Mice. All 5-week DBA/1 J mice were purchased from Sankyo Labo Service (Tokyo, Japan) and placed under specific pathogen-free conditions in animal facilities certified by the Keio University animal care committee. Animal protocols were approved by that committee. Animal experiments were carried out in accordance with the Guidelines of the Keio University Institutional Animal Care and Use Committee and Institutional Guidelines on Animal Experimentation at Keio University.
CIA model. Experimental CIA models were generated in 5-week-old male mice by injecting 200 µl of emulsion containing 100 µg of type II collagen (CII) (Collagen Research Center, Tokyo, Japan) intradermally at the proximal end of the tail. Emulsion composition was 1mg/ml bovine CII dissolved in PBS plus an equal volume of complete Freund’s adjuvant (CFA, Difco, Detroit, MI, USA). Following the first injection, a second immunization of CII/CFA was administered. Clinical symptoms of arthritis were evaluated visually in each limb and graded on a scale of 0–4: 0, no erythema or swelling; 0.5, swelling of one or more digits; 1, erythema and mild swelling of the ankle joint; 2, mild erythema and swelling involving the entire paw; 3, erythema and moderate swelling involving the entire paw and 4, erythema and severe swelling involving the entire paw. We defined the clinical score for each mouse as an overall “arthritis score” calculated as the sum of scores for all four limbs (maximum = 16).
Chemicals, drugs, reagents.
Drugs selected by in silico screening were purchased from following companies.
UZI/2602234 was purchased from Zelensky, Russia; P2000N-15185 and P2000N-29959 from Pharmeks, Russia; Z1082961568 from Enamine, Kyiv, Ukraine, AP-113/15379001 from SPECS, Germany; P2001S-254605, STK576506, STK587698, and STK591642 from VITAS-M LABORATORY, Russia; Z24581406, Z90677708 from Enamine, Kyiv, Ukraine; and F0648-0027, F0648-0476, F0648-0479 and F0648-0589 from Life Chemicals Inc, Canada.
In silico screening
We retrieved a crystal structure of Stat3 (PDB ID: 3CWG) from the Protein Data Bank (www.rcsb.org) . The protein structure was prepared for in silico analysis using Protein Preparation Wizard in Maestro of Schrödinger Suite 2015-2 (Schrödinger, LLC, New York, USA). In silico fragment mapping was then performed according to the previous studies[20, 21]. Briefly, residues in the dimerization interface of Stat3 were defined as the mapping surface. The surface was then searched for similar topographies to the protein substructures (subsites) stored in our in house fragment-subsite structure database. When a local topography similar to the subsite was found on the target protein, the corresponding fragment was mapped on its matching subsite. The 3D pharmacophore model was then constructed according to the physicochemical properties of the mapped fragments as well as characteristics of fragment-protein interactions, using the UNITY program of SYBYL-X 1.3 (Certara, Princeton, NJ, USA). This model was used to screen approximately 4.9 million compounds from a commercially available compound database (Namiki201404HTS; constructed in April 2014 by Namiki Shoji Co. Ltd., Tokyo, Japan) using the UNITY Flex Search program of SYBYL-X 1.3. For ligand docking, a receptor grid was generated using the grid generation feature of Glide (Schrödinger Suite 2015-2). Multiple conformations of compounds were then generated using LigPrep (Schrödinger Suite 2015-2), followed by ConfGen (Schrödinger Suite 2015-2). All docking calculations were performed with Glide and run in standard precision (SP) mode. The ligand van der Waals radius for non-polar atoms was scaled by a factor of 0.8. The conformers generated above were docked into the receptor grid, and the top pose for each compound, determined by the Glide docking score (cutoff: -5.0), was collected.
Cell culture and in vitro screening of drugs by a Stat3 inhibition assay.
The NIH3T3 cell line was maintained in DMEM (Sigma–Aldrich Co.) containing 10% fetal bovine serum (FBS) with penicillin G and streptomycin. Adherent cells were collected and cultured in 24-well plates (5 × 104 cells per well) and subjected to serum starvation in the presence or absence of IL-6 (100 ng/ml, R & D Systems) and sIL-6R (100 ng/ml, R & D Systems). Eight hours later, mRNA was collected to analyze IL-6 and RANKL expression, as described below in realtime PCR analysis.
RNA isolation and real-time PCR.
Total RNAs were isolated from cultured NIH3T3 cells using TRIzol reagent (Invitrogen Corp) using general procedures. cDNA synthesis was then performed with oligo(dT) primers and reverse transcriptase (Wako Pure Chemicals Industries). Quantitative PCR was performed using SYBR Premix ExTaqⅡ reagent and a DICE Thermal cycler (Takara Bio Inc.). Stored data was analyzed using a calibrator. β-actin (Actb) expression served as internal control. Primer sequences for realtime PCR were as follows.
Histopathology analysis in vivo.
Ankle tissues from CIA mice were fixed in 10% neutral-buffered formalin and embedded in paraffin, and tissue blocks were cut into 4-µm sections. For other analyses, lower ankles were decalcified in 10% EDTA, pH7.4, before embedding. Safranin-O staining was performed using standard procedures, and then the Mankin score, a histopathological arthritis score, was evaluated using a BioRevo microscope and corresponding software (Keyence, Tokyo, Japan).
Fluorescent immunohistochemistry analysis in vitro.
NIH3T3 cells were fixed in 4% paraformaldehyde (Wako Pure Chemical Industries, Osaka, Japan) and then incubated with PBS buffer containing 0.5%Triton. Cells were blocked using 5%BSA for 20 minutes at room temperature and then stained with rabbit anti-Stat3 (1:100 dilution; #12640, Cell signaling Technology, Danvers, MA, USA) or mouse anti-pSTAT3 (1:100 dilution; #4113, Cell signaling Technology, Danvers, MA, USA), followed by Alexa Fluor 488-conjugated goat anti-rabbit IgG (1:100 dilution; Invitrogen, Carlsbad, CA, USA) or Alexa Fluor 488 goat anti-mouse IgG (1:100 dilution; Invitrogen, Carlsbad, CA, USA), respectively. After staining in secondary antibodies for 1h at room temperature, DAPI (1:1000; Wako Pure Chemicals Industries, Osaka, Japan) was used as a nuclear stain.
Whole-cell lysates were prepared from cell cultures using RIPA buffer (1% Tween 20, 0.1% SDS, 150 mM NaCl, 10mM Tris-HCl (pH 7.4), 0.25 mM phenylmethylsulfonylfluoride, 10µg/mL aprotinin, 10µg/mL leupeptin, 1mM Na3VO4, 5mM NaF (Sigma-Aldrich Co.)). Equivalent amounts of proteins were separated by SDS–PAGE and transferred to a PVDF membrane (EMD Millipore Corp, Burlington, MA, USA.). Proteins were detected by using anti-pSTAT3 (#9134), anti-STAT3 (#4904) (Cell Signaling Technology, Danvers, MA, USA) or anti-actin antibody (Sigma-Aldrich Co., St Louis, MO, USA).
Results are expressed as means ± s.d. Statistical significance of differences between compared groups was evaluated using Student’s t-test (*p < 0.05; **p < 0.01; ***p < 0.001; NS, not significant, throughout the paper).