Farnesoid X receptor (FXR) is a nuclear receptor that functions as a bile acid sensor controlling bile acid homeostasis. We investigated the role of FXR in regulating bone metabolism. We identified the expression of FXR in calvaria and bone marrow cells, which gradually increased during osteoblastic differentiation in vitro. In male mice, deletion of FXR (FXR−/−) in vivo resulted in a significant reduction in bone mineral density by 4.3% to 6.6% in mice 8 to 20 weeks of age compared with FXR+/+ mice. Histological analysis of the lumbar spine showed that FXR deficiency reduced the bone formation rate as well as the trabecular bone volume and thickness. Moreover, tartrate-resistant acid phosphatase (TRACP) staining of the femurs revealed that both the osteoclast number and osteoclast surface were significantly increased in FXR−/− mice compared with FXR+/+ mice. At the cellular level, induction of alkaline phosphatase (ALP) activities was blunted in primary calvarial cells in FXR−/− mice compared with FXR+/+ mice in concert with a significant reduction in type I collagen a1(Col1a1), ALP, and runt-related transcription factor 2 (Runx2) gene expressions. Cultures of bone marrow–derived macrophages from FXR−/− mice exhibited an increased number of osteoclast formations and protein expression of nuclear factor of activated T cells, cytoplasmic 1 (NFATc1). In female FXR−/− mice, although bone mineral density (BMD) was not significantly different from that in FXR+/+ mice, bone loss was accelerated after an ovariectomy compared with FXR+/+ mice. In vitro, activation of FXR by bile acids (chenodeoxycholic acid [CDCA] or 6-ECDCA) or FXR agonists (GW4064 or Fexaramine) significantly enhanced osteoblastic differentiation through the upregulation of Runx2 and enhanced extracellular signal-regulated kinase (ERK) and β-catenin signaling. FXR agonists also suppressed osteoclast differentiation from bone marrow macrophages. Finally, administration of a farnesol (FOH 1%) diet marginally prevented ovariectomy (OVX)-induced bone loss and enhanced bone mass gain in growing C57BL/6J mice. Taken together, these results suggest that FXR positively regulates bone metabolism through both arms of the bone remodeling pathways; ie, bone formation and resorption. © 2013 American Society for Bone and Mineral Research.
Farnesoid X receptor (FXR, also known as NR1H4) is an adopted member of the nuclear receptor superfamily of ligand-activated transcription factors, originally known as a receptor of high concentrations of farnesol (FOH), an intermediate in the mevalonate pathway. Bile acids are identified as physiological ligands of FXR, which potently activate the transcription of FXR target genes.[2, 3] Chenodeoxycholic acid (CDCA), a hydrophobic bile acid, is one of the strong agonist of FXR, whereas hydrophilic ursodeoxycholic (UDCA) and muricholic acids are inactive in activating this receptor. Intracellular increase of bile acid concentrations leads to transcriptional activation of FXR by forming a heterodimer with the retinoid X receptor (RXR) and binding to the inverted repeat-1 (IR-1) responsive element.[4-7] This binding subsequently leads to the regulation of gene expression for a number of biosynthetic enzymes involved in bile acid synthesis, detoxification, and excretion. In this manner, FXR functions as a bile acid sensor, thereby playing a coordinative role in bile acid homeostasis. In addition, FXR also plays a crucial role in glucose metabolism and insulin sensitivity, and lipid metabolism and atherosclerosis. Therefore, FXR modulators have been extensively studied as a new therapeutic target for numerous metabolic disorders.
A growing body of evidence shows that several nuclear receptors regulate skeletal homeostasis through their direct effects on osteoblasts or osteoclasts. Haploinsufficiency of peroxisome proliferator activator γ (PPARγ) resulted in increased bone mass by stimulating osteogenesis from marrow progenitors and we have shown that in vivo overexpression of PPARγ in osteoblasts using the collagen type 1 promoter resulted in reduced bone mass gain in male and accelerated bone loss after an ovariectomy in female mice. In addition, Remen and colleagues has demonstrated that activation of the liver X receptor (LXR) inhibits the receptor activator of nuclear factor kappa B ligand (RANKL)-induced osteoclast differentiation. Furthermore, small heterodimer partner (SHP), which is known as one of the target genes of FXR, has been shown to stimulate bone formation through enhancing Runt-related transcription factor 2 (Runx2) transcriptional activity and augmenting osteoblast differentiation.
Given that FXR regulates diverse metabolic pathways and other nuclear receptors have important roles in the regulation of bone homeostasis, it would be reasonable to expect that FXR can also have a role in the regulation of bone metabolism. Although FXR is mainly expressed in the liver and intestines, which are the classical target organs of bile acid, recent studies have shown that its expression has also been identified in the kidneys, adrenal glands,[1, 16] bone marrow stromal cells, and SaOS2 osteoblast-like cells. Moreover, activated by CDCA, FXR has been shown to stimulate Runx2-derived osteoblastic differentiation while reciprocally inhibiting adipogenesis in SaOS2 cells in vitro. However, the functional role of FXR in bone mass regulation in vivo has not been studied. In this study, we examined the endogenous expression of FXR in skeletal tissues and analyzed its functional role using FXR-deficient (FXR−/−) mice.
