Introduction

Osteoporosis is a metabolic bone disease characterized by low bone mass, deterioration in bone microarchitecture and increased fracture risk. The main reason for bone loss in primary osteoporosis, but also in some secondary types of osteoporosis like in cancer treatment induced bone loss (CTIBL), is sex steroid withdrawal. CTIBL is a side effect of many anti-hormonal cancer treatments like anti-estrogens, aromatase inhibitors and androgen deprivation therapy [1]. In addition, regular chemotherapy causes ovarian failure in 40–70% of premenopausal patients, depending on age and type of chemotherapy, which may affect bone health [2]. Furthermore, it has been shown in several preclinical studies [3–5] that increased bone resorption can increase the incidence and severity of bone metastases. If this is true also in humans, the bone effects of new anti-tumor compounds need to be carefully evaluated.

Among the most commonly prescribed anticancer drugs are compounds that target microtubules. One of them is the taxane paclitaxel, used in treatment of metastatic breast cancer. It inhibits tumor cell division by inducing a mitotic block at the metaphase–anaphase transition, thereby leading to apoptosis [6]. In contrast, the epothilones isolated from myxobacteria comprise a novel class of microtubule stabilizers that are structurally distinct from the taxanes, yet have similar mechanisms of action, leading to cell cycle arrest and apoptosis. Sagopilone (ZK-EPO), the first fully synthetic epothilone in clinical development, was rationally designed to overcome limitations of other microtubule stabilizers [7]: Sagopilone is not a substrate of P-gp efflux pumps leading to more efficient retention in tumor cells and increased antiproliferative effects, it is water-soluble favouring formulation, it is able to cross the blood–brain barrier in humans and it is active in taxane-resistant models in vitro and in vivo [8]. Sagopilone has shown in vitro and in vivo activity at sub-nanomolar concentrations (IC50 <1 nM) with balanced tolerability against a broad range of tumor models, including superior activity compared with paclitaxel in many breast cancer cell lines [7, 9, 10]. Sagopilone has also showed good potential in recently reported phase I and II clinical trials [11–13]. Furthermore, sagopilone was shown to have a dual inhibitory effect on bone metastases in the MDA-MB-231(SA) mouse model of breast cancer bone metastasis [14], indicating an inhibitory effect of this compound on both osteoclasts and tumor cells. These data suggest that sagopilone might be efficacious also in other diseases with increased bone resorption. Thus, we here investigate its effects in a mouse model of ovariectomy induced osteoporosis and analyze its antiresorptive effects in vitro and in vivo.

Materials and methods

Osteoclast activity and differentiation assay

The osteoclast activity assay was performed by Pharmatest Services Ltd., Turku, Finland (www.pharmatest.fi). Briefly, human osteoclast precursor cells were differentiated for 7 days without test compounds. The amount of secreted TRACP 5b was determined at day 7 as an index of the number of osteoclasts formed [15]. Sagopilone (2.5–50 nM) or paclitaxel (2.5–200 nM) was added at day 7 and mature osteoclasts were cultured for an additional 3 days, allowing them to resorb bone. The level of carboxy-terminal cross-linking telopeptide of type I collagen (CTX) was measured at day 10 to determine bone resorption activity during days 7–10. Cytotoxicity was determined by the level of dying cells in the culture medium at day 10 using Toxilight® BioAssay Kit (Lonza, Verviers, Belgium). trans-Epoxysuccinyl-l-leucylamido-(4-guanidino)butane (E64), a cysteine protease inhibitor that shows selectivity for cathepsin B and is known to inhibit osteoclast activity [16], was used as a control compound at 1.0 μM.

Furthermore, we tested the short-term effects of sagopilone and paclitaxel on human osteoclast differentiation and activity to mimic the rapid clearance from blood in mice. The osteoclast activity assay was performed as described above with the difference that the compounds were added only for 2 h at day 7 and not for the whole resorption period. The following concentrations were tested: 10, 50, 20 and 50 nM sagopilone; 20, 50, 100, and 200 nM paclitaxel. After 2 h of incubation, the compounds were washed away by removing the culture medium and adding new medium without test compounds into the well. E64 (1 μM) was again used as a reference inhibitor of osteoclast activity and added into the cultures at day 7.

