Interleukin-17A upregulates receptor activator of NF-κB on osteoclast precursors

The pathologic outcome of the activated immune system interacting with the skeletal system is evidenced by articular bone erosion and joint loss of function seen in autoimmune joint diseases, such as rheumatoid arthritis (RA). Pathological bone resorption is the result of increased differentiation and/or activity of osteoclasts [1]. Osteoclast precursors are present in the circulating monocyte population [2], and these cells differentiate into multinucleated osteoclasts in the presence of macrophage-colony stimulating factor (M-CSF) and receptor activator for NF-κB ligand (RANKL) produced by osteoblasts [3]. Osteoprotegerin (OPG) is a soluble decoy receptor for RANKL that inhibits osteoclast formation and bone resorption [4]. Terminally differentiated osteoclasts polarize onto the bone surface by forming filamentous actin (F-actin) rings associated with α_vβ₃ integrin, acidify the enclosed extracellular space using a proton pump [5] to solubilize the inorganic calcium phosphate, and release organic matrix-degrading enzymes such as cathepsin K, matrix metalloproteinases and tartrate-resistant acid phosphatase (TRAP) that results in a continuous series of resorption lacunae on the bone. The receptor activator for NF-κB (RANK)/RANKL/OPG axis governs homeostatic bone remodelling, as mice deficient in RANK, RANKL, or OPG have severe bone phenotypes [3, 4, 6].

In addition to producing RANKL, activated T cells produce other proinflammatory factors, such as TNF, which stimulate osteoclastogenesis in RANK^-/- mice [6] and induce osteoclastogenesis in vitro by direct stimulation of murine bone-marrow-derived macrophages exposed to permissive levels of RANKL [7]. Haematopoietic precursors from RANKL^-/-, RANK^-/-, or TRAF6^-/- mice also become osteoclasts in vitro when they are stimulated with TNFα in the presence of cofactors such as transforming growth factor beta [8], and recently TNF was shown to induce osteoclastogenesis from spleen-cell-derived macrophages in the absence of RANKL [9]. Collectively these data demonstrate that immune-mediated osteoclastogenesis pathways can lead to joint pathology in inflammatory arthritis in both a RANKL-dependent and a RANKL-independent manner [10‐12].

The IL-23/IL-17A axis has recently been implicated in joint pathology and autoimmune diseases. The proinflammatory cytokine IL-23 promotes the differentiation of a novel memory T-cell subset (Th17) in mice characterized by the production of the signature cytokine IL-17A. IL-23 is comprised of the IL-23p19 subunit and the IL-12p40 subunit, (shared with IL-12), and IL-23p19^-/- mice have reduced IL-17A⁺-Th17 cells, are resistant to collagen-induced arthritis induction, and have no joint or bone pathology [13]. Th17 cells stimulate local inflammation and express RANKL as well as inducing RANKL expression in osteoblasts [14, 15].

Since both Th17 cells and osteoclast precursors are present in the peripheral blood of healthy adults and both are elevated in the synovial fluid of RA patients [2, 14, 16], we studied the role of IL-17A in the direct differentiation and activation of osteoclast precursors [17]. Specifically, we analysed the in vitro mechanisms by which IL-17A affects osteoclastogenesis, using mice deficient for IL-17A, and treating human peripheral blood monocytes with exogenous cytokines. We confirm that IL-17A does not play a role in physiological bone homeostasis, and show that IL-17A sensitizes osteoclast precursors to the key osteoclast factor RANKL by increasing RANK expression on osteoclast precursors [18]. These data provide a direct mechanism of IL-17A action on osteoclast precursors that is distinct from IL-17A's known action on osteoblasts [14], thereby providing an additional link between the immune and skeletal systems. Furthermore, the induction of RANK and RANKL expression make IL-17A a potent inducer of bone erosion under inflammatory conditions and its blockade may be used to combat disabling conditions such as RA.

IL-17A is the signature cytokine of the recently discovered Th17 memory T-cell subset. Th17 cells are the only T-cell subtype that express RANKL [15], which is the main osteoclast differentiation factor leading to the differentiation of osteoclast precursors to mature bone resorbing osteoclasts [3]. IL-17A-producing Th17 cells are present in the joints of RA patients [22, 23] and synovial membrane IL-17A gene expression was one factor that was predictive for subsequent bone erosion and joint damage [24]. IL-17A is present in RA synovial fluid [14], and RA synovial fluid macrophages can differentiate to fully functional osteoclasts [16]. In the collagen-induced arthritis model, IL-17A^-/- mice are protected from joint disease with less TRAP⁺ cells correlating with reduced bone resorption, suggesting that IL-17A plays a role in osteoclastogenesis [25, 26]. IL-17A can act independently of TNF under arthritic conditions [27] and stimulates cartilage destruction in the IL-1-deficient mice [28], supporting the hypothesis that IL-17A can act independently of IL-1 [29]. Moreover, the combination of IL-17A with IL-1 and TNF shows a marked increase in inflammation and bone destruction [30, 31]. Collectively the above experiments elegantly show that IL-17A acts independently of IL-1 and TNF but can also synergize with those factors.

