Targeting the IL-6–Yap–Snail signalling axis in synovial fibroblasts ameliorates inflammatory arthritis

Objective We aimed to understand the role of the transcriptional co-factor Yes-associated protein (Yap) in the molecular pathway underpinning the pathogenic transformation of synovial fibroblasts (SF) in rheumatoid arthritis (RA) to become invasive and cause joint destruction. Methods Synovium from patients with RA and mice with antigen-induced arthritis (AIA) was analysed by immunostaining and qRT-PCR. SF were targeted using Pdgfrα-CreER and Gdf5-Cre mice, crossed with fluorescent reporters for cell tracing and Yap-flox mice for conditional Yap ablation. Fibroblast phenotypes were analysed by flow cytometry, and arthritis severity was assessed by histology. Yap activation was detected using Yap–Tead reporter cells and Yap–Snail interaction by proximity ligation assay. SF invasiveness was analysed using matrigel-coated transwells. Results Yap, its binding partner Snail and downstream target connective tissue growth factor were upregulated in hyperplastic human RA and in mouse AIA synovium, with Yap detected in SF but not macrophages. Lineage tracing showed polyclonal expansion of Pdgfrα-expressing SF during AIA, with predominant expansion of the Gdf5-lineage SF subpopulation descending from the embryonic joint interzone. Gdf5-lineage SF showed increased expression of Yap and adopted an erosive phenotype (podoplanin+Thy-1 cell surface antigen−), invading cartilage and bone. Conditional ablation of Yap in Gdf5-lineage cells or Pdgfrα-expressing fibroblasts ameliorated AIA. Interleukin (IL)-6, but not tumour necrosis factor alpha (TNF-α) or IL-1β, Jak-dependently activated Yap and induced Yap–Snail interaction. SF invasiveness induced by IL-6 stimulation or Snail overexpression was prevented by Yap knockdown, showing a critical role for Yap in SF transformation in RA. Conclusions Our findings uncover the IL-6–Yap–Snail signalling axis in pathogenic SF in inflammatory arthritis.

the analysis stage. Mice were maintained on a 12:12 light-dark cycle with water and standard chow provided ad libitum. To induce Cre-mediated recombination, starting at 8-9 weeks of age, Pdgfrα-CreER;Confetti mice, or Pdgfrα-CreER;Yap fl/fl and genotype control mice, received 200 mg/kg tamoxifen dissolved in corn oil daily via oral gavage for 5 days, and 9 days later for a further 5 days.
Fluorescent protein expression was negligible in the absence of tamoxifen treatment. The arthritis induction protocol was started 11-12 days after the last tamoxifen dose.
The left knee (control) was injected with PBS only (Figs. [2][3] or not injected . Paracetamol (200 mg/kg) mixed with soft food was provided for 3 days following arthritis induction (Figs. 4,5). Mice were humanely killed between 3 and 9 days after AIA induction, and knees were collected for analysis.
For Yap KO experiments, mice were age-and sex-matched to their genotype controls (Supplementary   Tables 2 and 3). Arthritis induction was performed in no particular order by researchers blinded to genotypes, and mice were maintained in the same room. Randomisation was not applicable to this study since contralateral limbs served as controls, or group allocation was based on genotypes. For BrdU-labelling of proliferating cells, 2 mg BrdU was either injected intraperitoneally on day 2 after AIA induction and mice were killed 3 days post-AIA, or injected subcutaneously at the time of AIA induction followed by 1 mg/ml BrdU in drinking water until mice were killed at 6 days post-AIA.

Histological analyses
Mouse knees or human synovial tissue samples were fixed in paraformaldehyde, EDTA-decalcified as needed, and paraffin-or cryo-sectioned, as described. [2] For Yap KO experiments, DNA was extracted from paraffin tissue sections to confirm mouse genotypes by PCR. Immunohistochemistry (IHC) and immunofluorescence (IF) stainings were performed as described [3] using antibodies listed in Supplementary Table 5. To validate antibody stainings, donor-matched sections were stained, in parallel, with isotype negative control antibodies that were species-matched, concentration-matched, and matched for clonality to the antibody of interest. TRAP staining was performed using an Acid Phosphatase, Leukocyte (TRAP) kit (Sigma-Aldrich) according to manufacturer's protocol. Images were acquired on a Zeiss Axioscan Z1, Olympus upright microscope, or Zeiss LSM710 confocal microscope with ZEN software.