Subjects and Methods
Anti-mouse FXR and nuclear factor of activated T cells, cytoplasmic 1 (NFATc1), and axin antibody were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Anti-mouse extracellular signal-regulated kinase (ERK), phosphorylated ERK (pERK), and β-catenin antibody were purchased from Cell Signaling (Boston, MA, USA). The bile acid, CDCA, 6α-ethyl-chenodeoxycholic acid (6-ECDCA), bone morphogenic protein 2 (BMP-2), and macrophage colony-stimulating factor (M-CSF) were purchased from R&D Systems (Minneapolis, MN, USA). RANKL was purchased from Peprotech (Rocky Hill, NJ, USA). The synthetic FXR agonists, GW4064 and fexaramine, were purchased from Tocris Bioscience (Ellisville, MI, USA). Wnt-3a conditioned media (CM) were prepared from L-Wnt-3a cells as reported. All reagents for quantitative PCR gene expression assays were from Applied Biosystems (Carlsbad, CA, USA). Cell culture supplies were obtained from Invitrogen (Carlsbad, CA, USA) or Sigma-Aldrich (St. Louis, MO, USA), unless otherwise indicated.
All animal experiments were performed with approval from the Institutional Animal Care and Use Committee of Seoul National University (approval #SNU-110504-2). FXR−/− mice were generously provided by F. Gonzalez (NIH, Bethesda, MD, USA) and backcrossed to C57BL/6J background for 10 generations. Genotyping of the mice was performed as described. Littermate C57BL/6J wild-type (WT) (FXR+/+) mice were used as controls.
Four-week-old to 20-week-old FXR−/− male mice were fed a standard chow diet and age- and sex-matched FXR+/+ mice were fed either a standard chow diet or a high fat Western diet (HFD) containing 42% fat calories (saturated fat from anhydrous milk fat) + 0.2% cholesterol (TD88137; Harlan Teklad, Madison, WI, USA). After 16 weeks, 6-hour fasting total cholesterol or triglycerides were measured, and 16-hour fasting intraperitoneal glucose tolerance test (IPGTT) with glucose was performed as described. For in vivo treatment of FXR agonist, 12-week-old C57BL/6J WT female mice underwent an ovariectomy (OVX) or sham operation followed by treatment with standard chow or a FOH (1%)-containing diet, immediately after the operation.
Quantitative real-time PCR
Total RNA was isolated from the tissues of mice or cells using TRIzol reagent (Invitrogen) and subsequently reverse transcribed using the Reverse Transcription System kit (Promega, Madison, WI, USA). cDNA was amplified on an ABI Prism 7700 Sequence Detection System (Applied Biosystems) with TaqMan gene expression-specific primer/probes (Applied Biosystems) or SYBR Green PCR technology. Rodent glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (Applied Biosystems) was used as an endogenous control. Relative quantification of the data was carried out using the standard curve method or the comparative cycle threshold (CT) method. Primers used are listed in Supplemental Table 1. The relative gene expressions were measured to compare the ratios to the β-actin CT.
FXR+/+ calvarial cells were grown on glass coverslips and treated with CDCA (100 μM) or vehicle for 48 hours. FXR−/− calvarial cells were used without treatment. Cells were fixed with 4% paraformaldehyde after washing with ice-cold PBS and permeabilized with 0.05% Triton X-100 for 15 minutes. Cells were then blocked with 5% goat serum/1% bovine serum albumin at room temperature followed by 1 hour incubation with fluorescein anti-mouse FXR antibody (sc-13063; Santa Cruz Biotechnology). The secondary antibody (A-11012; Invitrogen) alone, omitting the primary antibody, was used as a negative control. Cells were visualized under a confocal scanning laser microscope using 400× objectives.
Radiological analysis of the bone
Bone mineral density (BMD, g/cm2) of the whole body excluding the head regions was measured using the dual-energy X-ray absorptiometry (DXA) instrument PIXIMUS (GE Lunar, Madison, WI, USA) adapted for measurements of small animals. The animals were placed in a prone position under anesthesia induced by intraperitoneal injections of xylazine (2.2 mg/kg; Rompun, Bayer, Monheim, Germany) and tiletamine/zolazepam (6.0 mg/kg Zoletil 100; Virbac, Carros Cedex, France). A phantom was scanned daily for quality control.
Three-dimensional measurement was performed by micro–computed tomography (µCT). Femurs obtained from 16-week-old mice were dissected free of soft tissue, fixed overnight in 70% ethanol and analyzed with a µCT scanner and associated analysis software (model 1076; Skyscan, Antwerp, Belgium) with a voxel size of 9 µm. Image acquisition was performed at 35 kV of energy and 220 A of intensity. The threshold was set to segment the bone from the background and the same threshold setting was used for all samples.
