Late-stage prostate cancer patients are refractory to hormone therapy and exhibit a high propensity to develop skeletal metastasis. In this regard, the role of a novel cytokine system belonging to the tumor necrosis factor (TNF) family that is critical for osteoclastic osteolysis and that consists of receptor activator of NF-κB ligand (RANKL), its receptor (RANK), and decoy receptor osteoprotegerin (OPG) is of potential interest.
Reverse-transcriptase polymerase chain reaction (RT-PCR) and immunohistochemical analysis was used to examine the expression of RANKL, RANK, and OPG in human prostate cancer cell lines and in 89 archival samples of primary and metastatic (lymph nodes, skeleton) prostate cancer patients. Expression of these proteins was correlated with clinicopathogic parameters of the prostate cancer.
Expression of RANKL/RANK/OPG was low in normal but markedly higher in prostate cancer cell lines. Analysis of surgical biopsy specimens showed the expression of RANKL (31%), RANK (38%), and OPG (19%) in primary carcinoma. The expression frequency was significantly higher (RANKL [44%], RANK [49%], and OPG [73%]) in metastatic prostate cancer. OPG (83%) production was more common in skeletal as compared with lymph node metastases (46%), whereas the expression of RANKL expression was less discordant in bone (47%) and lymph node metastases (36%). The increased expression of RANKL/RANK/OPG observed correlated with Gleason score, TNM stage, androgen status, and serum prostate-specific antigen (PSA) levels in the prostate cancer patients.
Metastatic prostate cancer cells often localize in bone to form secondary lesions responsible for the death of the patient.1, 2 The underlying molecular mechanisms of prostate cancer metastasis to and invasiveness in the skeletal, however, are still poorly understood. Recently, a novel cytokine system consisting of receptor activator of NF-κ B ligand (RANKL), its receptor, RANK, and the protein osteoprotegrin (OPG) was identified and extensively characterized for its role in bone remodeling. RANKL is expressed as a membrane-bound protein on the surface of osteoblasts and on bone marrow stromal cells3 and a soluble form of RANKL (sRANKL) can be secreted by activated T-cells.4 RANKL binds to RANK, present at the surface of osteoclasts precursors, to induce osteoclastogenesis and activation of mature osteoclasts in the presence of M-CSF.4, 5 OPG, also produced by osteoblast/stromal cells, is a decoy receptor of RANKL, which by binding to RANKL can prevent the interaction between RANKL and RANK and can inhibit osteoclastogenesis.6 In transgenic animal models overexpressing OPG7 or in RANK knockout (KO)8 mice, osteopetrosis was observed due to lack of osteoclasts. In these models administration of RANKL resulted in the development of osteoporosis associated with hypercalcemia.3 Moreover, OPG KO mice developed uncontrolled bone resorption and severe osteoporosis.3, 9 Therefore, the RANKL/RANK/OPG system represents a key regulatory mechanism in osteoclastogenesis. Most osteoclast-stimulating factors including interleukin (IL)-1, IL-6, 1,25-(OH)2D3 and PTHrP can also induce RANKL and inhibit OPG production.10
The RANKL/RANK/OPG system has been dysregulated in several tumors, such as breast cancer,11 malignant bone tumors (multiple myeloma,12 giant cell tumors of bone,13 and chondroblastomas14), neuroblastoma,15 squamous cell carcinoma,16 and Hodgkin disease.17 Furthermore, an altered RANKL/OPG ratio has been implicated in bone metastases.18 Inhibition of RANKL in vitro can prevent tumor cell-induced osteoclastogenesis15 and administration of Fc-OPG has been shown to have a therapeutic effect in animal models of several malignances, such as breast cancer,19 colon cancer,20 sarcoma,21 and myeloma.22 A Phase I study of recombinant OPG in patients with multiple myeloma and breast carcinoma-related bone metastases has also been recently reported.23
In prostate cancer, RANKL and/or OPG have been detected in several prostate cancer cell lines.24 RANKL has been reported to mediate cancer-induced osteoclastogenesis and OPG can prevent the development of skeletal metastasis in an animal model.25 sRANK-Fc and Fc-OPG have been shown to exert inhibitory effects on intraosseous growth of several prostate call lines.26 In addition, increased serum OPG in animal models and in prostate cancer patients with bone metastasis has been reported.27 However, characterization of the expression of RANKL/RANK and OPG in human prostate cancer is still limited. Accordingly, we hypothesize that the expression of RANKL/RANK/OPG may correlate with disease progression. Expression of RANKL, RANK, and OPG was examined in various human prostate cancer cell lines with different metastatic potential and compared with a normal human prostate cell line. Immunohistochemical analysis was carried out to assess the expression of RANKL, RANK, and OPG in primary human tumors of low and high Gleason score, and in lymph node and skeletal metastases. Expression of these proteins was correlated with clinicopathogic parameters of prostate cancer.
