Dr. Zheng is an employee of Biogen Idec. All other authors state that they have no conflicts of interest
Article first published online: 16 MAR 2009
Copyright © 2009 ASBMR
Journal of Bone and Mineral Research
Volume 24, Issue 8, pages 1434–1449, August 2009
How to Cite
Vincent, C., Findlay, D. M., Welldon, K. J., Wijenayaka, A. R., Zheng, T. S., Haynes, D. R., Fazzalari, N. L., Evdokiou, A. and Atkins, G. J. (2009), Pro-Inflammatory Cytokines TNF-Related Weak Inducer of Apoptosis (TWEAK) and TNFα Induce the Mitogen-Activated Protein Kinase (MAPK)-Dependent Expression of Sclerostin in Human Osteoblasts . J Bone Miner Res, 24: 1434–1449. doi: 10.1359/jbmr.090305
Published online on March 16, 2009
- Issue published online: 4 DEC 2009
- Article first published online: 16 MAR 2009
- Manuscript Accepted: 11 MAR 2009
- Manuscript Revised: 21 JAN 2009
- Manuscript Received: 3 JUL 2008
- TNF-related weak inducer of apoptosis;
We have recently shown that TNF-related weak inducer of apoptosis (TWEAK) is a mediator of inflammatory bone remodeling. The aim of this study was to investigate the role of TWEAK in modulating human osteoblast activity, and how TWEAK and TNFα might interact in this context. Recombinant TWEAK and TNF were both mitogenic for human primary osteoblasts (NHBC). TWEAK dose- and time-dependently regulated the expression of the osteoblast transcription factors RUNX2 and osterix. TWEAK inhibited in vitro mineralization and downregulated the expression of osteogenesis-associated genes. Significantly, TWEAK and TWEAK/TNF induced the expression of the osteoblast differentiation inhibitor and SOST gene product, sclerostin. Sclerostin induction was mitogen-activated protein kinase (MAPK) dependent. The SOST mRNA levels induced by TWEAK were equivalent to or exceeded those seen in steady-state human bone, and the TWEAK/TNF induction of SOST mRNA was recapitulated in fresh cancellous bone explants. TWEAK-induced sclerostin expression was observed in immature osteoblastic cells, both in cycling (Ki67+) primary NHBC and in the cell lines MC3T3-E1 and MG-63, as well as in human osteocyte-like cells and in the osteocyte cell line, MLO-Y4. Treatment of NHBC with recombinant human sclerostin mimicked the effects of TWEAK to suppress RUNX2 and osteocalcin (OCN). TWEAK, TNF, and sclerostin treatment of NHBC similarly altered levels of phosphorylated and total GSK3β and active and total levels of β-catenin, implying that the Wnt signaling pathway was affected by all three stimuli. Sclerostin also rapidly activated ERK-1/2 MAPK signaling, indicating the involvement of additional signaling pathways. Together, our findings suggest that TWEAK, alone and with TNF, can regulate osteoblast function, at least in part by inducing sclerostin expression. Our results also suggest new roles and modes of action for sclerostin.
TNF-related weak inducer of apoptosis (TWEAK) is a member of the TNF ligand superfamily(1) that has pleiotropic effects and has been shown to induce the production of pro-inflammatory mediators in a number of cell types, including fibroblasts and synoviocytes present in rheumatoid arthritis (RA) and advanced osteoarthritis (OA) patient tissues.(2) The unique receptor for TWEAK, fibroblast growth factor-inducible gene 14 (Fn14), is widely expressed(3) and is upregulated in response to tissue injury and inflammation.(4–7) Whereas neither TWEAK-(8) nor Fn14-deficient(9) mice display gross skeletal abnormalities, the effects of this signaling pathway on the biology of bone, and its response to physiological challenge and injury, have not been thoroughly examined to date. However, we(10) and others(11) have recently reported a significant role for TWEAK in the inflammatory bone remodeling seen in the mouse collagen-induced arthritis (CIA) model. Serum TWEAK was elevated in CIA mice(10) and a neutralizing TWEAK antibody significantly reduced disease severity. Together, the data suggest that TWEAK may regulate both joint inflammation and bone loss in inflammatory bone disease. The extent to which TWEAK and TNF, the latter also a mediator of inflammation-driven bone remodeling, each contribute to the loss of bone in CIA remains uncertain. We also reported that human osteoblasts express Fn14 and that TWEAK exposure inhibited their expression of the key osteoblast marker, osteocalcin,(10) suggesting a potential role for TWEAK in osteogenesis.
Ligation of TWEAK to Fn14 has been shown to activate a limited set of signaling pathways involved in cell proliferation and differentiation,(12) including NF-κB activation, which is associated with its pro-inflammatory effects.(3,13) TWEAK has also been shown to activate the mitogen-activated protein kinases (MAPK), JNK,(14) ERK,(14,15) and p38 MAPK.(5) Evidence for the direct involvement of these in the biological effects of TWEAK has thus far been shown for ERK1/2 in MC3T3-E1 cells(15) and p38 MAPK in astrocytes.(5)
The canonical Wnt signaling pathway fundamentally regulates osteoblast differentiation and bone formation.(16,17) Wnt ligands bind to frizzled (Fzd) and LRP5/6 co-receptors on target cells, preventing the proteosomal degradation of β-catenin and promoting the formation of transcription complexes with TCF/LEF transcription factors, resulting in the downstream transcription of osteogenesis-related genes. Several inhibitors of the Wnt pathway have been identified, including Dikkopf-1 (DKK1) and secreted Frizzled-related protein (sFRP).(16) Recent evidence, however, suggests that Lrp5 regulates bone mass indirectly, through its actions in the gut and on prevailing serotonin levels, rather than in the bone itself,(18) implying that the role of Wnts and their receptors in bone is not yet fully understood. The osteocyte product, sclerostin, is encoded by the SOST gene, mutations in which cause diseases with high bone mass in humans(16) as does deletion of SOST in mice,(19) indicating that this molecule has a key role in the regulation of bone mass. The mechanism of action of sclerostin is not fully elucidated, but there is evidence for its role as an atypical inhibitor of both the bone morphogenetic protein (BMP) and Wnt signaling pathways.(20–25) One activity of sclerostin seems to be similar to DKK1, to bind to and inhibit Lrp5/6,(20) although unlike DKK1, sclerostin does not antagonize exogenously added Wnt.(25)
In this study, we further examined the role of TWEAK in human osteoblast biology. In addition, we evaluated the interplay between TWEAK and TNF, because it seems that both of these cytokines influence osteoblast behavior and that both may be present under normal or pathologic conditions. Our results suggest that TWEAK may be both a physiologic and pathologic regulator of human osteoblast differentiation and activity.
