Concurrent antitumor and bone-protective effects of everolimus in osteotropic breast cancer

Everolimus (RAD001) (catalogue number S1120) was purchased from Selleck Chemicals (Munich, Germany) and dissolved in dimethyl sulfoxide (DMSO). The following primary antibodies were purchased from Cell Signaling Technology, Inc. (Beverly, MA, USA): mTOR (catalogue number 2983), Phospho-mTOR (p-mTOR; catalogue number 2974), p70 S6 kinase (p70; catalogue number 9202), and phospho-p70 S6 kinase (p-p70; catalogue number 9205). An antibody for the housekeeping gene glyceraldehyde 3-phosphate dehydrogenase (GAPDH) (5G4) was purchased from HyTest Oy (Turku, Finland). HRP-conjugated mouse immunoglobulin G (IgG; catalogue number HAF007) and rabbit IgG (catalogue number HAF008) secondary antibodies were purchased from R&D Systems (Wiesbaden-Nordenstadt, Germany). Recombinant murine RANKL (catalogue number 462-TEC-010) and recombinant murine macrophage colony-stimulating factor (M-CSF; catalogue number 416-ML-010/CF) were purchased from R&D Systems (Minneapolis, MN, USA). Factors used to induce osteoblastic differentiation included dexamethasone (Sigma-Aldrich, Darmstadt, Germany), β-glycerol phosphate (Sigma-Aldrich), and ascorbate phosphate (Sigma-Aldrich). Recombinant human bone morphogenetic protein 2 (BMP-2; catalogue number 120-02) and BMP-4 (catalogue number 314-BP) used for murine osteoblast differentiation were purchased from PeproTech (Rocky Hill, NJ, USA). Calcein for labeling bone turnover was also purchased from Sigma-Aldrich.

Breast cancer cell lines MDA-MB-231 and MCF-7 and the murine melanoma cell line B16-F10 were purchased from the American Type Culture Collection (ATCC; Manassas, VA, USA). MDA-MD-231 cells transduced with the firefly luciferase gene (MDA-MB-231-LucA12) were a kind gift from Dr. Sanjay Tiwari (University of Kiel, Kiel, Germany). Breast cancer cells were cultured in Gibco DMEM/F-12 medium (Life Technologies GmbH, Darmstadt, Germany), and B16-F10 cells were cultured in 1× Gibco DMEM (Life Technologies GmbH).

RAW 264.7 cells (ATCC) and bone marrow-derived mononuclear cells derived from C57BL/6 mice were cultured in α-minimal essential medium (Biochrom, Berlin, Germany) supplemented with 2 mM glutamine (Biochrom). Cell cultures were maintained in a humidified atmosphere at 37 °C in 5% CO₂/95% air atmosphere, and all culture medium conditions were supplemented with 10% fetal calf serum supreme (FCS; Biochrom) and Gibco 1% penicillin/streptomycin (P/S; Life Technologies GmbH). RAW 264.7 cells were seeded at a density of 1.25 × 10⁴ cells/cm² and bone marrow-derived mononuclear cells at a density of 1 × 10⁶/cm² when commencing osteoclast differentiation.

Mononuclear cells were isolated from donor bone marrow samples via density centrifugation before seeding in 1× DMEM supplemented with 10% FCS and 1% P/S. Nonadherent cells were removed 24 h later by washing with PBS. The resulting adherent human mesenchymal stem cells (hMSC) were cultured until confluency before seeding and again allowed to reach confluency before treatment. This study was approved by the local institutional review board ethics committee (EK245082010), and informed consent was obtained from healthy donors before bone marrow samples were collected at the Bone Marrow Transplantation Center of the University Hospital Carl Gustav Carus. Murine mesenchymal stem cells (mMSC) were isolated by flushing the bone marrow cells from the long bones of C57BL/6 mice into a culture of 1× DMEM. A medium change was performed 48 h later to remove nonadherent cells in a similar manner to the selection of hMSC. All adherent cell cultures were recovered using Gibco 0.25% trypsin-ethylenediaminetetraacetic acid (Life Technologies GmbH) before seeding and waiting until cells reached confluency before commencement of treatments.

