Regulation of Osteosclerosis by Inoculated Cd133+ PC3 Cells in Bone‐marrow Microenvironmental Niches

ABSTRACT Bone is the most common site of prostate cancer (PC) metastasis. Studies suggest that cancer stem cells (CSCs) are associated with stemness characteristics, providing some support for the concept that CSCs act as osteosclerotic precursors in bone microenvironmental niches. Here, we asked whether ectopic overexpression of CD133 maintains stability of CSCs in human PC cell lines and induces the changes of molecular features in the bone microenvironment. Ectopic overexpression of CD133 in PC3 or DU145 cells led to increased expression of ALDHA1, OCT4, and NANOG, enhanced colony‐forming ability, and increased ALDH activity. In addition, micro‐CT imaging, confocal microscopy, and H&E staining of mouse tissue confirmed that CD133 overexpression in PC3 and DU145 led to marked osteolytic bone tumor. However, expression of osteoblastic markers such as collagen type I, bone sialoprotein, and osteocalcin (OC) at the tumor margin of CD133‐overexpressing PC3 tumors in mouse tibiae was higher than that of CD133‐overexpressing DU145 tumors with osteosclerotic molecular features. In addition, expression of osteopontin (OPN) mRNA/protein by CD133‐overexpressing PC3 cells was higher than that by DU145 cells. Especially, conditioned medium (CM) from PC3CD133+ cells increased osterix (OSX) activity in bone marrow stromal cells (BMSCs), resulting in increased expression of OC mRNA/protein resulted in increased staining of mineralized matrix by Alizarin red. However, CM from OPN silenced PC3CD133+ cells led to a reduction of OC mRNA and protein expression through OSX activity resulted in reduced amount of mineralized matrix. In conclusion, these findings suggest that CD133 plays a functional role in regulating CSC characteristics in PCs and modulates their abilities in which induce the osteosclerosis of BMSCs. In addition, OPN from CSCs acts as a niche component that promotes osteosclerosis by supporting osteoblastic differentiation of BMSCs. © 2019 The Authors JBMR Plus published by Wiley Periodicals, Inc. on behalf of American Society for Bone and Mineral Research.


Introduction
A bout 90% patients with advanced prostate cancer (PC) have incurable bone metastases, with a mean survival of 1 year. (1,2) Bone metastasis results in significant complications, including bone pain, impaired mobility, pathological fracture, spinal cord compression, and symptomatic hypercalcemia. (3,4) In particular, the osteosclerotic nature of most bone metastases related to PC means that there is a marked increase in osteoblastic activity. (5) There is, however, clear histological evidence of increased bone resorption in these patients, even in the absence of overt osteolytic bone metastases. (6,7) Therefore, a better understanding of the early mechanisms underlying development of bone metastasis would allow therapies to be specifically targeted to early lesions as they arise.
The hypothesis that tumors depend on a small fraction of cells, called cancer stem cells (CSCs), for long-term survival is based on data demonstrating that subsets of human leukemic cells transferred tumor-initiating activity to SCID mice. (8) Recent studies show that PC stem-like cells originating from a primary tumor have a high potential for invasion and possess selfrenewal and clonogenic properties. (9) These cells can be identified by expression of stem cell markers such as aldehyde dehydrogenase 1 (ALDH1), CD133, c-Met, and prostate stem cell antigen. (10) It is considered that CSCs are typically dormant and that their later regrowth triggers metastasis. Also, the molecular machinery of the stem cell niches is designed to regulate stem cell quiescence and self-renewal; therefore, we hypothesized that CSC-like cells may alter the osteoblastic or osteoclastic progenitor stem cell phenotype via engagement with the niche, thereby establishing a future site of metastasis.
Here, we used murine models, in which PC cells would be inoculated in tibial bone marrow for establishment of interaction with bone marrow stromal cells (BMSCs) and metastatic PCs. (11) It will show that CSC-like cells affect the destination of pre-osteoblasts and pre-osteoclasts, which are significantly enriched for a mesenchymal and hematopoietic stem cell phenotype in the bone marrow. (12) CSCs in the bone marrow were maintained over time, a phenomenon not caused by effects on proliferation, homing, or cell survival in the circulation. Differentiation of osteocytes and osteoclasts occurred in the bone marrow following injection of CSC-like cells, resulting in establishment of metastases in the marrow. However, it is unclear which CSC characteristics affect differentiation of osteocytes or osteoclasts.
CD133 (prominin-1), a 5-transmembrane glycoprotein, was first identified as a marker of hematopoietic stem cells. (13) Consequently, CD133 is considered an important cell surface marker for a subpopulation of cancer initiating cells (CICs) in brain tumors, colon carcinoma, head and neck cancer, hepatocellular carcinoma, thyroid carcinoma, and PC. (14)(15)(16)(17) Previously, we demonstrated that upregulation of CD133 on head and neck cancer cells and in head and neck squamous cell carcinoma tissue correlates with chemoresistance to anticancer drugs such as 5-FU. (18,19) Recent reports suggest that expression of CD133 by tumor tissues may be a prognostic indicator for tumor regrowth, malignant progression, and patient survival. (20) Nevertheless, the CD133mediated molecular mechanisms that regulate differentiation of osteocytes and osteoclasts during bone metastasis of PC remain unclear.
The hypothesis presented herein show that CD133 in PCs plays a critical role in increasing stemness, thereby promoting formation of aggressive tumors in murine model when the cells were inoculated in tibial bone marrow. Double transfection of PC cells with green fluorescence protein (GFP) and DsRed2 revealed the distribution of CD133-overexpressing PC cells in the bone marrow. In addition, we tried to identify which niche components secreted by CD133 + PC play a critical role in osteoblastic differentiation of bone marrow-derived stromal cells (BMSC). Ultimately, we demonstrate that CD133 + PC play a critical role in bone remodeling during the progression of PC's bone metastasis.

