DNA extraction kit and Tamoxifen were purchased from Sigma–Aldrich (Sigma-Aldrich, St. Louis, MO, United States). 4% paraformaldehyde, 14% elhylene diamine tetraacetic acid (EDTA) and citrate buffer were purchased from Solarbio (Solarbio, Beijing, China). Normal goat serum, secondary biotinylated goat anti-mouse antibody and diaminobenzidine (DAB) solution were obtained from Invitrogen (Invitrogen, Frederick, MD, United States). COL-2, MMP-13 and pSMAD-2 primary antibodies were brought from Abcam (Abcam, Cambridge, MA, United States).

Yougui pills were produced by Henan Wanxi Pharmaceutical Co. Ltd. (Batch number: Z41022170) Small amount of sample was deposited as voucher specimen in institute of orthopedics and traumatology, Zhejiang Chinese Medical University with voucher specimen No. GYS-2015-0016. YGPs comprise 10 kinds of herbal medicines (Table 1). The production process was based on the standards of Chinese Health and Family Planning Commission. Briefly, Cervicornuscolla was firstly melted with wine by heat and the other nine herbal medicines were ground to fine powder. Every 100 g powder was mixed with 65 g molten Cervicornuscolla and honey (Ji et al., 2016). The high performance liquid chromatography (HPLC) detection was performed to analysis of and the quality control of YGPs. The analysis procedure was conducted according to the previous reports (Lin et al., 2012). Ten of the major constituents including betaine, sweroside, loganin, hyperin, quercitrin, chlorogenic acid, songorine, fuziline, neoline and cinnamaldehyde were determined in our study.

Male C57BL/6J mice (10 weeks old) were obtained from Shanghai Laboratory Animal Center of Chinese Academy of Sciences (Grade SPF II, SCXK 2012-0002). Col2-CreER and TGF-βRII^fx/fx mice (background: C57BL/6J) were donated from Rush University Medical Center (Chicago, IL, United States). In order to specifically knock out Tgfbr2 gene in articular cartilage, TGF-βRII^fx/fx mice were cross with Col2-CreER mice to generate Col2-CreER; TGF-βRII^fx/fx mice (TGF-βRII^Col2ER mice) (Shen et al., 2013). Genotyping was performed by polymerase chain reaction (PCR) using a DNA extraction kit tail biopsy tissues (Supplementary Figure S1). PCR primer sequences as follows: primer sequences of Cre, upper primer 5′-ATTGCTGTCACTTGGTCGTGGC-3′ and lower primer 5′-GAAAATGCTTCTGTCCGTTTGC-3′ (200-base-pair PCR product); primer sequences of Tgfbr2 loxP, upper primer 5′-TAAACAAGGTCCGGAGCCCA-3′ and lower primer 5′-ACTTCTGCAAGAGGTCCCCT-3′ (Wild-type, 420-base-pair PCR product; homozygotic type, 540-base-pair PCR product). TGF-βRII^fx/fx littermates were used as controls (Wang et al., 2013; Shen et al., 2014; Chen et al., 2017). After identification of genotypes, TGF-βRII^Col2ER mice were induced by tamoxifen (1 mg/10 g body weight/day, i.p.) for five successive days at age of 2-week-old (Shen et al., 2013). All mice were housed in a temperature (23 ± 2°C) and humidity (40 ± 5%) controlled room, exposed to a controlled 12 h cycle of light and darkness with a solid rodent chow and water ad libitum. All studies were approved by the Committee on the Ethics of Animal Experiments of Zhejiang Chinese Medical University.

Knee OA model was induced to C57 mice at age of 10-week-old by destabilization of the medial meniscus (DMM) as previous report (Iijima et al., 2014). Briefly, DMM surgery was performed on the right hind limbs of WT mice as follows: (1) After a 3 mm longitudinal incision was made on the medial part of the knee under anesthesia, blunt dissection of the knee extensor muscles and patellar ligament was performed to expose the medial meniscotibial ligament (MMTL); (2) The MMTL was transected to give DMM; (3) The medial joint capsule was sutured, and the skin was closed. The sham surgery was also performed by a similar surgical approach without manipulating the joint tissue in WT mice.

All mice were randomly divided into six groups: the sham group, the model group, the model YGPs group, the TGF-βRII^fx/fx group, the TGF-βRII^Col2ER group and the TGF-βRII^Col2ER + YGPs group (n = 10 in each group). YGPs were orally administered to both WT and TGF-βRII^Col2ER mice once a day after DMM surgery and tamoxifen inducement respectively for 12 consecutive weeks with dose of 3.5 g/kg body weight. The dose administered was determined according to the following formula: D_M (dose per kg body weight) = D_H × R × (W_H/W_M), as detailed in The Methodology of Pharmacological Experiment. D_M and D_H are doses for mice and humans, and W_M and W_H are body weights of mice and humans, respectively. R is the coefficient of 0.0026 for human mouse equivalent dosage conversion (Xu et al., 2002). The mice in sham and TGF-βRII^fx/fx group were given an equal dosage of physiologic saline. Knee joint samples from each group were harvested at the end of drug intervention period for the follow-up experiments.

