Introduction

Historically, bone was thought to be a stable and metabolically inactive organ with the sole function of supporting the locomotive activity of the body. However, bone marrow has long been known, alongside the thymus, as a primary lymphoid organ.1,2 Thus, the immune function of bone has been well recognized, but detailed studies of how bone tissue contributes to immune responses were nevertheless neglected for a long time. The difficulty of accessing hard bone tissue—now less of a problem with modern techniques—hampered molecular analysis and probably contributed to this delayed characterization.The past decade has, however, witnessed surprisingly rapid progress in osteoimmunology, a discipline that examines the interactions and shared mechanisms between the skeletal and immune systems.1,3

Research into the mechanisms of bone destruction in autoimmune arthritis has inevitably led investigators to study interactions between cells of the immune and skeletal systems. Advances in genetic engineering have revealed unexpected skeletal phenotypes in mice deficient in immune molecules, such as osteoporotic and osteopetrotic phenotypes (Table 1).1 Furthermore, studies of the intracellular signal transduction pathways triggered by bone-regulating cytokines such as receptor activator of nuclear factor κB ligand (RANKL; also known as TNF ligand superfamily member 11) have indicated that the immune and bone systems share a huge range of regulatory mechanisms in terms of cellular signalling pathways.1,4

Table 1 Skeletal phenotypes caused by deficiencies in immunomodulatory molecules in mice
Full size table

Increasing evidence suggests that not only does the immune system influence bone, but also that bone reciprocally regulates the immune system. Molecular mechanisms of bone-mediated regulation of the immune system are being elucidated, particularly in relation to how osteoblasts, osteoprogenitor cells and other nearby cells comprise a haematopoietic stem cell (HSC) niche that contributes to immune regulation.5,6 Indeed, in the bone marrow microenvironment, bone and immune cells are localized side by side and any resident cell can be influenced by surrounding cells of either bone or immune lineage.5,6 Thus, functions of the immune and skeletal systems should now be reconsidered in the context of the osteoimmune system as a whole.

In 2009, when I reviewed osteoimmunology in the context of rheumatic disease in this journal,7 research had focused on how the immune system, via osteoclastogenic T cells, mediates bone destruction. Just 3 years later, understanding of osteoimmunology has broadened to encompass the reciprocal effect of bone cells on cells of the immune system. In this Perspectives article, I summarize and update how the immune system is linked with bone remodelling, focusing on T cells and RANKL in the context of rheumatoid arthritis (RA), and describe advances in the understanding of the osteoblastic HSC niche.

Immune cell regulation of bone

Physiological bone remodelling seems to occur independently of signalling by immune cells, but the ability of the immune system to influence bone metabolism in pathological conditions is being increasingly recognized. Bone destruction in RA, which involves aberrant activation of osteoclasts in the absence of equivalent levels of osteoblast activity, is the best understood example of skeletal consequences of autoimmune inflammation. RANKL is a typical osteoimmunological molecule in that it is clearly essential in both the skeletal and immune systems, and critically mediates the immune cell regulation of bone.1 Therefore, I discuss here the pathogenesis of bone destruction in RA by focusing on type 17 T helper (TH17) cells and RANKL; nevertheless, these mechanisms of immune cell-mediated regulation of bone might be applied to various pathological conditions, including spondyloarthritis, systemic lupus erythematosus, periodontitis and bone cancer.

Osteoclast activation in RA by RANKL

RA is an autoimmune disease that is characterized by inflammation of the synovial joint, leading to severe structural damage including bone destruction. Investigations into the sites of bone destruction in patients with RA and animal models of the disease have suggested that osteoclasts (bone-resorbing cells) have a substantial role in RA pathogenesis.7 The osteoclast differentiation factor, RANKL, is highly expressed in the RA synovium, and inflammation-mediated bone damage is largely attributable to abnormally high expression of RANKL.1,7

