RANKL immunisation inhibits prostate cancer metastasis by modulating EMT through a RANKL-dependent pathway

Prostate cancer (PCa) morbidity in the majority of patients is due to metastatic events, which are a clinical obstacle. Therefore, a better understanding of the mechanism underlying metastasis is imperative if we are to develop novel therapeutic strategies. Receptor activator of nuclear factor kappa-B (NF-κB) ligand (RANKL) regulates bone remodelling. Thus, agents that suppress RANKL signalling may be useful pharmacological treatments. Here, we used preclinical experimental models to investigate whether an inactive form of RANKL affects bone metastasis in RANKL-induced PCa. RANKL was associated with epithelial–mesenchymal transition (EMT) and expression of metastasis-related genes in PC3 cells. Therefore, we proposed a strategy to induce anti-cytokine antibodies using mutant RANKL as an immunogen. RANKL promoted migration and invasion of PC3 cells through EMT, and induced a significant increase in binding of β-catenin to TCF-4, an EMT-induced transcription factor in PCa cells, via mitogen-activated protein kinase and β-catenin/TCF-4 signalling. Thus, RANKL increased EMT and the metastatic properties of PC3 cells, suggesting a role as a therapeutic target to prevent PCa metastasis. Treatment with mutant RANKL reduced EMT and metastasis of PC3 PCa cells in an experimental metastasis model. Thus, mutant RANKL could serve as a potential vaccine to prevent and treat metastatic PCa.


Effects of human RANKL (hRANKL) on EMT and metastasis of PC3 cells.
To explore the relationship between RANKL and EMT of PC3 cells, we performed cell migration and invasion assays with PC3 cells treated with hRANKL.
In the migration assay, wound healing in hRANKL-treated PC3 cells was significantly better than that in control cells (Fig. 1A). Cell invasion also increased significantly following hRANKL treatment (Fig. 1B), although cell proliferations were not increased or decreased (Supple 1).
EMT is closely related to tumour metastasis and progression. Therefore, to determine changes at the molecular level, we measured EMT markers in PC3 cells at the mRNA (Fig. 1C) and protein (Fig. 1D) levels following treatment with hRANKL. Expression of the EMT marker E-cadherin in hRANKL-treated PC3 cells fell significantly, but that of vimentin and β-catenin increased. Expression of MMP-9 and IL-6 (markers of metastasis) were higher in hRANKL-treated cells than in control cells. Expression of E-cadherin protein was significantly lower in hRANKL-treated PC3 cells than in control cells. By contrast, expression of vimentin and β-catenin proteins was significantly upregulated following treatment with hRANKL.
Next, we examined phosphorylation of extracellular signal-regulated kinase (ERK), protein kinase B (AKT), SRC and GSK-3B to investigate the effects of RANKL on mitogen-activated protein kinase (MAPK) and Wnt signalling in PC3 cells (Fig. 1E). PC3 cells treated with hRANKL showed a significant and time-dependent reduction in the level of phosphorylated GSK-3B. A time-dependent increase in SRC and AKT phosphorylation levels was also observed. www.nature.com/scientificreports/ www.nature.com/scientificreports/ Next, we performed co-immunoprecipitation of β-catenin/TCF-4 and TOP/FOP reporter assay to investigate the status of TCF-associated signalling in hRANKL-treated PC3 cells. The results showed a slight increase in TCF-4 levels (Fig. 1F) and TOP/FOP reporter luciferase activity (Fig. 1G) in hRANKL-treated PC3 cells.
Thus, RANKL treatment may trigger metastasis of PC3 cells by suppressing GSK-3B phosphorylation and facilitating Wnt/β-catenin pathway, resulted in EMT progression.

Overexpression of RANKL modulates EMT and metastasis of PC3 cells. To investigate whether
RANKL overexpression stimulates PC3 cell growth in vitro, cells were transiently transfected with an overexpression plasmid containing RANKL. GFP expression by PC3 RANKL cells was monitored by fluorescence microscopy (Supp. 2A). We found a significant increase in expression of RANKL mRNA and protein (Supple 2B and 2C).