MATERIALS AND METHODS
In the current study, surgical biopsy samples of 89 patients with prostate cancer, which included 48 cases of primary and 41 cases of metastatic prostate cancer (11 lymph node metastases and 30 skeletal metastases), were investigated. All patients with metastatic disease did not receive hormone therapy at the time of sample collection. Detailed clinical data on these patients is provided in Table 1. Tumors were staged using the American Joint Committee on Cancer tumor, lymph node, metastases (TNM) classification and were graded by Gleason score. Median PSA levels (10 μg/ml), mean age (65 years), and Gleason score (< or > 7) were used as patient classification criteria as previously reported for similar studies.28
Table 1. Correlation between RANKL, RANK, and OPG Expression and Clinicopathologic Parameters of Prostate Cancer Patients
Total RNA was extracted from PrEC, LNCaP, DU145, and PC3 cells by a single-step method using TRIzol reagent (Invitrogen, La Jolla, CA). The following primers were used to amplify RANK: forward 5′-CAGGGATCGATCGGTACAGT-3′ and reverse 5′-GTTTGAGACCAGGCTGGGTA-3′; RANKL: forward 5′-GCTTGAAGCTCAGCCTTTTGCTCAT-3′ and reverse 5′-GGGGTTGGAGACCTCGATGCTGATT-3′; OPG: forward 5′-GAACCCCAGAGCGAAATACA-3′ and reverse 5′-TATTCGCCACAAACTGAGCA-3′. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) primers were used as a control. The amplified PCR product was fractionated on a 1.3% agarose gel and visualized by ethidium bromide staining. Band intensities for RANK, RANKL, OPG, and GAPDH were quantified using densitometric software (Quantity One; Bio-Rad Laboratories, Mississauga, Ontario, Canada).
Cell Proliferation Assay
For growth curves, PrEC, LNCap, DU145, and PC3 cells were plated in 6-well plates (Falcon Plastics, Lincoln Park, NJ) at seeding densities of 5 × 103 cells/well. Cells from triplicate wells were cultured for 5 days in the presence of different doses of RANKL (10.0-100.0 ng/ml), trypsinized, resuspended, and counted in a model Z Coulter counter (Coulter Electronics, Beds, UK). The medium was changed every 2 days. RANKL was from R&D Systems (Minneapolis MN).
Cells and Immunohistochemistry
Normal human prostate cells, PrEC, were purchased from Clonetics (San Diego, CA) and the human prostate cancer cell lines LNCaP, DU-145, and PC-3 cells were obtained from the American TypeCulture Collection (Rockville, MD). PrEC cells were maintained in PrEBM cell culture medium and supplemented with suggested reagents, growth factors, and antibiotics according to the supplier's instructions. All prostate cancer cell lines were maintained in vitro in RPMI 1640 (Invitrogen, Burlington, Ontario, Canada) supplemented with 10% fetal bovine serum (FBS) and 100 U/ml penicillin-streptomycin sulfate (Invitrogen). Cells were incubated at 37°C in 5% CO2. For immunocytochemical analysis, all cell lines were cultured in 24-well plates and fixed with cold methanol for 10 minutes.
