Mechanical loading-related bone gain is enhanced by tamoxifen but unaffected by fulvestrant in female mice.

Accumulating evidence indicates that estrogen receptors (ERs) are involved in the mechano-adaptive mechanisms by which loading influences the mass and architecture of bones to establish and maintain their structural load-bearing competence. In the present study, we assessed the effects of the ER modulators tamoxifen and fulvestrant (ICI 182,780) on loading-related changes in the volume and structure of trabecular and cortical bone in the tibiae of female mice. Ten days after actual or sham ovariectomy, 17-wk-old female C57BL/6 mice were treated with vehicle (peanut oil), tamoxifen (0.02, 0.2, or 2 mg/kg · d), fulvestrant (4 mg/kg · d), or their combination and the right tibiae subjected to a short period of noninvasive axial loading (40 cycles/d) on 5 d during the subsequent 2 wk. In the left control tibiae, ovariectomy, tamoxifen, or fulvestrant did not have any significant effect on cortical bone volume, whereas trabecular bone volume was decreased by ovariectomy, increased by tamoxifen, and unaffected by fulvestrant. In the right tibiae, loading was associated with increases in both trabecular and cortical bone volume. Notably, the medium dose of tamoxifen synergistically enhanced loading-related gain in trabecular bone volume through an increase in trabecular thickness. Fulvestrant had no influence on the effects of loading but abrogated the enhancement of loading-related bone gain by tamoxifen. These data demonstrate that, at least in female mice, the adaptive response to mechanical loading of trabecular bone can be enhanced by ER modulators, in this case by tamoxifen.

In humans, estrogen deficiency is a major correlate of bone loss in both sexes (1,2). This loss is more severe in women than in men and is sufficiently severe to be associated with a high incidence of fragility fracture. This situation represents a failure of the mechanisms by which bone mass and architecture normally adapt to match the demands of habitual mechanical loading (3,4). The causal mechanisms underlying estrogen deficiency-related loss and loading-related adaptation of bone are both obscure, but a possible link between the two was suggested by the findings of Damien et al. (5,6) that the proliferative response to mechanical strain of female and male primary rat osteoblast-like cells in vitro was blocked by the estrogen receptor (ER) modulators tamoxifen and ICI 182,780 (fulvestrant). These compounds are used to block ER-mediated effects in breast cancer. The inhibitory effect of fulvestrant, but not that of tamoxifen, was confirmed in similarly derived osteoblast-like cells from humans (7). Since these initial reports, there has been accumulating evidence (8) that ERα is involved not only in osteoblasts’ and osteocytes’ responses to mechanical stimulation in vitro (9,10,11,12) but also in loading-related osteogenesis in vivo (13,14,15).

The study we report here is the first of which we are aware to address whether the ER modulators tamoxifen and fulvestrant affect bone’s adaptive response to mechanical loading in vivo, although tamoxifen has been shown to have partial protective effects against bone loss caused by immobilization in rats (16) and dogs (17). Because mechanical loading is the primary, natural, functional regulator of bone mass and architecture, influencing the potential contribution of ER to the adaptive mechanisms involved could have profound effects on the outcome of the process. To assess these in both trabecular and cortical bone, we used the mouse unilateral tibia axial loading model. The mouse has recently become the animal of choice for investigating bones’ adaptive responses to loading (13,14,15,18), and the externally loadable tibia model is well suited to study the responses of trabecular as well as cortical compartments (19,20,21).

Materials and Methods

Animals

Virgin, female C57BL/6 mice at 7 wk of age were purchased from Charles River Laboratories, Inc. (Margate, UK) and group housed in sterilized polypropylene cages with free access to water and a maintenance diet containing 0.73% calcium, 0.52% phosphorus, and 3.5 IU/g vitamin D (RM1; Special Diet Services Ltd., Witham, UK) in a 12-h light, 12-h dark cycle, with room temperature at 21 ± 2 C. All procedures complied with the United Kingdom Animals (Scientific Procedures) Act 1986 and were reviewed and approved by the ethics committee of the Royal Veterinary College (London, UK).

