The potential value of autologous mesenchymal stem cells (MSCs) derived from a patient's joint is now well recognized in osteoarthritis (OA) and rheumatoid arthritis (RA) for the repair of bone and cartilage damage. However, the fact that MSCs have been directly exposed to inflammation in the joint has not been given sufficient attention. We have shown recently that chondrogenesis was qualitatively diminished in MSCs derived from the synovial fluid of RA patients compared with OA [1]. In this report, we investigate the effect of the joint microenvironment on MSCs' repair capability in vitro and the possible molecular mechanism by which cells become deficient.

Synovial tissue was obtained during arthroscopy from 15 patients with RA and OA, fully documented with visual analogue scores, which were validated as measure of inflammation locally. High and low levels of inflammation were observed for both diseases. We observed an inverse relationship between levels of inflammation in the joint (VAS) and the chondrogenic potential of MSCs (R = -0.750, P < 0.01) using an in vitro differentiation assay. Such a relationship was not observed for bone differentiation.

The induction of two specific transcription factors (TF) is necessary for differentiation to take place: Sox9 for chondrogenesis and Runx2 for osteogenesis. We measured the levels of expression of these TF in MSCs isolated from nine synovial fluids of RA and OA patients and compared them with MSCs isolated from healthy bone marrow donors (n = 6) and skin fibroblasts (negative control, n = 4). Sox9 expression was higher in OA than in bone marrow but was reduced to negative control levels in RA. Runx2 expression was reduced in both RA and OA. Sox9 and Runx2 expression was also investigated in synovial tissue MSCs from six RA patients. Sox9 and Runx2 expression were inversely correlated with VAS scores for inflammation (R = -0.737, P < 0.01 and R = -0.843, P < 0.01, respectively).

To investigate the effect of proinflammatory cytokines on the ability of MSCs to differentiate we quantified the expression of Sox9 and Runx2 following treatment with TNFα, TGF-β₃, and IFNγ. TGF-β₃ is a trigger of chondrogenesis and accordingly increases the expression of Sox9 and Runx2 in bone marrow MSCs (n = 4), whereas TNFα and INFγ reduce the expression of both TF. In RA (n = 6), TGF-β₃ had lost the ability to induce the expression of Sox9 and Runx2 whereas TNFα and INFγ increased it. The cytokine effects were all directly correlated with previous exposure to inflammation (P < 0.05). Therefore, in vivo the joint 'milieu' (which includes these cytokines and several others) is unlikely to allow the induction of a differentiation pathway where TF must be expressed at the right time and amounts and in the right order. Finally, using tissues collected pre and post biologic treatment, we also demonstrated that this deficit can be corrected in vivo in a controlled joint milieu (but so far not in vitro) – raising the hope that one day cell-based therapies for cartilage repair will be available.

Altogether, these data demonstrate that exposure to inflammation has a profound effect on the ability of MSCs to respond to differentiation triggers. Our data also suggest that cytokine(s) are likely to be involved in the acquired deficit in chondrogenesis in RA. Controlling joint inflammation with anti-TNF is associated with the downregulation of inflammation-related parameters (including bone damage [2]) and would therefore be necessary, but may not be sufficient to allow MSCs to undertake cartilage repair.