Image analysis
Zen 2.1 Lite software was used to visualise slide scanned images of 6 H&E-stained sections from each knee, ≥75 µm apart, and arthritis severity scores were assigned at 0.25 increments to indicate the observed extent of synovial lining hyperplasia (0-3), immune cell infiltration (0-3), cellular exudate (0- 3), and erosions at the patellofemoral joint margins (0-3). To determine the erosion score, separate scores were assigned to medial and lateral sides of the patella and femoral condyles, and the average erosion score was calculated for each section. For each parameter, the average score per knee was calculated, and scores of the 4 parameters were then summed to give an overall arthritis score (0-12).
For TRAP+ cell quantification, cells were counted along the medial and lateral femoral periosteum, or endosteum of the femoral epiphysis, and the average per knee was calculated. Cell counting was performed using ImageJ 1.47v, QuPath 0.1.2 or Zen 2.1 Lite software and normalised to length of synovium analysed. For mouse IHC analysis, the average DAB intensity of 2 or 3 sections per mouse was calculated in a 600 µm-length area along the medial patellofemoral synovium using QuPath 0.2.0 BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance Supplemental material placed on this supplemental material which has been supplied by the author(s)  [4] Replicate sections were stained on separate occasions, and within each staining experiment, data were normalised to the average DAB intensity of the control group. For human IHC analysis, the average DAB intensity of 1-5 sections per donor was measured in 200-400 µm-length areas of quiescent and hyperplastic synovium using QuPath 0.2.0 software. [4] Hyperplastic areas were recognised by the more roundish shape and large nuclei of RA-SF, as previously described. [5] Sections were stained in two separate batches and data were normalised to the average DAB intensity of quiescent areas. For the Yap KO experiments, researchers were fully blinded to genotypes throughout all experiments, outcome assessments, and data analyses.

Cell isolation, sorting, and culture
Cells were isolated from mouse knee joints by digestion with 1 mg/ml collagenase type IV (Sigma) as described, [2] according to a protocol optimised for synovial cell isolation. [6] Cells were left unsorted, or were sorted according to Tom fluorescence in conjunction with CD45-APC and Pdgfrα-BV421 immunostaining (Supplementary Table 6), using an Influx Cell Sorter (BD Biosciences) as previously

Immunophenotyping of cells
Cells were stained with antibodies listed in Supplementary Table 6 and analysed by flow cytometry.
Data were acquired on a BD LSRII or Fortessa flow cytometer and analysed using FlowJo v10 software.

RNA extraction, cDNA synthesis, and quantitative PCR
RNA was extracted using TRIZOL reagent (Invitrogen), miRNEasy/RNEasy Micro kits (Qiagen) or PicoPure RNA Isolation Kit (ThermoFisher) using standard protocols, and quantified using a NanoDrop ND-1000 spectrophotometer (Labtech). cDNA was synthesised from up to 1 µg of total RNA using random hexamer primers and SuperScript IV Reverse Transcriptase (Invitrogen), according to the manufacturer's instructions. Quantitative PCR (qPCR) was performed with a Roche LightCycler 480 using SYBR Green Master (Roche). Primers were designed using Primer-BLAST (NCBI) and are listed in Supplementary Table 9. Amplification of a single product of correct size was confirmed by agarose gel electrophoresis and/or melting curve analysis. Relative concentrations were quantified by ΔΔCt or standard curve method, and normalised to expression of HPRT or GAPDH.

Cell invasion
Cells were serum-starved overnight and seeded at 20,000 cells/well in the upper chamber of a 6.

Statistical analysis
All data points on graphs, and n-numbers in text, indicate individual human donors or mice, or independent experiments. SigmaPlot v13 or v14 and GraphPad Prism v5 software were used for statistical analysis. Tests used to determine statistical significance (p<0.05) are indicated in figure legends. Normality and equal variances were tested in Sigmaplot using the Shapiro-Wilk and Brown-Forsythe tests, respectively. Ordinal data were analysed using a non-parametric Mann-Whitney U test.
Log-transformation was used to equalise variance prior to statistical testing as indicated. One Yap WT AIA knee was accidentally damaged during sample processing, and another showed patellar dislocation upon histological examination. These two mice were excluded. A Grubb's outlier test was used to identify and exclude 1 major outlier from the WT group of the Yap cKO experiments, with an arthritis score of 1.108 and a z-value of 3.137 (p<0.05). The Pearson's test was used for correlation analysis.