Treatment of collagenase-induced osteoarthritis with a viral vector encoding TSG-6 results in ectopic bone formation

Virus production

For cloning of the third generation self-inactivating lentiviral vector (sin), the full-length TSG-6 coding sequence was obtained from a cDNA sample of inflamed synovium of a C57BL/6 mouse by nested PCR. The first round of PCR was done by amplification using forward primer 5′-CGGCTCTGCAACCGAAGA-3′ and reverse primer 5′-ATCCAAAAGTATTTATTACAGCAAT-3′. For the second round of amplification, forward primer 5′-GTCGACGCCACCATGGTCGTCCTCC-3′ and reverse primer 5′-CATATGATCCAAAAGTATTTATTAC-3′ were used, introducing restriction sites for SalI and NdeI, respectively, after which the TSG-6 gene was cloned in the PCR-script CAM cloning vector (Agilent Technologies, Amstelveen, Netherlands) according to the manufacturer’s protocol. Subsequently, the TSG-6 gene was cloned in the pRRL-cPPT-PGK-luc-PRE-SIN vector (PGK-TSG6) using the strategy previously described (Broeren et al., 2016). As a control vector (PGK-luc), the Trifusion reporter gene was digested from the pcDNA3.1 vector (kind gift from Robert E. Reeves, Stanford University) using NheI and XbaI (New England Biolabs, Ipswich, MA, USA). For adenovirus cloning, TSG-6 was digested from the PCR-script CAM vector using SalI and ligated in the SalI pre-digested pShuttle-CMV (Stratagene, La Jolla, CA, USA) (CMV-TSG6). The CMV-luciferase adenovirus (CMV-luc) was used as control vector. Lentiviral and adenoviral vector production were performed as described previously (Blaney Davidson et al., 2007; Broeren et al., 2016).

Bone marrow-derived cell culture

Bone marrow-derived cells (BMDCs) of C57BL/6 mice were obtained from the femur and the tibia. Bone marrow cells were flushed from the bones using RPMI medium supplemented with 10% fetal bovine serum (FBS), 1 mM pyruvate and 1% penicillin/streptomycin (P/S) using a Microlance 3 needle (Becton Dickinson (BD), Breda, The Netherlands). Cells were passed trough a 70 μm cell strainer (Corning, NY, USA) and centrifuged for 5 min at 1,500 rpm/423 g in a Heraeus Megafuge 16R (Thermo Scientific, Waltham, MA, USA). In a 96-well plate (Greiner Bio-one, Alphen a/d Rijn, The Netherlands), 10⁵ cells/well were seeded on the surface of the plate for differentiation analysis or on an elephant dentin slice for the bone resoption assay. Osteoclast differentiation was induced in alpha-MEM medium, supplemented with 5% FBS, 1% P/S, 30 ng/ml murine M-CSF (R&D systems, Oxford, UK) and 20 ng/ml soluble murine RANKL (R&D Systems, Oxford, UK). For tartrate resistant acid phosphatase (TRAP) staining, after 72 h the the osteoclast medium was replaced to induce osteoclast fusion with or without 1 μg/ml recombinant murine TSG-6 (rmTSG-6) (R&D Systems, Oxford, UK) for 24 h. Subsequently, cells were fixed and stained with the TRAP staining kit (Sigma-Aldrich, Zwijndrecht, The Netherlands) according to the manufacturer’s protocol. For the bone resorption assay, after 72 h the differentiation medium was replaced and cells were transduced with 166 ng lentivirus per 10⁵ cells in each well or left untreated until day 4, at which an optimal dose 1 μg/ml rmTSG-6 (based on Mahoney et al., 2008) was added to the rmTSG-6 group to align with protein production in the TSG6-transduced cells. At day 7, medium was removed from all dentin slices. Subsequently, the cells were lysed with H₂O.