Osteoclasts were differentiated from RAW 264.7 cells in the presence of 50 ng/ml murine RANKL and increasing concentrations of everolimus for 5 days with replacement of culture medium, murine RANKL, and everolimus taking place every 48 h. On day 5, cells were washed and fixed in acetone/citrate buffer. The Leukocyte Acid Phosphatase TRAP Kit (Sigma-Aldrich, Vienna, Austria) was used to stain cells for tartrate-resistant acid phosphatase (TRAP) according to the instructions of the manufacturer. Cells that were positive for TRAP staining and showing three or more nuclei were counted as osteoclasts, and representative photographs were captured for each treatment condition. Murine bone marrow-derived mononuclear cells sourced by bone marrow aspiration from the long bones of C57BL/6 mice were differentiated in 25 ng/ml murine M-CSF for 2 days before continuing with 25 ng/ml murine M-CSF and commencing 50 ng/ml murine RANKL and everolimus treatment for 5 days and performing TRAP staining as described for RAW 264.7 cells.

Murine bone marrow-derived mononuclear cells were seeded onto bone slices (Immunodiagnostic Systems, Tyne & Wear, UK) with 25 ng/ml murine M-CSF and cultured for 2 days before medium was replaced with 25 ng/ml murine M-CSF and 50 ng/ml murine RANKL. Three days later, medium containing M-CSF and RANKL was replaced with the addition of everolimus. Everolimus-containing medium was changed 2 days later, and after a further 3 days of culture, supernatants were collected and analyzed for levels of collagen type I cross-linked C-telopeptide (CTx) using CrossLaps® for Culture (CTX-I) enzyme-linked immunosorbent assay (Immunodiagnostic Systems).

hMSC or mMSC were seeded into a 24-well culture plate, and upon reaching confluency (on day 1), cells were treated with osteogenic differentiation medium (DMEM containing 10% FCS, and 1% P/S supplemented with 100 μM dexamethasone, 10 mM β-glycerol phosphate, and 100 μM ascorbate phosphate) with increasing concentrations of everolimus. On day 21, cells were washed twice with PBS and fixed with 10% paraformaldehyde (PFA) for 15 minutes at room temperature. Following fixation, cells were washed twice with distilled water and incubated in a 40 mM alizarin red S solution, pH 4.2 (Sigma-Aldrich, Munich, Germany) for 20 minutes at room temperature. Stained cells were subsequently washed with distilled water until the excess dye was completely removed. Plates were dried at room temperature overnight before imaging. To quantify mineralization, the alizarin red S bound to the mineralized calcium was eluted in 0.1 M HCl/0.5% sodium dodecyl sulfate (SDS) solution for 30 minutes at room temperature. The resulting eluent was measured with a spectrophotometer at 540 nm. Each treatment was performed in triplicate for three different donors. In the murine osteoblast experiments, 100 ng/ml of BMP-2 and BMP-4 was added to the culture conditions.

Cancer cell lines and bone cells (differentiated RAW 264.7 cells and hMSC) were seeded onto 96-well and 24-well plates, respectively. The CellTiter-Blue® assay (Promega, Mannheim, Germany) was used according to the manufacturer’s instructions to evaluate cell viability at different time points following everolimus treatments at different concentrations (0, 1, 10, and 100 nM). Cancer cells were seeded at a density of 3000 cells per 96-well plate. hMSC were seeded at a density of 60,000 cells per 24-well plate and allowed to reach confluency before differentiation for 8 days and assessment of the effects of everolimus concentrations only after a treatment duration of 48 h. RAW 264.7 cells were seeded at a density of 25,000 cells per 24-well plate and differentiated with 20 ng/ml murine RANKL before being treated with everolimus for 48 h.