Mineralized bone histology and bone histomorphometry
All histological analyses were performed using 16-week-old FXR−/− and FXR+/+ mice. Mice were injected with calcein (16 mg/kg body weight) intraperitoneally at 7 and 3 days prior to sacrifice. Whole spines were isolated, cleaned and fixed in 4% paraformaldehyde at 4°C overnight. Undecalcified lumbar vertebrae were embedded in destabilized methyl methacrylate resin, after which 5-µm-thick sections were stained with van Gieson and von Kossa reagents. Static and dynamic histomorphometric analyses were conducted on the lumbar vertebrae using the Bioquant program (Bio-Quant Inc., San Diego, CA, USA). To analyze the osteoclast numbers and fractional surface, decalcified tibias were embedded in paraffin and serial sections were prepared following staining for TRACP activity.
In vitro osteogenic differentiation study
Primary calvarial cells were isolated from FXR+/+ or FXR−/− 4-day-old mouse pups with collagenase A digestion (Roche Diagnostics, Indianapolis, IN, USA) in α modified essential medium (α-MEM) at least four sequential times for 15 minutes each time. The first digestion was discarded and the second to fourth digestions were combined and the cells were expanded and plated (50,000/cm2) with α-MEM including 10% fetal bovine serum (FBS). Murine mesenchymal cells C3H10T1/2 (American Type Culture Collection, Manassas, VA, USA), which are pluripotent cells, were grown in DMEM containing 10% FBS. For the osteogenic differentiation study, cells were plated in 12-well plates and mineralization was initiated at confluence with osteogenic medium including 50 μg/mL L-ascorbic acid and 10 mM β-glycerophosphate (OM). For measurement of RANKL and osteoprotegerin (OPG) mRNA, the cells were treated with 10 nM of 1,25-dihydroxyviatmin D3 [1,25(OH)2D3] for 24 hours before harvest. For activation of FXR, natural (CDCA, 6-ECDCA) or synthetic (GW4064, fexaramine) ligands were added with OM during differentiation. Media was replaced with fresh media every 3 days and maintained for 10 days.
ALP assay and staining
To assess the ALP activity, cells were washed three times with ice-cold PBS (pH 7.4) and scraped immediately. Enzyme activity assays were performed in assay buffer (10 mM MgCl2 and 0.1 M alkaline buffer, pH 10.3) containing 10 mM p-nitrophenylphosphate as a substrate. The absorbance was read at optical density (OD) 405 nm. Relative ALP activity was defined as millimoles (mmol) of p-nitrophenol phosphate hydrolyzed per minute per milligram (mg) of total protein. ALP staining was carried out using an ALP kit according to the manufacturer's instructions (Promega, Southampton, UK).
Whole bone marrow culture and osteoclastogenesis
Whole bone marrow cells were flushed from 2-month-old FXR+/+ or FXR−/− mice. To induce osteoclast differentiation, unfractionated bone marrow cells were plated in a 96-well culture plate at a density of 2 × 104 cells/well and cultured in α-MEM with 10% FBS in the presence of 10 nM 1,25(OH)2D3. After 3 days of culturing, 90% of the medium was replaced, and 1,25(OH)2D3 was added again. After 6 days, the cells were fixed and stained for TRACP, a marker of osteoblast differentiation, according to the manufacturer's instructions (Sigma-Aldrich). Cells that had more than five nuclei under a light microscope were counted.
Bone marrow–derived or spleen-derived macrophage culture and osteoclastogenesis
Total bone marrow or spleen cells from FXR+/+ or FXR−/− mice were cultured for 1 day in α-MEM containing 10% FBS and M-CSF (10 ng/mL). Nonadherent cells were collected and further cultured with M-CSF (30 ng/mL) in α-MEM containing 10% FBS. After 6 days, adherent cells were cultured with M-CSF (30 ng/mL) and RANKL (100 ng/mL) to generate osteoclasts. For FXR activation, CDCA, 6-ECDCA, GW4064, or fexaramine were treated continuously during differentiation. Fresh mediums were replaced every two other days. After 5 to 7 days, cells were fixed and stained for TRACP and TRACP+ multinucleate cells with more than 5 nuclei were counted.
Transfection and reporter assay
HEK 293 (human embryonic kidney) cells were seeded into 24-well plates at 50,000 cells/well; 12 hours later, 0.1 µg of estrogen response element Luciferase (ERE-Luc) reporter plasmid, together with expression vectors for estrogen receptor β (ERβ) (pCS4-ERβ 0.4 μg) ± FXR (CDM8-FXR 0.4 μg) in the presence or absence of estradiol (1 × 10−8 M) was added. After 24 hours, cell lysates were collected using the Promega Luciferase assay system, and the reporter activity was measured by a luminometer (Lumat LB 9507; Berthold, Bad Wildbad, Germany) as described.