Next, we performed migration and invasion assays using PC3 RANKL cells to evaluate the effects of RANKL overexpression on EMT and metastasis. Wound healing was significantly better in PC3 RANKL cells than in control cells ( Fig. 2A). Also, PC3 RANKL cells were significantly more invasive than control cells (Fig. 2B). Analysis of mRNA encoding EMT-and metastasis-related factors in PC3 RANKL cells revealed significant downregulation of the gene encoding E-cadherin (Fig. 2C). Furthermore, expression of genes encoding vimentin, MMP-9, IL-6 and β-catenin was upregulated significantly in PC3 RANKL cells. Protein expression analysis revealed that PC3 RANKL cells expressed significantly lower levels of E-cadherin than PC3 Wild cells (Fig. 2D). By contrast, expression of MMP-9, IL-6, c-MYC and β-catenin was significantly higher in PC3 RANKL cells.
PC3 RANKL cells showed a significant increase in TOP/FOP luciferase reporter activity (Fig. 2E). Immunoprecipitation of β-catenin was carried out to investigate the signal transduction pathway associated with TCF in PC3 RANKL cells (Fig. 2F). PC3 RANKL cells overexpressing RANKL showed a significant increase in activation of the MAPK and β-catenin/TCF-4 signalling pathways owing to stronger binding between β-catenin and TCF-4 than in PC3 Wild cells. Thus, ectopic overexpression of RANKL may increase EMT and the metastatic properties of PC3 cells via the β-catenin/TCF-4 signalling pathway, suggesting the therapeutic potential of RANKL targeting for prevention of PCa metastasis.
To observe bone metastasis, luciferase activity in tumour-bearing tissues of Sham, PC3 Wild , PC3 RANKL+ and PC3 RANKL+ + IM mice was detected by IVIS. In PC3 RANKL+ mice, large and strong bioluminescence spots were detected throughout the body at 16 weeks post-cancer cell injection (Fig. 3C). However, no bioluminescence signals were detected in PC3 RANKL+ + IM mice. The photon flux values were significantly higher in PC3 RANKL+ mice than in PC3 RANKL+ + IM mice (Fig. 3D).
Survival rate analysis revealed a significant decrease in the survival of animals in the PC3 RANKL+ groups compared with that of animals from the PC3 Wild group. The survival rate was higher in the PC3 RANKL+ + IM group than that in the PC3 RANKL+ group (Fig. 3E). The metastasis rate in the PC3 RANKL+ group was higher than that in the PC3 Wild group; however, that in the PC3 RANKL+ + IM group was significantly less than that in the PC3 RANKL+ group (Fig. 3F).

Therapeutic effects of anti-RANKL antibodies induced by RANKL immunisation.
To demonstrate the effect of RANKL immunization on bone resorption, we examined 3D images in the trabecular bone architecture of the distal femur and serum calcium level. No significant changes in trabecular bones were observed in immunized mice (Fig. 4A), as evident from BMD (Fig. 4B). In addition, there is no significant differences between Sham and immunized mice on serum calcium level (Fig. 4C).
To examine the histological characteristics of metastatic tumour-bearing tissues, metastatic lesions from each mouse were stained with haematoxylin and eosin Y (H&E). As shown in Fig. 4D, gross examination of the excised tibiae from PC3 RANKL+ mice revealed a tumour mass in the primary spongiosum (trabecular epiphysis) and bone marrow cells; this was not observed in PC3 RANKL+ + IM mice. In particular, expression of IL-6 and RANKL increased markedly in the trabecular epiphysis region of bones from PC3 RANKL+ mice, but were undetectable in PC3 RANKL+ + IM mice.
Next, we investigated whether mRANKL-MT induces production of anti-RANKL antibodies. The concentration of RANKL (Fig. 4E) was highest in PC3 RANKL+ , and production of antibodies ( Fig. 4F) was highest in PC3 RANKL+ + IM mice. Also, we detected and measured anti-RANKL antibody levels in the PC3 RANKL + + IM group with bone metastasis to investigate generation of anti-RANKL antibodies after immunisation with mRANKL-MT. The anti-RANKL by mRANKL-MT immunization were detected in entire immunized mice (Supple 4), anti-RANKL titer in mice without bone metastasis was significantly higher than that in mice with bone metastasis (Fig. 4G). Also, serum RANKL levels in the PC3 RANKL+ + IM group without metastasis were significantly lower than those in mice with metastasis (Fig. 4H). These observations suggest that anti-RANKL antibodies generated by RANKL immunisation suppress metastasis of PCa cells. www.nature.com/scientificreports/

Effect of immunisation on EMT and metastasis.