Materials and Methods
Cell culture PC3, DU145, HT29 and 293 T cells were purchased in 2017, from the Korean Cell Line Bank (KCLB; no. 21435, 30081, 30038 and 21573) Korean Cell Line Research Foundation, Seoul, Korea. The cells were authenticated under short tandem repeat analysis by the cell bank. And then the cells were expanded and stored in liquid nitrogen upon receipt, and each aliquot was passaged for fewer than 6 months in our laboratory. The cells were maintained at 37°C/5% CO 2 in RPMI 1640 (Welgene, Daegu, Korea) supplemented with 10% heat-inactivated fetal bovine serum (FBS; Gibco BRL, Grand Island, NY, USA) and 0.1% antibiotic/antimycotic solution (Welgene). After seeding, the medium was replaced with fresh medium and adherent cells were allowed to reach about 70% confluence. The cells were then detached with trypsin/EDTA (Gibco BRL) and replated (subcultured) in 6well plates for each experiment.
Cloning of human CD133 HT29 colon cancer cells were used as a source of CD133. Cloning of CD133 was carried out as described previously. (18) Plasmid pcDNA3.1/NT-GFP lacking the CD133 insert was used as a transfection control. To check for the translated fusion product, 293 T cells were used as a positive control. Briefly, 293 T cells grown to near-confluence in a 60 mm dish were transfected transiently with pcDNA3.1/NT-GFP or pcDNA3.1/ NTGFPCD133 (Promega, Madison, WI, USA; ratio = 0.2 μg of plasmid to 0.6 μl of reagent) using the FuGENE HD transfection reagent (Promega). After 48 hours, cells were gently washed with ice-cold phosphate-buffered saline (PBS) and lysed.
Establishment of stably transfected PC3-GFP/DU145-GFP control cells and PC3 GFP-CD133 /DU145 GFP-CD133 cells expressing the fusion protein Stable cell lines overexpressing GFP and the CD133-GFP fusion protein were created by transfecting confluent PC3 or DU145 cells in 100 mm plates with 20 μg of pcDNA3.1/NT-GFP::CD133 or pcDNA3.1/NT-GFP plasmid using the FuGENE HD transfection reagent as described previously. (18) Briefly, cells stably expressing GFP-CD133 or GFP were selected using 0.4 mg/mL G-418 sulfate (A.G. Scientific INC., San Diego, CA, USA) in DMEM containing 4 mM glutamine, 1 mM pyruvate, 4.5 g/L glucose, and 10% FBS. After transfection, surviving green fluorescent cells expressing various levels of CD133-GFP or GFP were expanded into 60 mm wells. Stably transfected cells were further expanded and passaged onto 100 mm tissue culture plates using 0.05% trypsin/0.2 g/L EDTA/4Na in Hanks Balanced Salt Solution (Welgene) and cultured with 2 mg/mL G-418 to maintain expression of GFP-CD133 or GFP.

Confocal microscopic analyses
Cells were counterstained with 4′,6-diamidino-2-phenylindole provided in ProLong Gold antifade mounting medium (Invitrogen, Carlsbad, CA, USA) to visualize nuclear morphology. Digital images were captured at the Korea Basic Science Institute Gwangju Center using a TCS SP5 AOBS laser-scanning confocal microscope (Leica Microsystems, Heidelberg, Germany) fitted with a × 20 objective lens.

Colony-forming assays
Cells were seeded at a density of 1000 cells per well in nonadherent 24-well culture plates coated with a 10% polyHEMA (Sigma-Aldrich, St. louis, MO, USA) solution in absolute ethanol and then dried overnight. After seeding, cells were incubated in a serum-free DMEM-F12 medium supplemented with 200 ng/mL EGF (Sigma-Aldrich), 20 ng/mL basic FGF (Sigma-Aldrich), and B-27 supplement (Invitrogen). After 5 days of incubation, the number of spheroids in each well was counted under a light microscope (Zeiss, Zena, Germany).