MATERIALS AND METHODS
Recombinant cytokines and antibodies
Recombinant soluble human TWEAK and the TWEAK-neutralizing hamster anti-TWEAK monoclonal antibody (MAb) ABG11 were generated as previously described.(26,27) Mouse anti-TWEAK MAb P2D10 was generated as described previously.(1,10) Recombinant human TNFα, IL-1β, sclerostin, and mouse anti-human sclerostin MAb (MAB220902) were purchased from R&D Systems (Minneapolis, MN, USA). A sclerostin MAb suitable for Western blotting was kindly provided by Dr. J. P. Houchins (R&D Systems). Anti-human Fn14 MAb, ITEM-4, was purchased from eBioscience. Antibodies specific for phosphorylated or total isoforms of ERK1/2 (p42/44), JNK (p54/46), p38 MAPK, IκBα, NF-κB p65, AKT, GSK3β, and β-catenin were purchased from Cell Signaling Technology (Beverly, MA, USA). The MAb 8E7, which recognizes the active (dephosphorylated) form of β-catenin,(28) was purchased from Upstate/Millipore (Temecula, CA, USA).
Bone tissue, cells, and culture media
Adult human primary osteoblasts (normal human bone-derived cells [NHBC]) were isolated from femoral neck (trochanteric) trabecular bone and passaged, as described previously.(29) Equal numbers of male and female donors' cells were used (n = 8), whose mean age was 70.5 ± 7.5 yr. In addition, further similar bone samples from three men and three women, whose mean age of 72.5 ± 6.8 yr did not differ from that of the cultured NHBC (p = 0.61), were used for direct analysis of steady-state gene expression. The human osteoblast-like osteosarcoma (OS) cell line MG-63, shown previously to exhibit immature osteoblast characteristics,(30) was obtained from the American Type Culture Collection (ATCC, Rockville, MD, USA) and passaged in αMEM medium containing 10% FCS (αMEM-10). The mouse cell line MC3T3-E1, an immature osteoblast cell model,(31) was obtained originally from the American Type Culture Collection (ATCC) and was passaged as for NHBC. The mouse osteocyte cell line MLO-Y4(32) was kindly provided by Prof. Lynda Bonewald.
Immunofluorescence staining and flow cytometry
NHBC or cell lines were passaged as above and harvested using collagenase/dispase digestion. Cells were stained for surface Fn14 expression using the MAb ITEM4(33) or subclass-matched isotype control MAb, by immunofluorescence and flow cytometry, as described.(34) For staining of intracellular antigens, sclerostin and Ki67, cells were harvested as above and fixed in 4% paraformaldehyde/PBS for 10 min on ice, washed by centrifugation at 200g in 1 ml PBS, and permeabilized with 0.1% saponin for 10 min on ice. Cells were washed twice as above and resuspended in blocking buffer as above. Cells were incubated with anti-human sclerostin MAb or isotype-matched negative control IgG for 30 min on ice, and were washed three times by centrifugation at 200g in 1 ml PBS containing 0.1% BSA. Cells were incubated with a secondary antibody, goat anti-mouse-IgG-PE conjugate (1/50 dilution), and incubated for a further 30 min, whereupon they were washed three times as above. To each tube was added 5 μl of normal mouse serum to block excess IgG binding sites on the PE conjugate. After this, direct conjugates anti-Ki67-FITC (Becton Dickinson) or negative control IgG1-FITC (Becton Dickinson) were added to their respective tubes. After incubation for 30 min on ice and washing three times as above, the cells were finally fixed in FACS-fix for analysis by flow cytometry.
In situ immunofluorescence
NHBC were seeded into chamber slides (Laboratory-Tek; Nunc, Naperville, IL, USA) and grown to subconfluence for 3 days under the conditions indicated. The cells were rinsed with PBS, fixed with 4% paraformaldehyde in PBS for 10 min on ice, rinsed with PBS, and permeabilized with 0.1% Triton X100 for 5 min. Nonspecific binding sites were blocked with PBS containing 10% goat serum for 30 min at room temperature. The cells were incubated for 1 h with anti-TWEAK MAb (P2D10), mouse anti-sclerostin, or isotype-matched negative control MAb. After three washes in PBS, cells were incubated with anti-mouse IgG-FITC for 45 min in a dark humidified container. Cells were washed in PBS and mounted (Prolong Gold with DAPI anti-fade mounting media; Invitrogen). Samples were examined by confocal microscopy on a Radiance 2100 confocal microscope (Bio-Rad Microscience).
