Female Sprague-Dawley (SD) rats (180–200 g) and male C57BL/6 mice (6 weeks old) were purchased from the National Institutes for Food and Drug Control [animal license number: SCXK (Beijing) 2014-0013]. All animals were housed in the Experimental Animal Center of the Institute of Clinical Medical Sciences, China-Japan Friendship Hospital (Experimental Animal Center license number: SCXK (Beijing) 2016-0043). All experiments received ethic approval by the Institute of Clinical Medical Sciences, China-Japan Friendship Hospital, Beijing, China and were conducted within ethical limits. The animals were maintained in a specific pathogen-free environment with a temperature of 23°C (±2°C) and were subjected to a 12 h light/dark cycle.

The 40 female rats were randomly divided into two groups according to weight. Then, 32 of the rats were anesthetized, and their ovaries were removed according to standard surgical procedures. The other 8 rats were treated as the control group, in which fat tissue near the ovaries on both sides were removed. Twenty-eight days after castration, the 32 castrated rats were immunized intradermally at the tail root with 100 µl of a fully emulsified mixture of bovine type II collagen (Chondrex, Redmond, United States) and isopycnic incomplete Freund’s adjuvant (IFA; Chondrex) and received booster injections in the same way on day 7 after the initial immunization. Meanwhile, control group rats underwent the same method of injection with an equal volume of saline water.

On the 14th day after immunization, rats that successfully developed arthritis were divided into three groups as follows: 1) castrated CIA, 2) castrated CIA plus methotrexate (MTX; Shanghai Sine Pharmaceutical, Shanghai, China) treatment, and 3) castrated CIA plus YSJB (Jiangsu Zhengda Qingjiang Pharmaceutical, Jiangsu, China) treatment. The rats in the MTX group and YSJB group were treated with MTX (1.02 mg/kg body weight) and YSJB (3.87 g/kg body weight), respectively, by gavage twice a day. Meanwhile, rats in the control group and CIA group received an equal volume of distilled water (10 ml/kg). Rats in each group were given the treatments for 3 days. One hour after the last dose, all rats were anesthetized. Blood samples were collected from the abdominal aorta and centrifuged to obtain serum (quality control data are shown in Supplementary Material S1 and Supplementary Material S2).

OC progenitor cells (OPCs) were isolated and cultured as previously reported (Xu et al., 2016). The mice were sacrificed by cervical dislocation. Femurs and tibias were isolated, and the ends were trimmed. The bone marrow cavity was rinsed with alpha-modified Eagle’s medium (α-MEM; HyClone, Logan, United States), and bone marrow cells (BMCs) were collected by centrifuging the medium containing bone marrow. Subsequently, cells were cultured in α-MEM supplemented with 10% fetal bovine serum (FBS; PAN Biotech GmbH, Aidenbach, Germany), 1% penicillin-streptomycin (PS; Gibco BRL, Grand Island, NY, United States) and recombinant murine M-CSF (20 ng/ml; PeproTech, Rocky Hill, United States) after red blood cell (RBC) lysis buffer (Solarbio, Beijing, China) was used to deplete RBCs. Twenty-four hours later, nonadherent cells were collected and cultured in tissue culture plates at an appropriate density in α-MEM containing 10% FBS and M-CSF (20 ng/ml). Three days later, adherent cells were considered OPCs. To induce OPCs to differentiate into OCs, the cells were cultured for an additional 5 days in fresh differentiation α-MEM containing 10% FBS, 1% PS, M-CSF (20 ng/ml), and RANKL (50 ng/ml; PeproTech). Complete fresh differentiation medium was used every couple of days, and the concentration of each factor remained constant.

Suspensions of splenic cells were obtained by passing spleen tissue isolated from C57BL/6 mice through a 200-mesh metal mesh grid in RPMI-1640 (Gibco) medium, followed by filtration with a 70 mm cell strainer (BD Biosciences, San Jose, United States). Spleen cells were sorted using an EasySep™ Mouse CD4 Positive T Cell Pre-Enrichment Kit (STEMCELL Technologies, Vancouver, Canada) to obtain CD4⁺ T cells, and then negative selection was used to obtain CD4⁺CD25⁻ Tregs using an EasySep^® Mouse CD25 Positive Selection Kit (STEMCELL). CD4⁺CD25⁻ Tregs were resuspended in differentiation RPMI-1640 medium containing 10% FBS, 1% PS, recombinant murine IL-2 (1,000 U/mL; PeproTech), TGF-β1 (10 ng/ml; PeproTech) and anti-CD28 (2 μg/ml; BD Biosciences) and cultured in a round-bottom 96-well plate with anti-murine CD3 (10 μg/ml; BD Biosciences) at a density of 5 × 10⁵ cells/well (200 μl per well). Half of the differentiation medium was refreshed every couple of days and the concentration of each factor remained constant. CD4⁺CD25⁺ Tregs were successfully positively selected with an EasySep^® Mouse CD25 Positive Selection Kit after 7 days and were used for subsequent experiments. We explored the method during the early stage. Flow cytometry was used to identify the phenotype of Tregs (Xu et al., 2016).