Flow cytometric assessment of ALDH activity
This assay of tumor-derived cells was conducted using the ALDEFLUOR kit (Stem Cell Technologies, Seattle, WA, USA). Cells were suspended in ALDEFLUOR assay buffer containing ALDH1 substrate (BODIPY-aminoacetaldehyde, 1 M per 10 6 cells) and incubated for 30 minutes at 37°C. As a negative control, some batches of cells were incubated with a specific ALDH1 inhibitor, diethylaminobenzaldehyde (DEAB; 50 mM). Next, cells were washed twice with washing buffer and analyzed by flow cytometry (Beckman Coulter, CA, USA). Approximately 10,000 events were captured per sample.
Cloning of DsRed2 protein and establishment of stably transfected PC3 DsRed2 /DU145 DsRed2 control cells and PC3 DsRed2GFP-CD133 /DU145 DsRed2GFP-CD133 cells expressing the fusion protein To monitor the distribution of injected cells, wild-type or GFP-CD133 + cells were transfected with DsRed2. In brief, PC3/ DU145 and PC3-GFP-CD133/DU145-GFP-CD133 cells stably expressing DsRed2 were selected using 10 mg/L Blasticidin (Invivogen, San Diego, CA, USA) in RPMI1640 containing 4 mM glutamine, 1 mM pyruvate, 4.5 g/L glucose, and 10% FBS. After transfection, surviving red fluorescent cells expressing various levels of DsRed2 were expanded into 60 mm wells. Stably transfected cells were further expanded and passaged onto 100 mm tissue culture plates using 0.05% trypsin/0.2 g/L EDTA•4Na in Hank's Balanced Salt Solution (Welgene) and cultured with 10 mg/L Blasticidin to maintain expression of DsRed2 protein. Images of intracellular DsRed2 and GFP fluorescence were obtained by confocal microscopy (Leica Microsystems).

Animals
Five-week-old male athymic nude mice (BL-6/Nu, Orient Bio Co. LTD, Seoul, Korea) were housed under controlled light conditions and fed ad libitum. All experimental procedures involving animals were performed in compliance with institutional and governmental requirements and were approved by the Institutional Animal Care and Use Committee (CIACUC2015-A0032), Chosun University, Gwangju, Korea.

Intra-tibial xenograft model
Intra-tibial injection of PC cells was used to examine the ability of tumor formation in the bone marrow. The mice were anesthetized with isoflurane gas and then, PC3 DsRed2+ , PC3 DsRed2+CD133+ , DU145 DsRed2+ , or DU145 DsRed2+CD133+ cells (1 × 10 5 per mouse) were injected into the right tibia. Briefly explained, both knee joints were shaved and cells (in 20 μl of DMEM) were injected using a 100 μl Hamilton-type syringe with a 27 gauge needle. The needle was inserted into the intracondylar notch of the femur using a spinning motion. Cells were injected into the medullary space (approximately 0.3 cm proximal to the epiphyseal plate). To control for surgical effects, 5 μl of DMEM was injected into the contralateral limb. Following injection, 0.1 mg/kg buprenorphine was injected subcutaneously to minimize postprocedure pain and animals were returned to cages for recovery. Tumor-bearing tissues from mice at necropsy were obtained 4 weeks later.

Histologic analysis of mouse tissues
Tumor-bearing tissues were fixed in cold 4% paraformaldehyde. Bone tissue was first decalcified in sodium citrate solution before processing into histologic slides. Decalcified bones were cut at the midpoint and embedded in paraffin blocks. The fluorescence in serial paraffin sections was monitored by fluorescence microscopy (Leica Microsystems). Tissues were stained with H&E, and images were captured using a microscope slide scanner (3D-HISTECH Ltd., Budapest, Hungary).

Micro-CT analysis
The right femur was dissected, and CT imaging was performed using a Quantum GX μCT imaging system (PerkinElmer, Hopkinton, MA, USA) at the Korea Basic Science Institute (Gwangju, Korea). The X-ray source was set at 90 kV/88 mA, with a field of view of 45 mm (voxel size, 90 μm; scanning time, 14 minutes). The CT imaging was presented in the 3D Viewer within Quantum GX. Following scanning, image segmentation was performed using Analyze software (AnalyzeDirect, Overland Park, KS, USA). Briefly, segmentation of the leg was performed using both semi-automatic and manual tools; object extraction, region growing, and objector separator were performed using the Volume Edit tool. Subsequently, a 3D rendering of the leg and tumor was generated and tumor volumes were calculated using the ROI tool.
Immunohistochemical analysis of bone specimens Sections (3 μM thick) were deparaffinized in three changes of xylene and rehydrated in a graded series of ethanol solutions (ending with distilled water). For antigen retrieval, slides were placed in 0.01 M citrate-buffer, pH 6.0, and heated in a steamer for 30 minutes. Endogenous peroxidases were quenched by incubating with 3% hydrogen peroxide for 20 minutes at room temperature. Sections were incubated overnight at 4°C with a 1:50 dilution of primary antibody: anti-collagen Col1 (Santa-Cruz Biotechnology), anti-BSP (Santa-Cruz Biotechnology), anti-OC (Santa-Cruz Biotechnology), or anti-NFATC1 (Santa-Cruz Biotechnology). Subsequently, sections were incubated for 30 minutes with a biotinylated secondary antibody (LSAB; Dako Cytomation, Glostrup, Denmark), washed in PBS, and incubated for 30 minutes with a streptavidin-peroxidase conjugate (LSAB, Dako Cytomation). The reaction was developed for 5 minutes using 3,30-diaminobenzidine tetrahydrochloride (Sigma-Aldrich). Slides were briefly counterstained in hematoxylin, dehydrated, and cover slipped. Negative and positive controls were run simultaneously. Positive controls comprised mammary tissue. The slides were captured using a microscope slide scanner (3D-HISTECH Ltd.).

