Caveolin-1 regulates osteoclastogenesis and bone metabolism in a sex-dependent manner

Background : Caveolin-1 plays important roles in the regulation of diverse cellular responses. Results : Caveolin-1 knockdown reduced osteoclastogenesis in vitro while caveolin-1 knockout resulted in an increased and decreased osteoclastogenesis in male and female mice, respectively. Conclusion : Regulation of osteoclastogenesis by caveolin-1 was dependent on sex. Significance: This indicates complicated, but critical, role of caveolin-1 in the regulation of bone metabolism. Abstract Lipid raft microdomains have important roles in various cellular responses. Caveolae are a specialized type of lipid rafts that are stabilized by oligomers of caveolin proteins. Here we show that caveolin-1 (Cav-1) plays a crucial role in the regulation of osteoclastogenesis. We found that caveolin-1 was dramatically up-regulated by receptor activator of nuclear factor κB ligand (RANKL), the osteoclast differentiation factor. Knockdown of Cav-1 reduced osteoclastogenesis and induction of NFATc1, the master transcription factor for osteoclastogenesis, by RANKL. Consistent with in vitro results, injection of caveolin-1 siRNA onto mice calvariae showed reduction in RANKL-induced bone resorption and osteoclast formation. Moreover, Cav-1 -/- female mice had higher bone volume and lower osteoclast number compared to wild type mice. However, Cav-1 -/- male mice had both osteoclast and osteoblast numbers higher than wild type mice with no difference in bone volume. The sex dependency in the effect of Cav-1 deficiency was partly attributed to decreased RANK and increased cFms expression in osteoclast precursors of female and male mice, respectively. Taken together, these data demonstrate that Cav-1 has a complicated, but critical role, for osteoclastogenesis.


Introduction
Bone homeostasis is maintained by the balanced activity of bone-resorbing osteoclasts (OC) and bone-forming osteoblasts (OB) during bone remodeling, a physiological process of bone turnover. These two cell types are derived from different origin of stem cells via multiple steps governed by differentiation factors. The differentiation of OC from monocyte/macrophage lineage precursor cells requires two major cytokines, macrophage colony stimulating factor (M-CSF) and receptor activator of nuclear factor κB ligand (RANKL). M-CSF supports survival and proliferation of cells during osteoclastogenesis, while RANKL drives the differentiation program. RANKL interaction with its receptor RANK evokes recruitment of tumor necrosis factor receptor associated factor 6 to assemble signaling molecules that are required for the activation of mitogen-activated protein kinase signaling pathways (1,2). RANKL also stimulates the calcium signaling pathway and NFκB and c-Fos transcription factors (2)(3)(4). Ultimately, these signaling pathways are integrated by nuclear factor of activated T cells c1 (NFATc1), the transcription factor regarded as a master regulator of osteoclastogenesis, for the expression of osteoclast marker genes such as tratrate-resistant acid phosphatase (TRAP) (5,6).
Lipid rafts are membrane microdomains which consist of cholesterol, glycosphingolipids, and glycosylphosphatidylinositol-anchored proteins (7,8). These microdomains have been implicated in regulation of various cellular functions. Especially, lipid rafts in the plasma membrane serve as organizing center for assembly of signaling molecules that respond to extracellular stimuli. Caveolae, a specialized type of lipid rafts, are 50-to 100 nm flask-shaped cellsurface membrane invaginatiton (9). Like lipid rafts, caveolae have important roles in diverse cellular functions including endocytosis and signal transduction (10,11).
We found that Cav-1 was greatly increased by RANKL during osteocalstogenesis. Cav-1 knockdown using siRNA evidently inhibited the differentiation of OC and the induction of c-Fos and NFATc1 by RANKL. However, osteoclastogenesis and bone metabolism was differentially affected by in vivo Cav-1 deficiency between male and female mice.

Mice
Cav-1 -/mice in C57BL/6 background was purchased from Jackson Laboratory and backcrossed with wild type C57BL/6 for more than 6 generations. The heterogenic mice were mated and littermates were used for experiments. Animal experimental protocols were approved by the Committee on the Care and Use of Animals in Research at Seoul National University.

Reagents and antibodies
Anti-NFATc1 and anti-c-Fos were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-Cav-2 was purchased from Abcam (Cambridge, UK). All other antibodies were purchased from Cell Signaling Technology (Beverly, MA). Lipofectamine 2000 was purchased from Invitrogen (Carlsbad, CA). Human soluble RANKL and M-CSF were from Pepro-Tech (Rocky Hill, NJ).