To investigate the effects of mRANKL-MT on PCa metastasis, sera obtained from immunised mice were used to treat RANKL-overexpressing PC3 cells. The results of cell migration assays showed that wound healing was inhibited significantly in PC3 RANKL+ cells treated with immune serum (Fig. 5A). In addition, the invasive ability of PC3 RANKL+ cells declined following treatment with immune sera (Fig. 5B). Analysis of mRNA encoding EMT-and metastasis-related factors in PC3 RANKL+ cells treated with immune sera revealed significant upregulation of E-cadherin expression (Fig. 5C). By contrast, expression of vimentin, MMP-9, IL-6 and β-catenin was downregulated significantly in immune serum-treated PC3 RANKL+ cells. Protein expression analysis showed that E-cadherin expression was higher in immune serum-treated PC3 RANKL+ cells   www.nature.com/scientificreports/ than in control serum-treated PC3 RANKL+ cells (Fig. 5D). By contrast, expression of MMP-9, IL-6 and β-catenin was significantly lower in immune serum-treated PC3 RANKL+ cells, as was expression of c-MYC. In addition, protein expression of phosphorylated ERK, AKT, SRC and GSK-3B to investigate the effect of immune serum on MAPK and Wnt signaling in PC3 RANKL+ cells (Fig. 5E). Cells treated with immune serum showed a significant and time-dependent increase in GSK-3B phosphorylation and a time-dependent decrease in SRC phosphorylation. The luciferase activity of the TOP/FOP reporter in immunised serum-treated PC3 RANKL+ cells fell significantly (Fig. 5F). Finally, immunoprecipitation analysis revealed a significant decrease in binding between β-catenin and TCF-4 in cells treated with immune serum (Fig. 5G). These results indicate that the EMT and metastatic properties of PC3 RANKL+ cells were inhibited by treatment with immune serum.

Discussion
PCa, which is common among men in the western world, is associated with high mortality and morbidity with respect to advanced metastasis to the bone. Evidence suggests that the RANKL signalling cascade plays a key role in proliferation, metastasis, migration and invasion of PCa 6 . The RANKL-RANK interaction plays a pivotal role in PCa metastasis; indeed, RANKL expression induces osteoclast hyperplasia and bone destruction during PCa metastasis 26 . RANKL activates RANK directly on tumour cells, as evidenced by dysregulation of several biochemical signalling pathways in PCa cells.
High expression of RANKL facilitates PCa metastasis, an idea consistent with previous studies showing that signalling through the RANK/RANKL axis is related to bone metastases of solid tumours 20,27 . Especially, several previous study demonstrated that primary prostate cancer cells expressed the RANK/RANKL genes, which was further elevated in bone metastasis lesions [28][29][30] . Therefore, denosumab which is a specific human monoclonal antibody against RANKL, was found to be a new therapeutic option is expected to exert its antitumor effect by inhibiting RANKL.
We found that hRANKL-treated or hRANKL-overexpressing PCa cells showed a significant increase in expression of metastasis markers such as IL-6. In particular, in vitro experiments show that RANKL stimulation markedly increases the migration and invasion of PC3 cells, downregulates expression of the epithelial marker E-cadherin and upregulates the mesenchymal marker vimentin. EMT correlates with tumour metastasis and progression, which is consistent with impaired cell-cell adhesion following the loss of E-cadherin expression 31,32 . Furthermore, we show that GSK-3B phosphorylation was reduced significantly following RANKL treatment of PC3 cells due to the effect of RANKL on Src-ERK-AKT signalling. In the present study, Top/Fop luciferase activities and TCF4//β-catenin co-expressions were slight increased after RANKL treatment. As these results, RANKL treatment itself affect directly Src-AKT-GSK-3beta, but affect indirectly Wnt/beta-catenin signalling. Especially, GSK-3Beta is considered to be modulated by Wnt or RANK signal and it could affect the beta-catenin activation. Also, we observed altered expression of β-catenin and TCF-4 in PC3 RANKL+ cells, which resulted in a highly conserved developmental signalling pathway that includes the major effector protein β-catenin. Wnt signalling is an essential pathway involved in cell development, proliferation and differentiation; indeed, regulatory abnormalities in Wnt signalling are associated with metastasis of many cancers 33 . In particular, RANKL overexpression in PC3 cells led to a significant increase in expression of Wnt3a, suggesting that RANKL is a potential target of Wnt signalling in cancer cells 34 . RANKL plays a fundamental role in osteoclastogenesis by interacting with the RANK receptor on osteoclast progenitors during bone destruction by metastatic breast cancer, thereby driving osteoclast cell lineage commitment, monocyte cell fusion and osteoclast maturation via regulation of NF-κB-mediated gene expression; therefore, we were intrigued to find out whether catabolic Wnt signalling mechanisms exist alongside anabolic Wnt pathways to regulate osteoclast formation in bone. β-Catenin, the critical effector of the Wnt pathway, regulates a number of key processes during development, including proliferation, differentiation and cell fate determination 35 . Normally, β-catenin is localised to the cell adhesion junctions in epithelial cells and its abnormal cytoplasmic/nuclear stabilisation drives uncontrolled transcription of target genes (including c-jun, cyclin D1, c-myc, survivin and MMP-7) that regulate cell proliferation, survival and adhesion 36 . In view of cancer cell fate, it is not surprisingly that overexpression of RANKL by PC3 cells led to increased binding of β-catenin to TCF4 and to increased TOP activity. Regulation of β-catenin is linked to the pathogenesis of a number of human cancers, particularly those with an epithelial cell origin. Supporting its putative role as a Wnt signalling target, we confirmed that RANKL overexpression led to transcriptional activation of β-catenin in PC3 cells.
