, we confirmed that PF played a role in osteoclast differentiation through suppressed p65 phosphorylation. Moreover, we found the potential mechanism by which PF affected NF‐κB signalling pathway was by preventing p65 from translocating into the cell nucleus. Additionally, we used a vector containing an NF‐κB responsive element to verify the PF inhibitory effect on NF‐κB activation in RAW264.7 cells and the result was consistent with the former.

Collectively, we proved an important role of PF, which was capable of inhibiting CIA. We first provided evidence that PF has an inhibitory effect on osteoclast differentiation and clarified the potential mechanism systematically. Suppression of osteoclast‐specific genes by PF led to the inhibition of osteoclast differentiation through the attenuation of the NF‐κB signalling pathway. We clarify that the protective effect of PF in CIA was mainly through decreasing osteoclast formation and inflammatory infiltration in paws. Taken together, our data provide compelling proof for a previously undescribed role of PF in osteoclast differentiation and lay a solid foundation for PF as a potential treatment for RA and perhaps other autoimmune conditions with bone destruction.

Disclosures

The authors have declared that no competing interests exist.