IL-33 Shifts the Balance from Osteoclast to Alternatively Activated Macrophage Differentiation and Protects from TNF-α–Mediated Bone Loss

Interleukin-33 is a recently described member of the IL-1 family, which includes IL-1α, IL-1β, and IL-18 (1, 2). IL-33 is constitutively expressed in various tissues, particularly in endothelial cells and epithelial cells exposed to the environment, such as skin, gastrointestinal tract, and the lungs (3). Similar to IL-1β, IL-33 may act as both a cytokine and an NF (3–14). As a cytokine, IL-33 signals through its interaction with a heterodimeric receptor consisting of membrane-bound ST2L (member of the IL-1R family) and IL-1R accessory protein, leading to NF-κB and MAPK activation (1, 12, 15). ST2L is expressed on monocytes, macrophages, neutrophils, T cells, particularly Th2 (but not Th1 cells), and mast cells (2).

IL-33 is involved in the polarization of IL-5–producing T cells (12), migration of Th2 cells (16), and activation of basophils (17), mast cells (18), eosinophils (19), and alternatively activated macrophages (AAM) (20), contributing to allergic response and asthma. IL-33 also promotes chemoattraction of neutrophils to inflammatory sites, attenuates polymicrobial sepsis (21), and mediates mast cell-dependent arthritis (22). However, the role of IL-33 in bone metabolism is unclear.

Bone turnover is orchestrated by bone-producing and bone-resorbing cells, namely osteoblasts and osteoclasts, respectively. Osteoclasts are hematopoietic cells derived from the monocyte lineage and require specific signals, in particular receptor activator for NF-κB ligand (RANKL) and M-CSF, for their differentiation (23). Osteoclasts are of key importance in physiologic and pathologic bone remodeling for the adaptation of bone to individual demands but also for initiating pathologic bone changes such as osteoporosis and bone erosion. In vivo, osteoclast activity is regulated by cytokines, which can either stimulate osteoclastogenesis (such as TNF, IL-11, and IL-17) or inhibit osteoclast differentiation (such as IL-4, IL-12, IFN-γ, and GM-CSF) (24). IL-1 family members are prominent in regulating osteoclasts and bone resorption. IL-1 is a major enhancer of bone resorption by inducing the expression of RANKL and RANK (25). Moreover, IL-1 is a key mediator of TNF-driven bone loss (26). In contrast, IL-18 blocks bone resorption by inducing the expression of GM-CSF, which is a differentiation factor for dendritic cells (DC) rather than osteoclasts (27, 28). While this manuscript was in preparation, a physiological role of IL-33 in bone remodeling has been described by Schulze et al. (29). However, the exact mechanism of how IL-33 inhibits osteoclastogenesis and affects bone-resorptive disease has not been shown.

We report in this study that IL-33 profoundly inhibits osteoclast differentiation in vivo and in vitro. In transgenic mice overexpressing human TNF-α, which spontaneously develop local and systemic bone loss due to enhanced differentiation of osteoclasts, IL-33 markedly reduced osteoclast number and bone resorption. IL-33 appears to do so by inducing antiosteoclastogenic cytokines GM-CSF, IL-4, and IFN-γ and skewing the induction of osteoclast precursors toward AAM and DC. Furthermore, IL-33 also inhibits osteoclast formation from human bone marrow cells. Our data therefore suggest that IL-33 may be a novel option for preventing TNF-α–mediated bone resorption.

The human TNF-α transgenic (hTNFtg) mice (strain Tg197) have been described previously (30). hTNFtg mice were maintained in a specific pathogen-free facility. ST2^−/− mice (deficient for membrane-bound ST2L and soluble ST2) were originally provided by Dr. Andrew McKenzie (Laboratory of Molecular Biology, Medical Research Council, Cambridge, U.K.) (13). IL-4R–deficient mice (IL-4R^−/−) were a gift from Dr. A. Gessner (Institute of Clinical Microbiology, Immunology and Hygiene, University of Erlangen, Erlangen, Germany). All animal experiments were performed in accordance with the guidelines of the local experimental regulatory authorities.

For treatment with IL-33, mice (n = 12) received 2 μg rIL-33 (Biolegend) or vehicle (PBS) i.p. two times per week between weeks 6 and 10 of age. In addition, another group of mice (n = 12) were injected i.p. with an agonistic anti–IL-33R Ab (3E10) (31) or control Ig at a dose of 170 μg three times per week.