Enzyme-linked immunosorbent assay
The amount of Receptor activator of nuclear factor kappa-Β ligand (RANKL) in cell culture supernatants and serum of xenograft-bearing mice was measured using a commercially available enzyme immunoassay kit (R&D Systems, MN, USA), according to the manufacturer's protocol. Absorbance was measured at 450 nm in a colorimetric microplate reader (BioTek, Winooski, USA). The reading was then subtracted from the reading at 570 nm.

Cytokine profiling
Supernatants from PC cells were collected and assayed using a human inflammation antibody array (R&D Systems), according to the instruction manual. Table 1 lists the cytokines examined using this technique. The cytokine signal was detected using an ECL detection kit (Amersham) and quantified by densitometric analysis using Image J software (Scion Corp, MD, USA). The signal intensity of each spot was adjusted to that of the corresponding internal control provided for each membrane by the manufacturer (cytokine profiling). The ratio of expression was calculated by dividing the signal intensity of each cytokine by the signal intensity of the internal control for each sample.
Coculture assays CM was collected from PC cells. The medium was harvested, sterile filtered, and stored at − 20°C until needed. For coculture assays, BMSCs were seeded at a density of 5 × 10 4 /cm 2 and cultured in 10% α-MEM supplemented with 50% (v/v) CM in the presence of osteoblastic inducers (50 mg/mL ascorbic acid;AA and 5 mM β-Glycophosphate;β-GP). The medium was replaced every 48 hours, and differentiation was examined at the indicated times.
siRNA-mediated knockdown of OPN PC or BMSCs were plated in 6-well plates (2 × 10 5 cells per well) for 24 hours. The medium was removed, and cells were transfected with 30 nM control/hOPN1/hOPN2 siRNA oligonucleotide duplexes (for PC) or control/mOPN1/mOPN2 siRNA (for BMSCs) using the transfection reagent according to the manufacturer's instructions (Qiagen, Valencia, CA, USA). The cells were then incubated at 37°C/5% CO 2 in medium lacking antimicrobial agents for 48 hours. Next, siRNAs were removed and the medium was replaced by fresh medium for 24 hours. Irrelevant control siRNA (nonspecific control) was purchased from Dharmacon/Thermo Fisher Scientific (Lafayette, CO, USA). The hOPN1 sequence used for targeted RNA interference was 5′-CUUCUGAGAUGGGUCAGGGTT-3′, and the hOPN2 sequence was 5′-UUUCGUUGGACUUACUUGGTT-3′. (22) The mOPN1 sequence used for targeted RNA interference was 5′-GCUUU ACAGCCUGCACCCATT-3′, and the mOPN2 sequence was 5′-GCCAUGACCACAUGGACGATT-3′. (23) Alizarin red staining Cells were fixed in 95% ethanol and treated for 30 minutes with 40 mM Alizarin red stain (AR-S) solution (pH 4.2) to label calcium deposits. Stained cultures were photographed, and the AR-S was extracted with 10% (w/v) cetylpyridinium chloride in 10 mM sodium phosphate (pH 7.0). The AR-S concentration was determined by measuring the absorbance at 540 nm and reading off an AR-S standard curve.

Quantitative real-time PCR analysis
Total RNA was extracted from cells using Trizol (Invitrogen).
Complementary DNA (cDNA) was synthesized from 2 μg of total RNA using the Super-Script II First-Strand Synthesis System (Invitrogen). mRNA levels were measured using quantitative real-time PCR. The glyceraldehyde-3-phosphate dehydrogenase (Gapdh) gene was used as an endogenous control. The sequences of the primers used to target various genes are shown in Table 2. Details of the reaction system and amplification procedures have been described previously. (24) Luciferase reporter assay BMSCs were transiently transfected with a Runx2 or OSX target sequence-linked luciferase reporter plasmid using Lipofectamine (Invitrogen), according to the manufacturer's instructions. After transfection, cells were incubated in the designated medium for 7 days. Luciferase activity was measured using the Luciferase Assay System (Promega) and detected using a luminometer (GENios Plus; Tecan Group Ltd., Salzburg, Austria). The pCMV-β-gal expression vector was added to each transfection, and the β-galactosidase assay was carried out as described previously to normalize transfection efficiency. (25) Statistical analysis A two-tailed, paired the student t test and and one-way ANOVA followed by Sidak's multiple comparison test (unless specifically mentioned otherwise) when comparing more than two groups were used for statistical analyses. A P value < 0.05 was

