Everolimus restrains the paracrine pro-osteoclast activity of breast cancer cells

Background Breast cancer (BC) cells secrete soluble factors that accelerate osteoclast (OC) differentiation, leading to the formation of osteolytic bone metastases. In the BOLERO-2 trial, BC patients with bone involvement who received Everolimus had a delayed tumor progression in the skeleton as a result of direct OC suppression through the inhibition of mTOR, in addition to the general suppressor effect on the cancer cells. Here, we explored the effect of Everolimus, as mTOR inhibitor, on the pro-OC paracrine activity of BC cells. Methods Both MDA-MB-231 and MCF-7 BC cell lines were incubated with sub-lethal amounts of Everolimus, and their conditioned supernatants were assessed for their capacity to differentiate OCs from PBMC from healthy donors, as well as to interfere with their bone resorbing activity shown on calcium phosphate slices. We also measured the mRNA levels of major pro-OC factors in Everolimus-treated BC cells and their secreted levels by ELISA, and evaluated by immunoblotting the phosphorylation of transcription factors enrolled by pathways cooperating with the mTOR inhibition. Finally, the in vivo pro-OC activity of these cells was assessed in SCID mice after intra-tibial injections. Results We found that Everolimus significantly inhibited the differentiation of OCs and their in vitro bone-resorbing activity, and also found decreases of both mRNA and secreted pro-OC factors such as M-CSF, IL-6, and IL-1β, whose lower ELISA levels paralleled the defective phosphorylation of NFkB pathway effectors. Moreover, when intra-tibially injected in SCID mice, Everolimus-treated BC cells produced smaller bone metastases than the untreated cells. Conclusions mTOR inhibition in BC cells leads to a suppression of their paracrine pro-OC activity by interfering with the NFkB pathway; this effect may also account for the delayed progression of bone metastatic disease observed in the BOLERO-2 trial. Electronic supplementary material The online version of this article (doi:10.1186/s12885-015-1717-8) contains supplementary material, which is available to authorized users.


Background
The development of skeletal metastases, leading to hypercalcemia, fractures, and vertebral collapse with spinal compression and intractable pain, is a major cause of morbidity in patients with advanced breast cancer (BC) [1]. Their formation is primarily induced by a marrow environment fostering tumor growth, thanks to the functional cross-talk between BC cells and resident cells such as hematopoietic progenitors, stromal, endothelial and other marrow components [2]. Although a direct bone resorption activity by BC cells was postulated [3], it is currently believed that osteolytic lesions are produced by marrow osteoclasts (OC), whose differentiation from myeloid precursors is accelerated by the tumor cells [4]. BC cells secrete PTHrP (parathormone-related protein) [5], in addition to other soluble factors that are normally secreted by stromal and other marrow cells to regulate the OC maturation, and have M-CSF (macrophage-colony stimulating factor), IL (interleukin)-1β, IL-8, IL-6 and TNF-α (tumor necrosis factor) functions [6][7][8].
Specific signatures of molecular pathways in BC cells have been associated with a highly malignant phenotype [9] and worse clinical outcome [10]. For instance, hyperactivity of the phosphoinositide 3-kinase (PI3K)/Protein Kinase B (Akt)/mammalian target of Rapamycin (mTOR) pathway, is a molecular hallmark of accelerated proliferation and tumor progression in BC cells, since intrinsic molecular aberrations recur in approximately 40 % of patients with metastatic skeleton disease [11]. The constitutive phosphorylation of Akt, as well as mutations of the catalytic subunit of PI3K (PIK3CA), result in persistent activation of mTOR as the downstream serine/threonine kinase that primarily regulates proliferation, survival and apoptosis in BC cells [12]. Further work also suggests that this abnormal mTOR activity induces a peculiar osteotropic phenotype of BC cells, probably involved in skeleton colonization [13], that could ultimately be modulated by inhibiting this pathway, as demonstrated in other cancers [14].
Based on the results of the BOLERO-2 trial, the mTOR inhibitor Everolimus (RAD001) has been approved for use in postmenopausal women with hormone-receptor positive (HR + )/Her-2 − advanced BC, refractory to non-steroidal aromatase inhibitors [15]. In this study, a beneficial effect on bone health was documented in bone-metastatic patients receiving the drug, skeletal tumor progression being significantly delayed as compared to in the control group patients [16]. The bone resorption markers at 6 and 12 weeks revealed a significant decrease in patients treated with Everolimus [16], and this beneficial effect on bone health was attributed to the reduced OC function associated to the general anti-tumor activity induced by Everolimus on BC cells, as well as to a direct mTOR inhibition in BC-stimulated OCs, since this pathway drives their functional maturation [17]. However, this result in the BOLERO-2 trial may also be explained as an effect of suppressed pro-OC paracrine activity of BC cells in response to the mTOR inhibition, besides its modulatory role on hyperactive OCs. Another trial, the RADAR study, suggested that Everolimus as a single agent was beneficial in patients showing only bone involvement [18]. These effects have been further emphasized in a recent subanalysis of the BOLERO-2 trial confirming the beneficial effect of Everolimus in the sub-population of BC patients with bone metastases, as well as in other subgroups [19].
However, in OC progenitors, as in BC cells, the secretion of pro-OC factors is independent of the mTOR pathway since the nuclear factor kappa B (NFkB) pathway is primarily involved through transcriptional factors regulating their release and activity [20][21][22][23][24]. Therefore, assuming that mTOR inhibition restrains the whole tumor activity in BC cells, as described in BOLERO-2, it is conceivable that both the mTOR and NFkB pathways may be molecularly interconnected and that their interaction may ultimately bring about the inhibitory effect on the secretion of pro-OC factors [25].
Here, we investigated the potential of Everolimus to regulate the pro-OC paracrine activity of BC cells, and provide evidence that this mTOR inhibitor restrains the bone-destroying potential of BC cells both in vitro and in vivo, through a synergic negative efficacy on both mTOR and NFkB. Thus, the NFkB-mediated suppression of the pro-OC paracrine activity by BC cells in patients enrolled in the BOLERO-2 trial may account for the beneficial bone effect attributed to Everolimus.