Polymerase chain reactions (PCRs) were performed as previously described [18]. Briefly, the High Pure RNA Extraction Kit (Roche Applied Science, Mannheim, Germany) was used to isolate RNA following the manufacturer’s protocol. Purified RNA (500 ng) was reverse-transcribed using SuperScript II (Life Technologies GmbH) and underwent SYBR Green-based real-time polymerase chain reaction (RT-PCR) using a standard protocol (Applied Biosystems, Foster City, CA, USA). Primer sequences are listed in Table 1. The PCR cycling program ran at 50 °C for 2 minutes and at 95 °C for 10 minutes, followed by 40 cycles at 95 °C for 15 seconds and 60 °C for 1 minute. The melting curve was assessed at 95 °C for 15 seconds, 60 °C for 1 minute, and 95 °C for 30 seconds. The comparative cycle threshold method was used to calculate the results, which are presented as the x-fold increase relative to the housekeeping gene (human β-actin or murine β-actin) or as a percentage of the control.

The analysis of protein expression by Western blotting was performed as described previously [19]. In short, following treatment with everolimus, cancer cells were lysed and protein levels quantified. Protein samples of 20 μg were loaded onto a 6% SDS-PAGE gel and separated by electrophoresis. The separated proteins were then transferred onto a 0.2-μm nitrocellulose membrane. Blocking was performed in 5% nonfat dry milk in Tris-buffered saline with 1% Tween 20 for 1 h. Membranes were then washed in Tris-buffered saline with 1% Tween 20 and incubated overnight in 5% bovine serum albumin in Tris-buffered saline with 1% Tween 20 containing the primary antibody (mTOR, phosphorylated mTOR, p70 S6 kinase, phosphorylated p70 S6 kinase, or GAPDH). Membranes were washed before incubation for 1.5 h with the HRP-conjugated secondary antibody in 1% nonfat dry milk in Tris-buffered saline with 1% Tween 20. After another washing step, the membranes were developed and the protein visualized using SuperSignal substrate (Pierce Biotechnology, Bonn, Germany) enhanced chemiluminescence. Phosphorylated protein signals were quantified and normalized to GAPDH signals using ImageJ version 1.44 software (imagej.nih.gov/ij/).

Female immunocompromised NMRI nude and immunocompetent C57BL/6 mice were housed under institutional guidelines. The institutional animal care committees of the Technical University Dresden and the Landesdirektion Dresden approved all animal procedures (IRB TVV 61/2015). In subcutaneous tumor models, NMRI nude and C57BL/6 mice were inoculated subcutaneously at 6 weeks of age with 1 × 10⁶ MDA-MB-231 and 1 × 10⁴ B16-F10 cells, respectively, in 50 μl of a 1:1 Matrigel matrix dilution with PBS on day 1. The site of subcutaneous inoculation was dorsal, between the positions of the last rib and the hind limb, with two injection sites per mouse, left and right. Intraperitoneal injections (100 μl) of 1 mg/kg/day everolimus or control DMSO commenced on day 2 for 4 weeks in the case of the MDA-MB-231 model and for 2 weeks in the case of the B16-F10 model before mice were killed to assess tumor burden. Ten mice were allocated to each treatment group. To establish a model of bone loss similar to that of the human condition, an estrogen-deprived environment was induced in the C57BL/6 strain by performing OVX in 9-week-old mice. Four weeks postsurgery, OVX or sham (SHAM)-operated groups were treated intraperitoneally with 1 mg/kg/day everolimus or control for 4 weeks. Ten mice were allocated to each group initially. However, two mice did not survive the OVX procedure, resulting in both SHAM groups having only nine animals each at the time mice were killed. Calcein (15 mg/kg) labeling was performed 5 and 2 days before mice were killed. Killing was done after 8 weeks. In the bone metastasis model, 1 × 10⁵ MDA-MB-231-LucA12 cells were injected into the left ventricle of the heart of 6-week-old NMRI nude mice under ultrasound guidance. Intraperitoneal injections of 100 μl, 1 mg/kg/day everolimus or control DMSO, commenced on the day of tumor cell inoculation. Ten mice per group were initially included. One mouse from each group died following anesthetic administration at the first imaging session and were therefore excluded from the experiment. No adverse effects were observed for treatment with everolimus in any of the experiments performed.

