IL-6 has been shown to promote osteoclast formation and bone resorption in some studies, while others have shown an inhibitory effect. Mice administered IL-6 develop hypercalcemia, leukocytosis, and cachexia [33
]. Engrafting of human bone cells infected with an IL-6-expressing retrovirus into nude mice resulted in an increase in osteoclast-lined mineralized trabecular bone surfaces [34
]. Addition of IL-6 and its soluble receptor increased bone resorption in neonatal mouse calvaria organ cultures [35
]. In contrast, transgenic mice over-expressing human IL-6 have a reduction in both osteoclast number and bone resorption [36
]. In two studies using mouse neonatal calvaria and parietal bone organ cultures, IL-6 failed to induce bone resorption [37
]. Furthermore, blockade of IL-6 with antibodies did not prevent PTH, PGE2
, or 1,25-dihydroxyvitamin D3 mediated osteoclastogenesis or bone resorption [38
In this model we conclusively show by histomorphometry and micro-CT that osteolytic lesions are greater in IL-6 knockout mice indicating that IL-6 is anti-osteoclastogenic. Furthermore, the absence of IL-6 increased the number of OCP present in the spleens of titanium treated mice. Thus, IL-6 deficiency results in a systemic response to localized inflammation that increases the number of splenic OCP. We believe that IL-6 is part of a larger negative feedback loop targeting TNFα and osteoclastogenesis. Two possible downstream effects of IL-6 may be to inhibit the production of TNFα from particle-stimulated cells as well as inhibit early osteoclastogenesis (possibly by preventing the TNFα-mediated upregulation of early osteoclast differentiation genes). Our data (Figures and ) support both of these possible mechanisms by demonstrating the IL-6 treatment early in culture blocks osteoclastogenesis and TNFα production from splenocytes. Recently two other groups have also demonstrated an anti-osteoclastogenic effect of IL-6 on osteoclast precursors, one group showing that MAP kinase activation downstream of IL-6 inhibits RANK and JNK-mediated signals [22
]. Duplomb et al. have also shown an anti-osteoclastogenic effect of IL-6 that involves re-direction of the precursor pool toward a monocyte/macrophage phenotype [39
Interestingly, a new biologic therapy targeting IL-6 has emerged for rheumatoid arthritis (RA), called tocilizumab. This humanized IL-6 receptor monoclonal antibody has shown great promise at reducing swelling and radiographic evidence of joint damage in clinical trials for RA [40
]. Pre-clinical work with tocilizumab in collagen-induced arthrtitis in monkeys has demonstrated that IL-6 signaling blockade reduces RANKL expression and osteoclast numbers [42
]. Although the success of tocilizumab would appear to contradict the anti-osteoclastogenic effect of IL-6 that we have observed, both pro- and anti-osteoclastogenic effects may result from IL-6 signal transduction through the “signal orchestration model” as proposed by Kamimura et al. which postulates that individual cytokines acting on a single cell may use complex intracellular signaling cascades with opposing effects [43
]. IL-6 is also known to act synergistically with other cytokines such that the net effect on osteoclastogenesis may vary depending on the exact cytokine milieu that is present [44
]. In this way it might be possible for non-particle related diseases to experience a pro-osteoclastogenic effect of IL-6 that is mechanistically different from the anti-osteoclastogenic effect in particle-mediated osteolysis. We would therefore predict then, that the precise cytokine milieu present in various pathologies would be different, as would the response to tocilizumab therapy.
Our goal was to better define the role of IL-6 during particle-induced osteolysis. Using a well-established and highly quantitative murine model, as well as qualitative micro-CT scans, IL-6 deficient mice developed increased osteoclasts and had enhanced bone resorption following implantation of titanium particles onto the calvaria. Increased bone resorption, osteoclast numbers, and IL-1 expression were also observed in an osteomyelitis model using IL-6-/-
mice as well as in wild-type mice treated with anti-IL-6 antibody [21
]. Finally, IL-6-/-
mice treated with intra-articular zymosan had increased cartilage destruction compared with wild-type controls [45
]. Thus, these other models of inflammatory bone or cartilage destruction using a genetic approach also show an IL-6 anti-resorptive effect.
A series of in vitro experiments were performed in order to define mechanisms through which IL-6 inhibits inflammatory bone resorption, focusing on OCP as key target cells. Initial experiments confirmed that the difference in osteoclast number was not due to an increased pool of precursor cells in IL-6 deficient mice. Both wild-type and IL-6-/- mice spleens contained similar percentages of osteoclast precursor cells (CD11b+) by flow cytometry. Furthermore, wild-type and IL-6-/- spleen cells responded similarly to M-CSF and RANKL. Thus, differences in bone resorption are not related to differences in basal numbers or responsiveness of OCP.