Over the past decades, it has become clear that the RANK/RANKL axis exerts a broad range of functions during cancer cell fate. In the cancer setting, the RANK-RANKL pathway plays a role in every stage of tumorigenesis. Therefore, inhibition of RANKL by anti-RANKL antibodies is expected to be more far-reaching than simple inhibition of cancer cell activation. Denosumab, a drug used to treat metastatic prostate bone loss, has received FDA approval; this drug inhibits the RANK-RANKL pathway 37 . Denosumab is an effective and safe drug, and it is superior to zoledronic acid in terms of the prevention of skeletal-related events; this was borne out in a combined analysis that included three randomised phase III trials with a similar set-up 38 . These trials included patients with bone metastases due to advanced breast cancer 39 , prostate cancer 40 , other solid tumours or multiple myeloma 41 . However, despite medical and commercial success, passive anti-cytokine drugs such as OPG-Fc and denosumab have several limitations, including high production costs, the need for regular infusion, and a limited half-life 42 , indicating the need for a RANKL vaccine. In comparison with antibodies and other biologics, vaccines are better models for treatment of chronic disease because they are relatively cheap and small doses of protein can have a strong and long-lasting effect 43,44 . Here, we developed a novel vaccine targeting RANKL and examined its efficacy in a murine model of prostate cancer metastasis. To circumvent the problem of the immunogen triggering cytokine activity, mutants of RANKL were generated to prevent its interaction with RANK. A previous study shows that immunisation with mutant RANKL molecules generates anti-RANKL antibodies that block the interaction between RANKL and its receptor in an animal model of osteoporosis, thereby www.nature.com/scientificreports/ preventing proliferation and differentiation of osteoclasts and improving bone density 45 . Therefore, to block RANKL activation during PCa metastasis, we immunised mice with mRANKL-MT followed by intracardiac injection of PC3 cells. Inhibiting RANKL in animal models of metastases exerts therapeutic effects by inhibiting cancer cell metastasis. Currently, in vivo tumour models that are most commonly used to study the process of cancer metastasis rely on introduction of tumour cells directly into the systemic circulation by injection into the left ventricle of laboratory rodents 46,47 . Thus, we employed a mouse model of PCa metastasis that more accurately reflects the metastatic process of this type of cancer. Studies on the effects of mRANKL-MT in PC3 RANKL+ mice showed that tumour growth was completely inhibited. Immunisation with mRANKL-MT effectively inhibited metastasis of tumour cells by generating anti-RANKL antibodies.