Experimental design

An initial pilot experiment was performed to assess the dose-related effects of tamoxifen on trabecular and cortical bone’s response to loading. At 16 wk of age (d 1), the animals were weight matched into sham-ovariectomized (SHAM) and ovariectomized (OVX) groups. Ten days after the operation (d 11), each group was randomly subdivided into four groups receiving either vehicle or low, medium, or high doses of tamoxifen (n = 6–8 per group). Tamoxifen citrate (Tocris Cookson Inc., Ellisville, MO) or vehicle (peanut oil, 5 ml/kg; Sigma Chemical Co., St. Louis, MO) was administered by sc injection on the same 5 d as the right tibiae were loaded (Fig. 1A). The doses of tamoxifen citrate on each day (d 11, 13, 15, 18, and 21) were 0.1 (low dose), 1 (medium dose), and 10 (high dose) mg/kg, which are equivalent to doses of tamoxifen of 0.02, 0.2, and 2 mg/kg · d, respectively, during the last 2 wk.

Figure 1

A, Overview of the experimental design. Sixteen-week-old C57BL/6 mice were SHAM or OVX on d 1. Drug or vehicle was injected on d 11, 13, 15, 18, and 21. The right tibiae received external mechanical loading approximately 3–6 h after each injection. Left control and right loaded tibiae were collected for analysis on d 25. B, Direction of external mechanical loading in the mouse tibia and transverse μCT images at the trabecular (0.25–0.75 mm distal to the growth plate) and cortical (0.5-mm-long section at 37% of the bone’s length from its proximal end) sites analyzed.

Three to six hours after each injection, all animals were anesthetized with isoflurane and their right tibiae subjected to a single short period of mechanical loading externally applied through the flexed knee and ankle (Fig. 1B; see External mechanical loading for details). For the rest of the day, normal cage activity was allowed. The animals’ left tibiae were used as internal controls (21). Within the week after OVX, there is a transient reduction in bone formation (22), which is followed by the more sustained period of bone loss. Because such an acute response could modify the effects of loading (23), we started the loading regimen 10 d after the SHAM or OVX operation (Fig. 1A). At 19 wk of age (d 25), the mice were euthanized, and both left control and right loaded tibiae were collected for analysis.

Having obtained results from the pilot experiment that showed that the medium dose of tamoxifen induced significantly higher loading-related responses in trabecular bone compared with lower and higher doses, this dose was used for the main experiment, which also investigated the effects of fulvestrant (ICI 182,780; Tocris Cookson Inc.). This experiment was performed using the same protocol as in the pilot experiment (Fig. 1A) and included seven groups: 1) SHAM plus vehicle, 2) SHAM plus tamoxifen, 3) SHAM plus fulvestrant, 4) OVX plus vehicle, 5) OVX plus tamoxifen, 6) OVX plus fulvestrant, and 7) OVX plus tamoxifen plus fulvestrant (n = 7 per group). A SHAM plus tamoxifen plus fulvestrant group was not included because in female ovary-intact mice, fulvestrant treatment induces a marked increase in blood estradiol levels, which could modify bone’s response to loading (24). The dose of fulvestrant on each day (d 11, 13, 15, 18, and 21) was 10 mg/kg, which is equivalent to approximately 4 mg/kg · d during the last 2 wk. This dose of fulvestrant was based on what was shown to significantly inhibit trabecular osteogenesis induced by high-dose 17β-estradiol in female mice (25). Both left control and right loaded tibiae were collected for analysis, and uterine weight as well as body weight was measured.

External mechanical loading

The apparatus and protocol for dynamically loading the mouse tibia have been reported previously (19,21,26). In brief, mice are anesthetized for approximately 7 min during which time the flexed knee and ankle joints are positioned in concave cups; the upper cup, into which the knee is positioned, is attached to the actuator arm of a servo-hydraulic loading machine (model HC10; Zwick Testing Machines Ltd., Leominster, UK) and the lower cup to a dynamic load cell. The right tibia is held in place by a low level of continuous static preload, onto which is superimposed higher levels of intermittent dynamic load. In the present study, 10.0 N of dynamic load was superimposed onto the 2.0 N static preload in a series of 40 trapezoidal-shaped pulses (0.025 sec loading, 0.050 sec hold at 12.0 N, and 0.025 sec unloading) with a 10-sec rest interval between each pulse. Strain gauges attached ex vivo to the proximal tibial shaft of similar 17-wk-old female C57BL/6 mice showed that a peak load of 12.0 N engendered approximately 1200 microstrain in that region (27).