Cell sources of RANKL in RA

With the destructive role of RANKL established, the next question to arise was that of how autoimmune reactions induce RANKL expression and subsequent osteoclastogenesis. RANKL expression can be detected in both synovial fibroblasts and T cells in the inflamed joints of patients with RA.7 As I discussed in 2009,7 T-cell derived RANKL was initially proposed to be the main contributor to enhanced osteoclastogenesis in RA, but T cells also produce anti-osteoclastogenic cytokines, such as interferon (IFN)-γ, which counterbalance the action of RANKL.8 Thus, it is not easy for T cells to exert a positive effect on osteoclastogenesis. In fact, if activated T cells are co-cultured with osteoclast precursor cells, the T cells actually inhibit osteoclastogenesis rather than increase it.8 By contrast, synovial fibroblasts have a potent ability to induce osteoclastogenesis in cell culture. Therefore, it seems likely that synovial cell RANKL is the major driver of osteoclastogenesis in the synovium, and that T cells, although certainly contributing, do so indirectly.1,7

Traditionally, T-helper (TH) cells were classed as either type 1 or type 2 (TH1 or TH2) on the basis of the cytokines they produced. Nevertheless, as TH1 and TH2 cells inhibit osteoclastogenesis through the production, respectively, of IFN-γ and IL-4, it was necessary to identify a different TH-cell subset, one with the ability to increase osteoclastogenesis.1 In the course of these efforts, TH17 cells were shown, exclusively among T-cell subsets, to have the capacity to induce osteoclastogenesis. Interestingly, even TH17 cells do not induce osteoclastogenesis through their own expression of RANKL, but instead do so indirectly, via the IL-17-mediated induction of RANKL expression on synovial fibroblasts. In addition, TH17 cells promote local inflammation, causing exaggerated expression of TNF, IL-6 and IL-1, all of which increase RANKL expression on synovial fibroblasts and activate osteoclast precursor cells (Figure 1).7

Figure 1: Mechanisms of bone destruction in autoimmune arthritis.
figure 1

A variety of cell populations, including T cells, B cells, innate immune cells, synovial fibroblasts and osteoclasts have roles in the development of autoimmune arthritis. TH17 cells contribute to the development of arthritis in the initiation and inflammatory phases of disease, through production of autoantibodies as well as activation of innate immunity. Importantly, synovial fibroblasts contribute to TH17-cell-mediated immunity in the inflammatory phase by promoting TH17-cell migration to the inflamed joint, followed by TH17-cell proliferation with an increase in IL-17 production. TH17 cells, exclusively among T-cell subsets, are osteoclastogenic. TH17 cells secrete relatively large amounts of IL-17, which induces expression of RANKL, but not IFN-γ, on synovial fibroblasts. IL-17 stimulates local inflammation resulting in the production of proinflammatory cytokines such as TNF, IL-1 and IL-6 by synovial macrophages. These cytokines further enhance RANKL expression on synovial fibroblasts and act on osteoclast precursor cells. RANKL expression on TH17 cells might partly contribute to the enhanced osteoclastogenesis induced by these cells. Thus, the interaction of infiltrating CD4+ T cells with the synovium in joints has an essential role in the pathogenesis of rheumatoid arthritis in terms of both inflammation and bone destruction. Abbreviations: CCL20, CC chemokine ligand 20; DC, dendritic cell; MMP, matrix metalloproteinase; RANKL, receptor activator of nuclear factor κB ligand; TH0, naive T cell; TH1, type 1 T helper (cell); TH2, type 2 T helper (cell); TH17, type 17 T helper (cell); TREG, regulatory T (cell).

Full size image

Although accumulating evidence indicates that TH1 cells are not an osteoclastogenic T-cell subset,7 targeted depletion of TH1 and TH17 cells expressing lymphotoxin-α was shown to suppress inflammation and bone destruction in mice with collagen-induced arthritis;9 thus, whereas TH17 cells seem to be much more important than the TH1 subset in the context of osteoclastogenesis in RA, the role of TH1 cells in autoimmune inflammation nevertheless warrants further investigation.

Targeting RANKL and TH17 cytokines

Considering the exclusive role of TH17 cells in RA osteoclastogenesis, targeting TH17-related cytokines, IL-6, IL-23, IL-17A, IL-21 and IL-22 might be an auspicious strategy against inflammation-associated bone loss. Targeting RANKL in arthritis seems to be a promising method to specifically prevent bone damage,10 although the anti-RANKL antibody denosumab has minimal effects on cartilage damage or joint-space narrowing (Box 1).10 T-cell-specific knockout of Rankl in mice has shown that T-cell derived Rankl does not contribute to physiological bone regulation,11 but its role in pathological settings such as autoimmune arthritis is an intriguing issue to address in the near future. Rankl conditional knockout mice will be a powerful tool in the detailed analysis of cellular sources of human RANKL.11,12 Potentially, targeting RANKL in bone pathology may have unexpected immune consequences, as I discuss in the section Immune functions of RANKL, later.