In the osteoclast differentiation assay, sagopilone and paclitaxel were added into the cultures at day 0 and washed away after 2 h as described above. The following concentrations were tested: 0.1, 0.5, 1, 5, and 10 nM sagopilone; 0.5, 1, 5, 10, and 20 nM paclitaxel. Osteoprotegerin (OPG, 100 ng/ml) was used as a reference inhibitor of osteoclast differentiation and also added into the cultures at day 0. TRACP 5b activity was measured from the culture medium collected at day 7 and CTX was measured from the culture medium collected at day 10. TRACP staining to visualize osteoclasts and to evaluate cytotoxic effects of the compounds on osteoclasts was performed as described by Rissanen et al. [15].

OVX study

Animal studies were conducted in accordance to the German Animal Welfare Act of 1998 and with approval from the responsible authorities. Female C3H/HeN mice (Charles River) were maintained under pathogen-free, controlled conditions. Animals were randomized to groups (SHAM, OVX, OVX + E2, OVX + sagopilone, n = 10) according to their body weight. Surgical operations were performed on 3-month-old mice. Ovaries were removed together with oviducts and a small portion of uterus. Treatment was started at day 4 after surgical operations with either sagopilone (8 mg/kg i.v., single injection every 14 days) or estrogen (as pellet 0.01 mg/60 days; Innovative Research of America, Sarasota, FL, USA). Body weights were determined three times a week. At sacrifice (day 42) body weights and uterine weights were determined. In addition, hind limbs were collected to measure bone mineral density (BMD) by peripheral quantitative computed tomography (pQCT) measurements.

Peripheral quantitative computed tomography

BMD was measured from tibial metaphysis and diaphysis using pQCT. The pQCT measurements were performed ex vivo after sacrifice. For metaphyseal measurements, the site of CT scan was at the proximal end of tibia 1.6 mm distally from the articular surface. For diaphyseal measurements, the site of CT scan was at the tibial shaft 10 mm distally from the articular surface. Total BMD, trabecular BMD (metaphysis) and cortical BMD (diaphysis) were determined.

Statistical methods

The results are reported as mean ± standard deviation (SD), and the level of significance is P < 0.05. Statistical analysis was performed using one-way analysis of variance (ANOVA) with t-test as post hoc test for comparing all groups against the baseline group in the in vitro tests and all groups against the OVX group in the animal experiment.

Results

Anti-resorptive effect of sagopilone in vitro

Sagopilone inhibited osteoclast activity at 15 nM concentration in the in vitro human osteoclast activity assay (Fig. 1a) without cytotoxic effects on osteoclasts (Fig. 1c). In comparison, while paclitaxel also showed an inhibitory effect on osteoclast activity, although to a lesser extent (Fig. 1b), this was associated with significant cytotoxic effects at several dose levels (Fig. 1d). The reference inhibitor E64 significantly inhibited osteoclast activity and showed no cytotoxic effects, demonstrating that the results obtained are reliable.

Fig. 1
figure 1

a–d Sagopilone reduces the resorption activity of human osteoclasts with no cytotoxic effects in a human osteoclast activity assay. The mean osteoclast activity in sagopilone-treated (a) and paclitaxel-treated (b) cells expressed as the resorption index (CTX concentration at day 10/TRACP 5b concentration at day 7). Cytotoxicity in sagopilone-treated (c) and paclitaxel-treated (d) cells determined as the luminescence released from dying cells during the culture period. e–h Short-term sagopilone treatment inhibits human osteoclast differentiation but not activity in vitro. Effects of sagopilone (e) and paclitaxel (f) on osteoclast differentiation at day 7 shown as TRACP 5b activity (U/l) secreted into the culture medium during the differentiation period (days 1–7). Effects of sagopilone (g) and paclitaxel (h) on osteoclast activity at day 10 expressed as the resorption index (CTX concentration at day 10/TRACP 5b concentration at day 7). Mean ± SD values ate shown. BL baseline. *p < 0.05, **p < 0.01, ***p < 0.001

Full size image

Both compounds, sagopilone and paclitaxel, showed inhibitory effects after short-term treatment in the osteoclast differentiation assay (Fig. 1e and f), but had no inhibitory effects in the osteoclast activity assay (Fig. 1g and h). Sagopilone inhibited osteoclast differentiation at 5 nM and 10 nM concentrations, whereas paclitaxel showed inhibitory effects only at 20 nM concentrations. Based on microscopic evaluation neither of the compounds showed clear cytotoxic effects in the osteoclast differentiation and activity assay after short-term treatment of compounds (data not shown).

