Selective Estrogen Receptor Modulators Suppress Hif1α Protein Accumulation in Mouse Osteoclasts

Anti-bone resorptive drugs such as bisphosphonates, the anti-RANKL antibody (denosumab), or selective estrogen receptor modulators (SERMs) have been developed to treat osteoporosis. Mechanisms underlying activity of bisphosphonates or denosumab in this context are understood, while it is less clear how SERMs like tamoxifen, raloxifene, or bazedoxifene inhibit bone resorption. Recently, accumulation of hypoxia inducible factor 1 alpha (Hif1α) in osteoclasts was shown to be suppressed by estrogen in normal cells. In addition, osteoclast activation and decreased bone mass seen in estrogen-deficient conditions was found to require Hif1α. Here, we used western blot analysis of cultured osteoclast precursor cells to show that tamoxifen, raloxifene, or bazedoxifene all suppress Hif1α protein accumulation. The effects of each SERM on osteoclast differentiation differed in vitro. Our results suggest that interventions such as the SERMs evaluated here could be useful to inhibit Hif1α and osteoclast activity under estrogen-deficient conditions.


Introduction
Many developing countries show significantly increased numbers of aging individuals, resulting in a sharply increased number of osteoporosis patients and a parallel increase in the number of bone fragile fracture patients [1]. To prevent these conditions, several reagents have been developed such as either anti-bone resorptive or bone-forming drugs [2]. Among these, bisphosphonates, including alendronate and risedronate, the monoclonal antibody demosumab, and selective estrogen receptor modulators (SERMs) are anti-bone resorptive, while teriparatide is categorized as a bone-forming drug [2][3][4]. Mechanisms underlying bisphosphonate and denosumab activity are well-characterized [5]. For example, bisphosphonates are taken up by osteoclasts upon bone-resorption and inhibit the geranylgeranyl pathway, promoting osteoclast apoptosis [6]. By contrast, denosumab recognizes and inactivates the receptor activator of nuclear factor kappa B ligand (RANKL), blocking osteoclast differentiation and activation [7,8].
Analysis of post-menopausal development of osteoporosis indicates that ERα expressed in osteoclasts function to block osteoclast activation and bone loss [20][21][22]. Furthermore, our previous analysis of ovariectomized (OVX) mice indicated that hypoxia inducible factor 1 alpha (Hif1α) is required for osteoporosis development under an estrogen-deficient condition [23]. Specifically, Hif1α protein accumulation in osteoclasts was continuously suppressed by estrogen in pre-menopausal estrogen-sufficient conditions but accumulated in osteoclasts in estrogen-deficient conditions [23]. Moreover, treatment of OVX mice with a Hif1α-inhibitor completely abrogated estrogen deficiency-induced osteoclast activation and bone loss [23]. We have also shown that eldecalcitol, a vitamin D analogue and inhibitor of bone resorption in osteoporosis patients, functions as a Hif1α inhibitor [24]. Thus, Hif1α could serve as a therapeutic target to block osteoclast activation and bone loss under estrogen-deficient conditions.
Here, we define mechanisms underlying the anti-bone resorbing function of SERMs. To do so, we treated primary osteoclast precursor cells with SERMs in normal and estrogen-free conditions and evaluated osteoclastogenesis by multi-nuclear osteoclast formation and expression of osteoclastic genes such as Cathepsin K, nuclear factor of activated T cells 1 (NFATc1) and dendritic cell specific transmembrane protein (DC-STAMP). In addition, we assessed the effects of SERMs on Hif1α protein levels. We report that although the SERMs tamoxifen, raloxifene and bazedoxifene have varying effects on osteoclast differentiation, they all suppress Hif1α protein accumulation in osteoclasts grown under hypoxic and estrogen-free conditions. Thus, evaluation of Hif1α protein levels in osteoclasts grown under hypoxia in vitro may serve as a way to predict whether potential therapies will inhibit osteoclast bone-resorption activity in vivo.

Mice
Wild-type mice on a C57BL/6 background were purchased from Sankyo Labo Service (Tokyo, Japan). Animals were maintained under specific pathogen-free conditions and housed up to 5 mice per cage, and were kept in a 12 h light/dark cycle controlled rooms at the animal facility of the Keio University. Sterile distilled water and standard diet (CLEA Rodent Diet CE-2, Japan) was available ad libitum. Animal experiments were reviewed and approved by The Keio University Institutional Animal Care and Use Committee (Permit Number: 09092-14). Animal sacrifice was humanely performed by cervical dislocation for adult mice. This study was specifically approved by the Keio University animal care committee.
Osteoclastogenesis was evaluated by tartrate resistance acid phosphatase (TRAP) staining, and TRAP-positive multi-nuclear cells containing more than three nuclei were scored as osteoclasts [27].