B-cell RANKL effects on bone

The therapeutic effect of anti-CD20 antibody has highlighted the role of B cells in the pathogenesis of arthritis.13 Data from a study in ovariectomized mice available have now suggested a direct role of RANKL expression by B cells in the regulation of bone metabolism.12 T-cell specific knockout of Rankl in the mice did not alter ovariectomy-induced bone loss, but mice lacking Rankl protein expression in their B cells were partially protected from the bone loss via attenuated osteoclastogenesis.12 The role of B cells in promoting pathogenic osteoclastogenesis should be analyzed in other models of bone pathology in future.

Molecular bone–immune crosstalk

Shared signalling pathways

Studies of the signalling molecules shared by the immune and skeletal systems began, as I have mentioned, with analysis of the functions of known immune molecules in the skeletal system. In this regard, much has been learned from skeletal phenotypes of mice deficient in various immunomodulatory molecules (Table 1), though equivalent deficiencies have—mostly—not yet been reported in people. Although less is known about functions in the immune system of signalling molecules released by bone cells, accumulating evidence of immune phenotypes in mice and humans deficient in bone-regulatory molecules (Table 2) indicates that reciprocal regulation of the immune system by the skeletal system occurs. RANKL exemplifies a molecule identified as a signalling factor in bone that also signals in the immune system.

Table 2 Immunological phenotypes caused by deficiencies in bone-regulatory molecules*
Full size table

Immune functions of RANKL

As discussed above, RANKL expressed by synovial fibroblasts drives bone destruction in RA via stimulation of osteoclastogenesis. It has been suggested that osteoblastic cells or bone marrow stromal cells express RANKL to regulate bone remodelling. In adult mice, however, osteocyte-specific knockout of Rankl has shown that osteocytes are the critical source of Rankl in physiological bone remodelling in mice.11,14 Thus, the cell-type specific function of RANKL in bone has begun to be understood. What about the role of RANKL in the immune system? RANKL was originally suggested to have a role in the activation of dendritic cells by T cells, but RANKL also has a role in the regulation of regulatory T (TREG) cells, as discussed next.7

In a mouse model of diabetes, administration of CD4+CD25+ TREG cells prevented destruction of β cells in the islets of Langerhan. This effect was mediated by preferential accumulation of TREG cells in pancreatic lymph nodes and islets, where they inhibited development of CD8+ T cells into cytotoxic T cells. Inhibition of the RANKL–RANK interaction prevented the accumulation of TREG cells in the pancreatic islets, resulting in destruction of β cells by cytotoxic T cells.15 Similarly, inhibition of RANKL affected the expansion of TREG cells in the intestine of colitic mice thus worsening inflammation in the gut16. Furthermore, Langerhan cells of K14-Rankl transgenic mice, in which Rankl is overexpressed in the skin, elevated numbers of TREG cells in the spleen.17 Peripheral expansion of TREG cell numbers was demonstrated to be dependent on Langerhan cells, with depletion of the latter substantially reducing the peripheral pool of TREG cells.17 In humans and mice with breast cancer, RANKL expressed by T cells in the tumour has been associated with metastatic activity.18 RANKL on T cells is also critical for the development of the thymic medullary epithelial cells which are responsible for the negative selection of self-reactive T cells.19,20,21 Therefore, RANKL can be both a stimulator and a suppressor of immune responses, depending on the context, and the detailed implications for rheumatic diseases are not yet known.