BC cell lines and survival assay
MDA-MB-231 and MCF-7 BC cells (ATCC, Rockville, MD, USA) were cultured in RPMI 1640 supplemented with 10 % FCS (Sigma Aldrich, Milan Italy), 100 U/ml penicillin/streptomycin and 2 mM L-glutamine (PAA, Pasching, Austria) in 5 % CO 2 -incubator. Both cell lines were incubated for 48 h with increasing concentrations (10 −1 to 10 4 nM) of Everolimus and their viability was assessed by methylene blue. Briefly, after fixation with glutharaldehyde, the cells were incubated for 10 min with methylene blue [0.05 % (w/v) in water], then treated with HCl [3 % (v/v)] and read at 650 nM of absorbance in a microtiter reader (BIO-Rad, Milan, Italy). The inhibitory concentration 20 (IC 20 ), defined as the amount of Everolimus able to reduce cell growth by 20 %, was calculated according to BioDataFit 1.02 software by interpolation of the dose-response curves. IC20 was arbitrarily used as the sub-lethal dose and each experiment was performed in triplicate.

BC cell conditioned media (CM)
After 48 hr-treatment with control DMSO or Everolimus at IC 20 , BC cells (1 × 10 5 /well) were seeded in 6-well plates for 24 h. Then, supernatants were recovered, filtered, frozen at −20°C until use for in vitro and in vivo experiments (see below).