Knee joints of each group were successively fixed in 4% paraformaldehyde for 3 days and decalcified with 14% EDTA solution for 14 days. The samples were embedded in paraffin, and a series of histologic sections (3 μm) were stained with Alcian Blue Hematoxylin/Orange G and Toluidine Blue. The image of sections was collected under a light microscope (Axio Scope A1, ZEISS, Germany). Next, Histomorphometric analysis was performed using OsteoMeasure software (OsteoMetrics, Inc., Atlanta, GA, United States) to calculate and determine the area and thickness of cartilage. Moreover, according to Osteoarthritis Research Society International (OARSI) recommendations (Glasson et al., 2010), cartilage degeneration was scored by two blinded observers. The summed score for the knee joint (determined by summing the scores for the femoral condyle and tibial plateau) was used to evaluate the extent of cartilage degradation.

The representative knee joint images of mice were obtained from micro-CT equipment (Skyscan 1176, Bruker microCT N.V., Kontich, Belgium). The pictures were taken with a resolution of 4000 × 2672 pixels and an isotropic voxel size of 9 μm. In addition to the visual assessment of structural pictures, quantitative morphometry indexes were determined from micro-tomographic data based on the 3-D morphometry (Hildebrand et al., 1999). The region of interest was identified between the proximal tibia growth plate and tibial plateau. And the following indexes were evaluated subsequently: (1) bone volume fraction (BV/TV, %); (2) average trabecular thickness (Tb. Th, mm); (3) average trabecular separation (Tb. Sp, mm).

We also performed immunohistochemistry to analyze the expressions of type II collagen (COL-2), matrix metalloproteinase 13 (MMP-13) and phosphorylated protein mothers against decapentaplegic homolog 2 (pSMAD-2) in articular cartilage. The paraffin sections were heated at 60°C overnight, then deparaffinized and rehydrated. Next, the sections were incubated with 3% hydrogen peroxide to block endogenous peroxidase activity for 20 min. the sections were incubated in 0.1 mol/L citrate buffer (Zhongshan jinqiao, Beijing, China) at 95°C with water bath for 20 min as antigen retrieval. Immunohistochemistry analysis was achieved using primary antibodies of COL-2 (diluted 1:1000), MMP-13 (diluted 1:100) and pSMAD-2 (diluted 1:100) at 4°C overnight. After washing, the sections were incubated with the HRP-conjugated secondary antibody at room temperature for 30 min. After incubated with DAB, counterstained with hematoxylin for 2 s, positive staining of sections was observed under a light microscope (Axio Scope A1, ZEISS, Germany).

Routine blood test was performed at the end of the experiment. All mice were fasted but allowed to water freely for 12 h prior to collecting the blood sample. The blood samples were collected from the murine orbit, followed by the centrifugation in heparinized tubes. Hematologic indexes included white blood cell (WBC) count, red blood cell (RBC) count and platelet (PLT) count.

All the data were expressed as mean ± SD. One-way analysis of variance followed by Dunnett’s test when appropriate. P < 0.05 was considered to be statistically significant. The statistical analysis was performed using SPSS 18.0 software (SPSS, Chicago, IL, United States).

To evaluate cartilage degradation induced by DMM surgery, Alcian Blue Hematoxylin/Orange G and Toluidine Blue stainings were subsequently performed. As shown in Figure 1A, no pathological findings of OA were observed in mice from sham group. In contrary, DMM-induced mice exhibited localized damage of articular cartilage, early osteophyte formation and increases in subchondral bone mass. By treated with YGPs, cartilage surface erosion and joints degradation were noticeably decreased. Histomorphometric analysis showed a significant reduction in tibial cartilage area and thickness in OA model mice (both, P < 0.001). Likewise, prominent increase of tibial cartilage area (Figure 1B) and thickness (Figure 1C) also occurred after YGPs treatment compared with model group (P = 0.003; P = 0.002). Moreover, we analyzed the data using the histologic scoring system recommended by OARSI (Glasson et al., 2010), and we found that DMM-induced mice had significantly higher scores compared to sham group (P < 0.001), with significantly downregulated OARSI scores after treatment with YGPs (P = 0.005) (Figure 1D).