Preparation of total RNA and RT-PCR
Total RNA was prepared from human cancellous bone samples, as previously described.(35) Total RNA was extracted from bone, NHBC, and cell lines treated as above, and cDNA was prepared as described previously.(36) Gene expression was analyzed by real-time RT-PCR, as we have described previously.(37) Relative expression between samples was calculated using the comparative cycle threshold (CT) method (ΔCT), using GAPDH as the reference gene for comparisons within cell studies, as we have published.(37) Alternatively, 18S rRNA was used as the reference for comparisons between bone and cell gene expression because 18S rRNA expression mirrored total RNA quantities in bone samples more closely than GAPDH. Oligonucleotide primers were designed in-house to flank intron–exon boundaries and were purchased from Geneworks (Thebarton, South Australia, Australia). Real-time oligonucleotide primers for the amplification of osteocalcin and GAPDH (for both the mouse and human genes) were described previously.(37) Sequences of other primers for real-time PCR used were human 18S (102-bp product), GGAATTCCCGAGTAAGTGCG (sense), GCCTCACTAAACCATCCAA (antisense); human SOST (71-bp product), ACCACCCCTTTGAGACCAAAG (sense), GGTCACGTAGCGGGTGAAGT (antisense); mouse SOST (73-bp product), CCACCATCCCTATGACGCCAA (sense), TGTCAGGAAGCGGGTGTAGT (antisense); human DKK1 (97 bp product), ATCATAGCACCTTGGATGGG (sense), GACCGGTGACAAACAGAACC (antisense); human RUNX2/CBFA-1 (for the amplification of all splice variants giving a single 108-bp product), GATGTGCCTAGGCGCATTTCAG (sense), AGGGCCCAGTTCTGAAGCACC (antisense); human osterix (316-bp product), CTGCGGGACTCAACAACTCT (sense), GAGCCATAGGGGTGTGTCAT (antisense); human BSP-1 (123-bp product), ATGGCCTGTGCTTTCTCAATG (sense), AGGATAAAAGTAGGCATGCTTG (antisense); human LRP5 (139-bp product) GAATGTGGCCAAGGTCGTCGGA (sense), GCAGGTCTTCATGTCACTCAGCA (antisense); human LRP6 (76-bp product) CAACAGAGGCAGCCAAATGCCA (sense), GAGACATCAAACACAAATGGGAAC (antisense).
Cell proliferation experiments: carboxyfluorescein diacetate succinimidyl ester labeling of cells
Carboxyfluorescein diacetate succinimidyl ester (CFSE) was used as a fluorescence-based approach to track cell division, as we have described previously.(29,37) Briefly, NHBC were labeled with CFSE (Molecular Probes, Eugene, OR, USA) and cultured overnight before treatment with medium containing recombinant human TWEAK, TNF, or TWEAK/TNF, as indicated. Control cultures were treated with colchicine (300 ng/ml) to inhibit cell division, providing an input labeling index for the “parental” population. After 7 days, cells were detached by Trypsin-EDTA and analyzed by flow cytometry. Listmode data were analyzed using ModFit LT software (Verity Software House, Topsham, ME, USA) to calculate the number of cell divisions.
In vitro mineralization
To determine the ability of NHBC to form a mineralized matrix, a modification of a method reported previously(38) was used. NHBC were cultured in triplicate in wells of a 96-well plate (8 × 103 cells/well) in αMEM-10 containing dexamethasone (10−8 M) and KH2PO4 (1.8 mM) in the presence or absence of combinations of recombinant TWEAK, TNF, and neutralizing TWEAK MAb, ABG11. Media containing all supplements were replaced every 4 days, and cells were cultured for up to 6 wk before measurement of cell layer–associated Ca2+ levels, as described previously.(38) In some experiments, NHBC were cultured under the above mineralizing conditions with the addition of vitamin K2 (5 μM menatetrenone; Sigma) for 28 days, a culture condition that promotes the generation of mature osteocyte (OCy)-like cultures (G. J. Atkins and D. M. Findlay, unpublished data, 2008).
Western blot analysis
Replicate cultures of NHBC were serum deprived overnight in αMEM containing 0.1% FCS and incubated with recombinant TWEAK, TNFα, or both for the indicated times. Cells were lysed in PBS containing 0.1% Triton X100 and protease inhibitors (Complete Mini; Roche Diagnostics, Mannheim, Germany). Total protein was quantified using a commercial reagent (BCA Protein Assay Reagent; Pierce), and 50 μg total protein/track was electrophoresed on 4–12% SDS-PAGE gels (NuPage; Invitrogen, Mount Waverley, Victoria, Australia). Proteins were transferred to polyvinyl-difluoroacetate filters (PVDF; Geneworks, Adelaide, South Australia, Australia). After blocking for 2 h with 5% skim milk powder/0.05% Tween-20/PBS, filters were sequentially probed overnight at 4°C with antibodies with the indicated specificities at the recommended concentrations. After washing, filters were incubated with either mouse or rabbit IgG-specific antibodies conjugated to AP (Amersham Biosciences, Poole, UK) and binding detected with Attophos substrate (Promega, Madison, WI, USA) on a FluorImager (Molecular Dynamics, Sunnyvale, CA, USA). After this, filters were stripped using Western Blot Recycling Kit (Alpha Diagnostic International, San Antonio, TX, USA) and reprobed.
MAPK inhibition studies
NHBC labeled with CFSE, as described above, were cultured for 4 days in the presence or absence of TWEAK (100 ng/ml), TNF (5 ng/ml), or a combination of both, in the co-presence or absence of the MEK1 inhibitor PD89054 (10 μM) or the JNK inhibitor SP600125 (20 μM). After this time, total RNA was collected and gene expression tested by real-time RT-PCR, and cell proliferation was measured by flow cytometry using the methods described above.
Ex vivo human bone culture
Human cancellous bone, obtained from the proximal femur at hip replacement surgery, was dissected in tissue culture medium (αMEM) into ∼2- to 3-mm3 pieces, using a scalpel. The bone fragments were washed three times by centrifugation, and ∼10 pieces were added to each well of a 24-well tissue culture plate. Culture media consisting of αMEM-10, containing either TWEAK (100 ng/ml), TNF (5 ng/ml), or a combination of both, were added to replicate wells, and the bone samples were incubated for 24 h at 37°C. After this, samples were processed for real-time RT-PCR, as described above.