In vitro osteoclast differentiation
Bone marrow derived-macrophages (BMMs) were generated using long bones from 5-weekold female ICR mice and from 8-week-old WT and Cav-1 -/male and female mice. Flushed bone marrow cells were plated with α-MEM to remove adherent cells. Next day, non-adherent cells were transferred to new dishes and cultured in α-MEM with M-CSF (30 ng/ml) for 3 days. Cells at this stage (BMMs) were used for osteoclastogenesis by culturing with M-CSF (30 ng/ml) and RANKL (200 ng/ml) for 4 days.

RT-PCR and real-time PCR
Total RNA was extracted using TRIzol reagent (Invitrogen) according to the manufacturer's instruction and cDNA synthesis was carried out using SuperscriptII reverse transcriptase (Invitrogen). PCR was performed with target gene specific primers. Real-time PCR was performed with an ABI 7500 instrument (Life Technologies), for 40 cycles with SYBR Green Master Mix. Gene expression was calculated using the 2 −ΔΔC′T method and normalized to the level of β-actin. All samples analyzed in Fig. 1B were harvested at the same time (at day 3) and β-actin mRNA was stable in our experimental conditions.

Western blotting
Cells were washed with cold PBS and lysed in RIPA buffer (50 mM Tris-HCl pH 8.0, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, 0.5 mM PMSF, proteinase inhibitor cocktail (Roche, Germany), 100 mM sodium varnadate, 0.5 M sodium floride). After protein quantification, cell extracts were separated on polyacrylamide gel and transferred onto nitrocellulose membrane. After blocking with 5% skim-milk in TBS-T, membranes were incubated overnight with primary antibody at 4℃. Membranes were washed and incubated with HRP-conjugated secondary antibody, and the immune complexes developed were detected using enhanced chemiluminescence reagents.

In vitro TRAP staining
Cells were fixed in 3.7% formaldehyde solution and permeabilized with 0.1% Triton X-100. After washing with PBS, cells were stained using the Leukocyte Acid Phosphatase Assay Kit (Sigma, St. Louis, MO) following the manufacturer's instruction. TRAP-positive multinuclear osteoclasts containing three or more nuclei were counted under a light microscope.

In vitro resorption pit formation assay
BMMs were cultured on dentin slices for 9 days in the presence of M-CSF and RANKL. After removing the cells by sonication, dentin slices were stained with hematoxylin and observed under a light microscope.

Gene knockdown by small interfering RNA (siRNA)
Target-specific 21-mer nucleotide siRNA and control siRNA were purchased from Invitrogen. BMMs were transfected with siRNA (40 nM) using Lipofectamine 2000 (Invitrogen) following the manufacturer's instruction. At 2 days after transfection, the expression level of protein or mRNA was determined.

Retroviral transduction
Mouse Cav-1 gene was cloned into pMX-IRES vector. Retroviral particles were packaged by transfecting Plat-E cells with DNA plasmids using Lipofectamine 2000 according to the manufacturer's instructions. BMMs were infected with viral supernatants in the presence of polybrene (10 μg/ml) and M-CSF for 12 hr.

Confocal microscopy
Cells were fixed with 3.7% formaldehyde and permeabilized with 0.1% Triton X-100. Fixed cells were blocked with 1% BSA in PBS, incubated with primary antibody at room temperature for 1 hr and washed with PBS. Cells were then incubated with secondary antibody for 30 min and counterstained with 4´,6-diamidino-2-phenylindole. Cells were observed under a Zeiss LSM 5 PASCAL laserscanning microscope (Carl Zeiss Microimaging GmbH, Goettingen, Germany) with an X400 objective (CApochromat /1.2 WCorr).

Determination of C-terminal peptide of type I collagen (CTX-I) and N-terminal propeptide of type I collagen (PINP)
CTX-I levels in 8-week-old WT and Cav-1 -/mice serum were measured using CTX-I and PINP ELISA kits (Immunodiagnostic System, Boldon, UK) following manufacturer's instruction.

Calcein-xylenol orange double labeling
8-week-old WT and Cav-1 -/mice were injected intraperitoneally with calcein (25 mg/kg). After 7 days, 90 mg/kg xylenol orange was injected. Mice were sacrificed at day 10 and femurs were fixed and embedded in methyl metharcrylate. Using a LSM5 PASCAL laser scanning microscope (Zeiss), sectioned femurs (5 μm) were observed. The distance between calcein and xylenol orange deposition was determined from five different regions of a image using 5 slides per group.

