Article Text

Extended report
Neutralisation of Dkk-1 protects from systemic bone loss during inflammation and reduces sclerostin expression
  1. Gisela Ruiz Heiland1,
  2. Karin Zwerina1,
  3. Wolfgang Baum1,
  4. Trayana Kireva1,
  5. Jörg H Distler1,
  6. Mario Grisanti2,
  7. Frank Asuncion2,
  8. Xiadong Li2,
  9. Michael Ominsky2,
  10. William Richards2,
  11. Georg Schett1,
  12. Jochen Zwerina1
  1. 1Department of Internal Medicine 3 and Institute for Clinical Immunology, University of Erlangen-Nuremberg, Erlangen, Germany
  2. 2Amgen Incorporated, Thousand Oaks, California, USA
  1. Correspondence to Jochen Zwerina, Department of Medicine 3 and Institute for Clinical Immunology, University of Erlangen-Nuremberg, Krankenhausstrasse 12, D-91054 Erlangen, Germany; jochen.zwerina{at}


Introduction Inflammation is a major risk factor for systemic bone loss. Proinflammatory cytokines like tumour necrosis factor (TNF) affect bone homeostasis and induce bone loss. It was hypothesised that impaired bone formation is a key component in inflammatory bone loss and that Dkk-1, a Wnt antagonist, is a strong inhibitor of osteoblast-mediated bone formation.

Methods TNF transgenic (hTNFtg) mice were treated with neutralising antibodies against TNF, Dkk-1 or a combination of both agents. Systemic bone architecture was analysed by bone histomorphometry. The expression of β-catenin, osteoprotegerin and osteocalcin was analysed. In vitro, primary osteoblasts were stimulated with TNF and analysed for their metabolic activity and expression of Dkk-1 and sclerostin. Sclerostin expression and osteocyte death upon Dkk-1 blockade were analysed in vivo.

Results Neutralisation of Dkk-1 completely protected hTNFtg mice from inflammatory bone loss by preventing TNF-mediated impaired osteoblast function and enhanced osteoclast activity. These findings were accompanied by enhanced skeletal expression of β-catenin, osteocalcin and osteoprotegerin. In vitro, TNF rapidly increased Dkk-1 expression in primary osteoblasts and effectively blocked osteoblast differentiation. Moreover, blockade of Dkk-1 not only rescued impaired osteoblastogenesis but also neutralised TNF-mediated sclerostin expression in fully differentiated osteoblasts in vitro and in vivo.

Conclusions These findings indicate that low bone formation and expression of Dkk-1 trigger inflammatory bone loss. Dkk-1 blocks osteoblast differentiation, induces sclerostin expression and leads to osteocyte death. Inhibition of Dkk-1 may thus be considered as a potent strategy to protect bone from inflammatory damage.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Inflammation is a major contributor to bone loss. Rheumatoid arthritis (RA) is a typical example of such a disease often associated with osteoporosis and destruction of juxta-articular bone.1,,4 Inflammation per se is thus considered as an independent risk factor for bone loss, which is most likely mediated by the influence of mediators of inflammation on bone metabolism. Indeed, even tiny increases in inflammatory activity in humans enhance the risk for fracture.5

The pathophysiology of generalised bone loss in inflammatory conditions has not been completely elucidated but it is hypothesised that inflammatory mediators shift the balance between bone resorption and bone formation in favour of bone loss.6 One important mechanism of inflammatory bone loss is activation of osteoclastogenesis. Enhanced osteoclast activation may account for the increased bone resorption in patients with RA. Predominantly periarticular bone in RA is exposed to numerous osteoclasts, which appear locally within the inflamed synovial membrane.7 Many proinflammatory mediators expressed by synovial inflammatory tissue, in particular tumour necrosis factor α (TNFα) but also others such as interleukin 1 (IL-1) and IL-17, induce osteoclast activation by increasing the expression of osteoclast differentiation factors such as receptor activator for nuclear factor κβ ligand.8 9 This mechanism accounts for increased local bone resorption of articular bone in inflammatory arthritis. In addition, key balance mechanisms of bone remodelling appear to be blunted at a local level as osteoblasts are scarcely found along periarticular bone erosions in arthritis, which suggests that impaired bone formation contributes to increased bone damage.10 11 TNFα, a major proinflammatory cytokine involved in the arthritic process, has been shown to directly suppress bone formation as it can induce the expression of molecules which suppress osteoblast differentiation such as the Wnt antagonists Dkk-1 and sclerostin (SOST).12 13

