*4.4. Histochemical Analyses and Semi-Quantitative Analyses*

After the end of the treadmill running protocol, the right ankle joints of the rats were removed and set in 4% paraformaldehyde (Wako, Osaka, Japan). They were then decalcified with 20% ethylenediaminetetraacetic acid and embedded in paraffin. The center of each ankle joint was sliced in 6-μm thick sagittal slices, and stained with H&E and safranin O. Arthritic changes, such as infiltration of inflammatory cells, synovial proliferation, destruction of articular cartilage, and bone erosion, were then evaluated histologically, and we measured the histological score as described by Weinberger et al. [33]. In brief, the infiltration of mononuclear cells (0–3 points) and histiocytes (0–3 points) into the synovium, cartilage destruction (0–3 points), and bone erosion (0–5 points) were measured.

#### *4.5. Immunohistochemical Analyses*

For immunohistochemistry of TNF-α, Cx43, and cathepsin K, paraffin-embedded joint tissue sections were de-paraffinized in xylene, rehydrated with graded alcohol, and immersed in 0.4 mg/mL of phosphate buffered saline (PBS). Antigen activation was conducted in 8-min intervals with proteinase K, and endogenous peroxidase activity block was conducted at 5-min intervals with 3% H2O2. Rabbit polyclonal anti-TNF-α antibodies (ab6671, Abcam, Cambridge, UK) at a concentration of 1:150, and anti-Cx43 antibodies (#3512, Cell Signaling Technology, Danvers, MA, USA) at a concentration of 1:50 were incubated overnight at 4 ◦C. Rabbit polyclonal anti-cathepsin K (111239-1-AP, Proteintech, Chicago, IL, USA) at a concentration of 1:2000 was incubated for 50 min at room temperature. The sections were incubated in Histofine Simple Stain Rat MAX-PO for 30 min at room temperature, after extensive washing with PBS. Immunostaining was detected by 3,3 -diaminobenzidine (DAB) staining. Counter staining was done with Mayer's hematoxylin. We confirmed that there is no nonspecific staining. TNF-α and Cx43 staining areas of the synovial membrane interstitial cells were calculated using ImageJ® with modifying a method reported by Mane et al. [34]. In detail, three fields of view with a magnification of 400× were taken randomly. The images were deconvoluted and showed only DAB immunoreaction. A standard threshold was maintained without any adjustment. The percentage of the immunostained-positive area was averaged. Validation of ImageJ® analysis was performed by two expert orthopedists. Quantification of all the images was blinded.