Recipient hTNFtg mice aged 6 wk were irradiated at 11 Gy using orthovoltage irradiation (Stabilipan; Siemens), at 250 kV/15 mA/40 cm focus and surface distance, at a dose rate of 1.15 Gy/min. Mice were anesthetized by inhalation anesthesia (Forene; Abbott) performed during the irradiation process in a closed fixture made of Plexiglas. This fixture was mounted on a Plexiglas block (d = 50 mm) to achieve full reflection scattering. The next day, mice were reconstituted by i.v. injection of 5 × 10⁶ ST2^−/− or ST2^+/+ bone marrow cells in medium 199 (Sigma-Aldrich) containing 5 ml 1 M HEPES buffer (Life Technologies), 5 ml (1 mg/ml) DNAse (Sigma-Aldrich), and 40 μl (50 mg/ml) gentamicin (Sigma-Aldrich). Transplanted mice were analyzed 4 wk after the transplantation.

Murine bone marrow was isolated from mice by flushing femoral bones with complete anti-DMEM. Human femoral bone marrow samples were obtained with informed consent from patients with osteoarthritis during total hip replacement surgery. Osteoclast precursors were then isolated from bone marrow-derived cell suspensions using CD11b microbeads (Miltenyi Biotec) according to the manufacturer’s instructions. The purity of isolated precursors was assessed by flow cytometry analysis using CD11b-FITC–labeled Abs (Miltenyi Biotec). CD11b⁺ monocytes were plated in 96-well plates (2.5 × 10⁵/ well) or 48-well plates (5 × 10⁵/well) in Alpha MEM supplemented with 10% FCS, 30 ng/ml M-CSF, and 1–50 ng/ml RANKL (all from R&D Systems). TNF-α (10 ng/ml) was added to some cultures. To examine the contribution of IL-33 to osteoclast development, IL-33 (10, 50, and 100 ng/ml) was added to the culture at the same time as M-CSF and RANKL or 24 or 72 h later. In some cultures, 3E10 (20 μg/ml) or 3E10 cross-linked with IgG or IgG alone (10 ng/ml) instead of IL-33 were added. Medium was changed after 72 h. To examine the contribution of GM-CSF to IL-33–regulated osteoclast development, anti-mouse GM-CSF (1 or 5 μg/ml; eBioscience) or isotype control (5 μg/ml) were added together with IL-33. Osteoclast differentiation was evaluated by staining cells for tartrate-resistant acid phosphatase (TRAP) using a Leukocyte Acid Phosphatase Kit (Sigma-Aldrich) according to the manufacturer’s instructions.

For extracellular staining, 1 × 10⁶ cells/staining were washed with 1 ml PBS containing 0.1% BSA resuspended in 100 μl FACS buffer and incubated with saturating amounts of PE-, FITC-, PerCP-, PE-Cy7–, or allophycocyanin-labeled Abs against F4/80, CD11b, CD11c, CD14, CD206, TLR2, and Ly6G (all BD Biosciences) for 30 min at 4°C in the dark. For ST2 staining, 3E10 Ab was used. The cells were also stained with isotype control Ab and analyzed by FACSCanto Flow Cytometer (BD Biosciences).

RNA was purified using TRIzol (Invitrogen) or the RNease kit (Qiagen). cDNA was synthesized using a high-capacity cDNA transcription kit (Applied Biosystems). Quantitative PCR (Q-PCR) was carried out as described previously (32). Relative quantification was performed by calculating the difference in cross-threshold values (ΔCt) of the gene of interest and a housekeeping gene according to the equation 2^−ΔCt. In some experiments, the relative expression values were normalized to the expression values in the control condition. The following specific probe and primers were used: Hs00166156_m1* (hcathepin-K), Mm00484039_m1* (mcathepin K) (Applied Biosystems), or as provided previously (33).

Cultured cells were washed twice with PBS and subsequently lysed. Protein extracts were separated on 12% SDS-polyacrylamide gel, transferred on a nitrocellulose membrane, and stained with the following Abs: anti–phospho-p44/42 MAPK (Erk1/2) (Thr²⁰²/Tyr²⁰⁴), anti–phospho-SAPK/JNK (Thr183/Tyr¹⁸⁵), and anti–phospho-NF-κB p65 (Ser⁵³⁶) (Cell Signaling Technology). An Ab against β-actin (Sigma-Aldrich) was used as loading control.