To further confirm the action of RANKL immunisation, we assessed the effects of immune serum from immunised mice on PCa cells. Anti-RANKL antibodies blocked the RANKL-mediated chemotaxis of tumour cells. Furthermore, anti-RANKL antibodies inactivated RANKL on tumour cells directly. Treatment of RANKLoverexpressing PC3 cells with immune serum almost entirely abolished cancer cell migration and invasion. Wu et al. showed that a recombinant inactive RANKL vaccine (Y234pNO2Phe) induced high antibody titer and protected mice from collagen-induced arthritis by inhibiting osteoclast function and by preventing bone erosion 48 . However, no study has reported that these types of RANKL vaccine have been used to inhibit cancer metastasis. In addition, we found that EMT and metastasis-related genes were downregulated following treatment of RANKL-overexpressing PC3 cells with immune serum. The antisera obtained from mice immunised with mRANKL-MT almost entirely inhibited the EMT process in RANKL-overexpressing PC3 cells. EMT is characterised by the loss of cell-cell adhesion and by an increase in cell motility; it is a key process in cancer progression and metastasis, making EMT inhibition an attractive therapeutic strategy 49,50 . Deregulation of Wnt/β-catenin signalling is a hallmark of PCa metastasis 51,52 and β-catenin is a critical end component of the Wnt signalling pathway, which regulates cell growth, apoptosis and migratory behaviour in response to intercellular adhesion molecules 53 . Activation of β-catenin in PCa cells leads to transactivation of Wnt signalling target genes, including cyclin D1, HEF1 and matrix metalloproteinase 9 54 . Also, previous studies show that expression of Wnt-1 and β-catenin is increased in invasive PCa cell lines and in primary prostate cancer specimens 55,56 . In line with these previous reports, we demonstrated that treatment of PC3 cells with immune serum led to a marked decrease in RANK/SRC and GSK-3β signalling and β-catenin/TCF-4 transcription (Fig. 6). β-Catenin forms a cell adhesion complex with E-cadherin, raising the possibility that loss of expression or a change in β-catenin distribution in the cell alters downstream signalling, decreases intercellular adhesion and promotes metastasis. These results suggest that the inhibitory effect of immune serum on PCa cell metastasis may involve suppression of the Wnt/βcatenin signalling pathway.
The present study has several strengths and limitations. One potential advantage of a vaccination approach to RANKL inhibition compared with antibody-based approaches such as denosumab is that patients who discontinue treatments experience rapid increases in bone remodelling and an increased risk of osteolytic fractures during bone metastatic cancer. It is possible, though clearly unproven, that this vaccine approach may not induce rapid high-turnover bone loss, either by causing more durable RANKL inhibition or by allowing a more gradual resumption of remodelling when vaccinations are discontinued. That would be reasonable which vaccine-type approach could be tested for its ability to minimize the risk of high-turnover bone loss through in vitro and in vivo study.
One limitation of present study is a lack of validation of safety or efficacy beyond long-term rodent studies, despite their experimental reality approach. Because generated mutant RANKL is transformed from mouse RANKL, several amino acid sequences are different from human RANKL and the clinical relevance of the present results for human study thus remains unclear. Therefore, it needs to be carried out in human RANKL knock-in mouse model using by human RANKL mutant variant in further study.
Taken together, this study demonstrates the protective role of mRANKL-MT against RANKL-induced PCa in mice. This effect was mediated via induction of a high-titer antibody response and inhibition of EMT and metastatic functions. Our results highlight the potential application of an anti-RANKL vaccine for treatment of metastatic RANKL-induced PCa. Moreover, the results suggest that mutant RANKL could be used as a RANKL vaccine for the prevention and/or treatment of patients with metastatic PCa prostate cancer.

Real-time quantitative PCR (RT-qPCR).
Total RNA was extracted from PCa cells using Trizol (Invitrogen) and 1 μg was used for RT-qPCR along with oligo-dT primers (10 μg) and dNTPs (10 mM). Next, qRT-PCR was performed to analyse cDNA using SYBR Green SuperMix  BrdU cell proliferation assay. The BrdU cell proliferation assay was performed by means of BrdU cell proliferation assay kits (EMD Biosciences, Inc., Darmstadt, Germany) according to the manufacturer's instructions. Briefly, culture medium was removed, 100 mL of fresh culture medium and 20 mL of BrdU labeling solution were added to each well for 12 h. After 12 h of incubation with BrdU, cells were fixed and incubated with anti-BrdU conjugated with peroxidase. Subsequent to substrate addition, the optical density at 450 nm with a reference wavelength of 540 nm (OD450/540) was determined by microplate reader (BioTek Instrument, Winooski, VT, USA).
Cell migration assay. Cells were seeded in 6-well plates for the cell migration assay. After each treatment, a confluent monolayer was wounded using a 200 μL pipette tip. Images of wound closure were obtained under an inverted microscope after 48 h. The wound area was calculated using NIH ImageJ software.

Cell invasion assay.