The immunohistochemical analysis of the tumor samples was carried out as previously described29 using the avidin-biotin-peroxidase complex method. Briefly, the tumor samples were fixed in 10% buffered formalin (Fisher Scientific, Nepean, ON, Canada) and embedded in paraffin. The samples of bone metastases were decalcified with 10% EDTA (Fisher Scientific) in phosphate-buffered saline (PBS) for 2 weeks before embedding. Five-μm sections were cut and were deparaffinized in xylene and rehydrated through a graded ethanol into water. Antigen retrieval was done by microwaving slides in 10 mM sodium citrate buffer (pH 6.0) to boiling, then maintaining at a subboiling temperature for 10 minutes. Sections were treated with 0.3% hydrogen peroxide (H2O2) for 30 minutes to block endogenous peroxidase. To block the nonspecific binding, the sections were treated with 1% normal serum (Vector Laboratories, Burlingame, CA) for 30 minutes at room temperature before incubation with the primary antibody (polyclonal antibodies against RANKL or against OPG and monoclonal anti-RANK; R&D Systems; anti-RANKL, anti-OPG at 1:100, anti-RANK at 1:50) overnight at 4°C. Biotinylated goat antirabbit, goat antimouse, or rabbit antigoat IgG (Vector Laboratories) were used as the secondary antibodies at 1:200 for 30 minutes at room temperature. The slides were treated with Vectastain ABC-AP kit (Vector Laboratories) diluted 1:200 for 30 minutes at room temperature. The signals were visualized using DAB (3, 3′diaminobenzidine) kit (Vector Laboratories) as per the manufacturer's protocol. The slides were then counterstained with hematoxylin (Fisher Scientific), dehydrated through graded ethanol, cleared in xylene, and mounted with Permount mounting media (Fisher Scientific). After each step, all sections were washed 3 times for 10 minutes each with Tris buffer (pH 7.6). Negative controls included substitution of the primary antibody with PBS. Normal prostate tissue adjacent to the tumor was used for each specimen as an internal control.
Computer-Assisted Image Analysis
The immunostaining of all antibodies was quantitatively analyzed by using a computer-assisted image analysis system as previously described.29 Briefly, images of stained sections were captured with a Leica digital camera and processed using BioQuant image analysis software (version 6.50.10; BioQuant Image Analysis, Nashville, TN). The threshold was set by determining the positive staining of control sections and was used to automatically analyze all recorded images of all samples that were stained in the same session under identical conditions. The area of immunostained regions was calculated automatically by the software in each microscopic field. Pixel counts of the immunoreaction product were calculated automatically and were given as total density of the integrated immunostaining over a given area of the sections. This reflects the relative amount of proteins detected by the antibodies on the tumor sections. The ratio of OPG/RANKL was calculated from integrated immunostaining densities of OPG and RANKL in each group.
Data from image analysis of sections are presented as mean ± the standard error of the mean (SEM) of each group. Statistical analyses were performed using Student t-test and analysis of variance (ANOVA), with a P-value <.05 considered statistically significant. Chi-square test or Fisher exact tests were used to determine the correlation between RANKL/RANK/OPG expression and clinicopathologic parameters.28
Expression of RANKL, RANK, and OPG in Human Prostate Carcinoma Cell Lines
By RT-PCR, all 3 human prostate carcinoma cell lines LNCaP, DU145, and PC3 showed expression of RANKL, RANK, and OPG. In contrast, only low levels of RANKL, RANK, and OPG were seen in normal PrEC cells (Fig. 1). Semiquantitative RT-PCR analysis confirmed the differential levels of RANKL, RANK, and OPG expression in each group (Fig. 1). Two androgen-insensitive cell lines (DU145 and PC3) showed the strongest gene expression of RANKL, RANK, and OPG compared with the relatively lower levels of expression observed in the androgen-sensitive and less invasive LNCaP cells. Normal PrEC cells only showed minimal staining. Determination by immunostaining of OPG, RANKL, and RANK protein in these normal and malignant prostate cancer cell lines showed similar relative levels of protein expression as observed with RT-PCR (Fig. 2). However, significantly lower OPG and RANKL protein expression was seen in LNCaP cells than with the other 2 malignant cell lines.
In view of the fact that both RANKL and RANK were expressed in these cell lines, we next evaluated the action of RANKL on the prostate cancer cells per se. RANKL produced a dose- and time-dependent increase in cell proliferation in PC-3 cells (Fig. 3).
Increased Expression of RANKL, RANK, and OPG in Human Prostate Carcinoma
We next examined protein levels, by immunostaining, in human prostate tissue examples. Expression of RANKL, RANK, and OPG was low in normal (CTL) prostatic tissue (Fig. 4A), whereas all 3 proteins were up-regulated in the 89 specimens of prostate cancer (33%, 38%, and 19% respectively) (Table 2). Quantitative analysis of the immunohistochemical staining confirmed significantly higher levels of RANKL, RANK, and OPG production in prostate cancer tissues compared with normal prostatic tissue.
Table 2. RANKL, RANK, and OPG Expression in Primary and Metastatic Prostate Cancer
All 3 proteins were found in low levels in some tissue components such as lymphocytes and macrophages. The highest levels of production of RANKL, RANK, and OPG observed in these studies were at the invasive front of prostate cancer cells compared with the central portion of the cancer tissues (arrows, Fig. 4A).