OC differentiation and activity
Human OCs were obtained from the peripheral blood of healthy blood donors, after obtaining written informed consent, and approval by the Ethics Committee of the University of Bari. OCs were generated in vitro after 16day incubation of PBMCs with RANKL (50 ng/ml) and M-CSF (25 ng/ml) (Isokine, Iceland), as previously reported [26]. At day 8, PBMCs were supplemented with 20 % of CM from DMSO-or Everolimus-treated cells and after a further 8 days of incubation, both the morphology and function of OCs were assessed.
We arbitrarily considered as OC-like cells, polykaryons with at least three nuclei, that were counted in ten microscopic fields at 30× magnification after hematoxylin-eosin staining (Vector Labs, Sigma) and compared with tartrateresistant acid phosphatase positive (TRAcP + ) cells in parallel preparations using naphthol AS-BI 0.12 mg/ml, 6.76 mM tartrate, and 0.14 mg/ml Fast Garnet GBC (Sigma-Aldrich).
Functional OC activity was measured on experimental bone substrate. Briefly, pre-OCs obtained after 8 days of culture in the presence of RANKL and M-CSF were incubated for a further 8 days with and without CM on calcium phosphate discs (BioCoat Osteologic Discs; BD Biosciences). Then, the cells were removed by 5 % sodium hypochlorite and the substrates were stained by the Von Kossa method to reveal erosive pits. We also quantified both the number of pits and the percentage of the resorbed area by a dedicated software (Olympus) under light microscopy.

RT-PCR
After 48 hr-treatment with control DMSO or Everolimus at IC 20 , both the MDA-MB-231 and MCF-7 cell lines were measured for mRNA levels of M-CSF, RANKL, TNF-α, IL1-β, IL6, MMP (metalloproteinase)-9, MMP-13, MCP (monocyte chemoattractant protein)-1, MIP (macrophage inflammatory protein)-1α, as the major pro-OC factors [27][28][29][30][31][32]. Also, the OC-like phenotype of BC cells was assessed on the levels of TRAcP, cathepsin K (Cat K) and c-fms. RNAs were extracted by the RNeasy kit (Qiagen, Chatsworth, CA) and the relative concentrations calculated by a Bio Photometer (Eppendorf) at 260 nm. After evaluating RNA integrity by electrophoresis, cDNA was obtained by reverse transcription of 1 μg of total RNA using a commercial kit (Applied Biosystems, Milan) in a Mastercycler Personal (Eppendorf). Levels of the above genes were assessed by specific primers (Additional file 1: Table S1) designed on different exons to avoid amplification of contaminating genomic DNA. β-actin was selected as target gene for the RT-PCR using Sybr Green (Maxima® SYBR Green/ROX qPCR Master Mix, Fermentas), in the ABI Prism 7300 Sequence Detector (Applied Biosystems). Following an initial incubation at 50°C for 2 min, samples were denatured at 95°C for 10 min, then by 40 cycles at 95°C for 15 s, and 1 min at 60°C. All tests were completed in triplicate and gene expression levels were obtained as 2 -ΔΔCt [33].

Measurement of pro-OC cytokines released by BC cells
M-CSF, IL1-β, IL-6, TNF-α and MIP-1α were measured in CM from each cell preparation by the appropriate ELISA (R&D systems, Minneapolis, MN), as reported [34].

Generation and detection of BC bone metastases in vivo
To investigate the capability of BC cells to generate in vivo bone metastases and the effect of the 48 hr-treatment with sub-lethal doses of Everolimus, we utilized MDA-MB-231 as predominant bone metastasizing BC cell model [36] in 8-week old NOD.CB17-Prkdc scid /J mice (Charles River, Milan, I). All experiments were performed in accordance with the Italian Guidelines for the use of laboratory animals, following the European Union Directive for the protection of experimental animals (2010/63/EU), after receiving approval from the Animal Experimentation Ethics Committee (CESA) of University of Bari "Aldo Moro". Animals were maintained under standard environmental conditions and provided with feed and water ad libitum. Considering the animal ethical issues, all animals were kept under best hygienic conditions and were daily inspected for signs of pain or discomfort. Briefly, eight mice were anesthetized by Isofluorane, and 1 × 10 5 cells/20 μl of Everolimus-treated and untreated MDA-MB-231 were inoculated into the left and the right tibial cavity, respectively, of the flexed knees of each animal. After 4 weeks, the animals were euthanized by carbon dioxide and X-Rays were taken at 20 kV and 25 mAs for 5 s using a mammographic device (Model Flat E; Metaltronica, Rome). Films were then comparatively inspected for structural deformities and the size of visible tibial lesions was measured by ImageJ software, version 1.45 (National Institutes of Health, Bethesda, MD). The extent of osteolytic areas, as mm 2 of bone devastation, were compared in each mouse between the right and left tibias.