μCT analysis
Femurs of 8-week-old WT and Cav-1 -/mice were analyzed by micro-computed tomography (μCT) using SkyScan 1172 system (SkyScan, Aartselaar, Belgium; 40 Kv, 250 μA, 7.61 pixel size). Results were obtained with 1 mm thickness area of distal femurs, starting from 0.7 mm below the growth plate at thresholds minimum 70 and maimum 255. 3D images were constituted using the CT vol software.

Statistical analysis
All in vitro experiments were repeated at least three times. All quantitative experiments were performed at least in triplicate. Student's t test was used to determine the significance between two groups.

Cav-1 expression was induced by RANKL during osteoclastogenesis
In a microarray analysis to discover new genes involved in the regulation of osteoclast differentiation, we found that the Cav-1 mRNA level was increased by RANKL (data not shown). We next analyzed the expression level of all isotypes of caveolin by real-time PCR. The mRNA level of Cav-1 was strongly increased at day 1 after RANKL treatment and maintained at high levels for 3 days during RANKL-stimulation (Fig. 1A). Cav-2 mRNA level was slightly increased at day 3 after RANKL treatment. RT-PCR analysis also showed the same expression pattern of caveolin gene family (Fig. 1B). Cav-3 mRNA was not detected. Next, we determined protein levels of Cav-1 and Cav-2 during RANKL-mediated osteoclastogenesis. As shown in Fig. 1C, both Cav-1 and Cav-2 proteins were induced by RANKL. The up-regulation of Cav-1 protein was earlier and stronger than Cav-2. Since caveolins are components of caveolae, a special microdomain in the plasma membrane, we investigated whether Cav-1 localizes to plasma membrane microdomains. Immunostaining with a Cav-1 specific antibody showed that Cav-1 expression was induced by RANKL in prefusion osteoclasts (pOC) and mature OC. Moreover, Cav-1 was localized with FITCconjugated cholera toxin B subunit (CTxB) that stained plasma membrane microdomains (Fig.   1D). These results indicate that Cav-1 is induced by RANKL and may play a role as a component of membrane microdomains in pOC and mature OC.

RANKL-induced osteoclasogenesis was attenuated by Cav-1 silencing
To define the role of caveolin family during osteoclastogenesis, we designed siRNA oligonucleotides against Cav-1 and Cav-2. BMMs were transfected with scrambled or isotype-specific caveolin siRNA. Knockdown of Cav-1 strongly inhibited the formation of OC ( Fig. 2A). The resorption of dentin slice was also reduced by Cav-1 knockdown (Fig. 2B). The induction of NFATc1 and c-Fos is essential for OC differentiation by RANK (2,3). The RANKL-induced NFATc1 and c-fos expression was attenuated by Cav-1 silencing in BMMs (Fig. 2C). It has been reported that the Cav-2 expression is regulated by the presence of Cav-1 (34). Consistent with the previous report, we observed that both mRNA and protein levels of Cav-2 were reduced in Cav-1 knockdown BMMs (Fig. 2D). To examine whether the decreased osteoclastogenesis from Cav-1 knockdown BMMs was influenced by the accompanying reduction in Cav-2 reduction, we directly silenced Cav-2 and assessed the osteoclastogenic potential. Unlike Cav-1 siRNA, Cav-2 siRNA did not alter formation of OC or induction of NFATc1 and c-Fos (Fig. 2E). These results suggest that Cav-1 is the major caveolin involved in the regulation of osteoclastogenesis.

Cav-1 knockdown reduced osteoclastogenesis and bone resorption in vivo
We next examined whether the antiosteoclastogenic effect of Cav-1 siRNA could be confirmed under in vivo conditions. To this end, PBS-or RANKL-soaked collagen sponges were implanted onto mice calvariae that received injections of scrambled or Cav-1-targeted siRNA. In vivo knockdown was verified by RT-PCR (Fig. 3A). The RANKL treatment induced TRAP mRNA expression in vivo and this induction was significantly attenuated by Cav-1 silencing (Fig. 3A). When we stained the whole calvariae for TRAP activity, Cav-1 siRNAinjected mice showed reduced TRAP-positive area (Fig. 3B). In agreement with this result, micro-computed tomography (μCT) analyses of the calvariae from Cav-1 siRNA-injected mice revealed lower level of bone resorption compared to control siRNA-injected mice in response to RANKL (Fig. 3C). Taken together, these results showed that Cav-1 silencing prevents RANKL-induced bone loss in mice calvariae by inhibiting OC formation.