Results
Stable overexpression of CD133 in PC3 and DU145 cell lines To determine the effect of CD133 overexpression in vitro, we first established stable cell lines overexpressing CD133 under the control of a constitutive promoter. Specifically, we transfected the PC cell lines PC3 and DU145 with either GFP-tagged CD133 or an empty vector containing GFP only. A representative CD133-GFP clone was selected from each line (PC3 CD133+ or DU145 CD133+ ) and then compared with a control cell line transfected with the empty vector (PC3 Vec or DU145 Vec ). 293 T cells transiently transfected with GFP (293 T Vec ) or CD133-GFP (293 T CD133+ ) were used as a positive control. Basal expression of CD133 protein by PC3 Vec and DU145 Vec cells was very low; however, that by stable CD133-GFP-transfected cells (PC3 CD133+ and DU145 CD133+ ) and transiently transfected 293T cells (Fig. 1A) was apparently higher. Green fluorescence was detected in the cytosol and on the cell membrane of PC3 CD133 + and DU145 CD133+ cells (Fig. 1B). Next, to investigate the role of CD133 in cancer stemness we examined acquisition of CSC-like properties, such as an increased ability to form tumor spheres, increased ALDH activity, and increased expression of stem cell-like markers such as ALDHA1, OCT4, and NANOG. Expression of OCT4, ALDHA1, and NANOG proteins by PC3 CD133+ and DU145 CD133+ increased apparently, consistent with increases in properties associated with stemness (Fig. 1C). Stemness was also confirmed in a colony-formation assay. PC3 CD133+ and ◼ 6 of 14 KIM ET AL.
Finally, we used an ALDEFLUOR assay to examine the presence and size of the PC3 CD133+ and DU145 CD133+ cell populations exhibiting enzyme activity (Fig. 1E). Of the PC3 Vec and DU145 Vec cells examined, 1.5% and 3.4%, respectively, were ALDEFLUOR-positive. By contrast, 5.5% and 4.6% of PC3 CD133+ and DU145 CD133+ cells, respectively, were ALDEFLUOR-positive. These results demonstrate that ectopic overexpression of CD133 promotes stem-like properties, including increased ALDH activity, increased colonyforming ability, and increased expression of ALDHA1, OCT4, and NANOG.
First, we compared tumor formation following intra-tibial injection of PC3 DsRed2+GFP-CD133+ cells or DU145 DsRed2+GFP-CD133+ cells with that after injection of PC3 DsRed2+ or DU145 DsRed2+ cells, respectively. When we measured fluorescence generated by GFP and DsRed2 (Fig. 2C), we found that GFP fluorescence in PC3 DsRed2+GFP-CD133+ and DU145 DsRed2+GFP-CD133+ cells was strongest near the bone lesion and weaker within the tumor mass, suggesting that CD133-overexpressing PC cells attach to and interact with the bone surface.

Expression of osteosclerotic/osteolytic-related cytokines and their role in subsequent molecular events
To assess the osteolytic activity of PC cell lines, we measured the amount of RANKL released into the culture supernatant or mouse serum by PC3 Vec /PC3 CD133+ and DU145 Vec /DU145 CD133+ cells. We found that ectopic overexpression of CD133 increased the amount of RANKL in both cell culture supernatant and serum. This suggests that overexpression of CD133 by PC cell lines contributes to osteolytic bone fracture upon metastasis to bone (Fig. 4A). Next, we tried to identify other molecules that play a critical role in development of osteosclerotic or osteolytic characteristics by screening cytokines using a proteome array cytokine kit (Fig. 4B). Overexpression of CD133 by PC3 cells led to a marked increase in expression of CD40 ligand, vitamin Dbinding protein, and (especially) osteopontin (OPN); these increases were greater than those observed for DU145 cells. Indeed, levels of OPN mRNA and protein were higher in PC3 CD133+ cells than in PC3 Vec cells (Fig. 4C). However, overexpression of CD133 did not increase the amount of OPN mRNA and protein in DU145 cells (Fig. 4D).