Bone immunohistochemistry
The tibias were excised, fixed and decalcified in EDTA for paraffin-embedding; 4 μm-thick sections were stained with hematoxylin-eosin while parallel sections were prepared for TRAcP staining (Aviva Systems Biology, San Diego) using specific reagents and avidin-biotin (Vector Labs, Burlingame) to reveal TRAcP + cells [37].

Statistical analysis
We employed GraphPad Prism 6.1 software (Macintosh, La Jolla, CA) and differences were calculated by Student's t test. P < 0.05 was considered statistically significant.

Everolimus restrains the BC-mediated OC hyperactivity
The restraining effect of Everolimus on the pro-OC activity of BC cells was investigated at both RNA and functional levels.
As shown in Fig. 2, quantitative RT-PCR revealed that OCs had significantly increased mRNA levels of TRAcP, Cat-K and c-fms after incubation with CM from both the MDA-MB-231 and MCF-7 cell lines (p < 0.05). No variation was observed for RANK mRNA after adding MDA-MB-231 CM, whereas treatment with CM from Everolimus-treated BC cells resulted in a suppression of TRAcP, Cat-K and c-fms (p < 0.05). Functional tests supported these results. OCs showed a dramatically reinforced bone resorbing activity on calcium-phosphate slices after adding CM from both BC cell lines and, contrarily, a significantly inhibited activity when using CM from Everolimus-treated cells (Fig. 3). In particular, Fig. 3a (above) shows that OCs stimulated with CM produced higher numbers of erosion pits (MB-231: 45 ± 2.5; MCF-7: 32 ± 3) than control OCs (20 ± 3.1), with increased total resorbed areas, equal to 3.8 and 3.1 %, respectively, with CM from untreated MDA-MB-231 and MCF7 cells, compared to 1.5 % of the control OCs, as shown in Fig. 3a (below). By contrast, when using CM from Everolimus-treated BC cells, both erosion pits and resorbed areas were significantly reduced (p < 0.05 in all instances). Figure 3b depicts representative images of erosion pits on calcium-phosphate discs. These results emphasized the ability of Everolimus to restrain the BCmediated hyperactivity of OCs, at both molecular and functional levels. To verify whether the suppressed transcription of RNAs was paralleled by a concurrent reduction in the secretion of related proteins, we measured M-CSF, TNFα, IL-1β, IL-6, MIP-1α, and MMP-13 amounts in CM by ELISA. Figure 4b depicts the results, showing an apparent correlation of RNA amounts with lower secreted levels of those cytokines. MDA-MB-231 cells constitutively secreted consistent amounts of both M-CSF and TNF-α, whereas MCF-7 cells produced higher levels of IL-1β, IL-6 and MIP-1α than MDA-MB-231. However, Everolimus significantly inhibited this paracrine activity in both cell lines (p < 0.05 in all instances).

mTOR inhibition by Everolimus disables NFkB signaling in BC cells
Since the NFkB pathway regulates the transcription of M-CSF, IL-1β, IL-6, TNFα and MIP-1α for the relative protein secretion by BC cells, we investigated the interference exerted by Everolimus on mTOR and NFkB signaling. By Western-blot, we measured the amounts of major factors of both pathways, such as Akt, phosphorylated (p) mTOR, p-p70S6K, IKKα, p-IKKα, p65, and p-p65, in both BC cell lines before and after Everolimus treatment. Figure 5 (up) depicts the constitutively increased expression of p-mTOR and its downstream effector p-p70S6K in both cell lines, although at higher levels in MCF-7 (OD values are in, Additional file 1: Table S2). As expected, Everolimus decreased p-mTOR and p-p70S6K levels in both cell lines, particularly in MCF-7 cells, whereas no change in Akt expression was observed, confirming the ability of Everolimus to selectively target the m-TOR molecule and hence inhibit the m-TOR-dependent p70S6K factor.
The inhibition of the interaction between the mTOR and NFkB pathways exerted by Everolimus is shown in Fig. 5 (middle) on both cell lines. IKKα and p65 exhibited variably inhibited phosphorylated forms after Everolimus. The effect was more pronounced for p-IKKα in MCF-7 cells, due to the higher constitutive expression of this factor. Similarly, M-CSF, IL-6, IL-1β, TNFα, and MIP-1α were measured as NFkB-dependent factors. Figure 5 (below) shows their lowered intracellular content after Everolimus, especially in MCF-7 cells, supporting the hypothesized negative regulation of NFkB signaling by m-TOR inhibition.