Whether impaired bone formation is indeed responsible for systemic bone loss in the context of chronic inflammation is not well established. Skeletal sites distant from inflammatory tissue are not exposed to strongly enhanced osteoclastogenesis, which is found in the vicinity of joints and is based on a substantial influx of mononuclear cells to inflammatory sites serving as osteoclast precursors. Biomarkers of bone metabolism in chronic inflammatory disease do not yield an entirely consistent picture and increased bone resorption appears not to be the only mechanism leading to inflammatory bone loss. In fact, some studies show that markers of bone formation are rather low in patients with inflammatory disease and increase upon introduction of anti-inflammatory therapies such as TNF blockade.14 15 A substantial proportion of inflammatory bone loss might therefore result from blunted bone formation.

We tested this hypothesis and analysed whether blockade of a central negative regulator of bone formation, Dkk-1, can prevent inflammatory bone loss. In our analyses we used an animal model of chronic inflammatory arthritis, which is based on the overexpression of the proinflammatory cytokine TNF and is characterised by chronic inflammatory arthritis and generalised osteopenia. Hypothesising that low bone formation is a major precipitating factor in inflammatory bone loss and owing to the fact that the Wnt signalling pathway is instrumental in bone formation, we followed the strategy to foster bone anabolic signals by interfering with Dkk-1, a key natural inhibitor of the Wnt signalling pathway.

Material and methods

Details of the methods are included in the online supplement.

Animals and treatment

Heterozygous human TNF transgenic mice (hTNFtg, strain tg197; genetic background C57BL/6) used in the present study have been described previously.16 Six-week-old hTNFtg mice were divided into six groups (n=6/group) and treated for 4 weeks: group 1 received phosphate buffered saline (vehicle) as a control, group 2 received an anti-TNF antibody (Infliximab, Centocor, Leiden, The Netherlands), group 3 received a rat Dkk-1 antibody (10 mg/kg, Amgen, Thousand Oaks, California, USA), group 4 received anti-Dkk-1 (30 mg/kg), group 5 received anti-TNF (10 mg/kg) plus anti-Dkk-1 (10 mg/kg) and group 6 received anti-TNF (10 mg/kg) plus anti-Dkk-1 (30 mg/kg). All antibodies were administered by intraperitoneal injection three times per week. Animal procedures were approved by the local ethics committee of the University of Erlangen-Nurnberg.

Histological analysis and histomorphometry

Tibial bones were embedded undecalcified in methylmetacrylate (Technovit; Heraeus Kulzer, Wehrheim, Germany) and decalcified in paraffin for analysis. Bone histomorphometry was performed using a microscope (Nikon, Japan) equipped with a video camera and digital analysis system (OsteoMeasure; OsteoMetrics, Decatur, California, USA) and parameters were measured according to international standards.17 In vivo bone formation was assessed by dynamic histomorphometry: 0.3 mg Calcein (Sigma-Aldrich, St Louis, Missouri, USA) was injected subcutaneously 9 and 2 days before the mice were sacrificed. The mineral apposition rate (MAR; µm/day) was determined in undecalcified sections.


Paraffin-embedded sections were incubated with PH6 citrate buffer for antigen retrieval. Samples were incubated overnight with a rabbit anti-β-catenin antibody (Santa Cruz, California, USA) or goat anti-SOST antibody (R&D Systems, Minneapolis, Minnesota, USA).

Quantitative reverse transcription-PCR

RNA was isolated from left femurs and cultured primary osteoblasts using Trizol (Invitrogen, Carlsbad, California, USA). Quantitative reverse transcription-PCR was performed. The expression of the target molecule was normalised to the expression of β-actin. Primers for the following genes were used: osteocalcin, osteoprotegerin, Dkk-1 and SOST.

Cell culture

Primary osteoblasts were isolated from the calvariae of 4–6-day-old wild-type mice by collagenase digestion according to established protocols.18 For functional analysis, osteoblasts were stimulated with recombinant murine TNF (R&D Systems), Dkk-1 (R&D Systems), anti-Dkk-1 antibody (Amgen; Thousand Oaks, California, USA) or with a combination of these agents.

