Generation of PAR2 deficient animals
Generation of PAR2-deficient mice were previously described48. Breeding pairs were set up as heterozygous or homozygous pairs that were derived from heterozygous mattings, thus animals in the study were either litter mates or, at most, 1 generation removed. F2rl1flox (PAR2F/F) mice were generated in collaboration with Prof Neil Dear (Leeds University) from ES cells (C57Bl6 background) obtained from the Mouse Genetics Project (Sanger Centre, Cambridge). The L1L2_Bact_P cassette, consisting of an FRT site followed by lacZ sequence and a loxP site, was inserted at position 95514758 of Chromosome 13 upstream of the critical exon. Under the control of the human beta-actin promoter, neomycin resistance gene follows the first loxP site, further followed by SV40 polyA, a second FRT site and a second loxP site. Downstream of the target exon, a third loxP site is inserted at position 95511446. Thus, the target exon is flanked by loxP sites. A floxed allele is created by flp recombinase expression in mice carrying this allele to remove the lacZ sequence and neo selection cassette, leaving loxP sites flanking the critical exon 2. Exon 2 was targeted because it contains most of the coding sequence of the PAR2 protein. PAR2F/F mice were crossed with Tg(Col2a1-cre)1Bhr/J or Tg(BGLAP-cre)1Clem/J mice (Jackson laboratories) to knock out PAR2 specifically in chondrocytes (PAR2ch/ch) or in osteoblasts (PAR2ob/ob). Both tissue specific colonies were bred as homozygous PAR2F/F with a heterozygous cre. Thus, PAR2F/F controls are littermates. Background of tissue specific knockout mice was tested with MiniMUGA (Transnetyx, USA). Diagnostic SNPs indicated an outbred C57BL/6 background with sub-strains C57BL/6J (~ 53% of possible alleles), C57BL/6JOlaHsd (~ 14% of possible alleles).
Animals
Animals were housed in cages with enrichment and in a 12 h light cycle. Group sizes are stated in the supplemental table 1. A small pilot study (n = 4) on the cartilage damage comparing WT to PAR2−/− 12 months post induction, determined a signal to noise ratio of 2.4. The required sample size for 90% power assuming a 5% significance level and a two-sided test is a minimum of 5 mice per group. Considering the need to carry out microCT analysis, for which the PAR2−/− have only a mild phenotype in early development, we aimed to have 8 mice per time point to also allow for attrition, in addition to the samples on the pilot study. 4 week time point studies on the chondrocyte- and osteoblast-specific knockout did not show any significant changes in cartilage damage, thus we could not estimate a sample size as the long term effect may be different. Given the long-term design of the study, we aimed to have 10 to 12 mice per conditional knock out group. Supplemental table 1 describes the groups utilised in the study. No animals were excluded, although during the study 3 animals needed to be euthanised due to tumours or teeth malformation. Animals were randomly assigned to each group (excel) and assigned an ID which was only matched to the group after scoring and data analysis. Animal health was monitored based on a scoring system. Dynamic weight bearing was measured at end point. Blood was collected for serum assays and knees harvested for microCT followed by histology to determine cartilage score. Bone marrow was extracted from 6- to 8-week-old mice. Immature primary osteoblasts were extracted from 3 to 5-day old neonatal calvariae as previously described49. Mature osteoblasts were extracted from 8- to 12-month-old mice from a colleague’s ageing study controls. Osteoclasts were derived from enriched monocyte populations from the bone marrow.
All procedures were in accordance with Home Office regulations and approved by the University of Glasgow Animal Welfare & Ethics Review Body.
Induction of OA
OA was induced by DMM surgery on adult 10-week-old male mice. PAR2−/− mice were compared to wild type littermates. For tissue specific knockout studies PAR2f/f was used as a control. As previously described8,50, medial compartment OA was induced by DMM involving transection of the left medial meniscotibial ligament under aseptic conditions. The surgeon was unaware of the genotype and was told what surgery to conduct (sham or DMM) by the assistant who randomly allocated mice to the different groups. Buprenorphine (Vetergesic; 30 µg intraperitoneally) was administered postoperatively and animals maintained for up to a year. Our lab has optimised this surgery to allow a quick cut of the ligament without any artificial damage to the cartilage during the intervention50. This results in a slower progressing model due mainly to the destabilisation of the joint. All data is available upon request.
MicroCT
Knee joints were fixed in 4% paraformaldehyde solution for 24 h and subsequently stored in 70% EtOH, then analysed by µCT to examine the calcified tissues using Skyscan 1272 (Bruker, Belgium; 0.5 aluminium filter, 50 kV, 200 mA, voxel size 4.57 µm, 0.3° rotation angle). Scans were reconstructed in NRecon software (Bruker, Belgium), with stacks analysed as follows: (1) medial subchondral bone was analysed by selecting a volume of interest (450 x 450 x 900 µm) within the loaded region of the tibial epiphysis13,51 manually eliminating the growth plate and below, (2) a 900 µm thick slice from the metaphysis (225 µm from the growth plate) was selected to analyse trabecular bone parameters with an automated selection of the ROI (CTan, Bruker, Belgium) and (3) ectopic calcifications on the medial side around the knee, including the displaced meniscus, were manually selected to then shrink the ROI to the edge of the existing bone, hence measuring the volume occupied by these calcifications.