Everolimus-treated MDA-MB-231 cells reveal a restrained osteolytic activity in SCID mice
Everolimus-treated and untreated MDA-MB-231 cells were inoculated into the left and right tibia, respectively, of each 8-week old SCID mouse. After 4 weeks, X-Rays were taken to measure the extension of osteolytic areas. We observed smaller osteolytic areas in the tibias engrafted with treated MDA-MB-231 cells compared to the contralateral tibias, that uniformly showed variably enlarged proximal epiphyses due to the growth of metastatic lesions. Figure 6a (left) illustrates a representative X-Ray of a SCID mouse. As can be seen, the right tibia injected with untreated MDA-MB-231 produced a larger osteolytic lesion with blown cortical bone, compared to the smaller lesion produced by Everolimus-treated cells in the left tibia. Figure 6a (right) shows the average size of these osteolytic areas measured on X-Ray films of all mice, supporting the evidence of significantly larger lesions in the right tibias, injected with untreated MDA-MB231 cells (mean value: 1.7 ± 0.56 mm 2 ) compared to the left tibias, injected with Everolimus-treated cells (0.9 ± 0.31 mm 2 ) (p < 0.03). Finally, Fig. 6b and c shows the histology pattern of tumor engraftment by MDA-MB231 cells of the right tibia of a SCID mouse, displaying devastation of the bone structure, massive infiltration of the spongy bone, and a number of TRAcP + cells, so functional OCs, in close contact with the tumor cells and juxtaposed to the cortical bone.