Nonadherent BMCs were seeded in 24-well culture plates at a density of 5 × 10⁵ cells/well. Three days later, Tregs (2 × 10⁴) were added to each well. Cells were then maintained in differentiation α-MEM with a 1:25 ratio of Tregs: OPCs. Every 2 days, half of the medium was replaced with fresh differentiation α-MEM. The bottom was not touched when changing medium.

Each culture system was divided into four groups: DMSO, AG490, MOCK and siRNA. DMSO was used as a control for AG490, and MOCK was used as a control for siRNA. Each group was further divided into four subgroups as follows: 1) control, 2) CIA, 3) CIA+MTX, and 4) CIA+YSJB. OPCs and Tregs were cultured in the absence or presence of AG490 (MedChemExpress, New Jersey, United States) and siRNA (SyngenTech, Beijing, China). Twenty-four hours later, 10% drug-containing serum that had been previously obtained from rats was added to the differentiation medium.

The effect of AG490 on cell viability was measured using a CCK-8 assay. OPCs were cultured in 96-well plates at a density of 4 × 10⁴ cells/well in differentiation α-MEM (20 ng/mL M-CSF, 50 ng/ml RANKL, 10% FBS, 1% PS) with AG490 (0, 2.5, 5, 10, 20, 50 μM). The concentration of each factor was kept constant when adding fresh medium. Five days later, fresh α-MEM containing 10% CCK-8 was used to replace the differentiation α-MEM. The absorbance was measured at a wavelength of 450 nm.

siRNA was purchased from SyngenTech. All sequences are shown in Supplementary Material S3. Cells were transfected with siRNA using a Lipofectamine^® 3,000 transfection kit (Thermo Fisher Scientific, Waltham, United States) according to the manufacturer’s protocol. qRT-PCR was used to confirm the transfection efficiency.

Nonadherent BMCs were seeded in 96-well culture plates at a density of 1 × 10⁵ cells/well (200 μl per well). After the OPCs were cultured for 5 days, TRAP staining was performed according to the instructions of the TRAP staining kit (Sigma-Aldrich, Louis, United States) to assess osteoclastogenesis. OC formation was scored by the number of TRAP⁺ cells that contained 3 or more nuclei. OCs in the whole well were counted.

Bone slices (Immunodiagnostic Systems Limited, Boldon, United Kingdom) were placed in 96-well culture plates containing nonadherent BMCs at a density of 4×10⁵ cells/well (200 μL per well), incubated for 3 days and cultured with OPCs for 10 days. Then, the bone slices were fixed in 2.5% glutaraldehyde for 5 min and washed in 0.25 mol/L ammonium hydroxide for 10 min to remove the cells from the bone slices. The bone slices were sequentially dehydrated with 40, 75, 95, and 100% ethanol. After drying naturally, the bone slices were dyed with 1% toluidine blue (Sigma-Aldrich) at room temperature for 5 min. Using a Leica Qwin image analysis system (Leica Microsystem, Germany), the areas of the bone resorption pits were analyzed.

Culture supernatant was collected on day 5. According to the instructions of the ELISA kit (Diaclone, France), the levels of IL-10 and TGF-β1 were measured.

Total RNA was isolated using TRIzol reagent (Invitrogen, Carlsbad, CA, United States). A PrimeScript RT reagent kit (TaKaRa, Tokyo, Japan) was used to reverse transcribe RNA into cDNA. The levels of mRNA were analyzed using SYBR Premix Ex Taq (TliRNaseH Plus; TaKaRa) in a Quant Studio 5 real-time PCR instrument (Thermo Fisher Scientific). The thermocycling conditions were as follows: incubation at 95°C for 30 s, followed by 40 cycles of 95°C for 5 s, 60°C for 34 s, then dissociation stage. The results were normalized to β-actin and calculated using the 2^−ΔΔCt method. Primer sequences are shown in Supplementary Material S4.