Western blot analysis
Total cell lysates were isolated, separated by SDS-PAGE, and transferred onto Immobilon P membranes (Millipore, Billerica, MA, USA). The membranes were blocked with 5% nonfat-milk in PBS-T (phosphate buffered saline, 0.1% Tween 20), and then immunostained with relevant antibodies or β-actin antibody (Santa Cruz Biotechnologies) and immunoreactivity was detected with enhanced chemiluminescence reagent. The relative band intensity was assessed by densitometric analysis of the digitalized autographic images using Scion Image software (Scion Corporation, Frederick, MD, USA).
All data are presented as the mean ± SEM. Unpaired t tests were performed using GraphPad Instat Software (GraphPad Software, San Diego, CA, USA). A significance level of 5% was assigned to all tests.
FXR is expressed in bone tissues
To assess the functional role of FXR in bone metabolism, the endogenous gene expression of FXR was examined in various murine tissues and progenitor cells by real-time PCR. As shown in Fig. 1A, FXR mRNA was expressed in osteoblast rich-calvaria (4% of liver) and bone marrow (1% of liver) as well as the spleen (9% of liver) and fat tissues (70% of liver). In addition, mesenchymal C3H10T1/2 cells and osteoblastic MC3T3.E1 cells also expressed FXR mRNA (Fig. 1B) and the expression level was higher in MC3T3.E1 cells than in C3H10T1/2 cells. When we induced osteoblastic differentiation of C3H10T1/2 cells by OM including ascorbic acid and β-glycerophosphate, the mRNA of FXR was significantly increased compared to the cells treated with the control medium (Fig. 1C). To further verify the protein expression of FXR in osteoblasts, primary calvarial cells were analyzed by immunofluorescence staining. As shown in Fig. 1D, the calvarial cells showed clear expression of FXR mostly in the nuclear area. When the cells were treated with CDCA, a known ligand of FXR, the nuclear expression of FXR was significantly increased compared to the vehicle treatment in calvarial cells. As a control, we used calvarial cells from FXR−/− mice which showed no expression of FXR regardless of the CDCA treatment (Fig. 1D). Taken together, these results suggest that FXR might have a physiological role in osteogenic differentiation.
Whole-body FXR deletion decreased bone mass in male mice
To investigate the role of FXR in regulating bone mass in vivo, we next studied the skeletal phenotypes of FXR−/− mice. A previous study showed that FXR−/− mice have elevated serum bile acid, cholesterol, and triglyceride levels but develop normally and outwardly identical to WT animals. In males, the body weight and total body fat content analyzed by DXA were similar between the 8-week-old to 20-week-old FXR−/− and FXR+/+ mice (data not shown). However, the BMD of the FXR−/− mice was 4.3% to 6.6% lower compared with the FXR+/+ mice in these periods (Fig. 2A). The BMC of the FXR−/− mice also had corresponding reductions of 8.6% to 12.5% compared with the FXR+/+ mice (Fig. 2B). A gene-dosage effect on the BMD at the femur was observed, with the heterozygous mice (FXR+/−) exhibiting an intermediate BMD between the FXR+/+ and FXR−/− mice (Fig. 2C).
Because FXR−/− mice have been known to have dyslipidemia, we next asked whether the possibility that the low BMD in FXR−/− mice may be a result of indirect effects from persistent hyperlipidemia rather than the direct effect of FXR deficiency in these animals. We therefore decided to compare the skeletal phenotypes of three groups of mice: FXR−/− mice fed a standard chow diet (Chow); FXR+/+ mice fed a high-fat diet (HFD), and FXR+/+ mice fed a standard chow diet. The serum levels of cholesterol and triglyceride were 126 ± 11 mg/dL and 72 ± 14 mg/dL in the FXR+/+ mice on HFD and 108 ± 6 mg/dL and 84 ± 14 mg/dL in the FXR–/– mice fed a standard chow diet, respectively (Fig. 2D). Additionally, the IPGTT (2 g/kg) showed that both FXR+/+ on HFD and FXR−/− on a standard chow diet had a similar degree of glucose intolerance whereas chow-fed FXR+/+ mice had normal glucose tolerance (Fig. 2E). When we analyzed the BMD of these animals, the femoral BMD of the FXR−/− mice fed a standard chow diet was significantly lower than that of FXR+/+ mice on HFD (0.055 ± 0.002 versus 0.060 ± 0.001 g/cm2, p = 0.039; Fig. 2F). These results indicate that although a high serum level of lipids and insulin resistance may in part contribute to the low bone mass, the deletion of FXR in itself has an independent role in producing the skeletal phenotypes of FXR−/− mice.
Whole-body FXR deletion decreased the bone formation rate in male mice
Consistent with the DXA results, μCT analyses of femurs at 16 weeks of age also showed a reduction in bone volume (BV/TV) by 35.1% (p < 0.05), in trabecular number (Tb.N) by 25.0% (p < 0.05), and in trabecular thickness (Tb.Th) by 14.8% (p < 0.05) in the FXR–/– mice compared with the FXR+/+ mice (Fig. 3A). Analysis of the cortical bone at the mid-diaphysis of the femur also showed a significant reduction in the average cortical thickness by 8.6% (p < 0.05), although there was no significant difference in the cortical bone area (Fig. 3A).