Implications of targeting RANKL

As I have mentioned, the anti-RANKL antibody denosumab is in clinical trials and/or use for various bone pathologies (Box 1). Given the extensive functions of RANKL outside of the skeletal system, careful investigation will be required to guard against unexpected adverse effects of denosumab. It is worth noting, however, that serious immunologic problems have not been encountered in clinical trials of the antibody to date.22

Immunomodulation by bone cells

The role of bone marrow

Bone marrow is a primary lymphoid organs, harbouring HSCs, memory B cells and other B-lineage cells.23 These haematopoietic cells share their microenvironment with bone cells including osteoblasts, osteoclasts and possibly even osteocytes. Therefore, it is not surprising that bone cells, traditionally thought to be involved only in bone metabolism, participate in regulation of the immune system. Principally, this regulation relates to control of haematopoiesis—and, therefore, of the cellular composition of the immune system—by regulation of the HSC niche, the bone-marrow microenvironment in which HSC fate is controlled, by bone cells. In this section I discuss evidence for the influence of osteoblasts, osteoclasts and other potential cellular components of the HSC niche on HSC function.

Osteoblasts and the HSC niche

Groundbreaking research has suggested that osteoblasts are components of the HSC niche.24,25 These studies were based on genetically modified mouse models in which osteoblast numbers were increased. In one study, Calvi et al.24 used mice expressing constitutively active parathyroid hormone/parathyroid hormone-related peptide receptor (PPR) in osteoblastic cells. The activation of PPRs in osteoblasts not only induced an increase in osteoblast number but also upregulated expression of the Notch ligand protein jagged-1 (Jag1) in osteoblasts.24 This study thus implicated Notch signalling in increasing HSC numbers. In contrast to these findings, however, others have reported that Jag1 and Notch signalling are dispensable for the maintenance of HSC populations.23 For example, Notch signalling is not necessary for HSC maintenance at steady state and under certain situations of myeloablative injury.23 Therefore, the role of Notch signalling in the maintenance of HSC remains controversial.

Another study to support the concept of an osteoblastic HSC niche reported that angiopoietin-1 expressed in osteoblasts regulates HSC maintenance.26 Most HSCs in adult bone marrow are in the G0 phase of the cell cycle (quiescence). A balance between HSC quiescence and activation is critical for sustaining production of all mature blood cells and a normal HSC pool size. Angiopoietin-1 produced by osteoblasts activates angiopoietin-1 receptor on HSCs and promotes tight adhesion of these cells to the niche, resulting in HSC quiescence.26 However, the physiological significance of these findings has been questioned in subsequent studies that have shown that HSC number does not necessarily correlate with osteoblast number.27,28

Other cells constituting the HSC niche

As mentioned above, osteoblasts seem to be involved in regulation of HSC pool size and can be considered to constitute an endosteal HSC niche. However, subsequent reports have suggested that additional cell types contribute to the HSC niche (Figure 2). The vascular niche, which is largely populated by endothelial cells and perivascular cells, contributes to haematopoiesis by expressing factors that promote this process. Kit ligand, also known as stem cell factor (SCF), is one such factor and is mainly produced by leptin receptor-expressing perivascular stromal cells.29 In addition, areas adjacent to blood vessels are thought to constitute an HSC niche.23 In 2010, CXC-chemokine ligand 12 (CXCL12)-abundant reticular (CAR) cells, which are localized in closer proximity to vessels than osteoblasts, were proposed to be the cells that formed the reticular niche. As CAR cells in mouse bone marrow have the capacity to differentiate into osteoblasts and adipocytes in vitro and in vivo, it is possible that CAR cells are also involved in the endosteal niche as osteoprogenitor cells.30 Further studies have implicated other cell types in regulating different stages of HSC maintenance (Figure 2). Nonmyelinating Schwann cells maintain HSC quiescence by regulating activation of latent TGF-β.31 Nestin+ mesencymal stem cells (MSCs) express HSC maintenance genes, such as those encoding CXCL12 and angiopoietin-1.32

Figure 2: HSC maintenance in the bone marrow.
figure 2

The bone marrow provides microenvironmental niches that support the self-renewal ability, multipotency and quiescence of HSCs. Three types of HSC niches, endosteal (containing osteoblasts), vascular (containing Lepr+ perivascular stromal cells) and reticular (containing CAR cells) have been reported in mice, which mainly regulate the HSC pool size, stem cell factor production and HSC retention, respectively.24,29,30 Nestin+ MSCs and nonmyelinating Schwann cells are also suggested to be involved in HSC maintenance. CAR cells and nestin+ MSCs have the capacity to differentiate into osteoblasts and adipocytes.30,32 Nonmyelinating Schwann cells maintain HSC quiescence by regulating activation of latent transforming growth factor β.31 The role of osteoclasts in the regulation of HSC mobilization remains controversial.33,34 Abbreviations: CAR, CXC chemokine ligand 12-abundant reticular (cell); HSC, haematopoietic stem cell; Lepr, leptin receptor; MSC, mesenchymal stem cell.

