Glucocorticoids suppress Wnt16 expression in osteoblasts in vitro and in vivo

Glucocorticoid-induced osteoporosis is a frequent complication of systemic glucocorticoid (GC) therapy and mainly characterized by suppressed osteoblast activity. Wnt16 derived from osteogenic cells is a key determinant of bone mass. Here, we assessed whether GC suppress bone formation via inhibiting Wnt16 expression. GC treatment with dexamethasone (DEX) decreased Wnt16 mRNA levels in murine bone marrow stromal cells (mBMSCs) time- and dose-dependently. Similarly, Wnt16 expression was also suppressed after DEX treatment in calvarial organ cultures. Consistently, mice receiving GC-containing slow-release prednisolone pellets showed lower skeletal Wnt16 mRNA levels and bone mineral density than placebo-treated mice. The suppression of Wnt16 by GCs was GC-receptor-dependent as co-treatment of mBMSCs with DEX and the GR antagonist RU-486 abrogated the GC-mediated suppression of Wnt16. Likewise, DEX failed to suppress Wnt16 expression in GR knockout-mBMSCs. In addition, Wnt16 mRNA levels were unaltered in bone tissue of GC-treated GR dimerization-defective GRdim mice, suggesting that GCs suppress Wnt16 via direct DNA-binding mechanisms. Consistently, DEX treatment reduced Wnt16 promoter activity in MC3T3-E1 cells. Finally, recombinant Wnt16 restored DEX-induced suppression of bone formation in mouse calvaria. Thus, this study identifies Wnt16 as a novel target of GC action in GC-induced suppression of bone formation.

Several studies indicate that GCs suppress canonical Wnt signaling by reducing various signalling components such as cytosolic and nuclear abundance of β-catenin 16,39,40 and gene expression of low-density lipoprotein 5 16,41 . Concurrently, GCs increase the expression of Wnt inhibitors such as Dickkopf-1 16,42,43 and Wnt inhibitor factor 1 44,45 suggesting that the inhibition of Wnt signaling may be a key mechanism of GC-induced suppression of bone formation. Despite the crucial role of Wnt16 in the regulation of bone homeostasis, the modulation of Wnt16 by GCs has not been studied so far. Thus, in this study, we tested the hypothesis that GCs inhibit bone formation via suppressing Wnt16 expression.

Methods
Declaration of approval for animal experiments. All experiments involving animals were approved by the Landesdirektion Sachsen, Germany and were performed in accordance with the relevant guidelines and regulations (approval number: TVV 86/2017).