Effect of OPN silencing in PC3 Vec /PC3 CD133+ cells and in murine BMSCs on bone mineralization and ALP activity
To investigate the effect of OPN on osteoblastic differentiation of osteoblast progenitor cells upon metastasis of PC cells, PC3 Vec or PC3 CD133+ cells were transfected with control siRNA or with two different siRNAs targeting human (h)OPN. Figure. 5A shows that silencing of OPN by siRNA2 led to marked downregulation of OPN expression by both PC3 Vec and ◼ 10 of 14 KIM ET AL.
JBMR Plus (WOA) Fig. 5. Effect of knocking down OPN on bone mineralization and ALP activity. A, Expression of hOPN protein in PC3 Vec and PC3 CD133+ cells transfected with control siRNA, hOPN siRNA1, or hOPN siRNA2 was assessed by Western blot analysis. α-tubulin was used as a loading control. Similar results were obtained in three separate experiments. *, P < 0.05. B, Mineralized BMSCs were cultured in Mock medium, conditioned medium (CM) from control siRNA-treated PC3 Vec cells, CM from control siRNA-treated PC3 CD133+ cells, CM from hOPN siRNA-treated PC3 Vec cells, or CM from hOPN siRNA-treated PC3 CD133+ cells. BMSCs were then stained with 2% Alizarin Red S (AR-S) at 28 days. In addition, the AR-S concentration was assessed by measuring absorbance at 540 nm. Bar graphs show the mean ± the standard deviation (SD). *, P < 0.05. C, ALP activity of BMSCs. All data represent the mean ± SD (n = 3). D, Expression of mOPN protein by BMSCs treated with control siRNA, mOPN siRNA1, or mOPN siRNA2 was assessed by Western blot analysis. αtubulin was used as a loading control. E Mineralized BMSCs were cultured in Mock medium, or with control siRNA and CM from PC3 Vec cells, control siRNA and CM from PC3 CD133+ cells, mOPN siRNA and CM from PC3 Vec cells, or mOPN siRNA and CM from PC3 CD133+ cells. BMSCs were then stained with 2% Alizarin Red S (AR-S) at 28 days. F ALP activity of BMSCs. All data represent the mean ± SD (n = 3). PC3 CD133+ cells. Therefore, we used hOPN siRNA2 for further studies. To examine the effect of hOPN on PC-induced osteosclerosis, we cultured bone marrow stromal/pre-osteoblast cells (BMSCs) from mouse tibia for 28 days in conditioned medium (CM) from PC3 Vec cells or PC3 CD133+ cells treated with control siRNA or hOPN siRNA2, respectively. BMSCs exposed to CM from PC3 CD133+ cells transfected with control siRNA showed strongly staining when exposed to Alizarin red (Fig. 5B). However, exposure to CM from PC3 CD133+ cells transfected with hOPN siRNA2 led to a reduction in the amount of mineralized matrix stained with Alizarin red. CM from PC3 Vec and PC3 CD133+ cells harboring control siRNA induced a significant increase in ALP activity by BMSCs at Day 7 (Fig.  5C); however, this was reversed by exposure to CM from PC3 CD133+ cells harboring silenced hOPN.
To determine the autocrine effects of OPN on osteoblastic differentiation during metastasis of CD133 + PC cells, BMSCs were transfected with control siRNA or two species of siRNA specific for mouse osteopontin (mOPN). Figure. 5D shows that mOPN siRNA2 reduced OPN expression by BMSC cells. Therefore, BMSCs were transfected with mOPN siRNA2 in subsequent experiments. Control siRNA-and mOPN siRNA-treated BMSCs showed similar levels of Alizarin red-stained mineralized matrix and ALP activity upon exposure to CM from both of PC3 Vec and Fig. 6. Effect of knocking down hOPN on expression of OC, Runx2, and OSX. A Expression of OC mRNA and protein in BMSCs cultured in Mock medium, conditioned medium (CM) from control siRNA-treated PC3 Vec cells, CM from control siRNA-treated PC3 CD133+ cells, CM from hOPN siRNAtreated PC3 Vec cells, or CM from hOPN siRNA-treated PC3 CD133+ cells. BMSCs were then analyzed by Western blot analysis and real-time PCR at 28 days. β-actin was used as a loading control. Similar results were obtained in three separate experiments. *, P < 0.05. B Expression of Runx2 and OSX proteins in BMSCs after 7 days of culture in Mock medium/CM from control siRNA-treated PC3 Vec cells/CM from control siRNA-treated PC3 CD133+ cells, or in Mock medium/CM from hOPN siRNA-treated PC3 Vec cells/CM from hOPN siRNA-treated PC3 CD133+ cells. C BMSCs were transiently transfected with a pRunx2-Luc or pOSX-Luc reporter gene and then treated with the medium described in (a) for 7 days. Luciferase activity was measured as described in the "Materials and methods" section. Results are expressed as mean ± SD of triplicate tests. * and **, P < 0.05. PC3 CD133+ cells (Fig. 5E, F). These results suggest that mOPN from BMSC does not have autocrine effects on osteoblastic differentiation by stimulating CD133 + PC cells.
Effect of OPN-silenced PC3 CD133+ cells on expression of OC, Runx2, and OSX BMSCs cultured with PC3 CD133+ cells transfected with control siRNA showed a significant increase in expression of OC mRNA and protein (Fig. 6A). However, PC3 CD133+ cells transfected with hOPN siRNA showed decreased expression of OC mRNA and protein.
To determine the inhibitory effect of hOPN silencing on Runx-related transcription factor 2 (Runx2)-and osterix (OSX)-related pathways, both of which are master transcription factors for controlling osteoblast differentiation, we cultured BMSCs in CM from OPN siRNA-treated PC3 CD133+ cells and measured expression of Runx2 and OSX protein by Western blot analysis (Fig. 6B). Treatment with CM from OPN siRNAtreated PC3 CD133+ cells led to a decrease in expression of Runx2 and OSX by BMSCs. In addition, to investigate the effects of OPN silencing on Runx2 and OSX activity, BMSC cells were transiently transfected with a pRunx2-Luc or pOSX-Luc vector. Exposure to CM from PC3 CD133+ induced a significant increase in OSX activity in BMSCs (Fig. 6C). However, CM from silenced hOPN PC3 CD133+ cells decreased OSX activity in BMSCs. These results are consistent with the notion that hOPN (or the OPNstimulated signal transduction pathway) has a direct effect on osteoblastic differentiation via OSX activation by CD133 + PC cells.