ELISA analysis

Supernatants were analysed using a commercial ELISA kit for Dkk-1 (R&D Systems) and an ELISA for SOST established in our laboratory using antimurine SOST (AF1589; R&D Systems) as capture antibody and then with antimurine SOST (BAF1589; R&D Systems) as detector antibody. The plates were developed with tetramethylbenzydin substrate (R&D Systems). The absorbance was read at 450 nm using an ELISA plate reader (Molecular Devices, Spectramax 190).

Statistical analysis

Data are presented as mean±SEM. For group comparison, one-way factorial analysis of variance was used with the Dunnett test or the Mann–Whitney test. A p value <0.05 was considered significant.


Blockade of Dkk-1 protects against TNF-induced inflammatory osteopenia and preserves trabecular bone network by increasing bone formation

We performed histomorphometric analysis of trabecular bone in order to characterise the microarchitectural changes upon treatment with anti-Dkk-1 antibody. Representative photographs of animals treated with anti-Dkk-1 or with a combination of anti-Dkk-1 plus anti-TNF in comparison with wild-type or untreated hTNFtg mice are shown in figure 1A. Bone mass (bone volume/total volume: BV/TV) was significantly reduced in hTNFtg mice compared with wild-type littermates (figure 1B). Treatment with anti-Dkk-1 antibody significantly increased trabecular bone mass compared with vehicle treatment (+53% and +44% with 10 and 30 mg/kg anti-Dkk-1 antibody, respectively; both p<0.05 vs hTNFtg untreated mice), reaching the level of wild-type mice. The combination of anti-Dkk-1 and anti-TNF antibody was also effective, showing a tendency to a more pronounced increase in bone mass (+64% and +66% with 10 and 30 mg/kg anti-Dkk-1 antibody, respectively; both p<0.05 vs untreated hTNFtg mice). These effects were based on a higher number of bony trabeculae as well as an increase in trabecular thickness upon blockade of Dkk-1. Thus, trabecular number was significantly reduced to 2.1±0.5 in hTNFtg mice compared with wild-type mice (3.4±0.8, p<0.05 vs untreated hTNFtg mice) but increased upon treatment with anti-Dkk-1 antibody to 2.4±0.7 at a dose of 10 mg/kg and 3.3±0.7 at a dose of 30 mg/kg (both p<0.05 vs untreated hTNFtg mice). Trabecular thickness was significantly reduced to 26.7±8.4 µm in hTNFtg mice compared with 42.6±7.5 µm in wild-type mice (p<0.05), but increased upon treatment with anti-Dkk-1 antibody to 39.2±5.5 µm at a dose of 10 mg/kg and 37.7±3.6 µm at a dose of 30 mg/kg (both p<0.05 vs untreated hTNFtg mice; figure 1C, D). Dynamic histomorphometry showed a low MAR in hTNFtg mice (0.91±0.20 µm/day) compared with wild type mice (1.69±0.37 µm/day; p<0.05), which increased by 20% in mice treated with anti-Dkk-1 antibody at a dose of 10 mg/kg (1.44±0.22 µm/day; p<0.05 vs hTNFtg), reaching wild-type levels at a dose of 30 mg/kg (1.61±0.32 µm/day; p<0.05; figure 1E).

Figure 1

Anti-Dkk-1 treatment rescues tumour necrosis factor (TNF)-mediated osteopenia: bone histomorphometry analysis. (A) Von Kossa staining of tibial bones from wild-type mice and hTNFtg mice either left untreated or treated with anti-Dkk-1 (30 mg/kg) or anti-TNF+anti-Dkk-1 (30 mg/kg) (magnification 40×). Tibial metaphyses of wild-type mice, hTNFtg and mice treated with anti-TNF (10 mg/kg), anti-Dkk-1 (10 and 30 mg/kg) or a combination of both agents were analysed for (B) bone volume per total volume (BV/TV), (C) trabecular number (Tb. N) and (D) trabecular thickness (Tb.Th.) (N=6/group). (E) Dynamic histomorphometry was used to calculate the mineral apposition rate (MAR, µm/day). Data are mean±SEM; *p<0.05.