Correlation of RANKL, RANK, and OPG Expression with Clinicopathologic Parameters of Prostate Cancer
The relation between RANKL/RANK/OPG production and other clinicopathologic parameters were available in 48 of the 89 cancer patients from whom cancer specimens were obtained and the relation between RANKL/RANK/OPG and these parameters were determined in these tumors. As shown in Table 1, expression of RANKL, RANK, and OPG was not associated with age. However, a clear correlation was established between the expression of RANKL, RANK, and OPG production and tumor stage, grade, Gleason score, androgen receptor (AR) status, and serum prostate-specific antigen (PSA) levels of these patients. Therefore, higher expression of RANKL, RANK, and OPG was observed in more advanced metastatic tumors (Table 1). Fifty percent and 56% of highly malignant tumors as graded by Gleason score expressed positive RANKL and RANK staining, respectively, compared with 27% and 33%, respectively, of cases in the low Gleason score group. For OPG levels, 67% stained positive in the high Gleason group as compared with 30% in the lower Gleason score group (Table 1). When expression of RANKL and RANK were matched with serum PSA levels, it was observed that 40% and 49% in the high PSA group stained positive for RANKL and RANK, respectively, compared with 23% and 23%, respectively, in the low PSA group. In the case of OPG, 49% in the high PSA group stained positive compared with 31% in the low PSA group (Table 1). Analysis of the relation between AR status and expression of RANKL, RANK, and OPG also revealed that 46%, 54%, and 54% in the AR-negative group stained positive for RANKL, RANK, and OPG, respectively, in contrast to 20% (RANKL), 25% (RANK), and 30% (OPG) in the AR-positive group. The AR status of our patient population is similar to previous reports.30 Consequently, levels of RANKL, RANK, and OPG expression are associated with decreased AR production by tumor cells.
The level of expression of the 3 components of the system that were also greater in more malignant tumors of higher Gleason score compared with expression in lower Gleason score tumors (Fig. 4A). In addition, significantly higher ratios of OPG/RANKL immunostaining density was observed in the high Gleason group compared with the low Gleason group or the normal prostate tissue (Fig. 4B).
Expression of RANKL, RANK, and OPG in Prostate Cancer Associated with Metastases
Of 41 prostate cancer metastases to lymph nodes and bone included in the current study, 18 cases (44%) and 20 cases (49%) showed expression of RANKL and RANK, respectively, and 30 cases (73%) had expression of OPG. In comparison, in 48 primary cancers 31% and 38% of cases showed expression of RANKL and RANK, respectively, and only 19% of cases demonstrated expression of OPG (Table 2). Significantly stronger immunostaining for RANKL, RANK, and OPG was obtained in both lymph node and bone metastases as compared with the tumor cells at the primary site (Fig. 5A). This was confirmed by quantitative analysis of the immunostaining density of the proteins in each group.
Expression of RANKL, RANK, and OPG were also observed more frequently in skeletal metastases than in lymph node metastases (Table 3). Therefore, 36%, 46%, and 46% of lymph node metastases showed positive staining for RANKL, RANK, and OPG, respectively. In contrast, the percentage that stained positive for these 3 proteins in bone metastases was 47%, 50%, and 83%, respectively (Table 3). These differential levels of RANKL, RANK, and OPG expression were confirmed by quantitative analysis of the staining density of each group. In bone, RANKL and OPG could be detected in osteoblasts and RANK expression was seen in osteoclasts. RANKL and OPG expression were also found at low levels in some bone marrow stromal cells. In addition, in the skeletal metastases the highest levels of production of RANKL, RANK, and OPG were observed at the cancer cell–bone interfaces. RANKL and OPG expression were also seen in some lymphocytes in lymph nodes.
Table 3. RANKL, RANK, and OPG Expression in Bone and Lymph Node Metastasis of Prostate Cancer
No. of Positive (%)
No. of Patients
RANKL indicates receptor activator of NF-κB ligand; RANK, RANKL's receptor; OPG, osteoprotegerin; NS, not significant.
Analysis of the ratio of OPG/RANKL revealed much higher ratios in bone metastases as compared with lymph node metastases and primary tumor (Fig. 5B).