To further characterize the bone phenotypes in the FXR−/− mice, histomorphometric analyses were performed on the lumbar vertebrae of 16-week-old male mice. Consistent with the μCT analysis results, analysis of the mineralized bones of the lumbar spine stained with von Kossa confirmed the reduced fractional bone area and trabecular thickness (Tb.Th) to be about 22.9% (p < 0.05) and 25.3% (p < 0.05), respectively in the FXR−/− mice compared with the FXR+/+ mice (Fig. 3B) although the trabecular number and spacing were not significantly different. Moreover, a calcein double-labeling study, an assessment of the dynamic parameters of bone formation, showed that both the bone formation rate (BFR) and mineral apposition rate (MAR) were significantly reduced by 19.1% (p < 0.05) and 13.6% (p < 0.05), respectively (Fig. 3C). Interestingly, both the osteoclast number (Oc.N/BS, 26.5%; p < 0.05) and osteoclast surface (OC.S/BS, 71%; p < 0.05) were also significantly increased in the FXR–/– mice compared with the FXR+/+ mice (Fig. 3D).
FXR depletion leads to suppressed osteogenesis
In order to gain more insight into the cellular mechanism for the reduction of bone mass in FXR−/− mice, ex vivo culture of osteoblast-rich calvarial cells was done. As shown in Fig. 4A, the ALP activity of the calvarial cells from the FXR−/− mice decreased by 50% to 55% compared with the cells from the FXR+/+ mice when cultured in OM.
In addition, the mRNA expression levels of osteoblast-specific genes including Col1a1, ALP, and Runx2 were significantly decreased in the calvarial cells from the FXR−/− mice compared with that of the FXR+/+ mice (Fig. 4B). Of note, the expression levels of ALP and Runx2 in FXR+/− mice were intermediate between the FXR+/+ and FXR−/− mice (Fig. 4B). However, we were unable to demonstrate a significant change in the DNA binding activity of the Runx2 protein with the chromatin immunoprecipitation (ChIP) assay in primary calvarial cells from FXR−/− mice compared with that of the FXR+/+ mice (Supplementary Fig. 1).
Interestingly, the mRNA expression of PPARγ in response to rosiglitazone treatment was also significantly decreased in FXR−/− cells compared to FXR+/+ cells, although the expression levels were similar at the basal state (Fig. 4C), indicating that the attenuated osteogenic differentiation by FXR deprivation may not be accompanied by reciprocal enhancement of adipogenesis.
FXR depletion enhanced osteoclast differentiation
Because FXR−/− mice showed an increased osteoclast number in the femur (Fig. 3D), we next investigated whether the depletion of FXR can affect the potential to differentiate to osteoclasts in vitro. Whole bone marrow cells isolated from FXR+/+ or FXR−/− mice were first cultured and then osteoclast differentiation was induced using 1,25(OH)2D3. As shown in Fig. 4D, the whole bone marrow cells from the FXR−/− mice showed significantly increased osteoclast formation (approximately twofold, p < 0.05) compared with that of the FXR+/+ mice. Next, we examined the expression of RANKL and OPG from calvarial cells. The expression level of RANKL (approximately fourfold) and the RANKL/OPG ratio (∼2.3-fold) were significantly increased in calvarial cells from the FXR−/− mice compared to that of the FXR+/+ mice in the presence of 10 nM 1,25(OH)2D3, whereas that of OPG showed no difference (Fig. 4E). This result indicates that the depletion of FXR in a stromal/osteoblast lineage has resulted in an enhanced ability to support osteoclastogenesis.
Because FXR has also been reported to be expressed in macrophages cells, we next examined the endogenous expression of FXR in osteoclasts. As shown in Fig. 4F, FXR mRNA was expressed in both the bone marrow–derived monocyte/macrophages (BMMs) and osteoclasts. In addition, the culture of BMMs from the FXR−/− mice resulted in increased TRACP+ multinucleated osteoclast formation compared with that from the FXR+/+ mice in the presence of RANKL (Fig. 4G). RT-PCR analysis indicated that RANKL-dependent expression of TRACP, Calcitonin receptor, and Cathepsin K increased by FXR deletion (Fig. 4H). Furthermore, FXR deletion in spleen-derived macrophages significantly augmented RANKL-dependent NFATc1 induction (Fig. 4I), suggesting that FXR in itself may play an important role in the regulation of osteoclast differentiation from precursors.
Whole-body FXR deletion accelerated OVX-induced bone loss in female mice
In contrast to male mice, the BMD of FXR−/− female mice was not significantly different from that of FXR+/+ mice although the former was slightly lower than the latter (Fig. 5A). The lack of significant bone phenotypes in female mice led us to hypothesize whether estrogen has a role in the regulation of FXR-mediated signaling. We first studied whether the expression level of ER differs in calvaria cells of FXR−/− mice. As shown in Fig. 5B, the expression level of ERβ is significantly reduced in the FXR−/− cells whereas that of ERα is similar.