Full size image

Thus each of the many cell types described above interacts with HSCs in bone marrow. More evidence is required to clearly determine the relative contribution of bone cells to the regulation of haematopoiesis.

The role of osteoclasts

Less evidence has been obtained for the role of osteoclasts in the regulation of the immune system than for osteoblasts, but reports suggest that osteoclasts participate in the regulation of HSC mobilization by modifying the HSC niche,33,34 as well as having immunomodulatory roles in plasma cell maintenance35 and antigen presentation.36 Further studies are absolutely required for the detailed understanding of the immune regulation by bone cells.

Other topics

Bone marrow oedema, which may be at least partly attributable to TNF-mediated haematopoiesis, has emerged as an MRI-based biomarker of inflammatory arthritis.37 As alteration of the bone marrow microenvironment by TNF might be involved in an increase in the circulating osteoclast precursor number,38,39 this lesion is a novel example of inflammation-mediated regulation of osteoclastogenesis through ectopic haematopoiesis. Studies of bone marrow oedema might provide insights into the relationship between inflammation, osteoclastogenesis and haematopoiesis.

Future perspectives

The role of osteocytes

In addition to cells within bone marrow, bone matrix-embedded cells—osteocytes—seem to participate in the osteoimmune system. It has been demonstrated that targeted disruption of sclerostin results in a decreased number of B cells due to elevated B-cell apoptosis.40 As sclerostin is primarily expressed in osteocytes and is not expressed in any haematopoietic lineages, the B-cell defect in these mice was caused by the impaired production of sclerostin by osteocytes.40 These results suggest a novel role of osteocytes in the regulation of bone marrow environments that support immunological cells.

Osteoimmunological communication

All cells in bone tissue undergo intensive communication to maintain mechanical stability, calcium homeostasis and haematopoiesis in the skeletal system. Efforts are ongoing to identify molecules that mediate communication between cells in bone, and the involvement of several immunological mediators in this crosstalk has been shown. Osteoprotegerin (also known as TNF receptor superfamily member 11B), a decoy receptor for RANKL, and RANKL are among the first such molecules to be identified; they are expressed by osteoblasts and regulate osteoclastogenesis.1 Stimulation of bone formation after bone resorption is partly induced by classical coupling factors, such as insulin-like growth factor41 and TGF-β42 which, besides these skeletal roles, are also considered to be immunological molecules.

Much has been done to identify molecules other than these classic factors that link bone resorption to formation, although in vivo evidence remains scarce.41,42 Ephrin-B2 and ephrin B4, for example, are known to help mediate the transition of bone resorption to formation.43 Furthermore, since 2011 semaphorins have emerged as potent mediators of bone cell communication. Osteoclast semaphorin 4D, for example, maintains the bone resorption phase of bone remodelling by inhibiting osteoblastic bone formation in mice.44 As Sema4D is known to regulate the activation of B cells and dendritic cells and inhibits monocyte migration, this molecule can be considered as an osteoimmunological mediator. Semaphorin 3A, by contrast, not only inhibits bone resorption in mice but also enhances bone formation, suggesting that it could be a powerful bone-protecting factor.45

Conclusions

Osteoimmunology began with study of the mechanisms of bone pathology in arthritis, and these investigations have continued to drive advances in this field. RANKL is the most important and well studied molecule in the context of osteoimmunology and we will witness the clinical impact of RANKL inhibition in the next few years. Mechanistic insights obtained by research in osteoimmunology have already contributed to better understanding of the mode of action of antirheumatic drugs such as TNF blockers, and have provided molecular bases for new therapeutic strategies including antibodies against IL-17 and IL-23. The whole picture of regulation of immune cells by bone cells has not yet been uncovered, but may have important clinical implications in the future if immune responses can be modulated by pharmacological control of bone cell functions. In conclusion, detailed molecular understanding of the osteoimmune network will provide a novel framework for understanding both the immune and skeletal systems, as well as a molecular basis for developing new strategies against various diseases, which will surely benefit both clinical and basic research in broad disciplines.