Micro-computed tomography (μCT, vivaCT 75; SCANCO Medical, Brüttisellen, Switzerland) was performed on the excised femurs using X-ray energy of 70 keV, a resolution of 10.5 μm, and an integration time of 200 milliseconds. Calibration of the scanner took place weekly using hydroxyapatite (HA) phantoms. For the 3D visualization of bony tissue, we used the SCANCO evaluation software (SCANCO Medical). The threshold for bone absorption values was set to 285 mg HA/cm³, and 100 slices were measured commencing from 10 slices above the growth plate of the femur.

The femur was fixed in 4% PFA/PBS for 24 h, dehydrated in an ascending ethanol series, and embedded in paraffin. The tibia was separately embedded in methyl methacrylate (Technovit 9100 NEW; Heraeus, Wehrheim, Germany). Sections (2 μm) were used to stain for TRAP in the femur, allowing for the assessment of the number of osteoclasts per unit of bone surface. Sections (7 μm) were cut from the tibia for the assessment of calcein labels to determine the bone formation rate per unit of bone surface (BFR/BS). Analysis and quantification of the bone histomorphometric parameters were performed using OsteoMeasure software (OsteoMetrics, Decatur, GA, USA). Relevant units for histomorphometric measurements were consistent with those advised by the nomenclature committee of the American Society for Bone and Mineral Research.

Bioluminescence imaging was used to quantify tumor growth by correlating the tumor burden to the luminescence signal measured with a Xenogen IVIS 200 in vivo imaging system (PerkinElmer, Rodgau, Germany). Successful intracardiac injection was determined immediately after intracardiac inoculation. Imaging for the assessment of tumor growth commenced 2 weeks postinoculation, and continuous assessment was performed once weekly until mice were killed at the end of week 5. Living Image software (PerkinElmer) was used to obtain and quantify the bioluminescence data. Mice were anesthetized, and 5 minutes prior to imaging, each mouse was given an intraperitoneal injection with a dose of 10 mg/kg D-luciferin (PerkinElmer) in PBS. Mice were imaged individually for an exposure period of 2 minutes. The resulting bioluminescent images were analyzed by measuring individual, manually contoured signals with final measurement units in photons per second per centimeter squared per steradian.

Each in vitro experimental setup was repeated a minimum of three times, and using Prism 6 software (GraphPad Software, Inc., La Jolla, CA, USA), one-way analysis of variance (ANOVA) with the Bonferroni posttest or Student’s t test was performed to evaluate the equality of the mean. To analyze the effects of OVX and everolimus treatment, two-way ANOVA with Tukey’s posttest was performed. The results are presented as SD of the mean, and a p value <0.05 was considered statistically significant.

In the B16-F10, MDA-MB-231, and MCF-7 cell lines, everolimus exerted a potent negative effect on the growth of all cell lines tested as assessed by the CellTiter-Blue cell viability assay (Fig. 1a). In the murine B16-F10 melanoma cell line, concentrations of 10 and 100 nM were effective at significantly impairing cell viability when assessed at 72 h (p < 0.05 and 0.01, respec tively). In the human breast cancer cell line MDA-MB-231, everolimus at a concentration as low as 1 nM was sufficient to induce significant suppression in viability, most notably after 72 h of treatment, when a 45% reduction in viability was observed compared with control treated cells (p < 0.05). Antitumor effects of everolimus were also apparent and equally effective for all three concentrations used in the ER-positive MCF-7 cell line. Effective inhibition of the mTOR pathway was confirmed by Western blot assessment of mTOR phosphorylation and the downstream target of mTOR, p70 S6 kinase (Fig. 1b). Increasing everolimus concentrations inhibited the phosphorylation of mTOR in a dose-dependent manner, with 100 nM inducing a significant suppression in all three cell lines investigated (p < 0.01) (Additional file 1: Figure S1). Interestingly, all concentrations were sufficient to significantly suppress the phosphorylation of p70 S6 kinase by ≥50% (p < 0.01) (Additional file 1: Figure S1). In murine models of subcutaneous tumor growth, everolimus at a dose of 1 mg/kg/day was sufficient to significantly inhibit the growth of B16-F10 and MDA-MB-231 cells over a period of 2 and 4 weeks for each respective tumor model. Tumor weight was reduced by 71% (345 ± 66 mg to 103 ± 25 mg, p < 0.01) and 81% (34 ± 5 to 7 ± 1 mg, p < 0.001) in the B16-F10 and MDA-MB-231 models, respectively (Fig. 1c).