Rescue from inflammatory bone loss upon blockade of Dkk-1 is based on increased bone formation and blunted bone resorption

To further define the mechanism of how blockade of Dkk-1 prevents TNF-induced bone loss, we analysed osteoblast and osteoclast numbers in trabecular bone of wild-type mice and hTNFtg mice treated with vehicle, anti-TNF antibody or anti-Dkk-1 antibody. Osteoblast-covered bone surface was significantly reduced to 5.69±3.18% in hTNFtg mice compared with wild-type mice (15.5±5.8%; p<0.05 vs untreated hTNFtg mice) and increased upon treatment with anti-Dkk-1 antibody to 9.9±1.9% at a dose of 10 mg/kg and 12.5±3.6% at a dose of 30 mg/kg (both p<0.05 vs untreated hTNFtg mice; figure 2A). On the other hand, osteoclast-covered bone surface was significantly increased to 36.9±8.5% in hTNFtg mice compared with wild-type mice (13.1±4.2%; p<0.05 vs hTNFtg untreated mice) and decreased upon treatment with anti-Dkk-1 antibody to 19.3±4.7% at a dose of 10 mg/kg and 12.9±8.7% at a dose of 30 mg/kg (both p<0.05 vs untreated hTNFtg mice; figure 2B). A reduction in osteoclast numbers in hTNFtg mice treated with anti-Dkk-1 antibodies was additionally confirmed by staining for tartrate resistant acid phosphatase and representative microphotographs are shown in figure 2C.

Figure 2

Anti-Dkk-1 treatment rescues tumour necrosis factor (TNF)-mediated osteopenia: osteoblast and osteoclast assessment. Tibial metaphyses of wild-type mice, hTNFtg and mice treated with anti-TNF (10 mg/kg), anti-Dkk-1 (10 and 30 mg/kg) or a combination of both agents were analysed for (A) osteoblast surface per bone surface (Ob.S/BS) and (B) osteoclast surface per bone surface (Oc.S/BS) (N=6/group). (C) Representative tartrate resistant acid phosphatase stainings of tibial bones of wild-type or hTNFtg mice either left untreated or treated with anti-Dkk-1 (30 mg/kg) or anti-TNF+anti-Dkk-1 (30 mg/kg). Data are mean±SEM; *p<0.05.

Enhanced expression of bone formation molecules upon blockade of Dkk-1

To further substantiate the finding of increased bone formation and decreased bone resorption in hTNFtg mice treated with an anti-Dkk-1 antibody, we analysed the expression of key proteins involved in bone remodelling which are linked to the Wnt pathway. Protein expression of β-catenin was analysed by immunohistochemistry. Whereas hTNFtg mice showed virtually no expression of β-catenin at trabecular sites, expression was strongly increased upon treatment with anti-Dkk-1 antibody (figure 3A). In addition, significant increases in mRNA expression of osteoprotegerin as well as osteocalcin were observed upon blockade of Dkk-1, indicating suppression of osteoclast activity as well as enhanced osteoblast-mediated bone formation (figure 3B,C). These data indicate that blockade of Dkk-1 reverses the negative effects of TNF on bone by increasing Wnt activity, enhancing bone formation and blocking bone resorption.

Figure 3

Dkk-1 blockade restores osteoblast function in tumour necrosis factor (TNF)-mediated osteopenia. (A) Tibial bones of wild-type mice, untreated hTNFtg and mice treated with anti-TNF (10 mg/kg), anti-Dkk-1 (10 and 30 mg/kg) or a combination of both agents were stained for β-catenin by immunohistochemistry. Positive cells show nuclear staining in red. Representative photographs from wild-type mice and hTNFtg mice either left untreated or treated with anti-Dkk-1 (30 mg/kg) or anti-TNF+anti-Dkk-1 (30 mg/kg) are shown (magnification 200× and 800×). RNA was extracted from femoral bones and relative mRNA expression of (B) osteoprotegerin (OPG) and (C) osteocalcin was determined by quantitative reverse transcription-PCR. Data are mean±SEM; *p<0.05 vs untreated transgenic animals (N=6/group).

TNF enhances Dkk-1 expression of osteoblasts and its blockade rescues impaired osteoblast function upon TNF challenge

When osteoblasts cultured for 21 days in differentiation medium were challenged with TNF, mRNA and protein expression of Dkk-1 significantly increased within 24 h (figure 4A,B). Time kinetic analysis of mRNA expression of Dkk-1 and SOST at 0, 1, 6 and 24 h after TNF exposure showed that induction of Dkk-1 starts after 6 h whereas induction of SOST was slower, starting 24 h after stimulation with TNF (figure 4C). Moreover, blockade of Dkk-1 prevented TNF-mediated downregulation of alkaline phosphatase activity in osteoblasts (figure 4D), suggesting that Dkk-1 indeed plays a critical role in the suppression of osteoblast function induced by TNF.