To assess the microarchitecture of the bone, the three-dimensional image was carried out using micro-CT. As shown in Figure 2A, YGPs prominently reduced DMM-induced osteophyte formation. The parameters including BV/TV, Tb. Th and Tb. Sp of each group are shown in Figures 2B–D. The results showed that, following the DMM surgery, all the parameters for the model group were significantly different compared to sham group (P < 0.001). After YGPs treatment, the BV/TV and Tb. Th measurements of YGPs group were significantly lower than those of DMM-induced mice (P = 0.005, P < 0.001). Conversely, the Tb. Sp measurements of the YGPs group were significantly higher than those of DMM-induced mice (P = 0.027). These results suggest that YGPs have a positive effect on bone health in DMM-induced mice.

An important link in cartilage damage is the unbalance of synthesis and degradation of the ECM, an event marked by the expression of catabolic enymes such as MMP-13, and the key molecular of TGF-β/Smad signaling, pSMAD-2. Accordingly, it is possible that the chondroprotective effect on YGPs exhibited in Figure 1 is associated with TGF-β/Smad signaling. To address this question, immunohistochemistry was conducted to evaluate COL-2, MMP-13 and pSMAD-2 protein expressional levels in the articular cartilage. As expected, DMM-induced mice had reduced levels of both COL-2 and pSMAD-2, and elevated levels of MMP-13 in cartilage compared to sham mice, indicating the suppression of TGF-β/Smad signaling (all, P < 0.001) (Figures 3A–C). Mice treated with YGPs were protected from the DMM induced COL-2 and MMP-13 positive staining in cartilage (P = 0.006, P = 0.003), suggesting that the maintenance of cartilage architecture in YGPs group could be due to inhibition of matrix degradation (Figures 3A,B). Interestingly, pSMAD-2 positive cells were increased in YGPs group (P = 0.028) (Figure 3C), indicating that YGPs may have been from cartilage degradation in part due to reduced MMP-13 production by chondrocytes residing in cartilage via activation of TGF-β/Smad signaling.

The marked preservation of ECM produced by chondrocytes post-DMM in YGPs treated mice begged for subsequent investigation of the impact of YGPs on OA progression in TGF-βRII^Col2ER mice. Firstly, we performed Alcian Blue Hematoxylin/Orange G and Toluidine Blue staining respectively to further explore the role of TGF-β/Smad signaling in the chondroprotective action of YGPs. Comparison of the sections from all groups revealed a severity of articular cartilage damage, subchondral sclerosis and osteophyte formation in TGF-βRII^Col2ER mice. It should be noted that there were no discernable effects of YGPs treatment on inhibition of OA progression in TGF-βRII^Col2ER mice (Figure 4A). Additional analyses were also performed using the histomorphometry approach as well as standardized OARSI cartilage scoring. Consistent with the histopathologic analysis, reduction of tibial cartilage area and thickness in TGF-βRII^Col2ER mice were not mitigated significantly by YGPs treatment (P = 0.199, P = 0.328) (Figures 4B,C). Finally, the impact of YGPs on the articular cartilage in TGF-βRII^Col2ER mice was also no significant difference based on OARSI scoring (P = 0.514) (Figure 4D). As analysis performed above, it should be noted that there were no significant differences in TGF-βRII^Col2ER mice with or without YGPs treatment including tibial cartilage area and thickness, and OARSI cartilage scoring.

Moreover, our micro-CT results also showed that YGPs almost can not inhibit the osteophyte formation (Figure 5A). In addition, BV/TV and Tb. Th were increased, and Tb. Sp was decreased in TGF-βRII^Col2ER mice, whereas no significant improvement was observed in the animals treated with YGPs (Figures 5B–D).

To understand better the role of TGF-β signaling in the chondroprotective action of YGPs, we further performed immunohistochemistry to evaluate COL-2 and MMP-13 protein expressional levels based on TGF-βRII gene deletion in chondrocyte. As shown in Figure 6A, TGF-βRII^Col2ER mice had reduced levels of pSMAD-2 more than 80% in the cartilage compared to TGF-βRII^fx/fx mice, suggesting that TGF-β/Smad signaling was almost blocked in our study. Interestingly, TGF-βRII^Col2ER mice treated with YGPs were not protected from the COL-2 positive staining in the cartilage (P = 0.207) (Figure 6B). Accordingly, MMP-13 positive cells were not reduced in TGF-βRII^Col2ER mice treated with YGPs (P = 0.226) (Figure 6C), indicating that YGPs almost had no effect on inhibition of matrix degradation with suppressing of TGF-β/Smad signaling. To summaries, no differences were observed in the articular cartilage, as chondrocytes were all actively producing similar levels of both COL-2 and MMP-13 in TGF-βRII^Col2ER mice with or without YGPs treatment (Figures 6B,C).

To evaluate the effect of YGPs orally administered on hematologic values, routine blood test was performed at the end of the experiment, including WBC count, RBC count and platelet (PLT) count. As shown in Figures 7A–C, there were no significant differences in both healthy and YGPs treated mice. YGPs treatment did not cause a significant change in any of the hematologic parameters compared to the control groups.