Student's t-test was used to analyze differences in mineralization and cell proliferation experiments. One-way ANOVA followed by Tukey's posthoc analysis was used to examine differences in gene expression studies. p < 0.05 was considered significant.
Human osteoblasts and osteoblast-like cell lines express Fn14 and TWEAK
NHBC were used for this study because they have been extensively characterized to display the expected properties of osteoblasts, including the ability to differentiate into mature osteoblasts and form a mineralized matrix,(30,38,39) under certain culture conditions differentiating into cells that display qualities of osteocytes(40) (G. J. Atkins and D. M. Findlay, unpublished data, 2008), and also express genes associated with the regulation of osteoclastogenesis.(29,40–42) NHBC (Fig. 1A) expressed high basal levels of cell surface Fn14, as identified by flow cytometry. Similar results were obtained for the human osteosarcoma cell lines SaOS-2, HOS, MG-63, and G292 (data not shown). Immunofluorescence staining of NHBC showed intracellular staining for TWEAK protein (Fig. 1B), suggesting that TWEAK could have an autocrine role in osteoblast activity.
Exogenous TWEAK inhibited the ability of NHBC to form a mineralized matrix in long-term cultures, a key functional indicator of osteoblast activity and osteogenesis. This inhibitory effect was reversed with the co-addition of TWEAK neutralizing antibody, ABG11 (Fig. 1C). Conversely, TNF increased in vitro mineralization by NHBC. This effect was antagonized by simultaneous administration of TWEAK, in a dose-dependent fashion. Again, co-administration of ABG11 reversed the TWEAK effect on this action of TNF. These results suggest that TWEAK is a negative regulator of osteoblast differentiation and osteogenesis.
Effect of TWEAK and TNFα on osteoblast proliferation
Osteoblast proliferation is another important determinant of the osteogenic effect and is inversely related to osteoblast maturation. To examine the effect of TWEAK and TWEAK/TNF on osteoblast proliferation, we used the CFSE fluorescence approach(29,43) that measures cell proliferation and takes into account that not all cells in a given population divide at the same rate. Using this method, we found that TWEAK induced the proliferation of NHBC, because there was a reduction in the number of cells in the parental (P) population and increased numbers of cells having three or more cell divisions (Fig. 2). This effect was reversed with the co-addition of the neutralizing TWEAK antibody ABG11. TNF was also mitogenic for NHBC, and this effect was enhanced in the presence of TWEAK, similar to that seen with TWEAK alone (Fig. 2). These results are consistent with maintenance of an immature osteoblast phenotype in the presence of TWEAK and TNF.
Effect of TWEAK and TNF on osteoblast gene transcription
To determine the corresponding effect of exogenous TWEAK on osteoblast gene expression, alone and in the presence of TNF, we performed real-time RT-PCR analysis on mRNA extracted from cells treated for up to 3 wk with increasing concentrations of TWEAK and TNF. After treatment for 72 h, TWEAK at concentrations >10 ng/ml inhibited RUNX2 mRNA expression by ∼65% (Fig. 3A; Table 1). TWEAK in the co-presence of TNF decreased RUNX2 expression ∼2-fold (Fig. 3A; Table 1). TNF alone had a variable effect on RUNX2 levels, decreasing RUNX2 expression at 1 ng/ml and increasing it at 5 ng/ml, in the example shown (Fig. 3A). The effect of TNF alone on RUNX2 expression varied in a donor-, time-, and dose-dependent manner (Table 1). TWEAK dose-dependently increased osterix mRNA levels by up to 9-fold (Fig. 3B). TNF alone had no effect on osterix expression (Fig. 3B; Table 1). Interestingly, in response to various doses of TWEAK, a significant inverse relationship existed between RUNX2 and osterix expression (r = −0.73, p < 0.05; Fig. 3C). The combined effect of TWEAK/TNF on osterix was similar to the effect of TWEAK alone, suggesting that TWEAK has a dominant effect on this important osteoblast transcription factor. TWEAK also dose-dependently downregulated the expression of genes associated with osteoblast maturation, including osteocalcin (OCN) and BSP-1 (Figs. 3D and 3E). Type I collagen-α1, alkaline phosphatase (ALP), and osteopontin (OPN) expression was also reduced (data not shown). TNF was more potent than TWEAK in its inhibitory effect on OCN mRNA expression, with 1 ng/ml TNF having a greater effect than TWEAK at any concentration tested (Fig. 3D), whereas TNF and TWEAK similarly affected BSP-1 mRNA expression (Fig. 3E).