Discussion
Endocrine therapy is the standard care for patients with advanced HR + BC. However, after an initial response, most patients develop endocrine resistance. The underlying molecular mechanism is PI3K/Akt/mTOR hyperactivity, that drives the phosphorylation of the estrogen receptor (ER) by p70S6K, leading to a ligand-independent ER transcriptional activity. The involvement of mTOR in the cell growth of endocrine resistant BC cells is also supported by the convergence of signaling from different growth factor receptor pathways, such as the epidermal growth factor receptor (EGFR), insulin growth factor receptor (IGF-1R) and insulin receptor (InsR), toward PI3K/ Akt/mTOR [38]. Moreover, besides the expected improvement of progression-free survival obtained by mTOR inhibition in Everolimus-treated patients with endocrine resistant BC, results from the BOLERO-2 trial suggest that this drug significantly delays skeletal disease progression, exerting a definite bone-sparing effect in metastatic BC patients [16].
Skeletal metastases occur in BC as an effect of the osteotropism of this tumor, that primes OC hyperactivity to generate typical osteolytic lesions. Although BC cells themselves are suspected to degrade the bone matrix [3], their paracrine pro-OC activity is a major pathogenic event driving excessive OC differentiation and hyperactivation, resulting in the formation of bone metastases [39,40]. Therefore, we investigated whether Everolimus acts as a negative regulator of the paracrine pro-OC activity of BC cells.
The beneficial effect of Everolimus on bone health in the BOLERO-2 study has also been interpreted as a result of the inhibition of mTOR activity in OCs [16,17]. It has been demonstrated in transgenic mice, bearing human-TNFα to resemble a rheumatoid arthritis model, that mTOR signaling drives OC function, and that inhibition of this pathway can mitigate bone erosion [41]. In these animals, mTOR inhibition by Everolimus or Sirolimus resulted in a significantly decreased structural damage in joints.
We used the BC cell lines MDA-MB-231 and MCF-7, both capable of producing osteolytic lesions in animals [36,42]. In line with previous data [43], both cell lines secreted variable amounts of pro-OC factors and their CM accelerated the differentiation of PBMCs to TRAcP + polykaryons, namely OCs showing bone-resorption activity in vitro, with a concurrent RNA increase of pivotal OC-related genes, such as RANK and c-fms as phenotype markers, and cathepsin-K and TRAcP as functional markers.
Everolimus treatment of BC cells inhibited the release of pro-OC factors. Using their CM, we observed a decline in the formation of TRAcP + polykaryons, together with a lower transcription of major OC factors and Fig. 3 Measurement of bone resorbing activity of OCs stimulated by BC cells. The OCs capacity to produce erosive lacunae on calcium phosphate was measured after stimulation with CM from both untreated and Everolimus-treated BC cells. a Significant increases of the number of erosion pits (above) and percentage of resorbed area (below), were observed after incubating OCs with CM from both BC cells, whereas this effect was significantly suppressed by Everolimus (*p < 0.05). Results are means of three experiments. b Variation of erosive pit formation by von Kossa-staining of calcium phosphate slices seeded with OCs compared to culture conditions. CM from Everolimus-treated MCF-7 cells inhibited the bone resorbing activity (bottom, right) compared to the constitutive activity (above, right) and to untreated cells CM (bottom, left) (scale bar: 25 μm; 40x magnification) decreased bone resorbing ability. However, although these effects were similarly detectable in both cell lines, a higher inhibitory effect was recorded when PBMCs were stimulated with CM from Everolimus-treated MCF-7 cells. We interpreted this result as dependent on the higher sensitivity to Everolimus treatment of MCF-7 cells, as an effect of the intrinsic PIK3CA deregulation leading to a constitutive mTOR hyperactivity [44]. On the other hand, in both cell lines the RNA of M-CSF, TNFα, IL-1β, IL-6, MIP-1α and MMP-13, as major pro-OC factors, was variably lowered by Everolimus: the basal transcriptional profile implying a well-defined inherited pro-OC potential was negatively regulated by the mTOR inhibitor. However, transcriptional levels of other genes such as RANK-L, MCP-1 and MMP-9 remained undetectable, as reported by other investigators [45].
A major result in our work was the inhibition of M-CSF, IL-6, IL-1β, TNFα and MIP-1α induced by mTOR inhibition. Constitutive phosphorylation of p70S6K, 4EBP1 and eIF4E, as downstream mTOR effectors [46], has been described to contribute to cell growth, angiogenesis and metastasis in several tumors. mTOR inhibition restrains these effects without having any apparent impact on M-CSF, IL-6, and IL-1β secretion, since these factors are regulated by NFkB [20][21][22][23][24], whose activation is apparently independent of the mTOR pathway. This point is disputed, however, since previous works reported that a potential interaction between mTOR and IKKα triggers NFkB in tumor cells and constitutive activation of the PI3K/Akt/mTOR pathway [25,47]. We thus investigated whether NFkB signaling is affected by mTOR. After treatment of both cell lines with Everolimus, the secretion of These data, together with the increased phosphorylated levels of pIKKα and p-p65 at immunoblotting, support a synergic down-regulation of both mTOR and NFkB. We found, indeed, that higher levels of Akt, p-mTOR and p-p70S6K in MCF-7 cells were correlated with similarly increased levels of p-IKKα and p-p65 along the NFkB pathway. The inhibition of NFkB phosphorylated effectors after treatment with Everolimus definitely suggests a functional connection between the mTOR and NFkB pathways.
It has been suggested that the functional disablement of NFkB following proteasome inhibition may affect intracellular signaling propagated through the mTOR pathway [48]. In our study, however, we provide inverse evidence that, at least in BC cells, both p-IKKα and p-p65, as NFkB components, can be disabled by inhibiting mTOR. Although it has been reported that both MDA-MB-231 and MCF7 may acquire resistance to everolimus [49], in our hands, the minor intracellular content of pro-OC factors, namely M-CSF, IL-6, IL-1β, TNFα and MIP-1α, induced by NFkB in Everolimus-treated BC cells emphasizes the interactive negative regulation  Table S2, of phosphorylated mTOR and p70S6K components of the mTOR pathway, and concurrently, p-IKKα and p-p65 factors belonging to the NFkB signaling pathway. In both cell lines, Everolimus strongly suppressed the constitutive hyperphosphorylation of IKKα and p65, as well as of p-mTOR and p-P70S6K. This supports the potential of Everolimus to inhibit the NFkB pathway, since the intracellular expression of M-CSF, TNF-α, IL-6, IL-1β and MIP-1α, as major NFkB-dependent pro-OC factors, was suppressed as well as p-IKKα and p-p65. β-actin was the loading control Fig. 7 Potential inhibitory interactions between the mTOR and NFkB pathways activated by Everolimus. It is suggestive that Everolimus disables IKK by p-mTOR, thus restraining the phosphorylation of IKb. This event prevents the migration of p65, as NFkB component, to the nucleus where it normally acts as transcription factor for M-CSF, TNF-α, IL-6, IL-1β and MIP-1α. This ultimately leads to an inhibited secretion of the pro-OC soluble mediators coded by these genes between these pathways. A model of interactive molecular cross-talk between mTOR and NFkB signaling is proposed in Fig. 7.
To substantiate our hypothesis, we explored the effect of mTOR inhibition on BC paracrine pro-OC activity in a xenograft model of bone metastatic disease. In each experiment, Everolimus-treated and untreated MDA-MB-231 cells were intra-tibially injected into the left and right tibial marrow cavity, respectively, of SCID mice. This resulted in larger lytic lesions in the right than in the left tibias. This in vivo experiment supported our hypothesis that mTOR inhibition can inhibit the paracrine activity of BC cells.
Therefore, besides the general anti-tumor activity obtained by mTOR inhibition, the beneficial bone-sparing result in the BOLERO-2 trial may be explained by at least two effects. The first mechanism includes the known direct inhibition of OC activity [17,41], while the second acts through a negative regulation of the paracrine secretion of pro-OC factors by BC cells. An additional OC-independent effect is also related to the pro-osteoblast activity of Everolimus, as recently reported [50]. The definite effect of mTOR inhibition on bone metastatic cancers was recently explored in a collaborative review [51].
Finally, the potential of a sub-lethal dose of Everolimus to abrogate the paracrine pro-OC activity of BC cells acquires additional relevance when translated to the clinical setting. In fact, in disseminated tumor cells (DTC) PI3K/Akt signaling is reduced, thus suggesting that mTOR pathway inhibition might be linked to a biologic quiescence of DTC [52]. Thus, Everolimus at sub-lethal dose alone, or in association with other treatments, may delay or disable dormant bone metastatic disease by freezing the tumor cells living in a quiescent state in bone niches, which can generate metastases even after long periods of apparent disease remission [53].