Bone resorption assay

After incubation with the osteoclasts, dentin slices were incubated with 10% NH₃ and sonicated for 20 cycles 30 s on/off on a Bioruptor Next-gen (Diagenode, Seraing, Belgium) at “high” intensity. After washing, the slices were incubated for 10 min with 10% KAl(SO₄)₂ and stained with PhastGel Blue R-350 coomassie tablets (GE Healthcare, Eindhoven, The Netherlands) according to the manufacturer’s protocol. For the quantification, five pictures at 200× magnification were taken from every dentin slice using the Labovert FS (Leitz, Leica, Rijswijk, The Netherlands). The Leica application Suite (LAS) was used to analyse the average bone resorption percentage per dentin slice.

Synovial biopsy culture

Synovial biopsies were obtained from surplus C.B-17 mice after sacrifice. Using a 3 mm biopsy punch (Stiefel, Wachtersbach, Germany), synovial explants were obtained and combined from the lateral and medial synovium. The biopsies were kept in 200 μl RPMI medium supplemented with 1 mM pyruvate and 1% P/S and 10⁷ infectious units adenovirus CMV-luc or CMV-TSG6 was added. After 2 h, 20 μl FBS was added and after 24 h, total RNA was isolated using 500 μl Tri reagent (Sigma-Aldrich, Zwijndrecht, The Netherlands) according to the manufacturer’s protocol and processed and analyzed as described below.

RNA isolation and qPCR

Synovial biopsies in Trizol were first homogenated using Magnalyzer Green Beads (Roche Life Sciences, Almere, The Netherlands) and processed according to the provided protocol. DNA was removed by DNAse and cDNA was generated using Moloney murine leukemia virus reverse transcriptase, 0.5 μg/μl oligo(dT) primers and 12.5 mM dNTPs (Thermo Scientific). Quantative real-time PCR was perfomed as previously described (Broeren et al., 2016) using forward primer 5′-CAACCCACATGCAAAGGAG-3′ and reverse primer 5′-TACTCATTTGGGAAGCCCG-3′.

Collagenase-induced OA

A total of 30 female C57BL/6J mice (Janvier) of 10–12 weeks old were randomized and housed in groups of five in filtertop cages at DM-II level with 12 h light-dark cycles and water and standard diet (AB Diets, Woerden, The Netherlands) were provided ad libitum. The mice received an intra-articular (i.a.) injection of 10⁷ infectious units adenovirus CMV-luc or CMV-TSG6 dissolved in 6 μl 0.9% NaCl four days prior to the start of the induction of the model. All i.a. injections were performed during the day in the right knee using a BD microlance needle 30G 1/2′ (BD) under general anaesthesia using 2.5% isoflurane. The injections were performed by a technician without knowledge of the viral vector content in a laminar flow cabinet. CIOA was induced by two i.a. injection of one unit collagenase type VII (Sigma-Aldrich, Zwijndrecht, The Netherlands) in 6 μl 0.9% NaCl at day 1 and day 3 as described previously (Schelbergen et al., 2015). At day 21, a second injection of 10⁷ infectious units adenovirus of the same viral vector was given and mice were sacrificed by cervical dislocation at day 42. All animal experiments were approved by the local authority Animal Care and Use Committee and local ethics committee (RU-DEC 2014-080). Animal care was in accordance with the institution guidelines.

Prosense measurement

Six days after the first collagenase injection, five mice per group received i.v. injections of 1.33 nmol Prosense 680 probes (PerkinElmer, Groningen, The Netherlands) in 100 μl in the orbita plexus. At day 7, hair was removed from the knees and the mice were imaged with the IVIS Lumina (PerkinElmer, Groningen, The Netherlands) using the Cy5.5 filter. The data was analyzed using Living Image 3.0 (PerkinElmer, Groningen, The Netherlands). Regions of interest of the same size were drawn around the knees and the florescence intensity were determined for the experimental and contralateral knees.

X-ray imaging

After sacrifice, the mouse knees were removed and imaged using the Faxitron FX-20 (Faxitron, Tucson, AZ, USA) at 26 kV for 10 s. The images were blinded and randomized and the ectopic bone formation was given an arbitrary score of 0–5. 0 = No ectopic bone formation visible, 1 = ectopic bone formation just detectable, 2 = clear ectopic bone formation, 3 = bone formation stretching along femur and tibia, 4 = large ectopic bone formation, 5 = severe ectopic bone formation, similar to the most severe sample.