Animals and induction of glucocorticoid-induced bone loss.
Full GR knock-out mice (GR −/− ) and mice with impaired GR dimerization (GR dim ) were described previously 13 . Two different models of GC-induced bone loss were performed. In the first experiment, 24-week-old female FVB/N mice were used and received a 3.5 mg prednisolone (n = 6) or placebo (n = 6) pellet (60 day release; Innovative Research of America, Inc.) for 4 weeks. In the second experiment, 10-week-old female FVB/N mice received slow-release pellets with prednisolone (PRED, n = 3) or placebo (n = 3) for 14 days (15 mg; 60 day release; Innovative Research of America, Inc.). Pellets were implanted subcutaneously and released a calculated prednisolone dose of 12.5 mg/kg/day. This resulted in a calculated dose of 2 mg/kg/day. Bone loss and suppression of bone formation was confirmed in both models and has been reported previously 13,18 . Animal protocols were approved by the Landesdirektion Sachsen, Germany.
Bone analysis. Trabecular bone mineral density was assessed in the distal femur by peripheral quantitative computed tomography (pQCT; Stratec, Pforzheim, Germany). The measurements were made with a voxel size of 70 µm at the distal femur (two slices at the trabecular region located 1 mm from the growth plate, one at the mid-diaphysis located 6 mm from the growth plate). For the measurement of trabecular bone mineral density, regions of interest were set. For contouring mode 1 and peel mode 20 were used. Quality control measurements were performed using the standard and cone phantom, which was supplied by the manufacturer. In vivo calvarial model. week-old C57BL/6j wildtype mice were injected subcutaneously with 0.5 mg/kg dexamethasone (DEX, Sigma-Aldrich) or 0.5 mg/kg DEX and 500 ng recombinant human WNT16 (Peprotech) three times per week, respectively, in a total volume of 25 µl. One group received 25 µl PBS injections as vehicle control. Five and 2 days before sacrifice, the mice received two intraperitoneal injections of calcein (20 mg/kg, Sigma-Aldrich). The mice were sacrificed after three weeks. Calvariae were fixed in 4% PBS-buffered paraformaldehyde and dehydrated in ascending ethanol series. Subsequently, calvariae were embedded in methacrylate and cut into 7 µm sections to assess fluorescence labels using fluorescence microscopy. Bone formation rate per bone surface (BFR/BS) was determined using the Osteomeasure software (OsteoMetrics, Decatur, GA, USA).
Osteoblast cultures. Bone marrow was flushed from hind legs of C57BL/6 or GR mutant mice and seeded at a density of 1,000,000 cells per cm² in DMEM containing 10% fetal calf serum (FCS) and 1% penicillin/streptomycin (Pen/Strep) (all from Invitrogen, Karlsruhe, Germany). At 70% confluence, cells were switched to differentiation media with 5 mM β-glycerol phosphate and 100 µM ascorbate phosphate (all from Sigma-Aldrich) for 7 days. Thereafter, cells were switched to starving media (DMEM containing 1% FCS and 1% Pen/Strep) overnight and treated with DEX (Sigma-Aldrich) at concentrations as indicated in the figure legends for a maximum of 48 hours. In some experiments, cells were co-treated with 1 µM RU-486 (Sigma-Aldrich), an established GR antagonist.
Calvaria organ culture. Calvarial bones were isolated from 3-5 day old C57BL/6 pups and surrounding tissue was removed. Both halves were placed in a culture dish with DMEM containing 10% FCS and 1% Pen/ Strep. Calvariae were stimulated with 1 µM DEX or vehicle for 24 h. Thereafter, calvarial bones were used for RNA isolation as described below.
Alkaline phosphatase (ALP) assay. Cells were lysed in 100 µl lysis buffer (1.5 mM Tris-HCl pH 7, 1 mM ZnCl 2 , 1 mM MgCl 2 , 1% triton X-100) and centrifuged for 15 min at 6,000 rpm at 4 °C. Aliquots of each sample were incubated with 100 µl ALP substrate buffer (100 mM diethanolamine, 0.1% triton X-100 supplemented with 1:10 37 mM p-nitrophenylphosphate) for 30 min at 37 °C. The enzymatic reaction was stopped with 40 mM NaOH, measured at a wavelength of 405 nm, and normalized to the total protein content determined by the BCA method from the same protein extracts.
Promoter reporter assay. The GR binding site was identified as a transcriptional regulatory motif of the Wnt16 promoter region by the online search tool (http://RegRNA.mbc.NCTU.edu.tw/; http://tfbind.hgc.jp/) and the eukaryotic promoter database (http://epd.vital-it.ch/). In order to investigate activation of the Wnt16 promoter by GR signalling, a dual-reporter assay using secreted Gaussia princeps luciferase (GLuc) under the control SCIEnTIFIC RepORTS | (2018) 8:8711 | DOI:10.1038/s41598-018-26300-z of a Wnt16 promoter was applied (pEZX-PG04, Gene Copeia). The plasmid pEZX-PG04 also contains secreted alkaline phosphatase (SEAP) as a secondary reporter gene for transfection normalization. MC3T3-E1 cells were seeded in a density of 1.5 × 10 4 cells per cm 2 in basal medium (α-MEM, 10% FCS, 2 mM glutamine and 1% PenStrep). When cells were 70-80% confluent, medium was changed to basal medium without antibiotics one hour before the transfection. pEZX-PG04 was diluted to a final concentration of 1 μg plasmid DNA/100 μl medium (0.01 μg/μl) with the reduced serum medium Opti-Minimal Essential Medium (Life Technologies). Transient transfections were performed using the liposome-mediated X-tremeGENE HP DNA (Roche) transfection reagent. The medium was changed to fresh standard basal medium 24 h after transfection and DEX (1 µM) treatment was performed. Supernatant was harvested for the GLuc and SEAP assay 24 and 48 h after the treatment and stored at −20 °C until usage. Luciferase activity was measured by using the Secrete-Pair Luminescence assay kit (Gene Copeia) according to the manufacturer´s high sensitivity protocol.
Western Blot analysis. BMSCs were isolated from hind legs of wild type mice and seeded at a density of 1 Mio. cells per cm² in DMEM containing 10% fetal FCS and 1% PenStrep. At 70% confluence, cells were switched to starving medium overnight. Cells were then treated with DEX (100 nM) for 48 hours. After removal of cell culture media, cells were lysed in 1 × RIPA lysis buffer (Cell Signaling Technology, Danvers, USA) and frozen at −80 °C. After thawing, samples were vortexed and centrifuged at 25,000 × g for 30 min at 4 °C. For Western Blot analysis, lysates were boiled for 5 min at 98 °C. Twenty μg of proteins were loaded on a 10% SDS-PAGE and transferred onto a 0.2 μm nitrocellulose membrane. After blocking for 1 h with 5% milk powder in Tris-buffered saline with 1% Tween-20 (TBS-T), membranes were incubated with antibodies against Wnt16 (1:500, Abcam, #ab109437) and GAPDH (1:2,000, Hytest, #5G4) overnight. After washing 3 times with TBS-T, incubation with an appropriate HRP-conjugated IgG antibody (1:2,000) for 1 h at RT followed. Proteins were visualized using the ECL Western-Blotting Substrate (Thermo Scientific). The chemiluminescence signal was recorded using a MF-ChemiBIS 3.2bio imager system (Biostep).

