Accelerated post traumatic osteoarthritis in a dual injury murine model

Objective: Joint injury involving destabilisation of the joint and damage to the articular cartilage (e.g., sports-related injury) can result in accelerated post-traumatic osteoarthritis (PTOA). Destabilised medial meniscotibial ligament (DMM) surgery is one of the most commonly used murine models and whilst it recapitulates Osteoarthritis (OA) pathology, it does not necessarily result in multi-tissue injury, as occurs in PTOA. We hypothesised that simultaneous cartilage damage and joint destabilisation would accelerate the onset of OA pathology. Methods: OA was induced in C57BL/6 mice via (a) DMM, (b) microblade scratches of articular cartilage (CS) or (c) combined DMM and cartilage scratch (DCS). Mice were culled 7, 14 and 28 days post-surgery. Microcomputed tomography ( m CT) and histology were used to monitor bone changes and in ﬂ ammation. Dynamic weight bearing, an indirect measure of pain, was assessed on day 14. Results: Osteophytogenesis analysis via m CT revealed that osteophytes were present in all groups at days 7 and 14 post-surgery. However, in DCS, osteophytes were visually larger and more numerous when compared with DMM and cartilage scratch (CS). Histological assessment of cartilage at day 14 and 28, revealed signi ﬁ cantly greater damage in DCS compared with DMM and CS. Furthermore, a signi ﬁ cant increase in synovitis was observed in DCS. Finally, at day 14 osteophyte numbers correlated with changes in dynamic weight bearing. Conclusion: Joint destabilisation when combined with simultaneous cartilage injury accelerates joint deterioration, as seen in PTOA. Thus, DCS provides a novel and robust model for investigating multiple pathological hallmarks, including osteophytogenesis, cartilage damage, synovitis and OA-related pain. © Authors. Osteoarthritis


Introduction
Osteoarthritis (OA) is globally the most prevalent musculoskeletal disease. 1 OA is characterised by significant structural changes in joint tissues such as articular cartilage degradation, osteophyte formation (bony outgrowths), subchondral osteosclerosis, epiphyseal bone expansion, synovitis and pannus formation. Although OA is often associated with increasing age, an accelerated progression is recognised in individuals suffering previous (often sport-related) joint injury involving multi-tissue damage. Notably, 66% of patients presenting with articular surface damage in their joints progressed towards OA. 2,3 Furthermore, current data suggests that patients with chronic OA who suffered an earlier articular trauma are, upon diagnosis, >10 years younger than those patients with no history of joint injury. 4 Taken together, post-traumatic OA (PTOA), as characterised by accelerated joint degeneration, pain and dysfunction, accounts for 12% of all OA cases. 5 Lapine models used to investigate PTOA resulting from multitissue joint injury, usually involve traumatic rupture of the anterior cruciate ligament (ACL) and damage of surrounding tissues including the meniscus via mechanical impact on the patellofemoral joint. 6e8 These have yielded valuable insight into PTOA progression, but involve relatively undefined and uncontrolled damage to various joint tissues. In comparison, surgically-induced murine models that develop OA-like pathology include the destabilised medial meniscotibial ligament (DMM) model and anterior cruciate ligament transection (ACLT) murine model. While both recapitulate features of OA pathology, these do not arise directly from an initial multi-tissue injury, as occurs in PTOA. This may be an important limitation, as increasing evidence suggests that joint tissue interactions may play a significant role in driving disease progression. 9e12 Alternative murine models include articular injection of agents that induce inflammation and/or tissue toxicity/ damage, but questions remain as to how well these represent the joint damage seen in OA. 13 To better understand accelerated OA progression, what is needed is a murine model that combines joint destabilisation with defined/controlled joint tissue injury, and is accessible to genetic manipulation, while enabling comprehensive evaluation of disease-related changes in relevant joint tissues (e.g., cartilage, bone, meniscus). The present study is the first to present a dual injury murine model involving simultaneous cartilage damage and joint destabilisation, and was developed to test our focused hypothesis that concomitant cartilage damage and joint instability accelerates the onset of OA pathology as seen in PTOA.