TWEAK alone, and in combination with TNF, induces sclerostin expression
The gene expression profile of NHBC in the presence of TWEAK was consistent with inhibition of osteoblast differentiation. We therefore examined the expression of known inhibitors of osteoblast differentiation, in response to TWEAK. TWEAK was found to induce the expression of the SOST gene product, sclerostin (Fig. 3F), a reported inhibitor of the BMP(21,44) and Wnt(20,25) signaling pathways. TWEAK at low concentrations (1–5 ng/ml) also modulated the expression of the Wnt and reported sclerostin(20) receptors LRP5 (Fig. 3G) and LRP6 (Fig. 3H), providing evidence that TWEAK may also modulate the ability of osteoblasts to respond to Wnt ligands and/or Wnt antagonists. The positive effect of TWEAK on SOST mRNA expression was highly reproducible between NHBC isolated from different donors, the mean fold-induction of SOST mRNA expression after 72-h exposure to TWEAK at 100 ng/ml being 27.7 ± 4.7-fold (Table 2; n = 17, p < 0.0001). TNF alone stimulated SOST mRNA expression to a lesser extent than TWEAK; however, the combination of both cytokines resulted in a higher mean fold induction than either cytokine alone (Table 2). Conversely, we did not observe any consistent effect of TWEAK or TNF on the expression of DKK1 in NHBC. The changes in DKK1 mRNA levels in response to TWEAK (1.8 ± 1.6-fold), TNF (1.5 ± 1.3-fold), and TWEAK + TNF in combination (2.5 ± 2.7-fold) seen in eight donors' cells after a 72-h treatment were not significant. Further examination of the effect of TWEAK on SOST expression in NHBC showed that TWEAK also induced intracellular sclerostin protein expression, which was detected by both in situ immunofluorescence (Fig. 4A) and flow cytometry (Fig. 4B). Note that, whereas TWEAK gave increased immunostaining for sclerostin, TNF or the combination of TWEAK + TNF (Fig. 4A) did not do so to the same extent, implying that post-translational mechanisms, including secretion, potentially affect intracellular levels. Continuous exposure to TWEAK increased SOST mRNA expression for the duration of TWEAK treatment of NHBC, cultured under conditions otherwise permissive for osteoblast differentiation, as indicated by the time course of expression of osteocalcin mRNA in the control cells (Fig. 4C). TNF also upregulated SOST mRNA levels in a dose-dependent manner (Fig. 3F) but suppressed expression under mineralizing conditions (Fig. 4C), consistent with a positive effect of TNF on mineral apposition. However, TWEAK in the presence of TNF, synergistically upregulated SOST mRNA levels for the first 7 days of in vitro mineralization, expression returning thereafter to untreated levels (Fig. 4C). This is consistent with the observation that TWEAK in the presence of TNF was not as inhibitory of mineralization compared with TWEAK alone. SOST mRNA expression in response to TWEAK was dependent on new protein synthesis, because induction was completely abrogated in the presence of the protein synthesis inhibitor, cycloheximide (Fig. 4D). The levels of basal and induced SOST mRNA in NHBC were comparable to the steady-state levels observed in human bone samples, similar to those from which the NHBC were derived (Fig. 4E), supportive of the likely biological significance of these findings. To determine whether TWEAK/TNF could modulate SOST mRNA levels in human bone in situ, we cultured freshly isolated human cancellous bone in the presence or absence of each cytokine for 24 h and prepared total bone RNA for analysis by real-time RT-PCR. As shown in Fig. 4F, both TWEAK and TNF induced SOST mRNA expression in this model, providing evidence that SOST expression might be modulated by these cytokines in vivo.
The robust induction of SOST mRNA in subconfluent, nonmineralized cultures of NHBC indicated that immature osteoblasts were capable of expressing SOST in the presence of TWEAK, TNF, and TWEAK/TNF. Consistent with this, TWEAK and TNF/TWEAK also induced the expression of SOST mRNA, albeit at low absolute levels, in the immature osteoblastic cell lines, MC3T3-E1 and MG-63 (Table 2). To test the effect of TWEAK and TNF on SOST expression in mature, osteocyte-like cells, NHBC were cultured for 28 days under mineralizing conditions in the added presence of vitamin K2, a condition that we have found results in a postproliferative mature osteocyte-like phenotype, as shown by decreased expression of osteoblast markers ALP and type I collagen and increased expression of osteocyte markers such as E11/gp38,(45) dentin matrix protein (DMP)-1,(46) matrix extracellular phosphoglycoprotein (MEPE),(47) and sclerostin(48,49) (G. J. Atkins and D. M. Findlay, unpublished data, 2008). Cell populations generated under these conditions were exposed to TWEAK, TNF, or TWEAK/TNF for 3 days. This resulted in an induction of sclerostin transcription up to 3.5-fold (Table 2). This response was similar to the fold-change observed in ex vivo experiments (Fig. 4F), which likely measured the response, to a large extent, of mature cancellous bone osteocytes. Furthermore, TWEAK was able to induce SOST expression, from very low basal levels, in osteocyte-like MLO-Y4 cells (Table 2). These data suggest that mature osteocyte-like cells respond to TWEAK and TWEAK/TNF by upregulating sclerostin expression.
TWEAK induction of sclerostin expression is in proliferating cells
To further examine the maturation stage, at which human primary osteoblastic cells could express sclerostin, we examined sclerostin expression as a function of cell proliferation. NHBC were treated with TWEAK and double stained by immunofluorescence for sclerostin and Ki67, a nuclear marker for cells in cycle.(50) The majority of cells in these cultures were in cycle, evidenced by the high percentage of Ki67+ cells (Fig. 5). Intracellular sclerostin expression was evident in the Ki67+ cells. Note that, although high levels of sclerostin protein were detected at the highest concentration of TWEAK (200 ng/ml), this method does not take into account secreted protein and may therefore underestimate the increase in total sclerostin synthesis. The induction of SOST in immature, proliferating cells is consistent with the expression data in the immature cell line models MC3T3-E1 and MG-63 (Table 2). Together, our data suggest that TWEAK induction of SOST occurs in both immature osteoblasts and in mature osteocyte-like cells.
TWEAK induces sclerostin by phosphorylation of ERK1/2 and JNK
We next examined several signaling pathways, shown previously to be activated by either TWEAK or TNF in other cell types. Time course studies of serum-starved NHBC showed that both TWEAK and TNF induced the phosphorylation of ERK1/2 (p42/p44), consistent with both cytokines having growth and differentiation effects on osteoblasts. Whereas TNF alone induced more rapid and marked phosphorylation of ERK1/2, there was an additive effect of TWEAK in combination with TNF, with a sustained increase in the ratio of phosphorylated to total ERK at 3 h (p < 0.05; Fig. 6A). TNF also induced NF-κB activation as indicated by the increase in phosphorylated IκBα (Fig. 6A) and the appearance of the phosphorylated p65 subunit of NF-κB (data not shown). TWEAK did not induce detectable NF-κB activation in NHBC. Interestingly, TWEAK also induced the phosphorylation of JNK to a greater extent than TNF, whereas TWEAK and TNF combined gave rise to a more rapid and robust phospho-JNK response (Fig. 6A). We did not observe phosphorylation of p38 MAPK or activation of the Akt pathway in response to either TNF or TWEAK (data not shown).