Assessment of cartilage damage
Histological analysis of progression and severity of cartilage damage was undertaken on joints previously scanned, then decalcified (Formical 2000; Decal Chemical, New York, USA) overnight. Joints were embedded in paraffin wax and coronal sections (6 µm) cut then stained with haematoxylin, safranin-O/fast green. We used the new cartilage damage score described by Haubruck et al.52 applied to each quadrant of the joint. This newly published scoring system gives a more progressive score separating the damage to the uncalcified and calcified regions of the articular cartilage and is more suitable for severe OA. The scoring system is summarised as follows: 0 = Normal cartilage, 1 = Roughened surface AND/OR superficial fibrillation < 10% of cartilage depth (any % of joint surface area),, 2 = Fibrillation extending > 10% of cartilage depth but not reaching the calcified cartilage AND/OR loss of surface lamina (any % or joint surface area), 3 = Horizontal cracks/separations between calcified and non-calcified cartilage OR clefts down to calcified cartilage BUT no loss of non-calcified cartilage, 4 = Fibrillation to the calcified layer OR loss of non-calcified cartilage lesion for 1–25% of the joint surface, 5 = Fibrillation to the calcified layer OR loss of non-calcified cartilage lesion for 25–50% of the joint surface, 6 = Fibrillation to the calcified layer OR loss of non-calcified cartilage lesion for 50–75% of the joint surface, 7 = Fibrillation to the calcified layer OR loss of non-calcified cartilage lesion for > 75% of the joint surface, 8 = lesion extends through the calcified cartilage (1–25% joint surface area), 9 = lesion extends through the calcified cartilage (25–50% joint surface area), 10 = lesion extends through the calcified cartilage (50–75% joint surface area), 11 = lesion extends through the calcified cartilage (> 75% joint surface area). 6 sections from each mouse knee, between 25 and 100 µm apart, were graded by two scorers blinded to the specimens, with scores averaged. Poor histological sections (ripped, folded or wrong orientation) were not scored. The second scorer had no previous experience in scoring histological sections and was unaware of the aims of the study. There was good agreement between scorers with a Spearman correlation coefficient of 0.80 (95% CI 0.68 to 0.89), the mean difference in score being 1.6.
Dynamic weight bearing
As an indirect indicator of pain, limb weight bearing in mice was assessed at end point, using a dynamic weight bearing chamber (BioSeb, Marseilles, France). Animals were individually recorded for 5 minutes, of which a minimum of 1 minute was subsequently validated and analysed (Dynamic Weight Bearing 1.0, Bioseb). The parameters examined were the individual paw load in proportion to body weight as well as the % time spent on each paw.
Serum assays
To determine differences in bone remodelling, serum was collected from the mice at experimental end point. A sandwich ELISA for C-terminal telopeptide of type I collagen (CTX-I, RatLaps™; IDS) and N-terminal propeptide of type I procollagen (P1NP, IDS) ELISA assays were used and analyses performed according to the manufacturer’s instructions.
In vitro osteoblastogenesis and osteoblast cultures
Bone marrow was flushed out of the long bones of 6- to 8-week-old PAR2+/+ and PAR2−/− mice and resuspended in maintenance medium. 1x106 cells/well were plated in a 12 well plate and medium changed after 24 hours. Cells were then cultured in differentiation media (alphaMEM + 10% FBS + 1% Penicillin/Streptomycin + 10 µM Dexamethasone + 100 µM ascorbic acid + 5 mM β-glycerol phosphate). Immature primary osteoblasts (pOBs) were obtained by sequential digestion of excised 3 to 5 day old PAR2+/+ and PAR2−/− neonatal calvariae as previously described49. Mature long bone pOBs were extracted by digesting bone chips from femur and tibia of 8 to 12 month old mice (experiments were age and sex matched) in 2mg/ml Collagenase II for 2 hours and plating bone chips. Osteoblasts “climb out” the chips reaching confluency in 10 to 14 days. Assessment of pOBs growth was carried out by plating different concentrations of cells in a 96 well plate and changing maintenance media every 2 or 3 days. 7 days after seeding, Alamar blue was added to the medium and incubated for 4 hours. Medium was then colour measured in a spectrophotometer (540 nm). Prior to co-culture with osteoclasts, immature primary osteoblasts were matured to promote osteoclastogenic functions, culturing in 1µM Prostalgandin E2 (PGE2, Sigma,UK) and 10nM 1,25(OH)2D3 (VitD, Sigma, UK) for 8 days with media changed every 3 days and cells split if reaching 70 to 80% confluency.