Animals
Experiments were performed on 10 week old adult 22e28 g wild-type C57BL/6 male mice (Envigo, UK). Mice were housed at the University of Glasgow or University of the West of Scotland and maintained under standard animal husbandry conditions in accordance with UK home office regulations. All procedures were performed in accordance with UK Home Office Regulations, and with study approval by Animal Welfare Ethical Review Boards at the University of Glasgow and the University of the West of Scotland.

Induction of OA
OA was induced by one operator (to minimise surgical variability) under aseptic conditions via (a) transection of the medial meniscotibial ligament (DMM) 14  Buprenorphine (Vetergesic; 30 mg intraperitoneally) was administered pre-operatively. 2% oxygen and 2% Isoflurane (Isoflurane Vet, Merial) was used to anaesthetise animals throughout the surgical procedures. Animals were maintained for 7, 14 and 28 days, with knee joints subsequently harvested for mCT and histological analysis. Sham operated joints of OA-induced mice were used as controls.

Microcomputed tomography (mCT)
Knee joints were fixed in 4% paraformaldehyde solution for 24 h, then transferred to 70% EtOH and analysed by mCT to examine the calcified tissues using a Skyscan 1272 mCT scanner (Bruker, Belgium; 0.5 aluminium filter, 50 kV, 200 mA). Samples were scanned at voxel size 4.5 mm, 0.5⁰ rotation angle for quantification, and 2 mm, 0.2⁰ rotation angle for imaging [ Fig. 2 Scans were reconstructed in NRecon software (Bruker, Belgium). Osteophytes were identified in the resulting three-dimensional image stacks using CTvol software (Bruker, Belgium) and then number of osteophytes per joint counted. Osteophyte bone volume was measured by manually delineating the edge of osteophytes protruding from the subchondral plate as the region of interest (ROI) for analysis, and quantified for bone volume using CTan software (Bruker, Belgium). Subchondral bone was analysed by selecting a volume of interest (VOI) of 0.5 Â 0.9 Â 0.9 mm 3 in the centre of load in the medial and lateral tibial plateau. 15 ROIs (medial and lateral) delineating the trabecular structure within the tibial epiphysis were selected in the 2D coronal view of the stack to analyse the bone volume and microarchitecture using CTAn.

Histological assessment of cartilage damage and synovitis
Joints were decalcified overnight (Formical 2000; Decal Chemical, New York, USA) and embedded in paraffin wax. Coronal sections (6 mm) were cut and then stained with haematoxylin, safranin-O/fast green. Using a validated scoring system 14 ranging from 0 (normal) to 6 (>80% loss of cartilage), severity of cartilage damage was assessed histologically on joints previously scanned for micro computed tomography (uCT). Two scorers blinded to the specimens, scored the medial tibial quadrant in 4e6 sections from each mouse. We did not include initial cartilage lesions inflicted by microblade scratches in both cartilage damage and DCS models in our scoring. Synovitis was assessed by two blinded scorers using a recently developed scoring system 16 which we modified from the original to focus only on pannus formation, synovial membrane thickening and sub-synovial hyperplasia. For both scoring systems, there was good agreement between scorers with intraclass correlation coefficient for cartilage scoring of 0.96 (95% CI 0.92 to 0.98), and synovitis scoring of 0.83; (95% CI 0.65 to 0.92).

dynamic weight bearing
As an indirect measurement of pain, limb weight bearing was assessed in mice 14 days after surgery using the BioSeb Dynamic Weight Bearing (DWB) chamber (BioSeb, Marseilles, France). All animals were individually recorded for 5 min, 1 min of which was validated and subsequently analysed. The parameters examined were load on ipsilateral (injured) hind paw and load on front paws.

Statistical analysis
Prism 6 (Graphpad) was used to perform statistical analysis. For normally distributed data, analysis involved one-way analysis of variance (ANOVA) and multiple comparisons using Bonferroni correction. Student t-tests were also used. Data not normally distributed was analysed by KruskaleWallis one-way ANOVA and multiple comparisons using Dunn's corrections. Linear regression curves were generated and Spearman's correlation coefficient (r) was used to assess strength of relationship. P values less than 0.05 were considered significant.