Conclusions
Two major points emerge from an analysis of our work. The first is the capacity of Everolimus to restrain the progression of skeletal metastatic disease in BC patients through a specific mechanism that includes inhibiting the release by these cells of the most effective pro-OC factors such as M-CSF, TNFα, IL-1β, IL-6, MIP-1α and MMP-13. The second point concerns the molecular pathway activated in the release of several of these factors. M-CSF, IL-1β and IL-6 secretion is regulated both in OCs and BC cells by the NFkB pathway, whose signaling is apparently independent of mTOR. Instead, here we provide evidence that these pathways are interconnected and that the inhibitory effects on mTOR exerted by Everolimus include a negative regulation of NFkB, thus explaining the reduced secretion of M-CSF, IL-1β and IL-6 by BC cells after mTOR inhibition. Therefore, the beneficial effects of Everolimus on bone health reported in the BOLERO-2 trial are related to the inhibition of the pro-OC paracrine activity of BC cells. To our knowledge, this is the first report describing this specific effect of the mTOR inhibitor.

Additional file
Additional file 1: Figure S1. Measurement of Everolimus cytotoxicity on BC cell lines. Table S1. Primer sequences used for Real-Time PCR. Table S2.

Competing interests
All Authors declare no conflict of interest.
Authors' contributions VS designed the study, the experiments and wrote the manuscript; SC did in vivo studies; AS carried out the molecular biology experiments; OB performed the PBMCs cultures and cell stainings; MC and MT drafted the manuscript and performed statistical analysis; GO and EM completed and analyzed histological assays; FS designed the research and wrote the paper. All Authors read and approved the manuscript.