Next, we determined whether FXR signaling can alter the transcriptional activity of estrogen using a promoter reporter assay. Transfection of HEK 293 cells with the ERE-Luc reporter and ERα showed basal reporter activity, which was significantly upregulated by treatment with 17β-estradiol (1 × 10−9 M) as expected (Fig. 5C). Interestingly, cotransfection with the FXR expression vector under this condition significantly enhanced the luciferase activity by ∼1.8-fold, indicating the existence of crosstalk between FXR and ER signaling. Finally, to investigate the interaction between estrogen and FXR in vivo, we performed an OVX or sham operation on 12-week-old female FXR−/− mice or their WT littermates, and measured the BMD at baseline and 4 and 8 weeks after the operation. Interestingly, ovariectomy-induced bone loss was significantly accentuated in the FXR−/− mice from 4 to 8 weeks after the operation (at 8 weeks, −4.2% ± 0.9% versus −10.9% ± 1.2%, p < 0.01; Fig. 5D), indicating that the effects of FXR deletion were less prominent in female mice due to the presence of estrogen, and the phenotypes became more obvious after estrogen withdrawal in vivo.
FXR agonist enhanced osteogenesis in vitro
To further investigate whether activation of FXR regulated bone metabolism in vitro, we first studied the effect of FXR activation on osteogenesis using murine mesenchymal C3H10T1/2 cells. As shown in Fig. 6A, treatment with CDCA (50 or 100 μM), which belongs to natural bile acids with FXR agonist properties, resulted in a twofold to fivefold increase in ALP activities compared with the vehicle treatment in a dose-dependent manner. In addition, 6-ECDCA, a semisynthetic bile acid having potent FXR agonist activity,[24, 25] and two pure FXR agonists, GW6064 and fexaramine, also significantly increased the ALP activity in a dose-dependent manner (Fig. 6A). Furthermore, expression of osteoblast marker genes, such as Col1a1, ALP, BSA, Runx2, and Osterix, were also significantly increased by treatment with CDCA, 6-ECDCA, GW4064, or fexaramine (Fig. 6B). Taken together, these results clearly suggest that FXR-mediated signaling is positively regulating osteoblast differentiation. We next studied the signaling pathways underlying the enhancement of osteoblast differentiation mediated by bile acids and FXR agonists. Because there was no significant change in the DNA binding activity of the Runx2 protein, we studied the possibility that signaling pathways other than Runx2 are also involved in the FXR-mediated enhancement of osteoblast differentiation. As potential signaling pathways, we examined BMP-induced mitogen-activated protein kinase (MAPK) and Wnt/β-catenin signaling. When C3H10T1/2 cells were cultured in the presence of BMP-2, the phosphorylation of ERK was significantly increased (Fig. 6C, D), whereas that of p38 and JNK was not (data not shown). Cotreatment of the cells with GW4064 (at 10 minutes; Fig. 6C), CDCA or fexaramine (at 15 minutes; Fig. 6D) significantly increased the phosphorylation of ERK. Next, the induction of Wnt signaling in C3H10T1/2 cells by treatment with Wnt-3a conditioned medium (CM) resulted in a significant increase in the abundance of β-catenin in the Western blot analysis. Cotreatment with CDCA, GW4064, or fexaramine further increased the abundance of the β-catenin protein levels (Fig. 6E). Moreover, the addition of each of these agonists alone was also able to significantly increase the β-catenin protein levels (last three columns; Fig. 6E). The induction of β-catenin protein occurred at 3 hours after treatment, which peaked at 12 hours (Supplemental Fig. 2A). In addition, CDCA, GW4064, or fexaramine treatment also upregulated the protein levels of Axin2, one of the downstream molecules of Wnt signaling that regulates the stability of β-catenin (Supplemental Fig. 2B). Taken together, our results suggest that induction of ERK and β-catenin signaling may also contribute to the FXR-mediated enhancement of osteoblast differentiation.
FXR depletion enhanced osteoclast differentiation
Because FXR depletion not only suppressed osteogenesis but also enhanced osteoclast differentiation, we next examined the effects of FXR activation on RANKL-induced osteoclast differentiation. As shown in Fig. 7A, a culture of BMMs in the presence of M-CSF and RANKL resulted in the formation of TRACP+ multinucleated cells in 4 to 5 days after culturing as expected (Vehicle). Interestingly, the addition of CDCA, 6-ECDCA, GW4064, or fexaramine under this condition significantly reduced the number of TRACP+ multinucleated cells (Fig. 7A). In addition, the expression levels of genes related to osteoclast function, including TRACP, Calcitonin receptor, and Cathepsin K (Fig. 7B), were significantly decreased by treatment with these reagents, consistent with the results of osteoclast differentiation. Furthermore, RANKL-induced expression of NFATc1 (Fig. 7C) and c-fos (Fig. 7D) was also significantly decreased. Taken together, these results suggest that activation of FXR suppresses osteoclastogenesis not only by regulating the expression of RANKL and OPG from stromal cells/osteoblasts but also by directly regulating RANKL-dependent osteoclast differentiation.