High-resolution micro-computed tomography (μCT) analysis

The tibiae from both sides in each animal were collected after euthanasia, stored in 70% ethanol, and scanned by μCT (SkyScan 1172; SkyScan, Kontich, Belgium) with a pixel size of 5 μm. The images of the bones were reconstructed using SkyScan software. As shown in Fig. 1B, three-dimensional structural analyses were undertaken for trabecular (secondary spongiosa, 0.25–0.75 mm distal to the growth plate) and cortical bone (0.5-mm-long section at 37% of the bone’s length from its proximal end). At these trabecular and cortical sites, the loading regimen used in the present study stimulates increases in bone volume that are statistically significant (27). Because it has been shown by our group (21) and others (28) that axial loading of the mouse tibia induces significant bone gain through an increase in bone formation at both trabecular and cortical sites, three-dimensional high-resolution μCT rather than two-dimensional fluorescent histomorphometry was selected to quantify functional adaptation. This enabled us to analyze precisely comparative sites of the loaded and contralateral control tibiae. The parameters evaluated in the trabecular region included bone volume/tissue volume (BV/TV), trabecular number and trabecular thickness and in the cortical region, bone volume, periosteally enclosed volume, and medullary volume.

Statistics

All data are shown as the mean ± se. In the pilot experiment, six μCT parameters (trabecular BV/TV, trabecular number, trabecular thickness, cortical bone volume, periosteally enclosed volume, and medullary volume) were compared by one-way ANOVA followed by a post hoc Bonferroni or Dunnett T3 test. In the main experiment, changes in body weight and uterine weight were compared by one-way ANOVA followed by a post hoc Bonferroni test. Mixed-model analysis was performed on the six μCT parameters. The fixed effects in the model were operation (SHAM or OVX), drug treatments (vehicle, tamoxifen, fulvestrant, or tamoxifen plus fulvestrant), and mechanical loading (yes or no). Animal ID (n = 49) was included as a random variable to account for pairs of left and right tibiae belonging to the same animal. Final body weight was included as a fixed covariate due to its potential influence on bone size. Post hoc comparison after mixed-model analysis was carried out using Bonferroni adjustment. Statistical analysis was performed using SPSS for Windows (version 17.0; SPSS Inc., Chicago, IL), and P < 0.05 was considered to be significant.

Results

Pilot experiment

In the pilot experiment to determine the dose-related effects of tamoxifen, there were no significant differences in any μCT parameters analyzed between the vehicle and low-dose groups, either in contralateral control values or in loading-related changes. The medium- and high-dose groups had significantly higher trabecular BV/TV and trabecular number compared with the vehicle group in both SHAM and OVX mice (Fig. 2, A and C). In OVX, but not SHAM, mice, trabecular thickness was slightly but significantly lower in the medium- and high-dose groups than in the vehicle group.

Figure 2

Effects of tamoxifen at low (0.02 mg/kg · d), medium (0.2 mg/kg · d), and high (2 mg/kg · d) doses on trabecular bone in female SHAM or OVX mice. The SHAM or OVX operation was performed on d 1; drug injection and external mechanical loading on d 11, 13, 15, 18, and 21; and collection for analysis on d 25. A, Relative contralateral control values of μCT parameters; B, relative loading-related changes [(right loaded − left control)/left control] of μCT parameters; C, representative transverse μCT images of the left control and right loaded trabecular bone (0.5 mm distal to the growth plate) in the proximal tibiae of OVX mice. Values are presented as the mean ± se (n = 6–8 per group). *, P < 0.05 vs. vehicle in SHAM (set at a value of 1); #, P < 0.05 vs. vehicle in OVX by one-way ANOVA followed by a post hoc Bonferroni or Dunnett T3 test.