Cav-1 -/female BMMs, but not male BMMS, have defects in osteoclastogenesis in vitro
As the number of OC on bone was observed to be affected by Cav-1 knockout in different ways between male and female mice, we next compared the osteoclastogenic potential of BMMs from male and female Cav-1 -/mice. In in vitro cultures, OC formation from BMMs of Cav-1 -/male mice was greater than that from WT male mice (Fig. 6A). In contrast, BMMs of Cav-1 -/female mice showed strong defect in RANKL-induced OC differentiation (Fig. 6B). The induction of NFATc1 was also attenuated in BMMs from Cav-1 -/female mice (data not shown).
Whether this defect in osteoclastogenesis could be rescued by Cav-1 restoration was then tested. Cav-1 -/female BMMs were infected with control (pMX) or Cav-1 (pMX-Cav-1)-containing retroviruses. As shown in Fig. 6C, Cav-1 overexpression increased TRAP-positive OC formation. The RANKL induction of NFATc1 was also restored by Cav-1 overexpression. To investigate a potential mechanism for the gender difference in the effect of Cav-1 deficiency, we next determined the protein levels of cFms (the receptor for M-CSF) and RANK in Cav-1 -/-BMMs. In cells from male mice, the basal cFms levels, both at mRNA and protein, were higher in Cav-1 -/-BMMs than in WT BMMs (Fig. 6D, E). However, the expression pattern of RANK was not different between Cav-1 -/and WT cells (Fig. 6D, E). In female mice, the RANK level was increased during osteoclastogenesis from WT BMMs, which was not observed in Cav-1 -/-BMMs (Fig. 6D, E). cFms expression was not altered both in Cav-1 -/and in WT cells from female mice (Fig. 6D, E). These data suggest that Cav-1 deficiency affected the property of OC precursor cells in a manner dependent on the sex of mice.