A total of 1 × 10 5 transfected cells were seeded into the top chamber of a 24-well polycarbonate Transwell chamber (8.0 µm pore size; Corning Incorporated, Glendale, AZ, USA) and then treated for 24 h with indicated treatment. After 24 h of incubation, non-invading cells on the top of the membrane were removed by cotton swabs. Invaded cells on the bottom of the membrane were fixed with 4% paraformaldehyde for 10 min, followed by staining with 0.05% crystal violet for 4 h. The cells were taken pictured and all the cells on the entire membrane were counted. The relative invasion activity was calculated after normalization to cell migration.
Purification of mRANKL-MT. E. coli cells expressing mRANKL-MT were cultivated in 1 L of an autoinduction medium supplemented with kanamycin (50 μg/mL), as previously described 42 . After centrifugation at 6000×g for 20 min at 4 °C, the pelleted cells were resuspended in 10 mL of lysis buffer (20 mM sodium phosphate, 500 mM NaCl, 10 mM imidazole, pH 7.4) supplemented with 0.1 mg/mL lysozyme and 0.1 mM PMSF.
Glycerol (20% v/v; CARLO ERBA, France) was added to the cell suspension and the cells were sonicated and centrifuged at 15,000×g for 10 min at 4 °C. The supernatants were passed through 0.2 μm paper filters and applied to Ni 2+ -affinity chromatography HisTrap FF columns (1 mL; GE Healthcare Life Science, Piscataway, NJ, USA) equilibrated with binding buffer (20 mM sodium phosphate, 500 mM NaCl, 10 mM imidazole, 5 mM dithiothreitol, pH 7.4). The columns were subsequently washed using binding buffer supplemented with 20 mM imidazole.
After washing, bound protein was eluted using elution buffer (Qiagen). The eluted protein was dialysed against a dialysis buffer (20% v/v glycerol in PBS) in a 10,000 MW Slide-A-Lyzer Dialysis cassette (Thermo Fisher Scientific). The purified protein was vacuum concentrated (Savant Instruments, Holbrook, NY, USA) and analysed by sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE). Protein concentrations were calculated using the Bradford assay. For endotoxin removal, an additional washing step was introduced after the initial wash for chromatography.
Animal study. The animal experimental protocol was approved by the Institutional Animal Care and Use Committee, Chosun University, Gwangju, Korea (CIACUC2019-A0015). The study was carried out in compliance with the ARRIVE guidelines. All experiments were performed in accordance with relevant guidelines and regulations. Five-week-old male athymic nude mice (BALB-c/nu, Orient Bio Co. LTD, Seoul, Korea) were used to generate a xenograft model by intracardiac injection of PC3 Wild , PC3 +RANKL (RANKL overexpression), or PC3 +RANKL + IM (immunisation) cells. Following immunisation, mice were divided into an immunisation group and a non-immunisation group. The Sham group was immunised by a subcutaneous injection of PBS, while the immunisation group was injected subcutaneously with mRANKL-MT (100 μg/kg three times every 2 weeks). Mouse sera and tissue samples were collected according to the indicated schedule.
In vivo bioluminescence measurement. Tumour-bearing tissues were subjected to in vivo bioluminescence imaging using a Living Image 4.5.4 IVIS Imaging System (Perkin Elmer). For luciferase imaging, D-luciferin (Promega) was injected intraperitoneally before imaging. Quantitative detection of luciferase was performed as follows: regions of interest (ROIs) were drawn to capture detected fluorescence, and auto-regions ROIs were used to precisely outline the target region.
Micro-CT imaging data acquisition. Micro-CT scanning for the distal femur was distally initiated at the level of growth plate using a Quantum GX (PerkinElmer, Hopkinton, MA, USA) micro-CT imaging system located at the Korea Basic Science Institute in Gwangju, Korea. The scanning X-ray source was set to 88 mA and 90 kV, with a 10 mm field of view (scanning time, 4 min; voxel size, 20 μm). The 3D architecture images were acquired using 3D Viewer commercial software included with the Quantum GX system. The 3D images were obtained at 4.5 μm resolution. After scanning, the bone structure parameters were analyzed with Analyze 12.0 software (Analyze Direct, Overland Park, KS, USA) using the ROI tool. Femur bone mineral density (BMD) was estimated using hydroxyapatite (HA) Phantom (QRM-MicroCT-HA, Quality Assurance in Radiology and Medicine GmbH, Germany) scanning using the same parameters. Parameter values are shown as mean ± standard deviation (SD).