Cytokines in the serum and cell culture supernatants were assessed by the mouse FlowCytomix Kit (Bender MedSystems) according to the manufacturer’s instructions using an FACSCalibur System (BD Biosciences). RANKL and OPG (both R&D Systems) serum levels were measured by ELISA according to the manufacturer’s instructions.

Immunohistochemistry was performed on sections from decalcified paraffin-embedded paws and knees of wild-type (WT) and hTNFtg mice using goat anti-mouse IL-33 Ab (5 μg/ml; AF3626; R&D Systems). Briefly, endogenous peroxidase activity was blocked in the samples by incubating in 0.5% H₂O₂/methanol solution. After blocking nonspecific binding sites with rabbit serum, the samples were incubated overnight at 4°C with the primary Ab and the appropriate isotype control. The samples were then incubated with a biotinylated rabbit anti-goat Ab (BA1000; Vector Laboratories) followed by the addition of Avidin/Biotin complex (Vectastain Elite ABC Kit, PK6100; Vector Laboratories). Immunoreactive sites were visualized by the addition of 3,3′-diaminobenzidine chromagen substrate (Impact DAB, SK-4105; Vector Laboratories) counterstained with hematoxylin, dehydrated, and mounted in DPX.

Histomorphometry was performed on methacrylate-embedded undecalcified plastic sections following von Kossa and Goldner staining. Quantifications were performed by digital image analysis (OsteoMeasure; OsteoMetrics). Histological analysis was performed on formalin-fixed, decalcified, paraffin-embedded tissue sections stained with H&E, TRAP, and toluidine blue. Synovial inflammation, osteoclast numbers, and cartilage destruction were quantified by digital image analysis (OsteoMeasure; OsteoMetrics).

The tibia samples of hTNFtg mice treated with IL-33 or 3E10 were measured with a commercially available desktop microcomputed tomography (μCT) (μCT35; SCANCO Medical AG, Brüttisellen, Switzerland). The following acquisition parameters were used: voltage: 40 kV; x-ray current: 250 μA; exposure time: 5000 ms/projection, 720 projections; matrix: 1024 × 1024; and voxel size in reconstructed image: 9 μm. Images were analyzed using a plug-in programmed for Amira 4.1.2. (Mercury) with the following histomorphometric parameters: bone volume/total volume, trabecular thickness, and trabecular number.

Arthritis (paw swelling and grip strength) was assessed by a semiquantitative score as described previously (34).

All statistical analyses were performed with Mann–Whitney U test, one-way, or two-way ANOVA followed by Tukey’s test and are represented as means ± SEM unless otherwise stated using GraphPad Prism 4.0 Software (GraphPad).

We first sought evidence of the presence of IL-33 in the bone tissues by immunohistochemistry on paraffin-embedded knee and paw sections from C57BL/6 and hTNFtg mice. The overexpression of TNF leads to destructive arthritis and severe systemic osteopenia and thus provides an ideal model to study dysregulated bone turnover (30). IL-33–positive cells were detected in the bone marrow and chondrocytes in normal C57BL/6 mice (Fig. 1A, 1B). A similar pattern was observed in the bone of hTNFtg mice with an additional strong expression of IL-33 in the synovium invading bone (Fig. 1C, 1D). These data indicate that IL-33 is present in bone and thus may participate in regulation of bone metabolism in homeostatic and pathogenic condition.

To address the role of IL-33 in bone metabolism such as bone loss and cartilage destruction, we treated hTNFtg mice with IL-33 over a period of 4 wk. Clinical symptoms manifested as a progressive paw swelling and a loss of grip strength were not significantly affected by treatment with IL-33 (Supplemental Fig. 1A, 1B). We next performed histological analysis of the tarsal joints of IL-33–treated hTNFtg mice to determine whether IL-33 modulated local bone destruction and cartilage breakdown. Significant protection of the joint architecture was observed in hTNFtg mice treated with IL-33 with less bone erosion and decreased osteoclast numbers (Fig. 2A, 2B). In contrast, inflammation was only marginally but not significantly reduced (data not shown). This is consistent with the clinical analysis, which shows no significant difference of joint swelling in these mice (Supplemental Fig. 1A, 1B). Q-PCR analysis of bones from IL-33–treated hTNFtg mice shows a significant reduction in the osteoclast marker genes TRAP and NFATc1 (Fig. 2C). Cartilage destruction was also significantly ameliorated in hTNFtg mice treated with IL-33 (Fig. 2D, 2E).