Histological analysis

After X-ray imaging, knee joints were fixed in formalin, embedded in paraffin and cut in 7 μm sections. The sections were stained with Safranin-O/Fast Green, blinded and randomized. Cartilage damage was assessed using an arbitrary score of 0–30 based on the OARSI cartilage OA histopathology grading system (Pritzker et al., 2006), modified for the assessment of murine knee joints by scoring the staging 0–5. Several mice showed dislocation of the knee joint, but these were distributed equally between the groups and are incorporated in the analysis.

Statistical analysis

Statistical comparisons were performed by one-way analysis of variance (ANOVA) and Mann–Whitney U test as indicated in the text using GraphPad Prism 5.03. Results are depicted as mean ± 95% confidence interval (CI) and p-values below 0.05 were regarded as significant.

TSG-6 inhibits osteoclast activity

An increase in TRAP-positive osteoclasts is observed in the subchondral bone of OA patients (Prieto-Potin et al., 2015). We therefore first tested the functionality of viral expression of TSG-6 in vitro by assessing its effects on osteoclast activity. BMDCs were differentiated to multinucleated osteoclasts using M-CSF and RANKL in the absence or presence of recombinant murine TSG-6. The formation of multi-nucleated osteoclasts from BMDCs was observed in conditions with and without TSG-6 (Fig. S1).

We subsequently assessed the effects of TSG-6 on osteoclast resorption activity on dentin slices. Under positive control conditions with lentiviral control virus, resorption pits covered ∼20% of the dentin surface (Fig. 1). The resorption by the multinucleated osteoclasts was significantly reduced by both recombinant TSG-6 (∼72%) and the lentiviral expression of TSG-6 (∼49%). No statistical difference was observed between the effects of recombinant TSG-6 and lentivirus-expressed TSG-6 This shows that viral overexpression of TSG-6 can result in functional levels of TSG-6 that can inhibit osteoclast activity.

Overexpression of TSG-6 in synovium was first investigated in synovial explants. After transduction with adenoviral CMV-TSG6, the expression of TSG-6 was significantly increased compared to CMV-luciferase (Fig. 2A).

TSG-6 effects in CIOA

The protective effects of TSG-6 on inflammation and cartilage damage were tested in the CIOA model, an experimental OA that includes inflammation (Schelbergen et al., 2015). Mice received an injection with adenovirus to provide high expression levels of TSG-6 or control luciferase in the right knee four days prior to the first collagenase injection. The effects of TSG-6 on inflammation-associated protease activity was determined in a subset of the mice after injection of Prosense 680 probes at day 6 after the first collagense injection. At day 7, the fluorescence intensity was measured and compared to the naïve contralateral knee. No significant differences in protease activity were found between the TSG-6 adenovirus (average 2.0-fold compared to contralateral joint) and the luciferase control virus (average 1.6-fold compared to contralateral joint) (Fig. 2B). A second adenovirus injection was given at day 20 to provide TSG-6 expression for the second half of the model and mice were sacrificed at day 42. The cartilage damage was assessed in histological sections using the OARSI cartilage OA histopathology grading system. No significant differences were observed in cartilage damage between the control group and the TSG-6 treated group (Figs. 2C–2E).

Before the joints were processed for histological analysis, the bone structure was analyzed using X-ray imaging and scored using an arbitrary scoring method from 0 to 5. Surprisingly, mice treated with adenoviral TSG-6 showed significantly more ectopic bone formation (77%) compared to the control group (20%) located at the medial collateral ligament (Figs. 2F–2H). Safranin-O staining of histological sections of the joints shown in Figs. 2F and 2G shows the presence of cartilage around the ectopic bone, indicating that the ectopic bone formation might be the result of endochondral ossification (Figs. 2I and 2J).