Quantitative PCR analysis
In three independent analyses, total RNAs were extracted from bone marrow cultures using an RNeasy kit (Qiagen, Venlo, Limburg, The Netherlands). Complementary DNA (cDNA) was prepared by using oligo (dT) primers and reverse transcriptase (Wako Pure Chemicals Industries). Quantitative PCR was performed using SYBR Premix ExTaq II reagent and a DICE Thermal cycler (Takara Bio Inc.), according to the manufacturer's instructions. β-actin expression served as an internal control. Primers for realtime PCR were: β-actin-forward:

Replication of in vitro experiments
At least three independent experiments were performed for all in vitro experiments, and representative data are shown.

Statistical analyses
Statistical analyses were performed using the unpaired two-tailed Student's t-test ( Ã P<0.05; ÃÃ P<0.01; ÃÃÃ P<0.005; NS, not significant, throughout the paper). All data are shown as means ± SD.

Tamoxifen inhibits osteoclast differentiation
Tamoxifen reportedly inhibits osteoclast bone reorption [10]. To determine if tamoxifen can inhibit osteoclast differentiation in vitro, we isolated osteoclast progenitor cells from wild-type mice and cultured them in the presence of M-CSF and RANKL with or without tamoxifen. Osteoclastogenesis was evaluated by assessing formation of multi-nuclear TRAP-positive cells containing more than three nuclei and by expression levels of the osteoclast-related markers Cathepsin K, NFATc1 and DC-STAMP (Fig 1). Although mono-nuclear osteoclasts were formed in the presence of tamoxifen, tamoxifen treatment significantly inhibited multi-nuclear osteoclast formation induced by M-CSF and RANKL (Fig 1A and 1B) as well as Cathepsin K and DC-STAMP expression compared with untreated cells (Fig 1C), suggesting that tamoxifen inhibits osteoclast differentiation.

Other SERMs have varying effects on osteoclast differentiation
To determine whether other SERMs have effects comparable to tamoxifen, we first evaluated the effect of raloxifene on differentiation of cultured osteoclast progenitors. Unlike tamoxifen effects, formation of multi-nuclear TRAP-positive osteoclasts was not inhibited by raloxifene (Fig 2A and 2B). Raloxifene treatment did, however, inhibit expression of osteoclastic genes such as Cathepsin K, NFATc1 and DC-STAMP in osteoclasts ( Fig 2C). We then performed similar tests in osteoclast precursors using the SERM bazedoxifene and found that neither multi-nuclear TRAP-positive osteoclast formation nor osteoclastic gene expression was suppressed (Fig 3A-3C), although bazedoxifene has been shown to inhibit osteoclast bone resorption in post-menopausal osteoporosis patients [29,30].

SERMs have varying effects on differentiation of osteoclasts cultured in estrogen-free conditions
SERMs reportedly act via both ERα-dependent and -independent mechanisms [14,16]. The presence of phenol red in culture media reportedly promotes estrogenic effects in several cell types; likewise, fetal bovine serum contains estrogen [23]. Thus to assess activity of tamoxifen, raloxifene or bazedoxifene in estrogen-free osteoclast culture conditions, we utilized phenol red-free media and estrogen-depleted serum (Fig 4). Since osteoclast formation promoted by M-CSF and RANKL is strongly inhibited in estrogen-free relative to normal culture conditions, we evaluated osteoclast differentiation based on osteoclastic gene expression rather than multinuclear TRAP-positive cell formation. Expression of Cathepsin K, NFATc1 and DC-STAMP was significantly inhibited by tamoxifen in estrogen-free conditions (Fig 4A), although NFATc1 expression was not significantly changed by tamoxifen treatment in normal culture conditions (Fig 1C). Raloxifene treatment significantly elevated Cathepsin K, NFATc1 and DC-STAMP expression in osteoclasts grown in estrogen-depleted conditions (Fig 4B), although all three genes had been significantly inhibited in normal culture by comparable treatment ( Fig  2C). Moreover, NFATc1 expression was significantly inhibited by bazedoxifene in estrogen free-conditions (Fig 4C), although expression of NFATc1 is upregulated by comparable treatment in normal culture conditions ( Fig 3C). Overall, despite these variations, the effects of SERMs on osteoclast differentiation in estrogen-free conditions differed from those seen in normal culture conditions.

Hif1α protein levels in osteoclasts are suppressed by SERMs
Finally, given that Hif1α protein is reportedly a target of estrogen [23], we asked whether Hif1α protein levels in osteoclasts decrease following SERM treatment (Fig 5). To do so, we cultured Raw264.7 osteoclast progenitor cells in the presence of RANKL with or without SERMs in normoxic or hypoxic conditions, and evaluated Hif1α protein levels by western blots (Fig 5A-5C). We did not detect Hif1α protein in normoxic conditions, but Hif1α protein accumulated in hypoxic conditions, and that accumulation was suppressed by treatment with tamoxifen, raloxifene or bazedoxifene or by estrogen (E2) (Fig 5A-5C).