Proteins of OCs and Tregs were harvested using RIPA Lysis Buffer (Beyotime Biotechnology, Shanghai, China) that was supplied with protease and phosphatase inhibitor cocktails (Beijing ComWin Biotech Co., Ltd. Beijing, China). The protein concentration was measured using bicinchoninic acid (BCA; Solarbio) protein assay. Then, the proteins were mixed with loading buffer and denatured in metal bath (100°C, 5 min). Following, the target proteins were divided via sodium dodecyl sulfate–polyacrylamide gel electrophoresis with constant voltage of 100 V and transferred onto a polyvinylidene fluoride membrane (Millipore, Billerica, United States) with constant current of 300 mA in 1 h. In order to eliminate nonspecific protein binding, membranes were blocked at room temperature for 2 h with 5% skim milk (BD Biosciences) in 1 × TBST (Beyotime). And then, membranes were incubated overnight at 4°C with primary antibodies followed: phosphorylated (p-) JAK2 (1:1,000), p-STAT3 (1:2000), JAK2 (1:1,000), STAT3 (1:1,000) (Cell Signaling Technology, MA, United States). Appropriate secondary antibodies were incubated for 2 h and protein bands were detected using eECL western blot kit (Beijing ComWin Biotech Co., Ltd.). Band densities were quantified using Image Lab 3.0 software (Bio-Rad Laboratories, CA, United States).

Statistical analysis was carried out using SPSS 20.0 software. One-way analysis of variance (ANOVA) and Dunnett’s T3 test were used to determine significant differences. Nonparametric tests were used if the data were abnormally distributed. All data are expressed as the mean ± SEM. The significance level was set at p < 0.05.

As we all know, OPCs can differentiate into OCs under the stimulation of M-CSF and RANKL (Xu et al., 2016). That’s how we get OCs in this study. Since a paucity of evidence has been found on the toxicity of AG490 in OPCs, we first determined the dose with CCK-8 assays and TRAP staining. OPCs might have differentiated into OCs during CCK-8 assay, so the experiment reflected the effect of AG490 on the viability of overall cells, including OPCs and OCs. The results showed that the absorbance decreased with increasing AG490 concentrations. AG490 had almost no effect on the viability of cells at doses ranging from 2.5 to 10 μM (Figure 1A), which indicated that AG490 was not cytotoxic to cells (OPCs and OCs) at concentrations below 10 μM. Moreover, when the concentration of AG490 was greater than 5 μM, the number of TRAP⁺ cells decreased significantly (Figures 1B,C). The TRAP staining revealed a situation how many OPCs differentiated into OCs. It suggested that AG490 could inhibit the differentiation of OPCs into OCs at concentrations greater than 5 μM. Based on these results, 10 μM AG490 was used for subsequent experiments.

To determine the regulatory effect of YSJB on OCs differentiation whether via the JAK2/STAT3 pathway, we performed loss-of-function studies in OPCs. OPCs were treated with JAK2 siRNA (#1, #2, and #3) to decrease JAK2 expression. JAK2 mRNA levels were significantly reduced after gene knockdown (Figure 2A). JAK2 siRNA #3 showed the highest knockdown efficiency and was used in subsequent experiments.

The mRNA levels of JAK2 and STAT3 as well as the protein levels of p-JAK2/JAK2 and p-STAT3/STAT3 were measured in single culture systems of OC and Treg respectively (Figures 2B–I), following intervention with YSJB and other reagents. The results showed that the mRNA levels of JAK2 and STAT3 in the CIA group were significantly increased compared with control group in OCs and Tregs. These results suggested that CIA-containing serum promoted the activation of JAK2/STAT3 signaling pathway in OCs and Tregs. However, after treatment with MTX or YSJB, their expression levels were reduced compared with those in the CIA group (Figures 2B,C and Figures 2F,G). Western blot assay showed similar trends in protein expressions of JAK2, STAT3 in OCs and Tregs after treatment with JAK2 siRNA and YSJB (Figures 2D,H). As we all know that, JAK2/STAT3 signaling pathway plays the role mainly by phosphorylation, so the levels of p-JAK2/JAK2 and p-STAT3/STAT3 were used to illustrate the degree of JAK2/STAT3 signal pathway activation. The levels of p-JAK2/JAK2, p-STAT3/STAT3 were significantly higher in the CIA group compared with the control group. In contrast, the relative expression levels of p-JAK2/JAK2, p-STAT3/STAT3 were significantly lower in the MTX and YSJB groups compared with the CIA group. JAK2 siRNA treatments resulted in significantly lower protein levels of p-JAK2, JAK2, p-STAT3, and STAT3 compared with the Mock groups. Furthermore, after treated with the JAK2 siRNA, the relative expression levels of P-JAK2/JAK2 and P-STAT3/STAT3 in CIA group did not increase as much as treated with the Mock, and additional YSJB treatment didn’t exert strong inhibitory effect on p-JAK2/JAK2, p-STAT3/STAT3 as before (Figures 2E–I). These results illustrated that YSJB could block the JAK2/STAT3 signaling pathway in both Tregs and OCs, indicating the role of JAK2/STAT3 pathway in the protection of YSJB regulate bone immune microenvironment.