Figure 4

Tumour necrosis factor (TNF) upregulates Dkk-1 and induces sclerostin in differentiated osteoblasts. (A,B) Primary differentiated osteoblasts were treated with TNF (10 ng/ml), anti-Dkk-1 (1 µg/ml) or a combination of both agents. Expression of Dkk-1 mRNA (A) and protein (B) after 24 h was measured. Data are mean±SEM; *p<0.05 vs control group. (C) Primary osteoblasts were stimulated with TNF (10 ng/ml) for 0, 1, 6 and 24 h and kinetic expression of Dkk-1 and sclerostin mRNA was measured. (D) Primary differentiated osteoblasts were treated with TNF (10 ng/ml), anti-Dkk-1 (1 µg/ml) or a combination of both agents for 0, 6 and 24 h and alkaline phosphatase (AP) enzymatic activity (expressed in nM/µg/min) was determined.

Blockade of Dkk-1 reduces SOST expression of osteocytes in vivo and has a protective effect on osteocytes

Taking into consideration previous data12 and our own observation that TNF indeed leads to upregulation of SOST expression in differentiated primary osteoblasts, we further analysed the interactions between TNF, Dkk-1 and SOST. A significant increase in mRNA as well as protein expression of SOST was observed after challenge of primary osteoblasts with TNF (figure 5A,B). Moreover, blockade of Dkk-1 completely neutralised TNF-induced upregulation of SOST, suggesting that Dkk-1 itself may be involved in induction of SOST expression. On the other hand, addition of recombinant Dkk-1 to osteoblast cell culture dramatically increased SOST levels, and this effect could be prevented by the addition of an anti-Dkk-1 antibody (see figure 1 in online supplement).

Figure 5

Tumour necrosis factor (TNF)-induced sclerostin expression is dependent on Dkk-1 expression. (A,B) Primary osteoblasts were stimulated with TNF (10 ng/ml), anti-Dkk-1 (1 µg/ml) or a combination of both agents And sclerostin mRNA and protein expression were determined after 24 h. Data are mean±SEM; *p<0.05 vs control group. (C) Immunohistochemical osteocyte-related sclerostin expression was determined in tibial bones of wild-type mice, hTNFtg and mice treated with anti-TNF (10 mg/kg), anti-Dkk-1 (30 mg/kg) or a combination of both agents. The number of sclerostin-positive osteocytes is given as positive cells per high power field (HPF). (D) Osteocyte death was determined by counting empty lacunae in the same sections (N=6/group). Data are mean±SEM; *p<0.05. (E) Representative immunostainings for sclerostin in bone sections of wild-type, untreated hTNFtg and anti-TNF-treated and anti-Dkk-1-treated mice (magnification 1000×).

Since Dkk-1 affects SOST expression, we next analysed the expression of SOST in the bones of wild-type mice and hTNFtg mice by immunohistochemical staining in vivo. Compared with wild-type mice, the expression of SOST, which was limited to the osteocyte compartment, was increased in hTNFtg mice (figure 5C). In addition, numerous empty lacunae, indicating osteocyte death, were observed in hTNFtg mice (figure 5D). Upon neutralisation of Dkk-1 the expression of SOST was significantly reduced (p<0.0047). These observations suggest that Dkk-1 regulates SOST expression and its blockade may have a protective effect on mature osteocytes (figure 5E).


In the present study we have shown that Dkk-1 and SOST, both Wnt antagonists, are induced by TNF and block osteoblast-mediated bone formation. We observed that neutralisation of Dkk-1 by specific antibodies protects against inflammatory bone loss and neutralises the detrimental effects of TNF on the skeleton. This protective effect of Dkk-1 blockade was based on increased osteoblast activity and bone formation as well as a reduction in osteoclast differentiation and osteocyte death. Dkk-1 blockade is particularly effective in protecting against inflammatory bone loss since expression of the target molecule Dkk-1 is induced by TNF.