To determine the functional relevance of activation of ERK1/2 and JNK pathways to the induction of sclerostin by TWEAK, we incubated NHBC with the MEK inhibitor PD90859 or the JNK inhibitor SP600125. Cells were treated for 4 days in the presence of TWEAK or TNF, with the respective inhibitor at the concentrations indicated. TWEAK and TWEAK/TNF-induced sclerostin expression was strongly inhibited by SP600125 and partially inhibited by PD89059 (Fig. 6B). No inhibitory effect on cell proliferation was observed at the concentrations of each inhibitor used (Fig. 6C), arguing against nonspecific toxic effects. These results show, for the first time, that sclerostin expression can be induced by TNF-related pro-inflammatory cytokines in osteoblasts and that this expression is mediated mainly through the phosphorylation of the MAPKs, JNK, and ERK1/2.
TWEAK and sclerostin have similar intracellular effects on canonical Wnt signaling and promote ERK1/2 phosphorylation
To further explore the consequences of TWEAK induction of sclerostin, NHBC were treated with recombinant human sclerostin. Short-term exposure of NHBC to recombinant human sclerostin resulted in a qualitatively similar effect to that of TWEAK or TWEAK/TNF on both RUNX2 (Fig. 7A) and OCN mRNA (Fig. 7B) expression. This confirms that NHBC are able to respond to sclerostin and suggests that sclerostin could mediate at least some of the TWEAK effects. We also investigated the effect of TWEAK on parameters of Wnt signaling. Recombinant TWEAK, TNF, or TWEAK + TNF treatment of serum-starved NHBC for 24 h resulted in increased levels of both total and phosphorylated forms of the enzyme GSK3β and the levels of active and total β-catenin, relative to levels of β-actin (Fig. 7C). Increases in phospho-GSK3β and hypophosphorylated β-catenin are normally associated with the promotion of Wnt signaling. It is possible that levels of phospho-GSK3β increase as a compensation for the increase in levels of the active enzyme. Likewise, active β-catenin levels may increase as a function of total β-catenin levels. Consistent with this hypothesis, treatment of the same cells with human recombinant sclerostin also resulted in increased total and phospho-GSK3β and active and total β-catenin (Fig. 7C). The pattern of GSK3β and β-catenin induction by TWEAK and TNF coincided with a detectable increase in endogenous levels of the mature (∼30 kDa) form of sclerostin (Fig. 7C). Note that exposure of cells to exogenous sclerostin resulted in an apparent increase in intracellular sclerostin. However, we cannot rule out that this is simply detection by the antibody of cell-associated recombinant protein. Examination of early intracellular signaling events in the response to recombinant sclerostin by serum-starved NHBC showed an unexpected rapid phosphorylation of ERK1/2 similar to the kinetics of TWEAK phosphorylation of ERK1/2 (Fig. 7D), indicating that sclerostin may have previously unrecognized actions on intracellular signaling.
This study showed that TWEAK regulates several important activities of human osteoblast metabolism and indicated that TWEAK may play an intrinsic role in bone remodeling. We found that primary human osteoblasts and osteoblast-like osteosarcoma cell lines express high levels of cell surface Fn14, consistent with our previous findings,(10) studies in mouse MC3T3-E1 cells,(15) and studies in human mesenchymal stem cells.(51) Because human primary osteoblasts were also found to express abundant TWEAK mRNA and protein, it is possible that osteoblasts may respond to TWEAK in an autocrine fashion. Exogenous addition of TWEAK potently induced the proliferation of primary human osteoblasts, an effect seen in a number of cell types, including fibroblasts, epithelial, endothelial, and glial cells.(9,14,51–54) TWEAK also enhanced TNF-induced cell proliferation.
The stimulation of osteoblast proliferation by TWEAK might be expected to lead to a magnified osteogenic response because of an increase in osteoprogenitor numbers. However, TWEAK inhibited in vitro mineralization by human primary osteoblasts. Consistent with this, TWEAK suppressed the expression of osteogenic markers, such as type I collagen-α1, ALP, BSP-1, and OCN, the latter as we have reported previously.(10) In contrast, TNF promoted in vitro mineralization by NHBC over an extended culture period. This is despite a strong initial suppression of OCN mRNA by TNF. Interestingly, TWEAK antagonized the osteogenic effect of TNF, perhaps linked to the concomitant effects on TNF-induced cell proliferation.
