Osteoarthritis (OA) is a degenerative joint disease (DJD) that impacts multiple joint tissues (i.e. subchondral bone, synovium), but is typically associated with a deterioration of articular cartilage. Although numerous factors have been suggested to be important contributors to the development and progression of this disease, very few, with the possible exception of FRZB or GDF5, have been confirmed to have causal roles 1. OA is considered in many instances to result from years of wear and tear on the joint. In this scenario, as with many other structures and organs within the body, the cartilage is considered to wear out as a result of ageing. Therefore, OA usually develops over a protracted period, which can be accelerated in certain individuals because of an underlying genetic predisposition or various environmental factors.

In the past few years, genetic links to OA have been established, and the first mutations in the collagen type II gene involved in the disease described 23. More recently, other genes associated with the WNT and GDF signalling pathways have been implicated in OA susceptibility 456. With regard to environmental factors, the biggest contributor is most likely to be physical activity/trauma and underlying medical conditions that place a greater mechanical burden on articular cartilage. The net result of these various factors is the loss of the integrity of the cartilaginous extracellular matrix (ECM), leading to a decrease in mechanical strength. This increases the susceptibility of the articular cartilage to further damage, and because of its limited ability to repair itself the disease worsens. In OA, the structural integrity of the matrix is irreversibly lost, leading to joint dysfunction 7.

One class of molecules that is important in development and homeostasis are the metabolites of vitamin A, the retinoids 89. In the developing mammalian limb, retinoic acid has long been known to affect cells of mesenchymal and chondrogenic origin 10111213. The addition of retinoic acid to high-density cultures of limb bud mesenchymal cells (which form cartilage nodules in vitro) has been shown to decrease the number and size of cartilage nodules formed. More interestingly, treatment of mature chondrocytes with retinoic acid causes them to assume an immature phenotype 14151617. This is accompanied by a decrease in Col2a1 expression 14 and an increase in metalloproteinase expression 18 that leads to degradation of the ECM. In this regard, retinoic acid treatment of cartilage is commonly used to study cartilage degeneration 1920. In vivo, intra-articular injection of retinoic acid leads to chondrocyte dedifferentiation and DJD 21. More recently, antagonists of the retinoic acid receptors have been tested in a rheumatoid arthritis model in mice and rats and found to improve histological scores, and this was associated with decreased expression of Mmp13 22.

The changes in aggrecan metabolism seen in OA are similar to those produced by treatment of cartilage with retinoic acid. Bovine cartilage explant cultures treated with retinoic acid exhibit increased degradation of proteoglycans 23. In rat osteosarcoma cells and primary bovine chondrocytes, treatment with retinoic acid produces cleavage of aggrecan (ACAN) at the E373-A374 peptide bond that is also cleaved in OA 24. The retinoic acid-mediated degradation of ACAN is inhibited by metalloprotease inhibitors, but not by inhibitors of cathepsin B 23. Others have shown that the addition of retinoic acid to chondrocytes stimulates maturation and hypertrophy consistent with the effects observed in vivo 2526. A switch from type II expression to type X and a decrease in ACAN expression accompanied by an increase in catabolism of collagen type II and ACAN was observed. In this regard, retinoic acid has been shown to enhance chondrocyte hypertrophy both in vitro and in vivo, where retinoic acid was observed to promote premature closure of the growth plate 2728.

To examine the status of the retinoic acid signalling pathway in OA, we have used two murine DJD models and quantified the expression of a retinoic acid-regulated gene, including Crabp2 in the articular cartilage. In a recent study, Crabp2 was found to be elevated in DJD in a rat model of OA 2930. We found Crapb2 to be significantly increased in early OA, indicating that the retinoic acid pathway may play a role in OA pathophysiology.

In an earlier study of a rat surgically induced model of DJD, Appleton and colleagues reported that Crabp2 expression as determined from microarray analysis was elevated in the OA joint four weeks post-surgery 30. Crabp2 is regulated by retinoic acid and is often induced upon activation of this pathway 33. We were interested in confirming these findings and determining the kinetics of Crabp2 induction during OA. Further, we desired to develop and validate murine cartilage-sparing models of OA to facilitate the use of genetically modified lines and enable molecular analysis of gene expression in the articular cartilage. For these purposes, surgeries were tailored to produce less joint instability with the intent of producing a more slowly progressing disease than the standard anterior cruciate ligament (ACL) models as has been recently described for the DMM and other models 3134.

MTX and PMX surgeries were performed and joint histology was scored at various times up to 10 weeks post surgery. In the MTX group there was a significant increase in the joint score by 10 weeks (n = 7, p < 0.001), although increases at two and six weeks were not significant (Figure 2). In the recently reported DMM model, significant joint deterioration was observed at four weeks, and the differences between this model and the related MTX model may be a consequence of the different mouse strains used (129/SvEv versus the C57BL/6NCr1 strain utilised herein) and/or the different scoring methodologies employed 31. Consistent with the more aggressive nature of the surgery, the PMX mice displayed a significant increase in joint score by six weeks after surgery (n = 7, p < 0.01); at 10 weeks these animals had a joint score of 8.4 (n = 5) out of a possible total of 24 (Figure 2). The early histopathological findings are focal, superficial to deep fibrillation sometimes associated with variable degrees of matrix depletion in these early stages. In the more aggressive PMX surgical paradigm the lesions become more diffuse and are more likely to have vertical fissures through the midzone with less common delamination of the superficial layer (Figure 3). Together, these results demonstrate that the PMX and MTX surgeries give rise to moderate and mild forms of DJD that are associated with a slowly progressing joint disease.