As shown in Fig. 2, B and C, the medium-dose group showed significantly higher loading-related increases in trabecular BV/TV and trabecular thickness than in the vehicle group in both SHAM and OVX mice. In contrast, although the loading-related increase in trabecular thickness was higher in the high-dose group than in the vehicle group, there was no difference in trabecular BV/TV between these two groups.

In contrast to trabecular bone parameters, there were no apparent differences in cortical bone parameters among the three different dose groups, either in contralateral control values or in loading-related changes (data not shown).

Main experiment

Based on the results of the pilot experiment, we chose the medium dose of tamoxifen because this dose induced significantly higher loading-related gain in trabecular BV/TV compared with the low and high doses. In this experiment, uterine weight as well as body weight was also measured.

Effects of OVX on body weight, uterine weight, and contralateral control bone

Increase in body weight and uterine weight in the OVX plus vehicle group were higher and lower, respectively, than those in the SHAM plus vehicle group. The latter difference was significant (Fig. 3). The OVX plus vehicle group had significantly lower trabecular BV/TV compared with the SHAM plus vehicle group, whereas cortical bone parameters in these two groups showed no difference (Figs. 4, A–C, and 5, A–C).

Figure 3

Effects of tamoxifen (0.2 mg/kg · d) and fulvestrant (4 mg/kg · d) alone or in combination on body weight and uterine weight in female SHAM or OVX mice. The SHAM or OVX operation was performed on d 1; drug injection and external mechanical loading on d 11, 13, 15, 18, and 21; and collection for analysis on d 25. A, Changes in body weight; B, uterine weight. Values are presented as the mean ± se (n = 7 per group). *, P < 0.05 vs. vehicle in SHAM; #, P < 0.05 vs. vehicle in OVX; ‡, P < 0.05 vs. tamoxifen in OVX by one-way ANOVA followed by a post hoc Bonferroni test.

Figure 4

Effects of tamoxifen (0.2 mg/kg · d) and fulvestrant (4 mg/kg · d) alone or in combination on trabecular bone in female SHAM or OVX mice. The SHAM or OVX operation was performed on d 1; drug injection and external mechanical loading on d 11, 13, 15, 18, and 21; and collection for analysis on d 25. A, BV/TV; B, trabecular number (Tb.N); C, trabecular thickness (Tb.Th); D, loading-related changes [(right loaded − left control)/left control] of μCT parameters in SHAM mice; E, loading-related changes of μCT parameters in OVX mice; F, representative transverse μCT images of the left control and right loaded trabecular bone (0.5 mm distal to the growth plate) in the proximal tibiae of OVX mice. Values are presented as the mean ± se (n = 7 per group). *, P < 0.05 vs. vehicle in SHAM; #, P < 0.05 vs. vehicle in OVX; ‡, P < 0.05 vs. tamoxifen in OVX by mixed-model analysis followed by Bonferroni adjustment (A–C). *, P < 0.05 by mixed-model analysis followed by Bonferroni adjustment (D and E).

Figure 5

Effects of tamoxifen (0.2 mg/kg · d) and fulvestrant (4 mg/kg · d) alone or in combination on cortical bone in female SHAM or OVX mice. The SHAM or OVX operation was performed on d 1; drug injection and external mechanical loading on d 11, 13, 15, 18, and 21; and collection for analysis on d 25. A, Cortical bone volume (CBV); B, periosteally enclosed volume (PEV); C, medullary volume (MV); D, loading-related changes [(right loaded − left control)/left control] of μCT parameters in SHAM mice; E, loading-related changes of μCT parameters in OVX mice. Values are presented as the mean ± se (n = 7 per group). There were no significant differences in A–C. *, P < 0.05 by mixed-model analysis followed by Bonferroni adjustment (D and E).