Accelerated osteophyte development after dual injury
Evaluation of osteophyte formation at an early timepoint (7 days post-surgery) revealed that although osteophytes could be detected in some mice within each of the experimental groups [ Fig. 1 To determine how these osteophytes develop over time in these models, we investigated osteophyte morphology and structural characteristics at day 14. In the DMM group, arboreal-like osteophytes were observable in 6 out of 8 mice, whilst in the CS and DCS mice, 100% of mice exhibited osteophytes

Dual injury increases cartilage degradation and synovitis
While cartilage damage and/or synovitis were not normally detectable at day 14 post-induction, we sought to investigate whether these might emerge in the DCS model at this early timepoint as an indicator of accelerated PTOA. Histological analysis of the medial compartment of the knee joint at 14 days following DMM, CS or DCS surgery was performed. Over and above the initial scratch-induced cartilage damage, we found significant further cartilage degradation in the DCS model on the tibia when compared with DMM and CS models [  Fig. 6(C)]. Suggesting that the severity of OA-related pain may be related to the extent of osteophytogenesis. As we observed a positive correlation between osteophyte number and both cartilage damage and synovitis, we further investigated these parameters in relation to OApain related behaviour: however, neither cartilage damage nor synovitis exhibited a significant correlation with increase front paw load (r ¼ 0.35, P ¼ 0.09 and r ¼ 0.28, P ¼ 0.19 respectively).