To quantify gene expression in single knee joints we developed efficient RNA isolation methods that yielded about 100 to 150 ng of total RNA from individual stifle joints sufficient for analysis of five to seven genes (20 ng of RNA per gene). Previous studies have shown that Col1a1 and Col10a1 are both elevated in OA 35363738. We also examined the expression of Sox9, a transcription factor important in chondrocyte differentiation and matrix production 39. With the exception of the four-week time point, the expression of Col1a1 was found to be elevated more than three-fold at all times in the MTX/PMX joint in comparison to the sham control (Figure 4). In contrast, there were only slight changes in the expression of Col10a1 or Sox9 during the first six weeks after surgery (Figure 4), with a significant (n = 5, p < 0.05) two-fold increase in Col10a1 being observed in six-week PMX samples in comparison to the sham control (Figure 4). Interestingly, examination of Crabp2 expression revealed a large increase in expression at two weeks after surgery in both MTX and PMX of about 18-fold and 10-fold, respectively (n = 5; MTX, p < 0.05; PMX, p < 0.01; Figure 5a). This magnitude of induction declined over time; however, by six weeks in the PMX mice there was about a three-fold increase in Crabp2 expression in PMX knees in comparison to the sham or contralateral knee (n = 5, p < 0.05; Figure 5a), whereas at 10 weeks, there was still about a five-fold increase in Crabp2 expression in PMX knees in comparison to the contralateral control (n = 7, p < 0.05; Figure 5b). Together these results show that PMX and MTX surgeries lead to a slowly progressing arthrosis in mice and that Crabp2 represents an early and sustained marker of DJD.

Crabp2 expression is regulated by retinoic acid and its increase in expression is consistent with activation of this pathway. CRABP2 has been shown to function both to suppress retinoic acid receptor activity by sequestering ligands and also as a vehicle to deliver ligands to the retinoic acid receptors, thereby enhancing ligand-mediated retinoic acid receptor transcriptional activity 404142434445. More recent reports favour the latter function, indicating that increased expression of Crabp2 enhances retinoic acid receptor transcriptional activity 414244. Further, Crabp2 null animals present with minor limb defects including an extra post-axial digit that (based on more recent reports) would suggest decreased retinoid signalling in the absence of CRABP2 94647. In aggregate, increased expression of Crabp2 is generally linked to activation of the retinoid pathway either directly, because expression of Crabp2 is regulated by retinoic acid, or indirectly, because increased CRABP2 would increase retinoic acid receptor transcriptional activity. Inappropriate activation of the retinoic acid signalling pathway is expected to enhance cartilage degradation and/or chondrocyte dedifferentiation, both of which are observed in the osteoarthritic process.

As mentioned above, retinoic acid has been commonly used to promote degeneration in cartilage explants, and this has been associated with increased activity of various aggrecanases. Knockout animals of the gene encoding the aggrecanase ADAMTS5 are protected to a great extent from OA in joint instability models, indicating that this enzyme may play a major role in cartilage catabolism, at least in mouse models of OA 484950. Interestingly, in explants derived from double mutants of Adamts5 and another major aggrecanase Adamts4, addition of retinoic acid was still found to promote release of aggrecan through cleavage in the CS-2 domain 51. Retinoic acid has been shown to increase Adamts5 and Mmp13 expression 1922, and these new findings by Rogerson and colleagues 51 suggest that retinoic acid may also be promoting cartilage degradation through additional and as yet undefined aggrecanase(s). Further, in collagen-induced arthritis in the mouse and streptococcal cell wall-induced arthritis in rats, small molecule antagonists of the retinoic acid receptors were found to ameliorate pain and decrease cartilage loss 22.

In addition to Crabp2, other components of the retinoid signalling pathway were found to be significantly elevated in the joints of the aforementioned rat OA joint instability model, including genes encoding proteins involved in retinoic acid synthesis (Aldh1a3) and retinol transport and delivery (Lrat and retinol dehydrogenase) and a putative retinoic acid target gene (Stra3) 30. Together, these findings along with our observations of elevated Crabp2 expression in mouse models of DJD suggest that retinoic acid may play a fundamental and perhaps unappreciated role in the osteoarthritic process.

As Crabp2 is robustly expressed in early OA, it may represent a marker for detection of early OA. Further, polymorphisms in FRZB and GDF5 have been linked to OA, and similar to the retinoid signalling pathway, they all play a role in endochondral ossification and modulation of their activity may impact maintenance of the articular chondrocyte 1. In this regard, as has been previously suggested, antagonists of the retinoic acid signalling pathway may prove useful for maintaining the chondrocyte phenotype 8.

The joint instability models presented herein in contrast to more aggressive models involving for instance ACL transection, present with OA-like pathology but still appreciable articular cartilage 10 weeks after surgery, thereby enabling the use of molecular approaches to quantify gene expression changes in early OA 34.

The expression of the retinoic acid-regulated gene Crabp2 is significantly elevated in early DJD, and may be a useful marker to follow early changes in cartilage in response to joint instability or in OA. Manipulation of the retinoic acid signalling pathway may prove useful in modifying the clinical course of OA.

ACAN: aggrecan; ACL: anterior cruciate ligament; ANOVA: analysis of variance; DMM: destabilisation of the medial meniscus; DJD: degenerative joint disease; ECM: extracellular matrix; MCL: medial collateral ligament; MTX: meniscotibial transection; OA: osteoarthritis; PMX: partial meniscoectomy; RT-qPCR: reverse transcription quantitative polymerase chain reaction.

The authors declare that they have no competing interests.

IW performed experiments, contributed to experimental design, writing of the manuscript and data interpretation. MC contributed to experimental design and carried out the experiments. FM and TMU were involved in experimental design, data interpretation and writing of the manuscript.