Effects of tamoxifen and fulvestrant on body weight, uterine weight, and contralateral control bone

Increases in body weight were similar between the vehicle and tamoxifen groups in both SHAM and OVX mice (Fig. 3A). Uterine weight in the tamoxifen group was significantly higher than in the vehicle group in OVX, but not SHAM, mice (Fig. 3B). The tamoxifen group had significantly higher trabecular BV/TV and trabecular number compared with the vehicle group in both SHAM and OVX mice, but no differences in cortical bone parameters were detected between these two groups (Figs. 4, A–C, and 5, A–C).

There were no differences in the increase in body weight between the vehicle and fulvestrant groups in SHAM or OVX mice (Fig. 3A). The fulvestrant group had significantly lower uterine weight compared with the vehicle group in SHAM, but not OVX, mice (Fig. 3B). No differences in trabecular or cortical bone parameters were observed between these two groups (Figs. 4, A–C, and 5, A–C).

In OVX mice, the increase in body weight was similar between the tamoxifen and tamoxifen plus fulvestrant groups, whereas uterine weight in the tamoxifen plus fulvestrant group was significantly lower than in the tamoxifen group (Fig. 3). The tamoxifen plus fulvestrant group had significantly lower trabecular BV/TV and trabecular number compared with the tamoxifen group in OVX mice, but there were no differences in cortical bone parameters between these two groups (Figs. 4, A–C, and 5, A–C).

Effects of OVX, tamoxifen, and fulvestrant on bones’ response to mechanical loading

The mixed-model analysis confirmed significant loading-related increases in trabecular BV/TV and cortical bone volume and showed that OVX was associated with a slight but significant increase in loading-related gain in trabecular BV/TV. In the SHAM plus vehicle and OVX plus vehicle groups, loading-related increases in trabecular BV/TV were 12.6 ± 3.7 and 18.3 ± 4.1%, respectively. In contrast, no difference in the loading-related change in cortical bone volume was detected between SHAM and OVX mice.

The group receiving tamoxifen had significantly higher loading-related increases in trabecular BV/TV, trabecular thickness, cortical bone volume, and periosteally enclosed volume compared with the group receiving vehicle in both SHAM and OVX mice (Figs. 4, D–F, and 5, D and E). There was statistically significant synergism (P < 0.01) between the effects of loading and tamoxifen in trabecular BV/TV and trabecular thickness in OVX as well as SHAM mice.

No differences in loading-related changes in trabecular and cortical bone parameters were detected between the vehicle and fulvestrant groups in SHAM or OVX mice. In OVX mice, however, the tamoxifen plus fulvestrant group showed significantly lower loading-related increases in trabecular BV/TV and trabecular thickness, but not cortical bone parameters, compared with the tamoxifen group (Figs. 4, D–F, and 5, D and E).

Discussion

Probably the most significant finding from our present experiments is that the ER modulator tamoxifen enhances loading-related gain in trabecular BV/TV. In contrast to the effect of tamoxifen alone, which stimulates an increase in trabecular number, the increase in loading-related BV/TV was achieved by an increase in trabecular thickness. This effect was most evident in mice that had been OVX and thus had no ovary-derived estrogen. Fulvestrant, an antagonist of ER, by itself had no effect on loading-related changes in bone mass or architecture. However, in OVX mice, it abrogated not only the increase in trabecular BV/TV induced by tamoxifen in nonloaded control bones but also tamoxifen’s enhancement of loading-related gain in trabecular BV/TV on bones that were loaded. This suggests that both these effects of tamoxifen are mediated by the ER.

