Even in patients with osteoblastic bone metastases, an osteolytic reaction is also frequently present; however the mechanisms of stimulation of this osteolytic reaction have not been fully elucidated. It has been reported by us and others that CaP cells express RANKL mRNA and protein [14–16, 29–33]. Moreover, we have shown that CaP cells in the bone environment express higher levels of RANKL than in primary tumors or soft-tissue metastases , and Zhang et al. demonstrated activation of transcription of RANKL in C4-2B cells in the tibiae of mice . The critical role of RANKL/RANK signaling in CaP associated osteolysis has been demonstrated by studies showing that the inhibition of RANKL/RANK signaling decreases tumor associated osteolysis which subsequently results in decreased CaP growth in the bone environment [16, 18, 20, 35]. We have shown previously that administration of OPG or over-expression of OPG inhibits the osteolytic response of the bone to C4-2 cells in vivo, clearly demonstrating a critical role for RANKL in this process [18, 21]. Similar results were reported in breast cancer and myeloma bone metastases [23, 36–38]. In our present study we set out to investigate whether RANKL expressed by human CaP cells in the murine bone environment is involved in direct stimulation of osteoclastogenesis and osteoclast activity or whether the increases in osteolysis are associated with the ability of the tumor cells to elaborate factors that induce host production of RANKL.
In patients RANKL is expressed by both osteoblasts and CaP cells, and it is impossible at present to distinguish the relative contributions of each cell type to the osteolytic component of bone metastases. In the C4-2 tibial injection model of experimental bone metastases, the tumor cells express human RANKL, while the osteoblasts express murine RANKL. The use of a xenograft model and a species-specific RANKL inhibitor provides an elegant model to discern the relative contributions of RANKL derived from tumor cells versus other cell types in the bone microenvironment. This is not possible when using molecules such as OPG-Fc, or soluble RANK-Fc, which each inhibit the action of both mouse and human RANKL. The use of a neutralizing muRANKL MAb might be an alterative way to test this hypothesis; however, we are not aware of an antibody that would specifically recognize murine RANKL and possess neutralizing capabilities.
In our experimental model of CaP growth in the bone, administration of the anti-huRANKL MAb did not significantly inhibit osteolysis associated with the growth of C4-2 cells in the bone, suggesting that the RANKL expressed by these tumor cells is not the driving force for tumor-associated osteolysis in this model. We hypothesize that the interactions between tumor cells and cells within the bone results in increased expression of RANKL on osteoblasts. In support of our hypothesis it has also been reported that DU 145 cells produce soluble factor(s), which increased local RANKL expression and activated both osteoclasts/osteoclast precursors . Other groups have also determined that osteoblasts can upregulate osteoclastogenesis associated factors in response to breast and prostate cancer cells in vitro [40, 41]. Furthermore, breast cancer cells altered the phenotype of osteoblast cells, causing increased expression of various osteoclast stimulating factors . Potential candidate factors are IL-11 and osteopontin, which were shown to stimulate recruitment and activation of osteoclasts in breast cancer osteolytic metastases, and IL-7 that can stimulate spontaneous osteoclastogenesis in bone metastases [43, 44]. Both of these proteins are also secreted by CaP cells [45, 46]. IL-11 is an interesting candidate gene, as IL-11 has been shown to increase RANKL expression in bone, and OPG fully prevented IL-11 induced bone resorption .
Another anti-human RANKL neutralizing antibody, AMG162 (denosumab, Amgen, Inc.), is a potent anti-resorptive agent that is currently under clinical evaluation as a treatment for cancer-induced bone loss and other bone-loss disorders [48–50]. In patients, the tumor cells and osteoblasts express human RANKL and therefore denosumab will inhibit the actions of RANKL expressed by both cell types. In addition, the use of denosumab has potential advantages as a therapeutic agent over other RANKL inhibitors, such as OPG, because denosumab does not bind to TRAIL and it also has a longer half-life in vivo compared to OPG [51–53].
We detected significant decreases in serum levels of Ca2+ in the tumored animals associated with long-term administration of the anti-huRANKL MAb. This effect has not been observed in previous studies (our unpublished data). Since no other alteration consistent with effects on osteoclast number or activity was observed, we hypothesize that long-term administration of the anti-huRANKL MAb might exhibit weak effects on osteoclastogenesis caused by mouse RANKL or that administration of huRANKL MAb inhibits human RANKL/RANK signaling in the CaP cells, altering expression of other factors secreted by these cells that are involved in decreasing serum Ca2+ levels.
As discussed above, we observed no increases in serum Ca2+ levels in animals treated with the anti-huRANKL MAb compared to untreated tumored animals. Nor did we observe increases in TRACP 5b activity after anti-huRANKL MAb administration. As the bone destruction and osteoclastic activity was localized to the right tibia and not systemic, our assays may not have been able to detect minor changes in TRACP 5b activity or serum Ca2+ levels. These assays do suggest, however, that the anti-huRANKL antibody did not have a systemic effect on murine RANKL as osteoclast activity and serum Ca2+ levels did not alter between tumored animals and animals treated with anti-huRANKL MAb. This is a further indication that the antibody anti-huRANKL MAb does not interact with murine RANKL.
We are cognizant that these studies have certain limitations and are subject to alternative interpretations. For example, it could be argued that the dose of the anti-huRANKL MAb (5 mg/kg) resulted in insufficient levels of the MAb in the tumor-bone microenvironment to inhibit huRANKL-mediated osteolysis. However, we do not believe that low levels of the huRANKL MAb are responsible for absence of an effect on osteoclast numbers, since a lower dose of huRANKL MAb (3 mg/kg) inhibited the action of high-dose human RANKL (0.5 mg/kg) in the circulation for four days. It is also possible that the inhibition kinetics of membrane bound and soluble RANKL are different. Prostate tumor cells were shown to produce membrane-bound as well as soluble RANKL , but the proportion of these forms in vivo has not been determined.
We attempted, but were unable, to demonstrate immunoreactivity of the huRANKL MAb on the C4-2 cells in the bone microenvironment of the tumor-bearing treated animals (data not shown). Not all antibodies recognize the targeted protein after formalin fixation and embedding in paraffin. Therefore the lack of immunoreactivity does not necessarily mean the absence of the target protein. In this particular case, because we have shown expression of RANKL in C4-2 cells in the tibiae using other antibodies, as well as RANKL expression in cells grown in vitro, we have concluded that the lack of immunoreactivity of RANKL under immunohistochemical conditions with the huRANKL MAb is due to its inability to recognize the protein in paraffin-embedded tissues. We believe that our selection of this model for our studies is justified for following reasons: 1) we have shown previously that their growth in the bone results in increases in bone destruction; 2) these cells express RANKL; and 3) RANKL is involved in this process.
In this study, administration of huRANKL MAb did not inhibit growth of C4-2 cells within the bone. Since administration of this antibody did not inhibit osteolysis, our results are consistent with the literature showing that inhibition of RANKL/RANK signaling by OPG does not inhibit the proliferation of tumor cells injected subcutaneously in mice . Together these results support the hypothesis that the inhibition of tumor growth by OPG in the bone environment is due to the indirect effects of OPG, through the inhibition of osteolysis and the direct inhibition of RANK/RANKL signaling in tumor cells. However, RANKL signaling has been shown to be important for metastatic spread of tumor cells . These effects were not amenable to study by the methods reported herein, since we injected tumor cells directly into the bone and therefore could not investigate release and trafficking of tumor cells or seeding of tumor cells at secondary sites.