In order to explore the effect of JAK2/STAT3 signaling pathway on OC differentiation, AG490 or JAK2 siRNA was used in the process of inducing OPCs differentiation into OC. Osteoclastogenesis was evaluated by TRAP staining (Figure 3A and Figure 4A). Interestingly, the number of TRAP⁺ cells was significantly downregulated after treatment with AG490 or JAK2 siRNA in both the OC single culture system and the coculture system. These results suggested that blocking or knocking down JAK2 could inhibit osteoclastogenesis. As shown in Figures 3B,C and Figures 4B,C, the number of TRAP⁺ cells in the CIA group was significantly increased compared with that in the control group in the absence or presence of AG490 and JAK2 siRNA. Compared with that in the CIA group, the number of TRAP⁺ cells in the MTX group and YSJB group decreased. These results suggested that CIA-containing serum promoted osteoclastogenesis, while YSJB and MTX could inhibit the osteoclastogenesis. The combination of YSJB and AG490/JAK2 siRNA did not enhance the inhibitory effect of YSJB on OC differentiation. It might suggest that YSJB could inhibit the differentiation of OCs through CIA-induced JAK pathway. In addition, we found that the number of TRAP⁺ cells was lower in the coculture systems than in the single OC culture system (Figures 4D–G). The results showed that Tregs increased suppression of YSJB in OC formation. This might be related to the fact that YSJB inhibited JAK2/STAT3 signaling pathway in OCs and Tregs.

After OPCs were cultured for 10 days, the bone slices were stained with toluidine blue. The absorbed surfaces of the bone slices were measured in each group to elucidate the effect of YSJB on bone resorption (Figure 5A and Figure 6A). Consistent with the TRAP staining results, the adsorbed surface was smaller in the presence of AG490 and JAK2 siRNA than in the absence of these factors. It suggested that blocking JAK2/STAT3 signaling pathway could inhibit bone resorption. In addition, in the absence or presence of AG490 and JAK2 siRNA, bone resorption was obviously increased in the CIA group compared with the control group. In the presence of MTX or YSJB, the areas of bone lacunae were significantly reduced. The results illustrated that MTX and YSJB inhibited bone resorption. (Figures 5B,C and Figures 6B,C). Moreover, the presence of Tregs also reduced bone resorption (Figures 6D–G). Perhaps, these due to the inhibitory effect of YSJB on JAK2/STAT3 signaling pathway.

As shown previously, OC differentiation and bone resorption were significantly inhibited in the groups with low JAK2 expression. As specific genes associated with OC differentiation, RANK, NFATc1 and c-fos reflect the activity of OCs. To investigate the molecular mechanism by which YSJB, AG490 and JAK2 siRNA regulate the activity of OCs, the mRNA levels of RANK, NFATc1 and c-fos in the OC single culture system and coculture system were measured. We showed that YSJB and MTX significantly suppressed the mRNA expression of RANK, NFATc1 and c-fos both in OC single cultures and in the coculture system compared with those in the CIA group. JAK2 inhibitors and JAK2 knockdown inhibited the expression of these three genes (Figures 7A–L). Collectively, we demonstrated that YSJB restrained the JAK2/STAT3 signaling pathway to inhibit the expression of certain genes associated with OC differentiation.

In previous studies, we demonstrated that IL-10 and TGF-β1 secreted by Tregs inhibited osteoclastogenesis (Xu et al., 2016). In this study, the expression levels of IL-10 and TGF-β1 were measured in a Treg single culture system. As shown in Figures 8A,B, Figurse 8E,F, IL-10 was expressed at low levels in the CIA group compared with the control group. MTX and YSJB reversed this reduction. However, there was no marked difference in TGF-β1 (Figures 8C,D, Figures 8G,H). Consistent with these results, we also showed that AG490 or JAK2 siRNA promoted IL-10 expression (Figures 8A,B, Figures 8E,F). These results indicated that blocking the JAK2/STAT3 signaling pathway upregulated the expression of IL-10. Perhaps the Treg-mediated reduction in the differentiation and bone resorption of OCs is associated with this signaling pathway.