Synovitis and bony destruction are pathophysiological characteristics of RA, and marginal bony erosion, periarticular osteopenia, and joint space narrowing are the radiographic hallmarks of RA [22
]. Synovial inflammation and bony destruction are closely related processes [24
], but contrary to synovitis, the bony changes are usually irreversible and accumulate with time, and can bring about joint dysfunction and an unfavorable disease outcome [25
]. As a result, RA causes significant socioeconomic impact because of physically disabled and unemployed people [27
Both cellular mechanisms and various inflammatory mediators are involved in the pathogenesis of bone erosion in RA, forming complex networks [24
]. Among these process, OCs are the essential cells involved in the cellular mechanisms of the process of bony erosion [30
]. In RA synovium, OCs are found at the pannus-bone and pannus-subchondral bone junctions of arthritic joints, forming erosive pits in the bone [30
]. Two additional cells play important roles in osteoclastogenesis: synovial fibroblasts and activated T cells. They express RANKL in the inflamed synovium, which promotes osteoclastogenesis, and also express cathepsin K at sites of synovial bone destruction [33
]. RANKL is the key molecule in OC differentiation and the augmentation of activity and survival of these cells, and is often called OC differentiation factor (ODF). In the serum transfer model of arthritis in the RANKL knockout mouse, the synovial inflammation and cartilage erosions are similar to those in wild-type mice, but the degree of bony erosion is significantly reduced [34
]. This result confirms the essential role of RANKL in the pathogenesis of bone erosion, regardless of inflammation or cartilage damage. The expression of RANKL is regulated by proinflammatory mediators such as TNF-α, IL-1, IL-6, IL-17, and PGE2
]. These inflammatory molecules are abundant in RA synovium, so the inflamed synovium supplies an optimal environment for RANKL activation.
In this study, we determined the relation between bony erosion and MIF in human RA. In the previous studies, MIF induces TNF-α, IL-1, IL-6, and PGE2
, which in turn promote RANKL expression [1
], and the synovial MIF concentration is higher in RA patients with bony erosion than in those without [8
]. Based on these results, we hypothesized that MIF might have a role in the pathogenesis of bone erosion, that is, it could have a direct effect on OC differentiation and an indirect effect on the induction of other inflammatory mediators that induce RANKL expression. First, we measured the synovial concentrations of MIF and RANKL in RA patients. Synovial fluid MIF concentration was higher in RA patients than in controls, as in our previous study [8
], but the synovial RANKL concentration did not differ between RA patients and controls. In previous studies, serum and synovial RANKL levels were higher in RA patients than in controls [37
], but the RANKL level was not related to any measures for disease activity [38
]. In contrast, we found that the serum and synovial MIF concentration was well correlated with RA disease activity [8
]. Compared with previous studies, the patients enrolled in this study had longer disease duration and less active disease [37
], so MIF may reflect disease activity more closely than does RANKL. In this study, synovial RANKL concentration was significantly correlated with synovial MIF concentration, and this observation led us to investigate their close relation in the RA synovial tissues.
We investigated the effect of MIF on RANKL expression in RA synovial fibroblasts. Synovial fibroblasts, such as activated T cells, are major sources of the RANKL that promotes OC differentiation and bone erosion [33
]. Like other proinflammatory cytokines, MIF stimulates the expression of RANKL mRNA and protein in RA synovial fibroblasts, but there was no additive effect with other proinflammatory cytokines such as TNF-α and IL-1β. After blocking IL-1β, MIF-induced RANKL expression was partially decreased. This result suggests that RANKL expression was directly induced by MIF and also that it was indirectly stimulated by MIF-induced IL-1β. IL-1β has the potential to induce OC differentiation and RANKL expression, and overexpressed MIF could induce some inflammatory mediators, such as IL-1β in RA synovium, resulting in upregulation of RANKL and promotion of OC differentiation. Therefore, the MIF-IL-1β-RANKL interaction could be a major axis involved in RA bone erosion.
We investigated the effect of MIF on OC differentiation. We substituted MIF for RANKL in the traditional culture system for OC differentiation. After isolated PBMC were cultured with rhMIF and M-CSF, the numbers of TRAP-positive multinucleated cells were counted. OC developed in this new system without RANKL, but the degree of OC differentiation by MIF was less than that of RANKL. This result showed that MIF is one of the inflammatory cytokines involved in osteoclastogenesis, even if RANKL is the major molecule that induces OC differentiation. We also demonstrated that MIF-prestimulated RA synovial fibroblasts have a potential effect on osteoclastogenesis when the cells are co-cultured with PBMC. This culture system is more practical in an in vitro system similar to human RA synovium. RA synovial fibroblasts are exposed to a variety of cytokines that promote inflammation, and when these ailing cells encounter OC precursors, they could induce osteoclastogenesis by cytokine production or direct interaction between cells. This study was focused on the indirect osteoclastogenic effect mediated by RA synovial fibroblasts and RANKL, but MIF could directly enhance osteoclastogenesis from monocytes in the absence of additional RANKL. These two pathways imply more distinct and reinforced mechanisms for MIF-induced osteoclastogenesis, and a tipping point such as MIF production could be a potential therapeutic target.