Statistical analysis.
Results are presented as means ± standard deviation (SD). Statistical evaluations were performed using a Student's t-test, one-way or two-way ANOVAs (with Dunn's multiple comparison corrections) as indicated in the figure legends. P-values ≤ 0.05 were considered statistically significant. Data availability. No large datasets were generated or analysed during this study. All other datasets analysed during this study are available from the corresponding author on reasonable request.

Wnt16 is downregulated by glucocorticoids in osteoblasts. To investigate the effects of GCs on
Wnt16 in osteoblasts, we treated murine osteoblasts derived from bone marrow stromal cells (mBMSCs) with increasing concentrations of DEX for different time periods. DEX led to a dose-dependent reduction of Wnt16 expression, reaching a maximum at concentrations of 0.1 µM and 1 µM DEX, respectively (3.2-fold) (Fig. 1A). Furthermore, DEX suppressed Wnt16 expression in osteoblasts in a time-dependent manner with the strongest reduction after 48 h (2.5-fold) (Fig. 1B). The suppression of Wnt16 after DEX treatment was also apparent at protein level (Fig. 1C).
In order to further validate inhibition of Wnt16 expression by GCs we cultured calvaria of 3-5 day old mice with DEX for 24 h. Gene and protein expression analysis confirmed the inhibition of Wnt16 by GCs (Fig. 1D). DEX decreased Wnt16 mRNA levels and Wnt16 protein expression more than 2-fold than in the control.