Tamoxifen is a selective ER modulator. In contrast to its estrogen-antagonistic effect in human breast cancer, studies in the rat (29,30), confirmed in the human (31,32), revealed that in bone, tamoxifen is an estrogen agonist. The effects of different doses of tamoxifen observed in our pilot experiment are consistent with previous findings that in female ovary-intact mice, tamoxifen stimulates trabecular bone formation in a dose-dependent manner (33). The medium dose of tamoxifen in our present study increased trabecular BV/TV by increasing trabecular number, but not trabecular thickness, in both SHAM and OVX mice. This supports the inference from a previous report that in short-term experiments in mice, 4-hydroxytamoxifen, an active metabolite of tamoxifen, induces trabecular bone gain by increasing trabecular number (34). Tamoxifen’s enhancement of loading-related trabecular bone gain could theoretically be attributed to its direct effects on stimulating bone formation (33) and/or suppressing resorption (30). In our experiments, it is most likely to be the former, because the primary effect of the present short-term loading model is increased osteogenesis (21,28), and this has been shown to be unaffected by suppressed resorption (35,36). The medium dose of tamoxifen used in the main experiment is equivalent to a supraphysiological level of estrogen (33), which would be expected to have its effect on osteogenesis through an ER-mediated mechanism (25,37). Although the points at which estrogen intervenes in bone cell metabolism to produce its effects are unclear, they have been shown to involve at least one mechanical strain-sensitive response, because strain stimulates nitric oxide (NO) production by rapid activation of endothelial NO synthase in osteocytes (38), and the osteogenic effects of estrogen are reduced in mice lacking endothelial NO synthase (39). Strain-related activation of IGF-I receptor, in association with ERα, also requires NO production in osteoblasts in vitro (12). Thus, estrogen may enhance loading-related osteogenesis by this means. However, the effect of estrogen on the outcome of bones’ loading-related responses is unlikely to be simple, or act at one location, in the many processes involved. For instance, in trabecular bone, the presence of estrogen at the time of loading suppressed the activation of new bone formation sites but later enhanced the amount of bone formed (24).

Although it is tempting to extrapolate from the data presented here to suggest that tamoxifen and exercise-related loading in humans could act together to enhance osteogenesis in trabecular bone, species differences in target tissues’ responses to tamoxifen must be acknowledged. In contrast to its effect in humans, tamoxifen in the mouse is a pure estrogen agonist (40). Also, although it acts as an estrogen agonist in bone in both species, the changes in bone architecture in mice and humans are significantly different (41). Finally, there is clinical evidence that estrogen increases bone formation in humans (42,43,44), but the effect of tamoxifen is unclear. It is well established that tamoxifen decreases circulating bone turnover markers in women (45,46).

Fulvestrant is considered to be a pure estrogen antagonist, and in ovary-intact rats, it reduces bone mass (47,48,49). Our finding that fulvestrant itself did not affect loading-related responses of either trabecular or cortical bone was unexpected, because it is inconsistent with our previous in vitro studies in which it blocked strain-related proliferative responses in both rat and human primary osteoblastic cells (5,6,7). Because the dose of fulvestrant used in the present study (4 mg/kg · d) not only markedly decreased uterine weight in SHAM mice but also abrogated the tamoxifen-related effects on uterine weight as well as trabecular bone in OVX mice, the lack of effect on the loading-related response is unlikely to be due to an inadequate dose of the drug. In nonloaded control tibiae, fulvestrant also showed no significant effects on trabecular or cortical bone. This is in agreement with a previous report that fulvestrant at a similar dose had no effect on bone formation in intact female mice, although it significantly inhibited the osteogenic response to high-dose 17β-estradiol (25). Interestingly, in both premenopausal and postmenopausal women, recent pilot clinical trials reported that fulvestrant had no significant effects on bone turnover or areal bone mineral density (50,51,52,53). Because fulvestrant is an antagonist of both ERα and ERβ (54), the lack of any significant effect of fulvestrant on bones’ loading-related responses may be associated with these receptors’ opposing effects on bone adaptation to loading in vivo (13,15,18). This possibility may also explain why OVX alone did not affect loading-related effects on cortical bone, as reported previously (55).

In conclusion, our present data suggest that in female mice the ER modulator tamoxifen enhances trabecular bone’s response to mechanical loading by an increase in trabecular thickness. This effect is ER mediated. Fulvestrant (ICI 182,780) alone has no effects on bone architecture or the effects of loading on changes in bone architecture. A variety of ER modulators, each with different activities, have been developed. Appropriate modulation of ER activity to enhance loading-related regulation of bone mass by one of these could be an attractive prospect, because such enhanced loading-related responsiveness would involve distributing bone tissue in a structurally appropriate manner.