Treatment with FOH increased bone mass in vivo
Finally, to confirm the positive effect of FXR agonist in bone mass regulation in vivo, we administered FOH (1%) to 12-week-old female C57BL/6J mice after performing either a sham operation or OVX. After OVX, an average of 2.8% of BMD loss was observed in the chow-fed group. However, as shown in Fig. 7E, administration of FOH (1%) for 5 weeks marginally attenuated the OVX-induced bone loss compared with the chow diet group (ΔBMD, −1.8% versus −2.8%, p = 0.05). Notably, the FOH diet significantly enhanced bone mass gain compared with the mice fed a standard chow diet (ΔBMD, 9.2% versus 7.2%, p = 0.03; Fig 7E) in the sham operation group. In addition, serum N-terminal propeptide of type I collagen (P1NP) showed a significant increase in the FOH group compared to the chow group (Fig 7F). These data validate the skeletal phenotypes of the FXR–/– mice and suggest the potential of FXR regulation as a therapeutic target for metabolic bone diseases.
In this study, we have shown that in vivo deletion of FXR resulted in a significant decrease in bone mass in mice, which appears to be a result from reduced osteogenesis and increased osteoclast differentiation. Conversely, in vitro treatment with bile acids or FXR agonists enhanced osteogenesis in mesenchymal progenitors and suppressed osteoclastogenesis in BMMs. Furthermore, administering a FOH-rich diet showed weak but significant anabolic effects on bones in vivo.
In this study, we have demonstrated that endogenous mRNA expression of FXR was observed in bone marrow and calvarial cells. Although the expression of FXR in osteoblast was only recently documented, the uptake and accumulation of the secondary bile acid salt, sodium deoxycholate (DC), was initially suggested in the MG63 osteosarcoma cell line. In that study, the authors demonstrated that bone cells derived from DC were originally released from the liver and eventually taken up by osteoblasts, suggesting a possible link between bile acid and bone metabolism.
Interestingly, the expression level of FXR was higher in preosteoblastic MC3T3-E1 cells than that of mesenchymal C3H10T1/2 cells. In addition, a culture of C3H10T1/2 cells with OM containing L-ascorbic acid plus β-glycerophosphates led to a gradual increase in the FXR expression level. These results strongly suggest that FXR may have a role in osteogenic differentiation of mesenchymal cells. Furthermore, the increased expression of FXR in calvarial cells in response to CDCA in our study indicates that the osteoblast FXR is functioning as a genuine hormone sensor and a specific bile acid exerts a positive feedback effect on its receptor.
Using an FXR−/− mice model, we have clearly shown evidence that in vivo deletion of FXR results in a significant decrease in bone mass through decreased bone formation. First, FXR−/− mice exhibited a decreased BMD by 4.3% to 6.6% from 8 to 20 months, indicating that the deletion of FXR inhibited bone mass gain in this growing phase with a gene dosage effect because the FXR+/− mice showed an intermediate BMD. Second, histomorphometric analysis showed a decreased bone formation rate and bone volume in FXR−/− mice. Third, at the cellular levels, calvarial cells from FXR−/− mice exhibited significantly decreased ALP activities. Finally, calvarial cells from the FXR−/− mice showed decreased expression of the osteoblast-specific transcription factors, Runx2 and Osterix, and osteoblast marker genes including type 1 collagen, ALP, and osteocalcin. Taken together, our results now identify that signaling through FXR may have an essential role in the regulation of osteogenesis and bone mass in vivo.
The reduced BMD in FXR−/− mice does not appear to be an indirect phenomenon secondary to the high lipid levels or insulin resistance because the BMD of FXR−/− mice was significantly lower than that in FXR+/+ mice fed an HFD that had comparable serum cholesterol and triglyceride levels and similar glucose tolerance.
The positive regulation of bone formation by FXR in vivo was further validated by the increased osteoblast differentiation of mesenchymal cells by treatment with CDCA and 6-ECDCA as a natural and semisynthetic bile acid, respectively,[24, 25] in vitro. Our results are in agreement with a previous study that demonstrated a significant increase in ALP activity by treatment with CDCA in bone marrow stromal cells. In addition, we also confirmed the positive effects of two pure FXR agonists, GW4064 and fexaramine, on osteoblastic differentiation because both CDCA and 6-ECDCA also function as agonist for TGR5, a member of the rhodopsin-like superfamily of G-protein coupled receptor, which also participates in transducing signals from bile acids. Therefore, together with the suppressed osteoblastic differentiation potential of FXR−/− calvarial cells, our results suggest that FXR in itself could have an important role in osteoblastic differentiation.
Regarding the molecular mechanisms underlying the pro-osteogenic role of FXR, Id Boufker and colleagues have shown that FXR agonists stimulated the DNA binding activity of Runx2 in cells treated with ibandronate. Although we were unable to demonstrate the difference in DNA binding activity between FXR+/+ and FXR−/− cells, a significant downregulation of Runx2 expression was observed in FXR−/− calvaria cells, whereas treatment with FXR agonists increased its expression. In addition, we found that ERK and β-catenin signaling might also contribute to increased osteoblast differentiation, indicating that multiple pathways are activated by bile acids or FXR agonists, thereby contributing to the enhanced osteoblast differentiation of mesenchymal progenitors. However, the exact molecular mechanism by which FXR signaling regulates ERK or β-catenin signaling should be further studied.