However, it is interesting to note that our in vitro
results were much more dramatic in terms of the relative increase in osteoclasts (3.8 fold increase in osteoclast area compared to wild-type in vs. only a 58% increase in osteoclast number in vivo
, ). One explanation is that osteoclastogenesis in vivo
is a temporally more variable process and that osteoclast numbers may peak at different times. We chose 7 days as our endpoint as this is the point of maximal osteolysis. It is also worth noting that the in vitro
experiment was measured as osteoclast area and the in vivo
data as the number of osteoclasts; one explanation for the discrepancy in magnitude of the response is that the dramatic increase in osteoclast area in vitro
is due to a greater proportion of very large multinucleated cells. This would explain why we see a large osteoclast area but no significant difference in osteoclast number in vivo
. Indeed there are several TRAcP+ osteoclasts visible in the histologic section from the IL-6 knockout that are much larger than any of the TRAcP+ cells in the wild-type or sham sections. Examination of mice deficient in gp130, a critical element in the active IL-6 receptor complex, finds that osteoclasts in these animals too are substantially larger than normal [46
A unique aspect of our approach was the use of OCP in different stages of differentiation. Prior work by our laboratory has defined early osteoclast precursor cells to be CD11b+
by FACS [7
]. Treatment of this early precursor population with IL-6 inhibited stimulation of osteoclastogenesis by RANKL and M-CSF in a dose-dependent manner. Effects were similar in wild-type and IL-6-/-
mice and inhibition persisted following co-treatment with TNFα. Thus, IL-6 is a potent inhibitor of early osteoclast maturation.
Previous work from our lab has shown TNFα-Tg mice have an increased proportion of OCP in the spleen expressing high levels of CD11b [6
] and that these OCP form mature osteoclasts more rapidly in culture. Using CD11b as a marker of osteoclast maturation, here we show that in vitro
treatment with M-CSF for three days enhanced OCP differentiation, consistent with previous reports showing that administration of M-CSF treatment increased the population of CD11bhigh
cells in vivo
]. M-CSF increased the population of cells expressing CD11b and c-Fms; the effect was further enhanced by TNFα. M-CSF is also known to potently stimulate proliferation with the daughter cell population having a particularly high percentage of CD11bhigh
cells; this effect is also enhanced by TNFα [6
]. This synergistic enhancement of proliferation and differentiation by M-CSF and TNFα may be one mechanism responsible for the presence of more mature OCP in TNFα-Tg mice [6
]. In the more mature OCP cultures obtained from either TNFα-Tg mice or following pre-treatment with M-CSF, IL-6 failed to inhibit osteoclastogenesis. These findings show that IL-6 has differential effects on osteoclastogenesis, dependent on the stage of maturation of the OCP, and the presence of other cytokines.
IL-6 also has anti-inflammatory properties, and has been shown to reduce TNFα expression in numerous models. LPS stimulates production of TNFα in mice, which in turn induces IL-6 secretion. IL-6 then acts in a negative feedback loop to decreases further production of TNFα [20
]. Similarly, IL-6 deficient mice have a three-fold increase in TNFα production in response to LPS compared to wild-type mice [47
]. Expression of both TNFα and IL-1 was increased in IL-6-/-
mice with systemic viral infection [23
]. Administration of recombinant IL-6 reduced inflammatory responses, including TNFα and IL-1, and mortality in mice treated with LPS [20
]. Similar effects were observed in mice with exposure to aerosolized endotoxin. IL-6 production was induced by the endotoxin, and higher concentrations of TNFα and MIP-2 were noted in mice deficient in IL-6 [49
TNFα is a key mediator of inflammatory bone resorption. TNFα increases inflammation, induces pre-osteoclast differentiation, expands the osteoclast precursor pool, and makes OCP more sensitive to the effects of RANKL [8
]. TNFα also stimulates RANKL expression by stromal cells [10
]. Titanium-induced TNFα production by ANA-1 macrophages was markedly reduced in cultures treated with IL-6. In contrast, addition of anti-IL-6 antibody resulted in increased TNFα production. Spleen cell cultures from WT and IL-6-/-
mice demonstrated enhanced TNFα secretion by titanium treated IL-6 deficient splenocytes; this enhancement was decreased by addition of exogenous IL-6. Thus, TNFα inhibition may be one of the targets through which IL-6 inhibits osteolysis in vivo
The observed effect of IL-6 deficiency in vivo
is to increase inflammatory bone resorption. Further experiments suggest this occurs both through the suppression of early maturation of osteoclasts by IL-6, as well as IL-6 decreasing production of TNFα by macrophages. IFN-γ and IL-4 inhibit osteoclast formation through the STAT signaling pathway [50
]. IL-6 signals through STAT1 and STAT3, suggesting that perhaps IL-6 shares similar signaling pathways with other inhibitors of OCP differentiation [51
]. Recent studies implicate the MAPK ERK1/2 pathway in the inhibition of osteoclastogenesis by IL-6 [39
]. Further, there may be an interaction between the particles themselves and other pathways, including SOCS3, an inhibitor of IL-6 signaling [52
]. Elucidation of the complex, stage-specific downstream signals that control the differential responses to IL-6 is an important area for future investigation.
These experiments show that IL-6 has diverse effects on osteoclast progenitor cells at various stages of maturation. IL-6 inhibits the differentiation of early osteoclast progenitors (CD11blow, c-Fms-) treated with M-CSF, RANKL, and TNFα. In OCP at an intermediate stage of differentiation (CD11bhigh, c-Fms+), IL-6 neither inhibits nor enhances osteoclastogenesis.