Anti-resorptive effect of sagopilone in vivo

The effect of sagopilone on bone resorption in vivo in the absence of tumor was studied in ovariectomized mice. Body weights of OVX mice increased and uterine weights decreased, except in the OVX + E2 group, as expected (data not shown). Body weights of sagopilone treated animals were significantly lower compared to the OVX animals (P = 0.002), whereas uterine weights were similar in these groups (P = 0.157) (data not shown). Six weeks after surgery, osteoporotic changes were observed by pQCT and CT measurements. OVX caused significant bone loss compared to sham-operated animals (P = 0.001) indicated by reduced BMD in the tibial metaphysis as determined by pQCT measurements (Fig. 2a). The effects of OVX on trabecular bone were prevented by estrogen replacement therapy (P < 0.001 for OVX vs. OVX + E2). Sagopilone treatment resulted in significantly increased trabecular BMD in the tibial metaphysis compared to OVX mice (P = 0.009) suggesting an inhibitory effect of sagopilone on activated osteoclasts in vivo (Fig. 2a). No differences in cortical BMD in the tibial diaphysis were observed between the groups (Fig. 2b).

Fig. 2
figure 2

Trabecular and cortical bone mineral density 42 days after OVX. a Trabecular bone mineral density (BMD) in the tibial metaphysis significantly decreased after OVX as measured by pQCT, the effects were prevented by estrogen replacement. Sagopilone significantly increased BMD compared to OVX animals. b No differences in cortical BMD in the tibial diaphysis were observed between the groups. Mean ± SD values are shown. Statistics: All groups were compared separately to OVX group using t test. **p < 0.01, ***p < 0.001

Full size image

Discussion

Effect of sagopilone on bone resorption in vitro

The doses of the in vitro assay were chosen according to the pharmacokinetic data in mice. As we do not have any evidence that sagopilone accumulates in osteoclasts, it was assumed that the sagopilone concentration in bone is around the same as in the serum. The serum levels of the compounds in mice quickly decrease after i.v. application and the compounds accumulate in the tumor. Sagopilone incorporates more efficiently in the cells compared to paclitaxel. It was reported that paclitaxel also has in vitro effects on osteoclastic bone resorption at therapeutically relevant concentrations, and that 10 μM paclitaxel had cytotoxic effects on rat osteoclasts [17]. In contrast, the results presented here indicated cytotoxic effects of paclitaxel at 0.05 μM in the osteoclast activity assay. This might be a species specific difference. Furthermore, we had longer incubation periods with paclitaxel. Our in vitro assay with long-term treatment of compounds during the whole resorption period indicated that sagopilone inhibits human osteoclast activity and bone resorption to a greater extent than paclitaxel, but without the cytotoxic effects associated with paclitaxel treatment.

The inhibition of the growth and activity of non-proliferating end cells, such as osteoclasts, by sagopilone may partly be due to interference with non-mitotic microtubule-dependent functions, which include exocytosis, cell morphology, polarization and adhesion. Indeed, the inhibition of endothelial cell adhesion has previously been reported with paclitaxel [18]. Normal adhesion is required for adequate osteoclast function because osteoclasts are activated by contact with the mineralized bone matrix [19]. Moreover, under non-adherent conditions, the formation of multinuclear osteoclasts is decreased, thus leading to coverage of a smaller area of bone and further reduction in activity [20]. The exocytosis of extracellular vesicles, important for several processes during normal bone resorption, including degradation of the bone matrix [17, 21] may also be blocked by sagopilone. It might be especially interesting to clarify the mechanism of sagopilone on osteoclasts to reveal a new target which might be an approach for targeted therapy for osteoporosis.