DISCUSSION
Caveolin has been implicated in many disorders including cancer, cardiac diseases, diabetes, and atherosclerosis (35)(36)(37)(38). Our results obtained in the present study suggest a link between caveolin and bone metabolism, adding another disorder to the list. Cav-1, a major structural protein of caveolin, was selectively induced at a great extent among three caveolin isoforms by RANKL (Fig.1). Despite lower but constitutive presence of Cav-2 protein in OC precursors, in vitro knockdown of Cav-1 by using siRNA, but not that of Cav-2, led to a decreased potential of osteoclastogenesis (Fig. 2). These results suggest a specific role of Cav-1 for the regulation of OC differentiation. Results of these in vitro siRNA experiments performed with BMMs from female ICR mice are consistent with the osteoclastogenesis data obtained with cells from female Cav-1 -/mice (Fig. 6B). Therefore, Cav-1 plays a positive role in OC differentiation of female mice.
Radiological and histomorphometric analyses of Cav-1 -/mice in our study revealed that the regulation of bone cells, and consequently that of bone mass, was different between female and male mice. In 8-week-old Cav-1 -/female mice, femoral trabecular bone volume was significantly higher with lower number of OC in bone in comparison with WT control mice (Figs. 4 and 5). As OB number and activity (BFR and MAR) were not different between Cav-1 -/and WT (Fig. 5), the osteopetrotic phenotype of female Cav-1 -/mice could be attributed solely to decreased bone resorption by OC. Consistent with this, female Cav-1 -/mice had lower CTX-1 level in the blood than WT mice (Fig. 5). In a marked difference, male mice showed similar bone volume between Cav-1 -/and WT (Fig. 4). This indifference could be explained by a balanced stimulation of both OC and OB activities in Cav-1 -/male mice as evidenced by bone sections from Cav-1 -/mice displaying an increase in both OC and OB numbers and a higher BFR and MAR compared to those in WT mice (Fig. 5).
Some of our results are in agreement and disagreement with data shown in a couple of previous studies in which effects of Cav1 deficiency on bone were investigated (39,40). In the paper by Rubin et al. (39), distal femur metaphysis of 8-week-old Cav1 -/male mice showed increased trabecular BV/TV with an apparently paradoxical decrease in MAR compared to WT mice. This increase in BV/TV was postulated to be due to increased bone formation by OB at an earlier stage (5 week). As histological analyses were not shown in this paper, we do not know whether data on OC and OB numbers in femur sections of male mice were also different from our data. While it is not easy to answer the question that why the data of femur trabecular BV/TV and MAR are in discrepancy between our and Rubin's studies, our μCT BV/TV results are consistent with histology and MAR data. In vitro formation of OC from marrow cells of male mice was modestly increased in the Rubin's study, which is in agreement with our observation. As female mice bone phenotype was not analyzed in that study, we do not know whether the increased BV/TV of Cav-1 -/female mice found in our study is also in discrepancy with their study. In another paper by Hada et al. (40), in vitro osteoclastogenesis of precursor cells from Cav-1 -/male mice was not significantly different from WT cells unless Cav-2 was additionally muted by siRNA. In that study, the Cav-1/Cav-2 double deficient cells showed increased osteoclastogenesis, leading to a suggestion that Cav-1/Cav-2 complex may negatively regulate osteoclastogenesis. This result is also in disagreement with our data showing that Cav-1 deficiency was sufficient to increase osteoclastogenesis from male BMMs (Fig. 6A). Radiological and histomorphometric analyses of Cav-1 -/mice were not carried out in Hada's study.
Our study clearly indicates that Cav-1 deficiency affected bone phenotype in a sexdependent manner. What would be a potential mechanism for this difference between male and female mice in responding to Cav-1 gene deletion? Cav-1 has been suggested to regulate ER trafficking to cell membrane and estrogeninduced signaling processes (23,24). Estrogen has a proapoptotic effect on OC and ER deletion abrogates this effect in the female but not in the male (41), suggesting more impact of ER signaling in female OC. The expression levels of ER and Cav-1 were similar between male and female OC in our study (data not shown). Therefore, it may be possible that Cav-1 deletion displayed the sex dichotomy in bone phenotype due to stronger dependency of estrogen signaling on Cav-1 in female. However, it is unreasonable to think that only a few factors are accountable for sex-dependent bone phenotype of Cav-1 -/mice. As Cav-1 is expressed in many cell types and more and more systemic factors are being discovered to influence bone cell regulation, it is likely that many factors that directly or indirectly affecting bone cell responses are changed in the Cav-1 general knockout mice used in our study. Nonetheless, it can be pinpointed that sexdependency of Cav-1 -/bone phenotype is associated with cell-autonomous mechanisms.
As shown in Figure 6, BMMs from male and female mice had contrasting osteoclastogenesis potential in vitro and the expression patterns of c-Fms in BMMs and RANK in differentiating OC were dissimilar between males and females. Therefore, Cav-1 deficiency appears to drive changes in characteristics of OC precursors in different ways between males and females.
In summary, our results demonstrate the importance of Cav-1-mediated processes in OC differentiation and subsequently bone homeostasis. Additionally, this study arouses attention to the necessity of analyses of both sexes in studies on bone metabolism. Further investigations are required with bone cell typespecific Cav-1 knockouts to distinguish direct roles of Cav-1 in bone cells from indirect effects of Cav-1 deficiency on bone metabolism. Molecular details on how Cav-1 regulates differentiation of OC and OB are also to be unraveled.    Cav-2 silenced BMMs were incubated with M-CSF (30 ng/ml) and RANKL (200 ng/ml). After 4 days, cells were subjected to TRAP staining. After 2 days, protein levels of NFATc1, c-Fos, and Cav-2 were determined by Western blotting. *, p < 0.05 as compared to controls. Scr, control siRNA. Figure 3. In vivo Cav-1 knockdown suppressed osteoclastogenesis and bone resorption. Control or Cav-1 siRNA was injected onto calvariae of 5-week-old female mice, followed by implantation of a collagen sheet soaked with PBS or RANKL. At day 6, mice were sacrificed. (A) Calvarial tissues were analyzed by RT-PCR. (B) Mice calvariae were stained for TRAP. TRAP + area was measured by densitometry using Image J (C) Calvariae were subjected to μCT analysis. 3D-images are shown. The resorbed area was measured by densitometry using Image J. n = 5 per group. *, p < 0.005 as compared to controls. Scr, control siRNA; Cav-1, Cav-1 siRNA, The red box indicates the analyzed area.     sense  antisense  sense  antisense  sense  antisense  sense  antisense  sense  antisense  sense  antisense  sense  antisense   AGAAGACGGTGCTGGAGTCT  TAGGAGCAGTGAACCAGTCG  CTTCACTCCGGTGGTGGTGG  GCGCACCTGGTACTTCGGCT  CGACCATTGTTAGCCACATACG  TCGTCCTGAAGATACTGCAGGTT  TATGACGCGCACACCAAGGA  GCCCAGATGTGCAGGAAGGA  CGGGATCCTCACCAGCTCAA  TACCCGCAATGAAGGCCAAG  TGGAAGCTCGGATCATCAAG  AGATGTGGCTGATGCACTGG  TCTGGCACCACACCTTCTAC  TACGACCAGAGGCATACAGG