To confirm this finding, we use an ST2L agonistic Ab (3E10) (31). The agonistic effect of 3E10 on osteoclast precursors was shown by intracellular activation of the MAPK JNK and ERK as well as NF-kBp65, similar to that of IL-33 stimulation (Supplemental Fig. 2). hTNFtg mice were injected i.p. three times per week with 170 μg 3E10. Again, inflammatory signs of arthritis were not significantly changed after treatment with 3E10 Ab (Supplemental Fig. 1C, 1D). In contrast, histological analysis showed that bone erosion and osteoclast counts were markedly decreased in hTNFtg mice treated with 3E10 compared with controls (Fig. 2F, 2G). This effect was accompanied by a reduction in cartilage destruction (Fig. 2H, 2I). Together, these data indicate that IL-33/ST2L signaling inhibits TNF-α–driven osteoclastogenesis and prevent local bone destruction.

To investigate the contribution of the endogenous IL-33/ST2 pathway to bone metabolism and to identify the cellular mechanism responsible for IL-33–mediated bone protection, we reconstituted irradiated hTNFtg mice with bone marrow from ST2^−/− or WT (ST2^+/+) mice. Clinical parameters such as paw swelling and grip strength did not significantly differ between the ST2^−/− and ST2^+/+ chimeras (Supplemental Fig. 1E, 1F). In contrast, histological analysis of the tarsal joints clearly shows that the bone erosion and osteoclast numbers were significantly increased in hTNFtg/ST2^−/− chimeras compared with the chimeras with a functional IL-33R (hTNFtg/ST2^+/+) (Fig. 3A, 3B). Furthermore, cartilage destruction was enhanced in the hTNFtg/ST2^−/− chimeras compared with the hTNFtg/ST2^+/+ mice (Fig. 3C, 3D). These data therefore demonstrate that bone marrow cells expressing ST2L mediate an IL-33–induced bone-protective effect and underline an important role of the endogenous IL-33/ST2 pathway in bone metabolism.

To further analyze the role of the IL-33 in bone metabolism, we assessed the microarchitecture of the tibial bones distal from inflamed joints. μCT analysis and three-dimensional reconstitution of the trabecular bone architecture demonstrated a significantly increased systemic bone mass in IL-33– and 3E10 Ab-treated mice compared with controls (Fig. 4A). Quantitative analysis showed a significant increase in trabecular thickness in the IL-33– or 3E10-treated mice (Fig. 4B). Histomorphometry on TRAP-stained sections of the same tibial bones shows that both IL-33 and 3E10 Ab treatment resulted in a significant decrease in the number of osteoclast compared with untreated control (Fig. 4C). Thus, the IL-33/ST2 signaling pathway markedly attenuates TNF-α–induced bone loss by decreasing osteoclast differentiation in vivo. Importantly, the analysis of systemic bone mass in hTNFtg/ST2^−/− bone marrow chimera mice also showed a significantly decreased systemic bone mass compared with the hTNFtg/ST2^+/+ bone marrow chimeras (Fig. 4D). Together, these data demonstrate that IL-33 has a key protective effect on systemic bone mass.

We next analyzed the distribution of immune cells and the mediators that are known to regulate bone turnover. As expected (1), IL-33 treatment resulted in splenomegaly in hTNFtg mice compared with that of untreated control mice (data not shown). FACS analysis shows that IL-33 treatment led to an increased accumulation of ST2L⁺ cells, CD14⁺ (monocytes) CD11c⁺ (DC), and Ly-6G⁺ (neutrophils) cells in the spleen (Fig. 5A). There was also a modest increase in F4/80⁺ (macrophages) and CD11b⁺ cells. There was also an increase in the number of CD206⁺ macrophages (marker for the AAM) in the spleen of the IL-33–treated mice compared with untreated control (data not shown). The number of T and B cells remained unchanged (data not shown). IL-33 treatment in hTNFtg mice also resulted in an elevated level of serum IL-4 and IFN-γ (Fig. 5B). This was confirmed by the treatment of the mice with 3E10, which also significantly increased the level of serum GM-CSF (Fig. 5C). In contrast, there was no significant difference in the level of serum RANKL or OPG in the hTNFtg mice following treatment with IL-33 or 3E10 (Fig. 5D, 5E). Together, these data indicate that the bone-protective effect of IL-33 is not likely mediated through a change of RANKL/OPG balance, but is associated with an increase in the production of antiosteoclastogenic cytokines (IL-4, IFN-γ, and GM-CSF), which shift the mononuclear cell differentiation away from osteoclast lineage and toward the AAM and DC pathway.