Discussions
Bone is a common site of distant metastasis in PC; indeed, the association between bone and metastatic cancer is an important driver of this site-specific manifestation of PC progression. (26) During the last decades, research has identified CSCs as central players in cancer metastasis. CSCs are a small population of cells responsible for initiation of tumor burden and therapeutic refractoriness. (27) Acquisition of self-renewal capacity and stemness properties at the time of tumorigenesis is an essential prerequisite for further accumulation of oncogenic transformations and eventual development of cancer. (28) Here, we sought to identify CSC-like cells by inducing ectopic overexpression of CD133 in two PC cell lines (PC3 and DU145). It led the genetic changes in PC cells, culminating in altered expression of several factors related to stemness, including OCT4, ALDHA1, and NANOG, resulting in increased colony-forming ability and ALDH activity. CD133-expressing cancer cells with a stem cell-like phenotype are responsible for tumor recurrence. (29) CD133 is also a common marker of stemness and expressed by normal tissue-resident and hematopoietic stem cells; however, it is no longer detectable after differentiation. Likewise, cancer cells stop expressing CD133 upon differentiation. (30) Colon, lung, and hematopoietic CSCs are isolated by cell sorting based on expression of CD133, although the function of CD133 remains unclear. (31) In vitro, these cells grow indefinitely as spheres and acquire tumorigenicity in vivo. Here, we found that overexpression of CD133 was associated with stemness and migration of PC cells.
So, is it possible that the CSC properties of PC cells are responsible directly for bone metastasis? To answer this question, we cotransfected PC3 or DU145 PC cells with CD133 and fluorescent proteins DsRed2. These cells (along with their respective controls) were then inoculated into the right tibia of athymic nu/mice. Considering the hypothesis that CSCs give rise to both CSCs and non-CSCs, we expected to find CD133 + and CD133-cells. Hongo et al reported that CD133+ cells can give rise to CD133-cells, but not vice versa, indicating that CD133 is responsible for stemness. (32) CSCs expand through asymmetric cell division, the result of which is two daughter cell populations; one is similar to the mother cell (retaining stem cell properties) while the other is committed to another specific line of differentiation. (33) These characteristics of CSCs are responsible for tumor heterogeneity, meaning that expression of CD133 varies widely within a single cancer cell line or a single tumor. (34) Our findings indicate that CD133 is likely to be one of perhaps several stem cell-related molecules involved in PC metastasis to bone. Radiologic and histologic findings suggested that CD133+ PC3 cells generated larger and more aggressive tumors than wild-type cells. In addition, we conducted a comparative analysis of the biologic characteristics of CD133 + and wild-type PC cells. The results showed that PC3 DsRed2+GFP-CD133+ cells generated GFP and DsRed2 fluorescence when they formed a tumor mass in the mouse tibia; moreover, the GFP signal was observed in a region attached to bone. Thus, we believe that attachment of CD133 + cells, which have stemness properties, to bone causes a high rate of bone fracture and osteolysis. Detailed histomorphometric analyses revealed that fast-growing intra-tibial PC3 DsRed2+GFP-CD133+ /DU145 DsRed2+GFP-CD133+ tumors formed heterogeneous lesions in bone, thereby causing higher levels of lysis than PC3 DsRed2+ /DU145 DsRed2+ mock tumors at the same time points. The most apparent feature of PC3 DsRed2+GFP-CD133+ /DU145 DsRed2+GFP-CD133+ intratibial tumors was a marked increase in lysis of local bone structures. In particular, micro-CT and histologic analyses revealed that PC3 DsRed2+GFP-CD133+ tumors almost completely destroyed the tibial bone architecture after 3 weeks postcancer cell inoculation, whereas damage caused by PC3 DsRed2+ was much less severe. NFATC1 is an osteoclastic marker, expression of which is observed in highly active osteoclast cells. Here, staining for NFATC1 revealed a high number of osteoclasts at the intra-tibial tumor/bone interface; in addition, the amount of RANKL in cell culture supernatants and mouse serum increased. This may explain the massive local net bone lysis caused by PC3 CD133 + /DU145 CD133+ intra-tibial tumors.