RAW 264.7 osteoclastic precursor cells predifferentiated with RANKL for 5 days and exposed to everolimus for 48 h showed a significant reduction of cell viability with incremental decreases of 13%, 21%, and 28% for the increasing concentrations of 1, 10, and 100 nM everolimus, respectively (Additional file 2: Figure S2a). In addition, everolimus exerted potent negative effects on osteoclast formation. The number of TRAP-positive cells developing in the presence of RANKL was reduced by 58% at an everolimus concentration of 1 nM (p < 0.001) (Fig. 2a). In accordance with this, markers of osteoclast differentiation, including Trap and Oscar, were significantly reduced by all concentrations of everolimus (p < 0.001) (Fig. 2b). There was a trend that the expression of Cstk was also reduced at all everolimus concentrations; however, this observation was not significant. Comparable effects of everolimus were observed in osteoclasts differentiated from murine bone marrow-derived mononuclear cells, where concentrations of 10 nM were sufficient to completely block osteoclastogenesis and expression of osteoclast marker genes (Fig. 2c and d). Of note, primary murine cells were more resistant to the lowest concentration of 1 nM everolimus, where no inhibitory effects were observed on osteoclast differentiation. Osteoclasts derived from primary murine cells were also used to assess the effect of everolimus on functional bone resorption by mature osteoclasts in vitro. Here, mononuclear cells were differentiated into osteoclasts on bone slices, and after a treatment period of 5 days with everolimus, it could be clearly observed that concentrations of 10 and 100 nM significantly decreased the levels of the bone resorption marker CTx present in the culture supernatants by 34.2% and 33.4% (p < 0.01), respectively (Additional file 3: Figure S3).

In human preosteoblasts derived from hMSC, markers of osteoblastogenesis ALP, OPG, RUNX2, and OCN were quantified by qRT-PCR on day 7 following the addition of osteoblastic differentiation medium and increasing everolimus concentrations (Fig. 3a). Here, the messenger RNA expression of ALP, a reliable marker of osteoblast activity, was negatively affected only at higher concentrations of everolimus (10 and 100 nM). In fact, the expression of OPG, RUNX2, and OCN significantly increased at 1 nM and remained elevated at 10 nM for RUNX2 and OCN. No inhibitory effect of everolimus on mineralization as assessed by alizarin red staining was observed after 21 days of exposure to everolimus at concentrations of up to 100 nM (Fig. 3b). When assessing the same parameters in murine preosteoblasts derived from mMSC, we observed that the expression of osteoblast marker genes following differentiation for 7 days in the presence of everolimus showed increasing reductions in Alp and Ocn from a concentration of 10 nM and only at 100 nM for Opg and Runx2 (Fig. 3c). Whereas 1 nM of everolimus maintained the osteoblast expression profile of all the genes assessed, we could not observe any pro-osteoblastic effects as seen in the osteoblastic gene signature of differentiating hMSC. When the mineralizing ability of everolimus-treated murine osteoblasts was assessed at 21 days, a significant decrease of 29% between control treated cells and cells treated with 1 nM of everolimus was observed (p < 0.01). However, increasing concentrations of everolimus did not result in further impairment of mineralization. When investigating whether everolimus has implications on osteoblastic metabolism, we observed that concentrations increasing from 1 nM to 100 nM did not show any effects on the viability of human preosteoblasts (Additional file 2: Figure S2b)