In contrast to our results, a recent study suggests that MIF inhibits osteoclastogenesis [13
]. Although MIF enhances the expression of RANKL mRNA in murine osteoblasts and the expression of RANKL mRNA is enhanced in MIF transgenic mice, MIF inhibits OC formation in bone marrow cultures by decreasing fusion and decreasing the number of nuclei. The number of TRAP-positive OC is greater in MIF-deficient mice than in wild type mice, and the addition of MIF to the cells decreased TRAP-positive OC formation. Therefore, it appears that MIF plays an inhibitory role in bone resorption. The discrepancy between two studies could be explained by several differences in study systems. First, our study used human PBMC, whereas the former study used osteoclast precursor cells from MIF knockout mice. MIF inhibits osteoclast formation in vitro
in wild type mice bone marrow cell cultures and in the RAW264.7 macrophage cell line. Based on these data, MIF appears to directly inhibit osteoclastogenesis in vitro
but its effects on osteoclasts in vivo
are complex and may result from decreased RANKL expression in the osteoclast precursor cells from MIF knockout mice that were exposed to low levels of RANKL in vivo
and as a result these cells have increased sensitivity to RANKL in vitro
when cultured at high density.The MIF knockout mice that they used, had a marked resistance to lipopolysaccharide-induced endotoxic shock, and decreased TNFα production in response to lipopolysaccharide treatment. TNF-a also acts directly on the osteoclast precursor to potentiate RANKL-induced osteoclastogenesis, even in the absence of elevated levels of RANKL. MIF knockout mice were used in the previous paper, and had inhibited TNF production. Thus, osteoclast formation may have been inhibited. Second, we put the focus on an actual inflammatory disease of humans. In human RA synovial fibroblasts, the over-expressed MIF induces other inflammatory mediators, and then the inflammatory mediators, such as RANKL and IL-1β, enhance and potentiate osteoclastogenesis. Third, the former study treated RANKL with MIF in the OC differentiation system, but we did not treat RANKL in the culture system. More intensive study will be needed for explaining these conflicting results. We hypothesize that MIF might play an essential role in normal bone remodeling; however, over-expressed MIF might have an osteoclastogenic effect on bone metabolism in inflammatory diseases.
We found that MIF-induced RANKL expression in RA synovial fibroblasts was decreased by inhibition of NF-κB, PI3K, STAT3, AP-1, and p38 MAPK, but not ERK and calcineurin. Of the three MAP kinase pathways, only p38 MAPK was involved in MIF-induced RANKL production. In addition, MIF-induced osteoclastogenesis was suppressed by inhibition of NF-κB, PI3K, AP-1, and p38 MAPK, but not by inhibition of JAK/STAT3. These results suggest that there are different signal pathways involved in MIF-induced osteoclastogenesis. Considering that AP-1 is a downstream molecule, MIF seems to induce RANKL production by synovial fibroblasts mainly via NF-κB, PI3K, STAT3, and p38 MAPK, while it promotes OC differentiation from monocyte precursors via NF-κB, PI3K, and p38 MAPK. In recent years, numerous studies have attempted to define the signal transduction pathways of inflammatory cells activated by MIF in RA synovial fluid. MIF promotes cyclooxygenase-2, PGE2
, and IL-6 expression via p38 MAPK [39
]. MIF also upregulates IL-8 and IL-1β via tyrosine kinase-, protein kinase C (PKC)-, AP-1-, and NF-κB-dependent pathways [40
]. MIF controls the proliferation of RA synovial fibroblasts, mediated by ERK [36
]. The upregulation of MMP-2 by MIF is dependent on PKC, JNK, and Src signal pathways [4
]. MIF also upregulates other MMPs including MMP-1 and MMP-3 via tyrosine kinase-, PKC-, and AP-1-dependent pathways [6
]. Through the various intracellular signal transduction pathways, MIF activates RA synovial fibroblasts to promote inflammation, cartilage degradation, and bony destruction. In our previous study, we found the induction of MIF is mediated by p38 MAPK pathway when RA synovial fibroblasts are stimulated by conA, IFN-γ, CD40 ligand, IL-15, TGF-β, as well as IL-1β and TNF-α [41
]. Among these data, intracellular signal pathways are deeply involved in the pathogenesis of RA. Clinical studies for RA treatment using the inhibitors of different signal pathways such as Syk, p38 MAP, and JAK have been performed until now, and successful results are expected [42
]. Beyond the inhibition of cytokines or immune cells, oral inhibitors of intracellular molecules will be another choice for refractory RA.