To further investigate the mechanism of IL-33–induced bone-protective effect, we tested if IL-33 could directly interact with osteoclast precursors in vitro. First, we examined the expression of IL-33R on differentiating osteoclasts. Bone marrow mononuclear CD11b⁺ cells from C57BL/6 mice were cultured under osteoclastic conditions in medium supplemented with M-CSF and RANKL for up to 5 d. FACS analysis showed that ST2L is expressed on osteoclast precursor cells (up to 10% of CD11b⁺ cells) on day 1 of culture and was further upregulated during the osteoclast differentiation, reaching 70% of cells by day 3 (Fig. 6A).

To investigate the effect of IL-33 on osteoclast development, purified bone marrow CD11b⁺ cells were cultured as above in the presence or absence of IL-33. IL-33 added 24 h after the start of culture strongly inhibited osteoclast development (assessed by the number of TRAP⁺ multinuclear cells) even in the presence of high concentrations of RANKL (Fig. 6B). Confirming this observation, 3E10 Ab also showed a direct suppressive effect on osteoclast differentiation in vitro (Fig. 6C). Moreover, cross-linking 3E10 with IgG led to further reduction of the number of osteoclasts (Fig. 6C). In addition, we examined the effect of IL-33 on osteoclastogenesis in the presence of TNF-α. Again, IL-33 strongly inhibited osteoclast development from bone marrow precursors (Supplemental Fig. 3).

To confirm the specificity of the IL-33 effect, osteoclast precursors from BALB/c and ST2^−/− (BALB/c background) were cultured as described above. Again, IL-33 strongly inhibited WT osteoclast development but did not affect the development of ST2^−/− osteoclasts (Fig. 6D, 6E). Furthermore, Q-PCR analyses show a marked reduction of cathepsin K expression (a gene essential for functional osteoclasts) in ST2^+/+ but not in ST2^−/− cells after 3 d culture with RANKL in the presence of IL-33 (Fig. 6F).

Characterization of the phenotype of cells cultured under osteoclastic conditions with IL-33 shows a significant increase in the percent of CD206⁺F4/80⁺ macrophages, indicating a shift toward the AAM phenotype (Fig. 6G). This shift was ST2 dependent, as it was not evident in ST2^−/− cells (Fig. 6H). Because IL-33–treated mice showed an increase in serum levels of IL-4 and GM-CSF (both known AAM differentiation and antiosteoclastogenic factors) (Fig. 5), we investigated whether these cytokines are directly associated with the IL-33–mediated inhibition of osteoclast differentiation. IL-33 triggered high levels of GM-CSF and, to a lesser extent, IL-4 from CD11b⁺⁺ cells under osteoclast-polarizing conditions (Fig. 6I, 6J). To directly investigate the role of IL-4 and GM-CSF in IL-33–mediated inhibition of osteoclast development, we cultured CD11b⁺ cells from IL-4Rα^−/− or WT mice under osteoclast-polarizing conditions in the presence of IL-33. Cells from the IL-4Rα^−/− mice produced a modest but significantly increased number of osteoclasts compared with that of the WT cells (Fig. 6K). Next, we cultured WT osteoclast precursors with GM-CSF–neutralizing Abs. The GM-CSF–neutralizing Abs markedly reversed the inhibitory effect of IL-33 on osteoclast development (Fig. 6L). Together, these data demonstrate that IL-33 directly inhibits osteoclast differentiation by shifting the differentiation of the precursors toward AAM via GM-CSF and, to a lesser extent, IL-4.

We next investigated the effect of IL-33 on human osteoclast differentiation. Similar to the mouse experiments, CD11b⁺ mononuclear cells expressing markers for monocytes (31% CD14⁺), DC (40% CD11c⁺), and osteoclast precursors (98% RANK⁺) (Supplemental Fig. 4) were purified from human bone marrow and cultured with M-CSF and RANKL in the presence or absence of recombinant human IL-33. IL-33 significantly reduced total and mature human osteoclast number (Fig. 7A, 7B). This reduction was correlated with a decrease in Cathepsin K mRNA expression (Fig. 7C). In agreement with recently published studies (35, 36), isolated CD14⁺ peripheral blood monocytes (instead of bone marrow mononuclear CD11b⁺ cells) were much less responsive to the suppressive effect of IL-33 (Fig. 7D), suggesting that in humans, IL-33 inhibits the differentiation of human osteoclasts by preferentially interfering with early precursors in bone marrow compartment.