However, the inhibition of osteoclast activity might be important only in humans whose pharmacokinetic of sagopilone is higher compared to mice (data not shown). Therefore, osteoclasts might be exposed for a longer time to sagopilone, so that it can exert effects on osteoclast activity without cytotoxic effects as demonstrated in vitro. In contrast, a short loading dose of compound of 2 h, as performed in the second setting of the osteoclast assay (Fig. 1e–h), indicates no effects on osteoclast activity but on osteoclast differentiation. Consequently, in mice sagopilone appears to inhibit earlier processes of osteoclast fusion and differentiation. The counting of nuclei per osteoclast would clarify if sagopilone inhibits fusion of osteoclast precursors into mature osteoclasts and might also explain the higher sensitivity of sagopilone in inhibiting osteoclast differentiation than activity.

Effect of sagopilone on bone resorption in vivo

The in vivo inhibition of bone resorption has been reported in the MDA-MB-231 mouse model of breast cancer bone metastasis by decreased number of activated osteoclasts in TRACP stained bone sections after sagopilone compared to vehicle treatment [14]. However, it is still unclear if this effect is mainly due to the anti-proliferative effect of sagopilone on tumor cells and consequent interruption of the vicious cycle at the tumor cell stage or to an additional, similarly strong inhibition of osteoclastic bone resorption or differentiation. First evidence of a direct anti-resorptive effect of sagopilone was obtained by the osteoclast activity assay which was performed with human osteoclasts in vitro. A short loading dose of compound only inhibited osteoclast differentiation and not activity. Finally, the possibly anti-resorptive effects of sagopilone were confirmed in vivo using the OVX model of osteoporosis in mice. Here, sagopilone showed efficacy in preventing bone loss in OVX mice indicated by significantly increased trabecular BMD compared to OVX animals. These data suggest sagopilone as an agent with direct inhibitory effects on osteoclast activity or differentiation and thus confirm the hypothesis of mode of action that sagopilone’s efficacy in treating bone metastases is due to the interruption of the vicious cycle on both tumor cell and osteoclast stages. However, in the OVX study a sagopilone dose (8 mg/kg) was used which is effective in tumor-bearing animals. This dose might be too high and possibly inhibiting osteoblasts so that the effect on bone mineral density is not optimal which might explain the low magnitude of the effect of sagopilone on BMD and the lower BMD in sagopilone treated mice compared to control mice. According to the in vitro assays, osteoclast differentiation is inhibited at 5–10 nM (short loading dose of compound) whereas the lowest osteoclast activity inhibiting concentration is 15 nM (long loading dose of compound). Considering the very fast pharmacokinetics of sagoplione in serum (data not shown), these concentrations of sagopilone correspond to a concentration of 3–5 mg/kg in vivo compared to the 8–10 mg/kg dose effective in tumor-bearing animals. The possibly too high dose is also reflected by a significant decrease in body weight in sagopilone treated animals compared to OVX mice caused by drug-related toxicity which is more evident in non-tumor bearing animals as no accumulation of the compound in tumor tissue occurs. As a consequence, in non-tumor-bearing mice fast-dividing somatic cells are possibly more affected by the compound compared to cells in tumor-bearing animals. However, the result of the pQCT measurements suggests that sagopilone is not at least reducing the bone density like many other cancer drugs (CTIBL) [1, 22] and appears to compensate the hypogonadal state by a simultaneous inhibitory effect on osteoclasts. Thus, by optimizing the sagopilone dose in the OVX model, it could be possible to obtain potent inhibition of osteoclast activation with less unwanted side effects and probably better preservation of osteoblast function.

To conclude, not only the OVX study in mice presented here but also previously published data [14] showed evidence of a direct inhibitory effect of sagopilone on osteoclasts in vivo. In light of recent preclinical studies showing increased bone metastasis due to increased resorption [3–5], the effect of a compound on bone metabolic balance becomes very important. Although the effect on BMD in this model is rather low due to not optimized dose regimens, the possibility that sagopilone is beneficial for bone, will strengthen the status of sagopilone as an anti-cancer compound compared to other microtubule stabilizing agents. Therefore, it is considered to be worthwhile to further evaluate the potential of sagopilone in the treatment of diseases associated with increased bone resorption.