Discussion
Trauma to the joint is known to promote early osteoarthritis, but the mechanisms for accelerated disease are not well understood. This study presents a novel dual injury murine model to address the important question of whether concomitant joint destabilisation and defined cartilage damage leads to early OA characterised by progressive cartilage destruction, osteophytogenesis, synovitis and pain.
Critical to model selection and/or development, is the specific research hypothesis being addressed; and how translatable these findings are to human OA. There are various well-established and widely used experimental models of OA across a range of species. While none are ideal in modelling human disease, Little and colleages have identified preferred characteristics, which include recapitulating human joint disease in a manner amenable for measurement and monitoring, and within a reasonable time frame. 19 Model types range from surgicial induction and traumatic injury, to chemical induction methods, each with their strengths Murine models are important tools for research on various human diseases, and for OA, one of the most widely used is the DMM surgical induction model generated by Glasson et al.. 14 This induces joint instability and recapitulates the cartilage damage, osteosclerosis, osteophytogenesis and pain that characterise OA in humans, although is technically demanding with respect to surgical skill and need for aseptic conditions. Nevertheless, an experienced researcher can achieve good reproducibility with minimal surgical artefact, and thus we consider DMM to be a robust model in our OA research programme. However, DMM does not reproduce the acute trauma-induced damage to surrounding soft tissue (including cartilage) that is observed in human sport-related injuries. 20,21 PTOA is characterised by accelerated pathology post-injury, and there is a recognised subcohort of OA patients with symptoms limiting life-style at an earlier time point than normally seen. While this is likely to be a consequence of the synergistic interaction of injury to multiple joint tissues, the exact mechanisms driving this accelerated disease progression are not fully understood. In the context of traumatic injury, altered joint loading and initial cartilage damage as a consequence of meniscal damage has been linked with OA development. 22 Rabbit ACL rupture models involving blunt impact trauma are available to study PTOA. 7,8,23 Whilst these models can imitate and reproduce multi-tissue damage similar to that of human knee joint impact traumas, they are limited in not being able to target and confine this trauma specifically to defined tissues within the joint in a reproducible fashion. To this end we have developed a novel dual-injury murine model specific for investigation of OA arising from simulataneous cartilage damage and joint destabilisation. Our DCS procedure now enables generation of a controlled multi-tissue injury murine model mimicking significant trauma to a healthy knee. The DCS model is thus highly relevant to human disease, offering significant potential in meeting a currently unmet scientific need for a model that more accurately represents the subcohort of young otherwise healthy individuals suffering accelerated PTOA resulting from sporting injuries. Such injuries involve simultaneous damage to various joint tissues, including ligaments, bone and cartilage. With this model we observed accelerated PTOA via combination of defined cartilage micro-lesions (Supplemental Fig. 1(D)) with altered loading due to destabilisation of the medial meniscus, allowing us to investigate mechanisms known to be key OA triggers in human disease.
Osteophyte formation is believed to be a response mechanism triggered by a loading imbalance within the joint, expanding the surface area of the tibial plateau to better support the alteration in biomechanical loading. 24,25 We have previously shown DMM alone can induce osteophytogenesis as early as day seven post-surgery in C57BL/6 mice. 18 This is consistent with the present study, in which we report the presence of small osteophytes, low in number, in the DMM model at day seven [ Fig. 1(B) and (C)]. Few, small osteophytes were also present in the CS model at day 7 [ Fig. 1(A) and (B)]. However, the DCS model presented with quantifiably larger, and in most mice, more numerous osteophyte formation at day 7 [ Fig. 1(A)  and (B)]. These data would suggest dual trauma consisting of cartilage damage and joint destabilisation can accelerate the formation of osteophytes as early as 7 days post-injury. Previous studies have reported that secondary osteophyte growth is initiated by cartilage damage, 26e28 however the factors influencing the formation and growth of osteophytes are poorly defined. Cartilage lesions have been associated with the formation of larger osteophytes, 29 supporting our day 14 findings of both cartilage damage and DCS models presenting with larger, more numerous osteophytes compared with the DMM model [ Fig. 2(C) and (D)], where no initial cartilage damage was induced. This provides evidence that osteophytogenesis is accelerated by cartilage damage when combined with simultaneous joint loading imbalance. The specific site of cartilage damage may itself alter the type of pathologic outcome. Our study focussed on cartilage damage within the medial compartment however, there could be value in future studies focussing on the impact of cartilage damage on alternative sites (i.e., non-weight-bearing areas of cartilage) to determine how this affects pathologic outcome. Whilst the DCS model was developed to test a specific hypothesis, it has considerable potential to extend beyond OA, to investigation of osteophytogenesis in various pathological scenarios. Examples include ankylosing spondylitis and degenerative disk disease. Additionally, the study of osteophytogenesis could provide a controlled tool to investigate the initiation of endochondral ossification.