Glucocorticoids reduce bone formation and Wnt16 expression in a model of GIO.
To assess the effects of GCs on Wnt16 expression in vivo, we treated wildtype mice with the clinically used GC prednisolone for 4 weeks. As expected, bone mineral density of femoral bone was significantly reduced by 25% (Fig. 2A). In line, prednisolone exposure inhibited bone formation by more than 70% as determined by dynamic histomorphometry following dual calcein labelling (Fig. 2B). Accordingly, gene expression levels of osteoblastic markers in the femur of prednisolone-treated mice revealed a more than 2-fold downregulation of alkaline phosphatase (Alpl) and osteocalcin (Bglap) (Fig. 2C). In addition, expression of Wnt16 mRNA was diminished by more than 70% in the femur of prednisolone-treated mice (Fig. 2C). A similar suppression of bone formation and Wnt16 expression was observed in mice that were treated for 2 weeks with a higher dose of prednisolone (data not shown). Thus, these data indicate that Wnt16 expression is inhibited by GCs in vitro and in vivo. Wnt16 regulation by glucocorticoids is mediated by the GR and requires intact dimerization of the GR. We next elucidated the underlying mechanisms of GC-induced Wnt16 regulation. Since GCs mediate their effects via the GR, we first co-treated osteoblasts with DEX and the GR-antagonist RU-486. Application of RU-486 completely abolished the DEX-induced reduction of Wnt16 expression (Fig. 3A). RU-486 alone had no effect on Wnt16 mRNA levels. To confirm the involvement of the GR in mediating GC-induced effects on Wnt16 expression, we treated osteoblasts derived from GR-deficient mice (GR −/− mice) with DEX. Whereas  DEX reduced Wnt16 expression in osteoblasts derived from wildtype mice, DEX failed to suppress Wnt16 in the absence of the GR (Fig. 3B). The GR alters gene expression via different modes of action, including binding as a homodimer to GREs in the promotor of GC-target genes and interaction as a monomer with DNA-bound transcription factors such as NF-κB, AP-1, IRF-3 or STAT5 46 . To further study the underlying mode of GR action in the regulation of Wnt16, we analysed mice with attenuated GR dimerization (GR dim mice 47 ) that show reduced DNA binding as revealed by genome wide studies 48 . GR dim mice were treated with prednisolone for 14 days. In contrast to wildtype littermate control mice, in which prednisolone treatment reduced Wnt16 expression in femoral bone, prednisolone did not reduce skeletal Wnt16 expression in GR dim mice (Fig. 3C). Thus, intact GR dimerization and likely DNA binding of the GR is required for mediating GC effects on Wnt16 expression.
As these results suggested an involvement of direct DNA-binding mechanisms of the GR, we next searched for transcriptional GR regulatory motifs in the promotor of Wnt16 with the help of online search tools. Using three different websites, we identified multiple putative GR binding sites. Further, we performed a luciferase dual-reporter assay to verify the regulation of Wnt16 promoter activity via GCs. Following lipid-mediated transfection of the Wnt16 promotor construct in the osteoblastic cell line MC3T3-E1, we treated the cells for 48 h with DEX. Luciferase activity was reduced up to 30% in DEX-treated cells (Fig. 3D) indicating that the GR directly binds GRE motifs in the Wnt16 promoter region.

Wnt16 rescues GC-induced suppression of bone formation. To assess the functional role of Wnt16
in GC-induced suppression of osteoblast function and bone formation, we treated osteoblasts with DEX and/ or recombinant Wnt16 and measured alkaline phosphatase (ALP) activity. As shown in Fig. 4A, ALP activity was reduced more than 2-fold upon DEX treatment, but addition of Wnt16 abrogated GC-induced suppression of osteoblast function. To further investigate the involvement of Wnt16 in GC-mediated inhibition of bone formation in vivo, we treated wildtype mice with DEX and recombinant Wnt16, respectively, for 3 weeks by subcutaneous injection over the calvaria and determined the bone formation rate by dual calcein labelling. Histomorphometric analysis revealed a strong reduction in bone formation in DEX-treated animals. The concomitant treatment with Wnt16, however, rescued the DEX-induced repression of bone formation (Fig. 4B,C). Thus, Wnt16 is involved in GC-induced suppression of bone formation in vivo.