A previous study shows that PC3 cells generate a predominantly lytic response within bone; however, actual deposition of new bone was also observed. (26) Similarly, clinical studies show that in most PC patients bone lesions involve both bone destruction and bone formation at metastatic foci. (35)(36)(37) The tropism of PC cells to bone suggests that they interact preferentially with specific cells in the bone microenvironment; the most likely candidates are osteoblasts. (38) Radiologic and histomorphometric evidence indicates that sites of PC metastases in bone often show microscopic evidence of increased bone production, including a greater osteoid surface, an increased osteoid volume, and higher rates of mineralization. (4) The present histological findings are consistent with experimental evidence of increased expression of osteoblastic markers such as Col1, BSP, and (especially) OC in PC3 DsRed2+GFP-CD133+inoculated murine tibiae, especially, suggesting that PC (characterized by increased CSC-like properties) induces bone production via an overall increase in bone remodeling. In particular, undifferentiated carcinomas are more aggressive and metastatic; thus increased CSC-like properties have an effect on osteoblastic differentiation of BMSCs. (39) In the case of PC, it appears that induction of osteoblast-mediated mineralization eventually outweighs the increase in osteoclast resorption, resulting in formation of osteoblastic lesions. (26) Various regulatory factors play a role in proper development and maintenance of bone structure during cancer metastasis to bone. So, which factors play a critical role in CD133+PC cellinduced bone remodeling? Here, we found that PC3 CD133+ cells showed higher expression of OPN than DU145 CD133+ cells, suggesting that OPN elicits essential biological functions and may play a critical role of osteosclerosis during PC metastasis to bone. (40,41) OPN, also called secreted phosphoprotein 1 (SPP1), was initially identified as an extracellular matrix protein that inhibited formation and growth of hydroxyapatite crystals within the bone matrix and other organs. (42) Osteoblasts are the major source of OPN in bone tissue. Generally, OPN is recognized as a middle-stage marker of osteoblastic differentiation, (43) suggesting that it plays an essential role in regulating bone metabolism. In addition, OPN is overexpressed by various cancers and is regarded as a novel cancer marker. OPN is secreted by malignant tumor cells and promotes self-adhesion, invasion, metastasis, neovascularization, and formation of new tumor tissue. The findings presented herein indicate that OPN affects osteoblast physiology in an endocrine manner by increasing the ALP activity and OC expression during PC3 CD133 + -induced BMSC differentiation. In addition, we found that a lack of OPN in CM led to reduced ALP activity and OC expression by BMSCs, resulting in a decrease in Alizarin red-positive mineral density. These changes suggest that OPN upregulates osteoblastic differentiation and influences the extent of tissue mineralization. This is consistent with previous studies pointing to a role for increased OPN deposition in the pathologies underlying mineralization and bone metastasis of breast cancer cells. (44,45) In addition, Runx2 and OSX (osteogenesis-specific transcription factors) are expressed during osteoblastic differentiation of BMSCs. (46,47) Runx2 is required for differentiation of mesenchymal cells into pre-osteoblasts. As a molecule down-stream of Runx2, Osx is required for differentiation of preosteoblasts into mature osteoblasts. In particular, Osx is necessary for bone formation and mineralization in vivo, and mature functioning osteoblasts strongly express later osteoblast marker genes such as OC and BSP in response to OSX signals. (48) Here, CM from PC3 CD133+ cells induced a significant increase in OSX activity in BMSCs; however, silencing of OPN decreased OSX activity in BMSCs. We identified a close relationship between increased expression of OC in the PC3 CD133+ -inoculated mouse tibia and increased OSX expression and activity; also, had a direct effect on osteoblastic differentiation via activation of OSX by CD133 + PC cells. The data demonstrate that CSC-like properties play a major role in bone remodeling in marrow due to increased release of OPN. Bone metastasis of PC resulted in mixed, heterogeneous osteoblastic and osteolytic lesions. Osteoblastic lesions are characterized by excess deposition of new bone, whereas osteolytic lesions are characterized by destruction of bone. These two types of PC metastasis disrupt the bone microenvironment. In particular, CSC-like properties accelerate secretion of factors that stimulate proliferation and differentiation of osteoprogenitor cells, or induce osteoclast precursors to differentiate into mature osteoclasts in the tumor environment. Here, we demonstrated that release of OPN from CSC candidate cells triggered bone formation, suggesting that PC bone metastases initiate osteoblastic lesions. OPN acts as a stimulator of osteogenesis by regulating OSX, which can function in an autocrine, paracrine, or endocrine manner (Fig. 7).
In light of significant advances in our knowledge of metastasis and stem cell biology, we propose a CSC-based model of osteosclerotic remodeling during bone metastasis of PC. The findings indicate that CD133 plays a central role in regulating CSCs in PC, resulting in acquisition and maintenance of osteolytic and osteosclerotic characteristics. Furthermore, this hypothesis substantiates the role of stem cells in cancer metastasis to bone. Therefore, the results should facilitate development of a novel classification system and provide new avenues to developing therapeutic strategies for bone metastasis.

Disclosures
The authors declare that they have no conflict of interest.