Discussion
Prevention of bone fragility fractures and control of osteoporosis are global health issues in developed countries. To date, most reagents used to prevent bone fragility fractures in  osteoporosis patients are anti-bone resorptive (among them SERMs), although a few activate bone formation. Mechanisms underlying SERM activity in this context remain unclear. Under sex hormone-depleted conditions, Hif1α was demonstrated to be a therapeutic target in conditions of post-menopausal [23] and male [25] osteoporosis. SERM treatment also inhibits osteoclast bone resorption in post-menopausal osteoporosis patients [9,11,13,31]. To date, there has been no definitive mechanistic explanation for why SERMs act as anti-bone resorptive agents. Our finding that SERM treatment blocks Hif1α accumulation provides a possible explanation for this outcome. Our observations also imply that the fact that the effects of any drugs on osteoclast differentiation in vitro is not sufficient to suggest that that drug would have antibone resorptive effects in vivo or in patients, as the three drugs tested here had different effects on in vitro osteoclastogenesis, although all of the SERMs tested were reportedly acting as inhibitors of osteoclast bone resorption [3,4,9,11,13], but all blocked Hif1α accumulation in osteoclasts. More studies are required to determine whether tamoxifen alters activities other than Hif1α protein suppression in osteoclasts. Moreover, the effects of SERMs on osteoclastic gene expression differed in normal and estrogen-free conditions, an outcome that we cannot yet explain. However, in normal conditions, following addition of estradiol to cultures of osteoclast precursors, we found that only Cathepsin K, but not NFATc1 and DC-STAMP, expression was upregulated in osteoclasts and multi-nuclear osteoclast formation was not accelerated (Fig 6A-6C). By contrast, in estrogen-free conditions Cathepsin K and NFATc1 expression was significantly inhibited by estradiol (Fig 6D). Thus, none of the SERMs tested here recapitulate the effects of estradiol in vitro in terms of osteoclastic gene expression. Since estradiol reportedly suppresses Hif1α protein accumulation in osteoclasts grown in hypoxic, estrogen-free conditions [23], suppression of Hif1α protein in osteoclasts is an activity shared by both SERMs and estradiol in vitro.
Since post-menopausal, estrogen-deficient conditions promote osteoclast activation leading to bone loss, estrogen administration is considered a means to reverse these conditions. Indeed, hormone replacement therapy (HRT) increases bone mass in post-menopausal osteoporosis patients, although treatment can have adverse effects [32]. Other agents such as bisphosphonates and denosumab can block osteoclastic bone resorption in osteoporosis patients, but these, too, can have adverse effects such as osteonecrosis of the jaws (ONJ) and severely suppressed bone turnover (SSBT) [33][34][35]. Comparable adverse effects have not been reported in patients treated with SERMs [3,31]. However, SERMs have been found to be less effective in inhibiting osteoclastic bone resorption than bisphosphonates [4,36]. This observation may support the idea that SERMs are Hif1α inhibitors, since treatment of normal mice with Hif1α inhibitors does not block physiological osteoclast activity in estrogen-or testosterone-sufficient conditions [23,25]. In contrast, administration of bisphosphonates or an anti-RANKL antibody to comparable, hormone-sufficient mice strongly inhibits physiological osteoclast activity, increasing bone mass [37].
Unlike individuals treated with bisphosphonates or denosumab, patients treated with HRT and SERMs can exhibit thrombosis development in deep veins [38][39][40], although the cause of these complications remains unclear. ERα is reportedly required to suppress Hif1α protein in osteoclasts [23], while SERMs are thought to act via ERα-dependent or independent mechanism [14][15][16]. Hif1α has also been considered a therapeutic target in some malignancies [41]. Tamoxifen and raloxifene are effective in inhibiting tumor burden in some breast cancers, while HRT promotes tumor development [42]. Further studies are needed to clarify molecular mechanisms underlying thrombosis and tumor development by HRT and SERMs.
We previously demonstrated that Hif1α could be a therapeutic target in osteoporosis [23], leading us to test the effects of candidate anti-bone resorptive agents on Hif1α protein suppression in osteoclasts in vitro. Eldecalcitol, which is used to inhibit bone resorption in osteoporosis patients [46], lowers Hif1α protein levels in osteoclasts [24], comparable to our observation following SERM treatment. To date, anti-osteoporosis agents have been tested in animal models such as OVX mice in vivo, but this system does not allow efficient testing of numerous candidate agents due to costs and time frame of testing. Screening in vitro for Hif1α protein suppression in osteoclasts could substitute for animal models and reduce expenses and time required to evaluate anti-bone resorbing agents.

Conclusions
SERMs act as inhibitors of Hif1α, a therapeutic target of post-menopausal osteoporosis, in osteoclasts under an estrogen-deficient condition. Testing inhibitory effects on Hif1α protein in osteoclasts in vitro is useful to screen candidate anti-bone resorptive agents before animal models.