Mineralization assay
Primary osteoblasts (immature or mature) were plated at 2×104 cells per well in 24 well plates and cultured in maintenance medium (alphaMEM + 10% FBS + antibiotics) for two days before changing to mineralization medium (maintenance medium + 50 µg/ml ascorbic acid and 5 mM β-glycerol phosphate, βGP; Sigma). Medium was changed every 3–4 days for up to 21 days. Cells were fixed for staining or lysed for mRNA extraction. Initial fixation was with 4% paraformaldehyde for 30 minutes at RT, then cells were stained with 2% alizarin red, pH 4.2. Extracellular matrix was stained for 30 minutes with 1% Alcian blue in 3% acetic acid and washed twice in 3% acetic acid. RNA was extracted with RNeasy kit (Qiagen, UK). RNA was quantified and reverse transcribed (Primer Design, UK). qPCR was carried with Precision PLUS SYBR Green Master mix (Primer Design, UK) on a Step One-Plus machine (Applied Biosystems). All gene expression data were normalized against Atp5B (Primer Design, UK; sequence not disclosed) and B-actin. Primer sequences are shown in supplemental table 2.
Bone marrow monocyte quantification
Freshly isolated BM cells were washed in PBS and then live dead staining was conducted using CyStain DNA 2 Step DAPI stain (Sysmex; UK) diluted 1:2 in PBS, incubated for 15 minutes at room temperature. The samples were topped up with FACS buffer and centrifuged at 400g for 5 minutes at room temperature. Cells were incubated with 1:10 dilution of CD16/32 block (BD Pharmingen; UK) to prevent non-specific binding of antibodies via Fc regions for 10 minutes at room temperature. Following this, an antibody cocktail against markers used to identify BM cell populations in mice (Supplemental table 3) was added to the samples for 30 minutes at room temperature. Monocytes were identified as Ly6G−, CD3−, B220−, CD11b+ and Ly6C+ (Supplemental table 3). Quantification was expressed as the percentage of gated populations from total bone marrow cells (from single cells). Data analysis was subsequently completed using FlowJo software.
Osteoclast in vitro cultures
Freshly isolated bone marrow was used to obtain osteoclast precursors for culture. In brief, total BM was cultured in a 75cm2 tissue culture flask overnight (12–16 hours) in 37°C and 5% CO2. Post overnight incubation the non-adherent bone-marrow cells (NA-BMCs) were collected; this population should be depleted of adherent stromal cells and enriched for monocyte populations. The NA-BMCs were resuspended at 1x106 cells/ml in complete α-MEM. These cells were cultured in flat bottom 96 well plates, at a concentration of 1x105 cells/well in the presence of 30ng/ml of recombinant murine M-CSF overnight (approximately 18 hours). The following day media was half changed by removing 50µl of media and adding 50µl of a-MEM containing 100ng/ml of both murine M-CSF and RANKL (to give a final concentration of 50ng/ml of each). Cells were checked daily to monitor progress and after 4 days media was refreshed again by half changing. Negative control for OC formation was NA-BM cultured in M-CSF alone, with no RANKL. The following day (day 5 of the culture) media was removed, cells were fixed, and TRAP stained. Analysis of whole well images of TRAP-stained murine cultures was conducted with ImageJ software. Analysis included counting total osteoclast numbers (TRAP+, 3 or more nuclei), as well as quantification of the area of each well that contains TRAP + multinucleated osteoclasts, used as a surrogate of osteoclast size. OC area was investigated by manually outlining OC cells on ImageJ and calculating the percentage of the total well area covered by outlined OCs.
In vitro resorption assay
As well as TRAP staining of murine osteoclasts, activity of these cells was also measured by culturing osteoclasts on osteo-assay surface plates (Corning; UK). Culture was extended to 12 days, with culture media half changed every 3 days. On day 12, the media was removed and 60µl of 10–15% sodium hypochlorite solution (Sigma-aldrich; UK) added to each well for 3 minutes to remove the cells. Wells were subsequently washed 3 times in distilled water and left to dry. Images of wells were taken on the EVOS FL Auto Cell Imaging System (Life Technologies; UK) light microscope. The proportion of the cell culture well with resorbed mineral was calculated on ImageJ software.
In vitro co-culture assays
To determine the osteoblast contribution to osteoclast differentiation, osteoblasts were co-cultured with monocytes. 2.5x103 osteoblasts were plated per well onto flat bottom 96 well plates. The following day, monocytes were isolated from bone marrow suspensions following EasySep mouse monocyte isolation kit manufacturers instructions (StemCell, UK). Monocytes were plated at 1x105 cells/well onto the osteoblasts. Co-cultures were cultured in maintenance media containing 1µM Prostalgandin E2 (PGE2, Sigma,UK) and 10nM 1,25(OH)2D3 (VitD, Sigma, UK) or 50 µg/ml ascorbic acid and 2 mM β-glycerol phosphate Media was changed every 3–4 days and cultured for 9–10 days. Cells were then TRAP stained and quantified.
Statistical analysis
Data were tested for normality (Graphpad Prism v10, Shapiro-Wilk test) and expressed in graphs as mean ± SD unless stated otherwise. Genotype comparisons were carried out with a one-way analysis of variance (ANOVA), unless otherwise stated. Temporal comparisons dependent on genotype were analysed with a two-way ANOVA. Specifics on statistical tests are included in the figure legends.