The direct link of IL-17A to osteoclastogenesis, however, remained unknown. In this manuscript we focused on IL-17A's direct role in the regulation of osteoclastogenesis in inflammatory arthritis. To investigate how pathologic IL-17A expression could impact bone remodelling we studied the effects of exogenously added IL-17A on PBMCs and CD14⁺ cells. As mentioned previously IL-17A acts on stromal cells to induce the expression of RANKL. To eliminate this component of IL-17A biology we employed a culture system where there are no stromal cells but only CD14⁺ cells. Moreover, no RANKL was detected in the conditioned medium of CD14⁺ cells stimulated with 1, 10, 50 and 100 ng/ml IL-17A after 1 to 24 hours.

IL-17A stimulated the formation of multinucleated TRAP⁺ from PBMC cultures at concentrations as low as 1 ng/ml (Figure 1a). In order to study the direct effect of these cytokines on osteoclast precursors, we used the CD14⁺ fraction of PBMCs and confirmed that IL-17A induced TRAP⁺multinucleated cells. The use of TRAP⁺ cells as an osteoclast marker has been overrated in recent literature, especially since macrophage polykaryons, immature dendritic cells, and mononuclear macrophages all stain positive for TRAP [15, 16, 32‐34]. Osteoclast functional assays were therefore performed to validate the findings. IL-17A-treated CD14⁺ cells formed F-actin rings and induced low-grade resorption, but notably did not induce lacunar excavation as in the RANKL-treated cultures (Figure 1b, c).

Cell morphology was studied by scanning electron microscopy. Cultures treated with IL-17A contained numerous small cells (<30 μm) rounded or flattened and spread over the dentine surface, to which they were attached by fine microvilli. Some cells had cytoplasmic processes that extended up to 20 μm over the dentine surface. These cells had numerous surface ruffles and were able to perform low-grade resorption (Figure 1c). The area of low-grade resorption was not similar to the lacunar excavation observed in RANKL-treated cultures and was generally small, discrete, round or ovoid areas that did not possess a well-defined margin and did not coalesce to form large areas of lacunar excavation (Figure 1c).

Moreover, IL-17A-stimulated CD14⁺ cell fusion was distinct from the cell fusion observed in RANKL-treated cultures, which formed uniform multinucleated giant cells (Figure 2a). Cell fusion in IL-17A-treated cultures was incomplete, with cell membranes only partly coalesced together and individual cells being distinguishable (Figure 2a). The exact mechanism of IL-17A on cell fusion remains to be elucidated; however, it is noteworthy that IL-17A induces dendritic cell fusion [35]. IL-17A addition to monocyte-derived dendritic cells induced a semi-mature, mixed monocyte-macrophage-dendritic cell phenotype with cells expressing CD14, CD68, CD1a, MHC II and CCR6 and induced dendritic cell fusion. The dendritic cell fusion was characterized as less efficient and the number and size of nuclei was smaller (four to eight nuclei) when compared with the previously described M-CSF and RANKL fusion pathway [35]. We confirm the above data and show that cell fusion occurred in IL-17A-stimulated CD14⁺ cells cultured for 18 days, and also also show that cells contained more than 20 nuclei as evidenced by DAPI staining (Figure 1b) and that complexes of partially fused cells were as large as 60 μm (Figures 2a).

We initially reasoned that IL-17A's low-grade resorption was due to the upregulation of matrix metalloproteinase-9 [36]; however, no significant increase in matrix metalloproteinase-9 message was detected (data not shown). More importantly IL-17A synergized with RANKL-treated cultures to increase bone resorption by 30% (Figure 2b, c). IL-17A could not synergize with TNF to increase osteoclastogenesis and its synergy with RANKL was blocked by OPG, further confirming the fact that it is a RANKL-mediated effect (Figure 2c). The only osteoclast-related genes (out of a panel of 60 genes) that were significantly upregulated were TRAP (data not shown), c-fms and RANK (Figure 3a).

RANKL addition to IL-17A-treated CD14⁺cultures showed increased bone resorption that correlated with increased RANK, as shown by flow cytometry (Figure 3b). To measure the osteoclastogenic potential of IL-17A on the c-fms and RANK receptors, we performed a quantitative TRAP assay where cells were cultured under standard concentrations of MCSF (Figure 3c, left) or RANKL (Figure 3c, right) and the addition of increasing dose of RANKL or MCSF, respectively, was tested in the presence or absence of 1 ng/ml IL-17A. Addition of IL-17A sensitized the pre-osteoclasts to both MCSF and RANKL, leading to increased osteoclastogenesis (Figure 3c).

Bone marrow macrophages from Wt mice and IL-17A^-/- mice were not statistically different under homeostatic conditions and the osteoclasts that develop in in vitro cultures function normally in their ability to form multinucleated TRAP⁺ cells capable of actin ring formation and lacunar resorption (Figure 4). Moreover, there was no detectable modulation of physiologic bone remodelling in IL-17A-deficient mice; serum RANKL and OPG levels were similar to control mice, and no differences in cortical or trabecular bone mineral density were observed (Figure 4). These data are in agreement with previous studies where using 40 μm resolution dual-energy X-ray absorptiometry revealed no obvious abnormality in skeletal development and bone morphometric analyses, concluding that bone resorption and formation was normal in IL-17A^-/- mice [15].

IL-17A does not play a role in physiological bone remodelling as shown by our IL-17A^-/--deficient mice studies; however, IL-17A is secreted by Th17 cells under inflammatory conditions. We have shown that IL-17A upregulates RANK in osteoclast precursors, sensitizing them to RANKL-induced bone resorption. We therefore propose IL-17A as a suitable target to combat bone loss in inflammatory arthritis and autoimmune diseases such as RA.