Another prominent feature of OA is the development of osteosclerosis in subchondral bone. When the cartilage becomes damaged there is an increased transfer in load to the subchondral bone, 30 as well as the release of inflammatory and osteoclast stimulation factors, which can lead to an increase in bone remodelling. 31,32 Osteosclerosis was measured as an increase in bone volume (% BV/TV) in the medial subchondral region of the tibia. At both 7 and 14 days post-surgery, all disease model groups exhibited evident osteosclerosis in the loaded medial compartment of the joint, where the destabilisation and/or cartilage damage was induced. After the initial increase at day 7, osteosclerosis in the CS model remained relatively constant at 14 days, whereas both the DMM and DCS groups significantly increased, indicating altered biomechanical load plays a key role in the development of osteosclerosis. Furman et al. showed that intra-articular fractures led to subchondral bone thickening and sclerosis of the knee in C57BL/6 mice. 33 The observation that cartilage degradation is more extensive and progressive in the DCS model than either of the other models, suggests that interactions arising from the dual injury exacerbates cartilage breakdown. Trauma to the articular cartilage results in the activation of genes encoding for morphogens and the secretion of soluble inflammatory mediators into the joint cavity. 34e37 The formation of microcracks and increased vascularisation associated with abnormal bone remodelling in joints during OA facilitate molecular transport between cartilage and bone 9 , therefore soluble inflammatory mediators present in the joint cavity as a result of cartilage damage and joint destabilisation may potentiate further cartilage destruction/bone remodelling, thus accelerating joint pathology in PTOA. Interestingly, we observed a significant correlation between cartilage damage and osteophyte number [ Fig. 4(H)], raising the intriguing possibility that soluble factors secreted by the damaged cartilage may contribute to accelerated osteophytogenesis. Cartilage damage also exhibited a positive correlation with synovitis [ Fig. 4(G)] suggesting a link with inflammatory factors.
While traditionally considered a 'wear and tear' disease, inflammation is increasingly recognised to have a role in degenerative changes occurring during OA pathogenesis 38 . Histological features consistent with synovitis have been observed in synovium from OA patients 39e42 as well as expression of proinflammatory cytokines, 41,42 and more recently we have reported considerable heterogeneity of various inflammatory indices in OA synovia. 43 Indeed knee arthrocopic studies of OA patients found synovial inflammation to be associated with OA structural severity, 44 whilst another study reported that serum C-reactive protein levels in early osteoarthritis predicted likely progressive disease; 45 c-reactive protein (CRP) also correlates significantly with disease joint count and radiographic score in OA. 46 Whilst the DMM model does recaptitulate many of the characteristic features of OA, it is generally considered a non-inflammatory model. 47 Our study, however, supported by previous findings, 16 detected histologically the presence of synovitis at 14 days post-surgery. Nevertheless, given the low-level nature of this synovitis, DMM is not the model of choice for investigation of inflammation in OA joints. While the MIA model evokes joint inflammation, 48 this inflammatory response is not intiated in a manner resembling that of OA. Our DCS model yielded a significant and consistently higher level of synovitis than seen in either the DMM or CS models and, similarly to cartilage degradation, synovitis was positively correlated with increased osteophyte number [ Fig. 4I]. Thus DCS presents value in enabling mechanistic interrogation of the putative contribution of synovitis to OA pathogenesis.
Pain is one of the most common and physically limiting symptoms of OA, however the origin of this pain is still poorly understood. DWB presents a behavioural non-invasive means of assessing pain, although other factors (e.g., mechanical impediment) may influence the observed changes. Ipsilateral (injured) hind paw and front paw loading were analysed. There was a significant decrease in injured hind paw load with a corresponding increase in load on the front paws in the DCS model compared with all other groups [ Fig. 6(A) and (B)], suggesting the combination of cartilage damage and DMM enhances pain-associated pathology. There have been conflicting studies regarding the relationship between pain and osteophytes in OA. One previous study presented a significant correlation between knee joint pain in OA patients with osteophytes present in the medial tibial condyle. 49 However, more recently, Sengupta et al. challenged this finding by reporting that the presence of high-signal osteophytes detected on magnetic resonance imaging (MRI) were not associated with pain or pain severity. 50 Our data demonstrates a significant and strong positive correlation between osteophyte number and front paw load at 14 days post-surgery [ Fig. 6(C)]. Synovitis is known to lead to heightened pain sensitivity in knee OA 51 and we therefore elected to further explore this finding in relation to pain-associated behaviour in the DCS model. We did not observe a significant correlation between front paw load and synovitis or an increase in cartilage degradation. The association of osteophytogenesis, cartilage damage, synovitis and pain implicates a causal relationship, and this merits future research to define the nature of any such relationships; this may help identify possible molecular mechanisms driving osteophyte formation, as well as joint pain.
We conclude that the concomitant joint instability and cartilage injury that typify knee joint traumatic injury predispose to accelerated OA progression. To establish this we developed and validated a murine model of acute trauma involving simultaneous controlled damage to multiple joint tissues. This robustly mimicked the accelerated OA pathogenesis observed in otherwise normal individuals with PTOA, as characterised by osteosclerosis and (compared with DMM) enhanced cartilage degradation, synovial inflammation, osteophytogenesis and pain. Importantly, this model also provides a valuable research tool for investigating osteophytogenesis in various contexts, as the DCS model generates an accelerated and more extensive formation of osteophytes at an early time-point. The reproducible synovitis associated with this model also presents means to elucidate the role of synovitis in OA pathology. Similarly, the enhanced pain-related behaviour in DCS provides a robust model for in vivo investigation of joint pain.

Competing interests
The authors declare no competing financial interests.