Because the majority of patients with breast cancer are postmenopausal when they present for diagnosis, or undergo hormone depletion therapies over the course of treatment, we wanted to recapitulate the hormone-deprived microenvironment in an animal model. To this end, 9-week-old wild-type C57BL/6 mice underwent OVX to induce an environment of high bone turnover and bone loss. Treatment with everolimus commenced 4 weeks post-OVX, and mice were treated with 1 mg/kg/day. Assessment of bone was performed after 4 weeks of treatment. As expected, there was a decrease in bone mineral density (BMD) in the OVX group compared with the SHAM-operated group by 27.49% (45.00 ± 24.85 vs. 32.63 ± 14.58) at the femur. Treatment with everolimus had a significant effect on BMD, restoring OVX-induced bone loss (Fig. 4a). BMD of the everolimus-treated OVX group was 38.44% (53.00 ± 16.57 vs. 32.63 ± 14.58) higher than that of the OVX control group. The average bone volume over total volume (BV/TV) of everolimus-treated OVX mice was also 37% higher than in the control OVX mice (2.62 ± 0.85 vs. 1.67 ± 0.75). These results were echoed by an increase in trabecular number (2.62 ± 0.43 vs. 2.07 ± 0.41, p < 0.01) and a decrease in trabecular separation (0.40 ± 0.07 vs. 0.51 ± 0.11, p < 0.01) in everolimus-treated OVX mice versus control OVX mice (Fig 4a). Analysis of bone histomorphometry demonstrated a 25% reduction in the number of osteoclasts in contact with the bone surface (14.54 ± 5.08 to 10.87 ± 2.89) in the everolimus-treated OVX mice when compared with control OVX mice (Fig. 4b), also depicted as whole sections (Additional file 4: Figure S4). Correspondingly, the BFR/BS increased in control OVX mice by 30% (from 0.64 ± 0.26 to 0.91 ± 0.11) when compared with the rate of control SHAM animals, and everolimus was able to reverse this increase by 41.5% (0.91 ± 0.11 to 0.53 ± 0.23, p < 0.01) (Fig. 4b). This demonstrates that everolimus prevents the high bone turnover and bone loss that is induced by OVX.

Having established the bone-protective effects of everolimus at a concentration capable of exerting antitumor potential, the ability of everolimus to inhibit the development of osteolytic breast cancer bone metastases was assessed. Firefly luciferin-labeled MDA-MB-231 breast cancer cells (MDA-MB-231-LucA12) were intracardially injected into 6-week-old NMRI nude mice. Images of all injected mice with the observed bioluminescent signal at sites of tumor burden are shown prior to their being killed at day 36 (Fig. 5a). The number of overt lesions per mouse in animals with bioluminescent signals >1 × 10⁷ photons/second/cm²/sr were counted and compared between the groups. Everolimus-treated animals had a significantly reduced number of overt lesions compared with control animals (−70.4%, 7.33 ± 5.32 to 2.17 ± 1.17, p < 0.05) (Fig. 5b). Each metastatic signal was also individually quantified, and this was reduced by 45.4% (8.62 ± 8.69 to 4.71 ± 3.86, p < 0.01) in the everolimus group compared with the control group (Fig. 5c). Osteolytic lesions corresponded with bioluminescent signals and could be visualized by reconstructing μCT scans of analyzed bones. Representative images of affected and unaffected femurs and tibiae are shown (Fig. 5d). Trabecular bone quality in the femurs of these animals was assessed by μCT analysis (Fig. 5e). Animals in the everolimus-treated group had an increased BMD and BV/TV of >50% (p < 0.001) compared with the placebo-treated group. Trabecular parameters reflected this observation, with the control group having 24.25% (2.33 ± 0.52 to 3.07 ± 0.75, p < 0.01) less trabeculae and 31.52% (0.46 ± 0.12 to 0.35 ± 0.11, p < 0.01) more separation between the trabeculae than the everolimus group. Interestingly, the total number of osteoclasts in the femurs of everolimus-treated mice were decreased by 42.88% (98.30 ± 39.21 to 58.11 ± 19.68, p < 0.05) compared with control-treated mice (Additional file 5: Figure S5). This experiment confirmed combined antiosteoclastic and antitumor effects of everolimus in the metastatic bone microenvironment.