Quantitative analysis of RANKL and calcium level in serum.
The amount of RANKL in mouse serum was measured using a commercially available enzyme-linked immunosorbent assay (ELISA) kit (R&D Systems, USA), while colorimetric assays were used to assess calcium (Biovision, San Francisco, CA, USA) according to the manufacturer's protocol. Absorbance was measured in a colorimetric microplate reader (BioTek) at 450 nm. www.nature.com/scientificreports/ Immunoreactivity blot. The mRANKL-WT and mRANKL-MT were separated by SDS-PAGE and electroblotted onto nitrocellulose membranes (Bio-Rad). Membranes were blocked with 5% (wt/vol) nonfat milk powder in TBST [10 mM Tris (pH 7.5), 150 Mm NaCl, 0.1% (vol/vol) Tween 20] and probed with mouse serum as primary antibodies in the blocking solution. The membranes were washed three times with Tris-buffered saline. Horseradish peroxidase-conjugated secondary antibodies were diluted 1:5000 in 1% (wt/vol) nonfat milk powder in TBST. The membranes were developed using the ECL system (Amersham Pharmacia Biotech).
Histological analysis in tumor bearing mice. Tumor bearing tissues were dissected, immersed in 4% formaldehyde, and decalcified in 7% EDTA with 0.5% paraformaldehyde for 40 days before processing. To analyze longitudinal sections of distal femurs, decalcified tissues were paraffin-embedded and 2-3-μm-thick sections were cut, mounted on glass slides, and rehydrated using graded alcohol. The tissue sections were stained with hematoxylin/eosin (Shandon Varistain 24-4, Histocom, Vienna, Austria), and images were acquired using an ECLIPSE Ts2R inverted microscope (Nikon). Paraffin sections were deparaffinized in three xylene washes and rehydrated in graded ethanol solutions. For antigen retrieval, the slides were placed in 0.01 M citrate buffer (pH = 6.0) and heated in a steamer for 30 min. Endogenous peroxidases were quenched by incubating the samples with 3% hydrogen peroxide for 20 min at room temperature. The sections were incubated overnight at 4 °C using 1:50 anti-IL-6 (Santa-Cruz Biotechnology Inc.) or anti-RANKL (Santa-Cruz Biotechnology Inc.). Sections were then incubated for 30 min with biotinylated secondary antibody (LSAB system HRP kit; Dako Cytomation, Glostrup, Denmark), rinsed in PBS, and incubated for 30 min with a streptavidin-peroxidase conjugate (LSAB; DakoCytomation). The reaction was developed for 5 min using 3,30-diaminobenzidine tetrahydrochloride (Sigma-Aldrich). The slides were counterstained in hematoxylin, dehydrated, and coverslipped. Negative and positive controls were simultaneously analyzed. The positive controls were mammary tissues. The slides were imaged using an inverted microscope (Nikon).

Measurement of anti-RANKL antibody titers.
Serum samples obtained from immunised mice were serially diluted with PBS containing 0.02% sodium azide and 2% bovine serum albumin (BSA), and then applied to ELISA plates (Sigma-Aldrich) coated with mouse recombinant tumour necrosis factor ligand superfamily member 11 (TNFSF11; 10 μg/mL, R&D Systems). Reactivity of serum antibodies to the target protein was determined using an HRP-conjugated goat anti-mouse IgG secondary antibody (Thermo Fisher Scientific) at a dilution of 1/1000 in PBS/0.02% sodium azide/2% BSA. After development with 1,2-phenylenediamine dihydrochloride (0.4 mg/mL in 0.066 M disodium phosphate, 0.035 M citric acid and 0.01% hydrogen peroxide), absorbance was measured in an ELISA plate reader at 450 nm.
Statistical analysis. Data are expressed as the mean ± standard deviation (SD) from three independent experiments. GraphPad Prism version 6.0 software for windows was used to analyse in vitro and in vivo data. Statistical significance for pairwise comparison was evaluated using an unpaired t-test or one-way analysis of variance (ANOVA) with Tukey's post-hoc test. Results were considered significant at *p < 0.05. Ethics approval. The animal experimental protocol was approved by the Institutional Animal Care and Use Committee, Chosun University, Gwangju, Korea (CIACUC2019-A0015). www.nature.com/scientificreports/