We also observed direct effects of TWEAK on the expression of the master transcription factors, RUNX2 and osterix, both of which are essential in the development and differentiation of the osteoblast lineage.(55) TWEAK dose- and time-dependently increased osterix and decreased RUNX2 transcription, whereas TNF by itself had only marginal effects on the expression of these genes. This effect of TNF, and our finding that TNF induced in vitro mineralization in NHBC, is in contrast to the findings by Gilbert et al.,(56) who showed inhibition of RUNX2 expression in fetal mouse calvarial osteoblasts and in MC3T3-E1 cells, and of MC3T3-E1 differentiation, in response to TNF. Thus, this study seems to have identified species-specific differences in the action of TNF, although this requires further investigation to account for possible age- and site-specific influences. Our findings implicate TWEAK in particular as a key controller of human osteoblast differentiation. The effect of TWEAK on RUNX2 expression is consistent with a reported role of RUNX2 in controlling osteoblast proliferation.(57,58) Also consistent with the mitogenic effect of TWEAK is the stimulatory effect on osterix expression, which has itself been shown to promote cell proliferation and decrease expression of osteoblast genes such as OCN and ALP.(59) The effect of TWEAK on cell proliferation may therefore explain the observed reciprocal relationship between RUNX2 and OSX mRNA expression in response to TWEAK (see Fig. 3C). The induction of osterix by TWEAK discounts the possibility that TWEAK diverts cells from the osteoblast lineage as an explanation for its inhibitory effects on differentiation. Consistent with TWEAK inducing an immature osteoblast phenotype, we observed increased expression of RANKL mRNA in response to both TWEAK and TWEAK/TNF, dose-dependently and during the first 7 days of culture under mineralizing conditions (data not shown), consistent with our previous findings that 1α,25-(OH)2vitamin D3 induction of RANKL occurs preferentially in immature osteoblasts.(29)
Strikingly, TWEAK induced the expression, at both the mRNA and protein levels, of the negative regulator of bone formation, sclerostin,(17,48) in a dose-dependent fashion. Sclerostin mRNA levels increased on average 27.7 ± 4.7-fold within 72 h of exposure and remained chronically high in the presence of TWEAK relative to levels under control mineralizing conditions. TNF by itself increased expression to a lesser extent than TWEAK and early in culture combined with TWEAK to further increase sclerostin expression. The induction of sclerostin expression by TWEAK and TWEAK/TNF was highly dependent on JNK and ERK1/2 phosphorylation. The evidence to date suggests that sclerostin expression in normal bone is restricted to mature osteocytes embedded in mineralized bone,(24,48) with some expression by osteoblasts.(21) Our unpublished observations concur with this, that sclerostin expression normally coincides with in vitro mineralization by NHBC (G. J. Atkins, unpublished data, 2008). However, the expression pattern of sclerostin in vivo has not been tested under inflammatory conditions. The levels of sclerostin mRNA achieved in NHBC in response to TWEAK were similar to, and in some cases exceeded, the levels seen in steady-state bone sampled from the proximal femur of donors matched for sex and age with the NHBC donors. We also showed that TWEAK and TNF could upregulate sclerostin mRNA levels in ex vivo cultures of human bone. This argues strongly that the response to TWEAK results in biologically relevant levels of sclerostin in vitro and implies that TWEAK may induce physiologically relevant levels in vivo. Whereas in this study we found that TWEAK induced sclerostin expression several-fold in heavily mineralized cultures of human osteocyte-like cells, we also observed sclerostin induction in proliferating cells (NHBC) in nonmineralized, subconfluent cultures. Closer study showed that NHBC expressing sclerostin in response to TWEAK were actively in cell cycle, evidenced by co-staining for the antigen Ki67. Interestingly, sclerostin mRNA expression was also inducible by TWEAK and synergistically by TWEAK/TNF in several human and mouse cell line models. These included MG-63 human osteosarcoma cells, an immature human osteoblast model,(30) undifferentiated mouse MC3T3-E1 cells, an immature osteoblast cell line model,(31) albeit from very low starting levels, consistent with previously reported findings,(60) and in MLO-Y4 cells, a cell line model of the mouse osteocyte.(32) It is noteworthy that this well-used osteocyte model has not been previously reported to express sclerostin mRNA. Together, these data indicate that both immature osteoblasts and mature osteocytes may be induced to express sclerostin under inflammatory conditions, perhaps contributing to the osteoblast and bone formation defects seen in conditions such as RA.(61) Whereas the downstream effects of sclerostin on osteoblast biology in vitro are not well characterized, our preliminary data indicated that the exposure of NHBC to recombinant sclerostin has some effects in common with treatment with TWEAK and TWEAK/TNF, viz suppression of osteocalcin and RUNX2 expression. Sclerostin is a member of the DAN family of secreted glycoproteins that share the ability to antagonize the activity of BMPs.(62) Whereas there is no doubt as to the effect of sclerostin, and the loss of sclerostin, on bone mass in both humans(16) and mice,(19,63) there is conjecture in the literature as to the precise mechanism of action of sclerostin because it does not behave as a classical antagonist of either the BMP or Wnt signaling pathways.(23–25) Sclerostin was found to form a complex with the BMP antagonist noggin, which appeared to prevent binding of either antagonist to BMP.(22) Expression of recombinant sclerostin in cells transfected with Wnt1 revealed an apparent inhibitory activity of sclerostin on Wnt signaling. Co-expression of LRP5, a Wnt co-receptor, overcame this inhibitory effect, suggesting that sclerostin behaved as a Wnt antagonist.(20) Recombinant sclerostin was found to bind both LRP5 and LRP6 with affinities comparable to the binding of the Wnt inhibitor DKK1.(20) However, work from Winkler's group(23) showed that sclerostin could not block Wnt signaling as measured by β-catenin accumulation, but suggested that sclerostin antagonized Wnt induced BMP signaling. Gene microarray analysis of myoblastic KS483 cells treated with either recombinant BMP, sclerostin or both revealed that BMP and sclerostin induced gene expression profiles that in some cases overlapped, both in terms of specific genes affected and the direction of expression.(25) Because sclerostin appeared to affect Wnt target genes more so than BMP-target genes, it was suggested that sclerostin might inhibit BMP-induced Wnt signaling.(25) Further suggestion that sclerostin does not behave as a classical Wnt antagonist arises from the observation that sclerostin could only inhibit canonical Wnt signaling if Wnts were expressed by means of transfection, as opposed to exogenously added Wnt.