Discussion
GCs are frequently used to treat chronic inflammation. However, long-term GC treatment causes severe side effects on the skeleton and leads to an increased fracture risk mainly due to the sustained suppression of bone formation 11 . Recently Wnt16 has been linked to cortical bone thickness and fracture risk [28][29][30][31]33 and has been recognized as a positive regulator of bone mass and strength 32,[34][35][36] . In this study, we investigated whether Wnt16 is a target of GC action. We show that GCs suppress Wnt16 expression in vitro and in vivo and that Wnt16 exposure counteracts the inhibitory effects of GCs on osteoblast function. Using primary mouse osteoblasts, we found that DEX reduces Wnt16 expression in a time-and dose-dependent manner. Moreover, DEX treatment also reduced Wnt16 protein levels in osteoblasts and calvaria organ cultures. These results are in line with other studies that have shown that increasing GC doses oppose Wnt signalling thereby attenuating osteoblast differentiation and function. In a study with murine osteoblasts, high doses of corticosterone reduced the expression of Wnt7b and Wnt10 42 . In contrast to Wnt7b and Wnt10b, Wnt16 was not modulated by GCs in a biphasic fashion as also low doses negatively affected Wnt16 expression.
In order to investigate the effects of GCs on Wnt16 in vivo, a GIO mouse model was used. Treatment of mice with prednisolone for four weeks caused a reduction of BMD and inhibition of bone formation as described previously 13,18 . In this model, Wnt16 expression was reduced in the bone tissue of GC-treated mice, and coincided with the reduced expression of the pro-osteoblastic markers alkaline phosphatase and osteocalcin.
After showing that Wnt16 is a novel target of GC action in bone, we analysed the underlying molecular mechanism. GCs bind to the GR and mediate their function by binding as a homodimer to palindromic DNA sequences (so called GRE, glucocorticoid response elements), binding to GRE half-sites as a monomer or binding as a monomer to DNA-bound transcription factors and thereby modulate target gene transcription 49 . Whereas for most anti-inflammatory GC effects GR dimerization is essential 50-52 , suppression of bone formation is proposed to be mediated by GR monomers interacting with the transcription factor AP-1 13 . Our data show that the suppression of Wnt16 expression by GCs is dependent on GR dimerization and DNA binding of the GR complex. This is in contrast to GC-induced suppression of other osteoblastic genes and cytokines such as Runx2, Col1a1 or Il11, which is likely mediated by monomeric GR transrepression mechanisms 13 . However, a distinct mechanism of GC-induced direct transrepression via binding of GR to evolutionary-conserved cis-acting negative GC-responsive elements (IR nGREs) was described by Surjit et al. 53 . In contrast to tethered transrepression, nGRE-mediated gene repression is abolished by the GR dim mutation. Moreover, co-treatment with RU486 prevented GC-induced IR GRE-mediated direct transrepression 53 . Thus, it is tempting to speculate that the Wnt16 gene may also contain IR nGREs, especially as also other GC-regulated Wnt genes such as Wnt1, Wnt5a and Wnt10b contain putative IR nGREs 53 and the number of genes regulated by nGREs has been proposed to be higher than previously thought. Additional evidence for the requirement of DNA-binding mechanisms of the GR is provided by in silico analyses, which identified GR transcriptional regulatory motifs in the Wnt16 promoter region, and a promoter assay, which revealed a suppression of Wnt16 promoter activity after GC treatment. Finally, considering that Wnt16 expression was already regulated after 3 hours, our data suggest that Wnt16 is a primary response gene of GC signalling via direct GR DNA-binding mechanisms in osteoblasts. Further studies are necessary to explore the exact GR binding motifs within the putative negative GREs on the Wnt16 promoter. After having established the regulatory role of GCs on Wnt16 expression, we addressed whether Wnt16 is able to rescue GC-induced suppression of bone formation. Interestingly, single treatment of both, primary osteoblasts and mice did not significantly alter osteoblast function, indicating that short-term treatment with Wnt16 may not be sufficient to enhance osteoblast function in the same way as long-term exposure as in WNT16 overexpressing mice 36,37 . Similarly, osteoblasts derived from Wnt16KO mice did not show alterations in their differentiation capacity 34 However, co-treatment of osteoblasts with DEX and recombinant Wnt16 abrogated the GC-induced suppression of osteoblast function in vitro and in vivo. This novel finding suggests that restoring Wnt16 expression can at least partly abolish GC-induced suppression of bone formation in vitro and in vivo. One limitation of this study is the local treatment with DEX and Wnt16, despite GIO being a systemic disease. However, considering that Wnt ligands act in a local, paracrine fashion, we chose the local treatment of calvarial bone with DEX and Wnt16. This approach has also been taken in a previous study, in which inflammation-induced calvarial bone loss was rescued using recombinant Wnt16 34 . Future studies using Wnt16 overexpressing mice are underway that may provide more insights into the systemic rescue of GIO via Wnt16 54,55 .
Collectively, our data suggest that Wnt16 is a direct target of GC action and that restoration of Wnt16 alleviates GC-suppressed bone formation. Thus, Wnt16 may represent a novel target for therapeutic intervention in GIO.