How Does the Immune System Break and Protect Bone?

Hiroshi Takayanagi*

Immune and skeletal systems have various regulatory molecules in common, including cytokines, which constitute the intricate signaling network that maintains the homeostasis of the two systems. Furthermore, immune cells form in the bone marrow, where they interact with bone cells. Therefore, one can imagine that the physiology and pathology of one system may affect the other. However, the molecular and cellular mechanisms of interaction of the two systems are poorly understood. I attributed the bone loss in autoimmune arthritis to the defective control of bone metabolism by the immune system, and revealed the molecular mechanism of T cell-mediated regulation of osteoclast formation (1). In addition, I found that IFN-β, which plays a pivotal role in the immune system, also functions in the regulation of bone remodeling (2). Thus, the efforts to understand the molecular interactions between bone and immune systems proved to be critically important. Here I would like to introduce a series of works, which are related to a new discipline termed ±osteoimmunology± (3).

Why do abnormal immune responses break bone? This question arose during the course of my study on bone loss in autoimmune arthritis. In the field of arthritis, many researchers have focused on the mechanism of autoimmunity. I began to analyze pathological sections of the degraded bone and became convinced that the main effector cells responsible for bone destruction are bone-resorbing osteoclasts. I hypothesized that these osteoclasts are differentiated from synovial cells, and I successfully established a culture system of human rheumatoid synovial cells that generates osteoclasts in vitro (4). However, the idea of osteoclastogenesis from synovial cells was regarded as heresy at the time, because it was believed that osteoclast formation occurs only in the bone. After an essential osteoclast differentiation factor, RANKL (receptor activator of NF-κB ligand), was cloned (5, 6), I and other groups confirmed that RANKL is overexpressed in synovium (7, 8). Thus, it is now widely accepted that RANKL-induced osteoclast formation in the synovium plays a critical role in arthritic bone destruction (6). I also found that RANKL-expressing synovial fibroblasts are responsible for osteoclastogenesis in my synovial culture system (4, 7). Therefore, I believed that synovial fibroblasts were the major source of RANKL, a concept which may be called ±synovial cell theory.±

In contrast, Kong et al. proposed that RANKL-expressing T cells were the primary inducers of osteoclastogenesis in arthritis (8). This ±T cell theory± was based on the fact that formaldehyde-fixed T cells could induce osteoclastogenesis. Which is more important, synovial cells or T cells? This issue has long been controversial regarding the pathogenesis of chronic synovitis in rheumatoid arthritis (9), but it seems also to be a question in the mechanism of arthritic bone destruction. Since abnormal bone resorption is not observed during normal T cell responses, I suspected that T cells have a negative regulatory mechanism to counterbalance the action of RANKL.

Using mice lacking a receptor component for IFN-γ, I revealed that T cell production of IFN-γ strongly suppresses osteoclastogenesis by interfering with the RANKL/RANK signaling pathway (1). I have shed light on a new biological function of IFN-γ, which is to protect against calcified tissue destruction upon T cell activation, demonstrating that activated T cells not only positively regulate but also negatively affect osteoclastogenesis. Thus the effect of T cells on osteoclastogenesis may depend on the balance between the positive action of RANKL and the negative action of IFN-γ. Furthermore, IFN-γ inhibition of osteoclastogenesis is rescued by overexpressing RANK adaptor protein TRAF6 in precursor cells, indicating that TRAF6 is the critical target for IFN-γ action.

Taken together, the mechanism of arthritic bone destruction initiated by T cells can be summarized as shown in the figure. In addition to RANKL on T cells, RANKL is abundantly expressed on synovial fibroblasts stimulated with inflammatory cytokines such as IL-1 or TNF-α. Importantly, despite a significant T cell infiltration in the arthritic joints, IFN-γ expression is known to be suppressed (9). Thus, the paucity of IFN-γ and the enhanced expression of RANKL may contribute to the activation of osteoclastogenesis in arthritis. In this context, ±synovial cell theory± and ±T cell theory± may not be mutually exclusive, but RANKL on both cell types synergistically contributes to the bone destruction.

Figure 1 Mechanism of arthritic bone destruction: the critical role of RANKL on synovial fibroblasts and T cells. Activated T cells stimulate the macrophages to secrete proinflammatory cytokines such as TNF-α and IL-1, which strongly induce RANKL on synovial fibroblasts. In addition, T cells themselves also express RANKL. In constrast, there is a very low level of IFN-γ, a potent RANKL inhibitor produced by T cells. This imbalance may be responsible for the aberrant activation of osteoclast formation in arthritis.

These results led me to examine whether osteoclast-targeted therapy is effective in the treatment of autoimmune arthritis. I constructed an adenovirus vector carrying the csk gene, which negatively regulates c-Src, an essential tyrosine kinase for osteoclast function. This gene transfer not only inhibited the osteoclastic bone resorption in vitro, but also suppressed bone destruction in rat adjuvant arthritis (10). Furthermore, it was reported that OPG, an inhibitory receptor for RANKL, also suppressed bone loss in the arthritis model (8). These findings in toto indicate that osteoclasts can be a good therapeutic target in arthritic bone loss.

Exploring the regulatory mechanism of RANKL-induced osteoclastogenesis, I found that RANKL induces IFN-β, a critical cytokine for antiviral defense. Mice deficient in IFN-β signaling exhibited severe osteopenia accompanied by enhanced osteoclastogenesis, suggesting that IFN-β is essential for normal bone remodeling by suppressing excessive osteoclast differentiation (2). IFN-? inhibited the differentiation by interfering with the RANKL-induced expression of c-Fos, an essential transcription factor for osteoclastogenesis. The mechanism of IFN-β gene induction by RANKL was distinct from that by virus, and was dependent on c-Fos itself. This study placed the IFN-β system in a novel context (11), unveiling further the intimate relationship between bone and immune systems.

Genetically engineered mice have provided evidence that each gene contributes to complex regulatory mechanisms, some of which are not easily expected from the original context of the gene discovery. We should be free from the conventional categorization of disciplines and make full use of genomic information to form new paradigms of experimental biology. Osteoimmunology is one of the good examples of such interdisciplinary research, and will become more important for understanding the normal and abnormal biological processes involving both systems.


    • H. Takayanagi et al., Nature408, 600 (2000).
    • H. Takayanagi et al., Nature416, 744 (2002).
    • J. R. Arron, Y. Choi, Nature408, 535 (2000).
    • H. Takayanagi et al., Biochem. Biophys. Res. Commun.240, 279 (1997).
    • T. Suda et al., Endocrine Rev.20, 345 (1999).
    • L. E. Theill, W. J. Boyle, J. M. Penninger, Annu. Rev. Immunol.20, 795 (2002).
    • H. Takayanagi et al., Arthritis Rheum.43, 259 (2000).
    • Y. Y. Kong et al., Nature402, 304 (1999).
    • G. S. Firestein, N. J. Zvaifler, Arthritis Rheum33, 768 (1990).
    • H. Takayanagi et al., J. Clin. Invest.104, 137 (1999).
    • T. Alliston, R. Derynck, Nature416, 686 (2002).
      *The author is at the Graduate School of Medicine, University of Tokyo, Japan. E-mail: taka{at}