Data reported in this study show that IL-33 is expressed in the bone tissue and acts as a bone-protective cytokine by effectively blocking osteoclastogenesis and local bone erosions in hTNFtg mice. Exogenously administered IL-33 blocks TNF-α–mediated local and systemic bone loss in vivo. Furthermore, an IL-33R agonist Ab had similar effects, leading to reduction of osteoclast numbers and protection against bone erosion. Conversely, genetic deletion of the IL-33R in bone marrow cells enhanced bone loss in the hTNFtg mice. These data therefore support an important endogenous role of the IL-33/ST2 signaling pathway in regulating osteoclast formation and providing protection against TNF-α–induced skeletal damage.

The mechanism by which IL-33 exerts its inhibitory effect on osteoclast differentiation is not likely by the regulation of OPG or RANKL synthesis (23). Instead, in vitro studies show that IL-33 directly acts on human and mouse bone marrow CD11b⁺ cells by inhibiting their development toward mature osteoclasts. IL-33 appears to shift the osteoclast precursor differentiation toward AAM. IL-33–activated AAM produce elevated levels of IL-4 and GM-CSF, which are known inhibitors of osteoclast differentiation (24). Our finding therefore places IL-33 upstream of IL-4 and GM-CSF in the osteoclast inhibition pathway.

Interestingly, IL-33 was not able to affect osteoclast development when added to committed mouse immature osteoclast (day 3 of RANKL/M-CSF–stimulated culture) (data not shown), suggesting that IL-33 acts on the very early step of cell commitment. This phenomenon may also explain the lack of an inhibitory effect of IL-33 on osteoclast development from human peripheral blood CD14⁺ cells as observed by us and by others (35, 36). This finding is in agreement with an increasing number of studies suggesting the existence of various subsets of human and murine circulating monocytes that differ in their cytokine profile, migratory properties, surface markers, and function. For example, ∼5% of the human peripheral blood CD14⁺ that express CD16 are believed to be proinflammatory (37–39). Thus, it is likely that IL-33 is not able to influence the function of circulating monocytes that are already committed to specific lineages. Unlike Mun et al. (33), we were not able to enhance osteoclastogenesis in human peripheral blood monocytes using IL-33 in the presence of M-CSF and RANKL. The reason for this discrepancy is unclear and may be due to difference in culture conditions. In our hands, osteoclasts were not generated in the absence of RANKL.

We and others have shown previously that IL-33 and ST2L are present in rheumatoid arthritis (RA) synovium (22, 29, 40, 41), and increased levels of IL-33 were also detected in sera and synovial fluids from RA patients (42, 43). Given the proinflammatory role of IL-33 in collagen-induced (22) and Ab-induced (44) arthritis in mice, the protective role of IL-33 in skeletal damage observed in this study is counterintuitive. It should, however, be noted that the proinflammatory effect of IL-33 in the collagen-induced and Ab-induced arthritis models was mediated principally by IL-33–induced mast cell degranulation. In the current hTNFtg model, IL-33 had little or no effect on the TNF-induced inflammation. Instead, IL-33 markedly reduced local and systemic bone loss. The present finding is consistent with the dual pathophysiological role of IL-33 (2). Depending on the cytokine environment or location (synovium or bone marrow), IL-33 could have a proinflammatory or bone-protective role via distinct molecular pathways. Our finding also highlights the different mechanisms that are involved in inflammation and bone resorption.

Bone erosion as a consequence of TNF-α activity closely mimics the skeletal pathology observed in human inflammatory arthritis, particularly RA, which is one of the most severe forms of bone loss in humans (24). To this end, it is important to note that IL-33 also inhibits osteoclast differentiation from human bone marrow. Thus, IL-33 may represent a potential therapeutic option against bone loss.

We thank Prof. George Kollias (Alexander Fleming Research Institute, Vari, Greece) for providing the hTNFtg mice, Dr. Andrew McKenzie for providing the ST2^−/− mice, Dr. A. Gessner for providing the IL-4R^−/− mice, and Lynn Crawford for technical support.

The authors have no financial conflicts of interest.