(25) Relevant to this study, the effect of sclerostin on primary osteoblasts, in particular those of human origin, has not been previously described. We found evidence that TWEAK may have effects on canonical Wnt signaling, evidenced by regulation at the mRNA level of LRP5/6 and an increase in endogenous levels of total and phosphorylated GSK3β, the latter event being usually associated with activation of canonical Wnt signaling.(64) In addition, TWEAK and TNF appeared to increase the levels of dephosphorylated β-catenin, being the form that is capable of translocation to the nucleus and activating β-catenin-responsive gene transcription in response to Wnt(28) and also the GSK3β inhibitor, LiCl.(65) The increase in phospho-GSK3β and active β-catenin coincided with an increase in levels of intracellular sclerostin, consistent with this event mediating the TWEAK/TNF effect. Unexpectedly, treatment of cells with recombinant sclerostin also resulted in an increase in levels of both total and phospho-GSK3β and the level of total and active β-catenin, implying that the mechanism of action in human primary osteoblasts is not straightforward, as the current state of knowledge suggests.(66) In support of our findings, Li et al.(20) observed that, under some circumstances, co-transfection of cells with SOST and Wnt1 resulted in increased Wnt activity. Thus, it is possible that sclerostin, rather than simply blocking Wnt signaling, is capable of modulating the signal depending on the local concentrations of Wnt ligands, Lrp receptors and of sclerostin itself. In support of this, sclerostin did not appear to inhibit the binding of Wnt3a to Lrp5.(20) Also strongly suggestive that sclerostin signaling is more complex than is currently recognized is our observation that exogenously added sclerostin, in addition to TWEAK, activated ERK1/2 MAPK signaling. Similar findings to ours were recently reported in abstract by Caverzasio,(67) who found that sclerostin induced ERK1/2 phosphorylation but in the absence of the expected (inhibitory) effect on proximal Wnt3a signaling in mouse osteoblastic cell lines. The role of MAPK signaling in osteoblast differentiation, and how this pathway interacts with the Wnt and BMP signaling pathways, is also poorly understood. Transient ERK1/2 phosphorylation has been reported in response to Wnt3a stimulation and seems to be involved in the proliferative response to Wnt.(68,69) Phosphorylation of ERK1/2 has also been reported to positively influence TGF-β and BMP-induced osteoblastic differentiation(70,71) and may be required for osterix expression.(72) Other studies have suggested that activated ERK1/2 may inhibit or delay osteoblast differentiation in response to BMP.(73,74) In light of our findings, further work will be required to elucidate the role of sclerostin induced MAPK signaling, as well as the involvement of Wnt and BMP signaling, with respect to human osteoblast differentiation and bone formation.
This is the first report linking the activity of pro-inflammatory mediators to sclerostin expression. A recent report(75) identified TNF mediated induction of DKK1 expression in a mouse model of inflammatory arthritis and in human RA. In the current study, we did not observe any consistent significant effect of TWEAK or TNF on acute expression of DKK1 by NHBC. However, DKK1 mRNA was expressed by the human primary osteoblasts and so we cannot discount that sclerostin and DKK1 may both participate in the regulation of Wnt signaling in this system, because both seem to share LRP5/6 as cellular receptors.(20) In addition, DKK1 may act at a later stage of the response because of the positive effect of TWEAK on osterix expression, which has been shown recently to inhibit Wnt signaling by inducing DKK1.(76) These possibilities remain to be studied. The induction of expression of inhibitors of the Wnt and/or BMP signaling pathways, such as sclerostin, represents a novel potential mechanism, by which TWEAK alone, or in concert with TNF, might regulate physiologic osteoblast differentiation and mineralization, and suppress these processes in chronic inflammatory disease states.
There is an emerging concept that TNF ligand family members are capable of modulating each others' activities. For example, TNF synergizes with RANKL in the induction of osteoclast formation.(77,78) Given that the microenvironment encountered by cells during physiologic or pathologic bone remodeling consists of a complex cytokine milieu, such interactions are likely highly relevant and important. For example, we recently reported that TWEAK is a mediator of joint erosion in the mouse CIA model, a model in which TNF is also expressed and is implicated in the mechanism.(10) Our data indicate that TWEAK and TNF modulate each other's activities with respect to osteoblast behavior, and the functional outcome likely depends on the relative level of expression of each cytokine and their respective receptors.
Taken together, the data presented here strongly suggest that TWEAK is potentially anti-anabolic in bone. In an inflammatory setting, TWEAK or TWEAK/TNF may act to inhibit the osteogenic activity of osteoblasts, through regulation of RUNX2, osterix, and sclerostin expression. These observations give new insight into the mechanism of physiologic bone remodeling and the bone loss and lack of repair that are observed in a number of clinically important bone loss pathologies, including osteoporosis and RA.
This work was supported by the National Health and Medical Research Council of Australia (NHMRC). G.J.A. was supported by a NHMRC R Douglas Wright Fellowship. The authors thank the surgeons and nursing staff of the Department of Orthopaedics and Trauma, Royal Adelaide Hospital, for the provision of bone samples, Lena Truong for preparing bone RNA, Dr. Gethin Thomas (Diamantina Institute, University of Queensland) for providing us with MC3T3-E1 cells, and Prof. Lynda Bonewald (University of Missouri-Kansas City) for providing us with MLO-Y4 cells. The authors also thank Dr. J. P. Houchins (R&D Systems) for provision of novel MAbs.
- 352001 The ratio of messenger RNA levels of receptor activator of nuclear factor kappaB ligand to osteoprotegerin correlates with bone remodeling indices in normal human cancellous bone but not in osteoarthritis. J Bone Miner Res 16: 1015–1027., , , ,
- 422004 Isolation of a human homolog of osteoclast inhibitory lectin that inhibits the formation and function of osteoclasts. J Bone Miner Res 19: 89–99., , , , , , , , , , , , , , ,Direct Link:
- 432001 Flow Cytometric Analysis of Cell Division History Using Dilution of Carboxyfluorescein Diacetate Succinimidyl Ester, a Stably Integrated Fluorescent Probe. Academic Press, San Diego, CA, USA., ,
- 492006 Vitamin K analogues promote the differentiation of human primary osteoblasts. Calcif Tissue Int 78: S44., , ,
- 672008 Wnt/LRP5-independent inhibition of osteoblastic cell differentiation by sclerostin. J Bone Miner Res 23: S72.
- 702002 Bone morphogenetic proteins, extracellular matrix, and mitogen-activated protein kinase signaling pathways are required for osteoblast-specific gene expression and differentiation in MC3T3-E1 cells. J Bone Miner Res 17: 101–110., , , , ,