Interestingly, in addition to suppressed bone formation, FXR−/− mice also showed increased bone resorption parameters that included the osteoclast number and fractional surface. Although the increased RANKL/OPG mRNA expression ratio may partly account for this increased bone resorption status, the increased osteoclast formation in spleen derived macrophage cultures from FXR−/− mice, together with the appreciable expression of FXR in osteoclast precursors, suggest the possibility that FXR in these cells can directly regulate the osteoclast differentiation potential. Indeed, the lack of FXR resulted in the reduced expression of NFATc1 expression, the key transcriptional factor for osteoclastogenesis. Although the crosstalk between FXR and NFATc1 signaling is still unknown, our data suggest that FXR signaling can modulate osteoclast differentiation by regulating the master switch for RANKL-induced osteoclastogenesis. Consistent with the in vivo data, our study showed that activation of FXR with bile acids (CDCA and 6-ECDCA) or pure FXR agonists (GW4064 or fexaramine) potently inhibited RANKL-induced osteoclast differentiation in vitro. In addition, FXR activation also inhibited the expression of c-fos and NFATc1. Although we have not dissected the molecular mechanisms, our data are analogous with a previous study that showed that exposure of lipopolysaccharide (LPS)-activated macrophages to 6-ECDCA resulted in the downregulation of NF-κB dependent-genes including c-fos and exert counter-regulatory effects on innate immunity.
In female mice, there was no significant bone phenotype in the female FXR–/– mice during the growing phase. Interestingly, however, bone loss was significantly accentuated only after an ovariectomy in FXR–/– mice compared with that of the FXR+/+ mice. Although the reason for this difference is not clear, we found that FXR is positively regulating ER signaling in vitro, suggesting the existence of crosstalk between FXR and ER signaling. In support of our notion, previous studies have demonstrated a physical interaction between CDCA-activated FXR and ER.[30, 31] Furthermore, although FXR activation by CDCA has no effects on the ER-positive MCF-7 cell line in estrogen-containing medium, it stimulates the proliferation of the cells in estrogen-free condition. The unmasking of the bone phenotype in FXR–/– mice after an ovariectomy in our study seems to be in agreement with our data and suggests that the crosstalk between FXR and ER may be dependent on the estrogen status.
The evidence presented in this report has clinical implication in that FXR could be used as a therapeutic target in the treatment of osteoporosis. Indeed, FXR became an attractive target in drug development for metabolic syndrome such as diabetes mellitus, dyslipidemia, and atherosclerosis. Selective bile acid receptor modulators have been extensively studied as a potential drug for metabolic diseases.[6, 33] In our study, we tested the possibility of using FXR agonist for preventing OVX-induced bone loss. A FOH (1%) diet marginally attenuated OVX-induced bone loss compared with the chow diet group. In addition, administering the FOH diet to growing C57BL/6J mice significantly increased the bone mass gain, suggesting an anabolic action of FXR in skeletal tissue. With the addition of our study, it might be possible to include osteoporosis as one of the target metabolic diseases for which modulation of FXR can have therapeutic potential. However, several limitations need to be solved for the current drug development strategy of FXR. First, the molecular mechanisms underlying the FXR metabolic actions are very complex and discrepant for individual pathways or different tissues (reviewed in Cariou and Staels). Therefore, the systemic use of simple FXR agonists or antagonists could cause undesired adverse effects from a single therapeutic point of view. Another major issue is the species specific differences in FXR biology. For instance, guggulsterone, a natural product from a tree resin, which acts as an FXR antagonist, increased the high-density lipoprotein (HDL)-cholesterol levels and decreased the serum triglyceride (TG) levels in rats,[34, 35] whereas it showed no effects in dyslipidemic patients.
In conclusion, our study shows that FXR positively regulated bone metabolism in vivo by controlling both arms of bone remodeling; ie, bone formation and resorption. The present study can shed light on potential therapeutic targets for managing metabolic bone diseases, including osteoporosis.
All authors state that they have no conflicts of interest.
This study was supported by a grant from the Ministry of Education, Science and Technology of Korea (2010-0026991). We are grateful to Dr. David D Moore at Baylor College of Medicine for his suggestions and review of the manuscript.
Authors' roles: Study design: SWC, JHA, CSS. Study Conduct: SWC, JHA, HP, JYY, HJC, MY, WYB, JEK. Data interpretation: SWC, JHA, HP, SWK, YJP, SYK, MY, JEK, CSS. Drafting manuscript: SWC, JHA, CSS. Revising manuscript content: SWC, HP, JYY. Approving final version of manuscript: SWC, JHA, HP, JYY, HJC, SWK, YJP, SYK, MY, WYB, JEK, CSS. CSS takes responsibility for the integrity of the data analysis.