Canonical Wnt signalling is pivotal for the regulation of bone mass, which was first discovered by assessing gain or loss of function mutations of low density lipoprotein-related protein (LRP)-5, which are associated with high and low bone mass, respectively.19 20 Wnt proteins bind to plasma membrane receptors of the Frizzled family and their co-receptors LRP5 and LRP6, inducing the stabilisation of intracellular β-catenin which leads to stimulation of target genes involved in bone formation.21 This process is regulated by endogenous inhibitors of the Wnt signalling pathway such as Dkk-1 and SOST, which bind LRP5 ligand and thereby prevent its activation by Wnt proteins.22 23 Dkk-1 is considered as a potent downregulator of bone formation.24 Whereas Dkk-1 null mice are embryonically lethal, heterozygous Dkk-1+/− mice show increased bone formation and a high bone mass phenotype.25 26 In addition, osteoblasts derived from these mice show enhanced activity and proliferative potential. Increased expression of Dkk-1 is observed in bone resorptive conditions such as RA, multiple myeloma and thalassaemia-induced osteoporosis.12 27 28

Blockade of TNF, however, may not suffice to foster the Wnt-mediated bone formation and to overcome bone loss at the systemic level, which is supported by the fact that bone mass does not increase after TNF blockade and β-catenin expression is not restored. The best strategy to overcome inflammatory bone loss still needs to be identified. Inflammatory cytokines are considered to be key inducers of osteopenia in the context of inflammation, and cytokines such as TNF indeed induce a shift of bone turnover from bone formation towards bone resorption.12 29 Low bone formation during inflammation may arise from impaired Wnt signalling, which could also explain enhanced bone resorption since active Wnt signalling is linked to the expression of osteoprotegerin, a potent inhibitor of osteoclastogenesis and bone resorption.30 Cytokines such as TNF increase the expression of inhibitors of the Wnt signalling pathway such as Dkk-1 and SOST and could thus precipitate a detrimental combination of low bone formation and high bone resorption.13 15 26

However, there appears to be a cross-talk between TNF, Dkk-1 and SOST, another key Wnt antagonist, which is exclusively expressed in late stage osteoblast differentiation and confined to the osteocyte compartment in vivo.22 Induction of SOST by TNF and a related molecule, TWEAK, has recently been reported.13 In our experiments we found that TNF induces SOST in mature osteoblasts and that this process is primarily mediated by Dkk-1. Blockade of Dkk-1 therefore completely neutralised induction of SOST expression by TNF. These data point to a tight interplay between TNF, Dkk-1 and SOST and suggest that TNF induces SOST, which is among the most potent inhibitors of bone formation, via Dkk-1. In vivo, the consequences of neutralisation of Dkk-1 in hTNFtg mice were very similar: SOST expression in osteocytes of bones of hTNFtg mice decreased upon blockade of Dkk-1. In addition, the number of empty lacunae marking osteocyte death31 32 significantly decreased upon blockade of Dkk-1, which indicates that Dkk-1 is responsible for SOST expression in vivo and that Dkk-1 may act as a death signal for osteocytes.

In summary, this study shows that Dkk-1 blockade protects against inflammatory bone loss. These data point to a major contribution of low bone formation and impaired Wnt signalling activity in the pathogenesis of inflammatory bone loss. TNF-mediated downregulation of bone formation is associated with upregulation of potent Wnt signalling inhibitors such as Dkk-1 and SOST. In this way, TNF induces SOST expression in vitro and in vivo by activating Dkk-1 and leads to enhanced osteocyte death. Dkk-1 blockade may thus be considered as an interesting therapeutic target to protect against inflammatory bone loss, to reset pathological upregulation of SOST and to promote osteocyte survival.


The authors thank Dr George Kollias (Alexander Fleming Biomedical Sciences Research Center, Var, Greece) for providing hTNFtg mice and Barbara Happich, Isabell Schmidt and Cornelia Stoll for their technical assistance.


Supplementary materials


  • Funding This work was supported by the Interdisziplinäres Zentrum für Klinische Forschung Erlangen Project A34 (to GS and JZ), Deutsche Forschungsgemeinschaft Grant FOR 661 (to GS) and SFB 423 (to JZ and GS) and the European Union projects Masterswitch, Adipoa and Kinacept. This study was also supported by the Focus Programm SPP 1468 ‘Osteoimmunology - IMMUNOBONE – a program to unravel the mutual interactions between the immune system and bone’ of the Deutsche Forschungsgesellschaft.

  • Competing interests GS and JZ received speaker's fees from Amgen Incorporated. MG, FA, XL, MO and WR are employees of Amgen Incorporated.

  • Provenance and peer review Not commissioned; externally peer reviewed.