Tumor cells can secrete soluble factors, such as PTHrP, to stimulate RANKL production by osteoblastic stromal cells that can then enhance osteoclastic osteolysis. However, it is well established that tumor cells can also directly produce RANKL and/or OPG.12, 24, 31 Our in vitro results exhibiting higher levels of RANKL and OPG in poorly differentiated, androgen-independent prostate cancer cell lines compared with the lower levels of expression of these proteins in androgen-dependent cells is in agreement with previous observations.24 RANKL increased the proliferation of exponentially growing cells. Effects have also been reported in human breast cancer cells.32 However, we cannot rule out the effect of RANKL on cell cycle, which will be the subject of future studies. Furthermore, we have demonstrated that human prostate cancer cells express all 3 components of the RANKL/RANK/OPG system and that the levels of these proteins are significantly increased in advanced prostate cancer, with the highest OPG/RANKL ratio in bone metastases compared with the primary cancer samples. RANKL has been found as a necessary factor for cancer-associated osteolytic lesions.10, 11 Recently, it has been reported that RANKL and MMP-7 colocalize at the tumor–bone interface, in which MMP-7 can promote prostate cancer-induced osteolysis by processing RANKL.33 Our studies confirmed this localization of RANKL but also demonstrated RANK expression at these sites.
Previous studies have also reported dysregulation of the RANKL/RANK/OPG axis in a number of cancers and the levels of these components seems to be associated with differing tumor characteristics. Multiple myeloma, a bone malignancy with pure lytic lesions, has been shown to exhibit high levels of RANKL and low levels of OPG.12 Similarly, in breast cancer with osteolytic metastases higher levels of RANKL have been reported, whereas OPG expression was found to be negatively correlated with increasing tumor grade.34 In the case of prostate cancer, RANKL and OPG expression were determined in animal models, but limited data is available in human prostate cancer cases.33 In 1 human study OPG and RANKL were shown in bone metastases; however, the relation of this expression with clinicopathologic parameters of prostate cancer could not be established due to limited sample size.31 In this context, although results from our current study are significant, they reflect a trend in 48/96 patients for which complete data of clinicopathologic parameters of cancers were available. Furthermore, quantitative analysis of immunostaining used in the current study provided a more objective and reliable method to evaluate the accumulated expression of target proteins in each group. Our observation that expression of all components of this system is increased in more advanced cancers is consistent with other cancers where dysregulation of the RANKL/RANK/OPG system was studied.
Serum OPG has been used to monitor the skeletal progression of prostate cancer in an animal study. Several clinical studies in various malignancies (gastric, prostate, and bladder carcinomas and myeloma) have reported increased serum OPG and RANKL levels in cancer patients with poor prognosis.27, 35, 36 The current results indicating that overexpression of RANKL/RANK/OPG was closely correlated with advanced tumor grade represented as Gleason score, tumor stage, or serum PSA levels provides histopathologic support for these clinical observations.
Skeletal metastases have been identified at autopsy in 90% of patients dying of cancer.2 These bone lesions exhibit histologically mixed bone resorption and formation; however, osteoblastic lesions uniformly predominate.37 New bone formation in osteoblastic metastases seems to occur at sites of previous osteoclastic resorption that may be a prerequisite for such tumor-induced bone formation. This suggests that an overall increase in bone remodeling accompanies the metastases and is initiated by osteoclastic activation.25 In normal bone, osteoclast activation and loss of activation are tightly controlled by a balance between RANKL and OPG, which act as a positive and negative regulator, respectively. Cancer cells may shift the balance of osteoclastic osteolysis through production of RANKL and/or OPG directly, or by the production of other factors that indirectly stimulate osteoblast/stromal cells to produce RANKL or OPG. This may then create a favorable local environment in bone for tumor cell seeding and development of metastasis.10 Previous studies in skeletal metastases of various tumors have indeed shown that RANKL/OPG ratio is increased in severe osteolysis.18 Our current results showing increased OPG/RANKL ratio in human prostate cancer skeletal metastasis may reflect the predominant osteoblastic characteristic of the bone lesion in prostate cancer, especially as the lesions progress. Indeed, in an animal model, overexpression of OPG was reported to decrease the lytic lesions of human prostate cancer cells C4-2 and increase bone volume.38
Whereas expression of these proteins have been studied in human cancers, limited reports are available where expression of RANKL/RANK/OPG were examined simultaneously.11, 39 RANKL can also directly induce chemoattractive and invasive factors to enhance tumor cell growth and invasion. In our studies, RANKL was shown to have a dose-dependent effect on PC-3 cells that also expressed RANK. Overexpression of OPG has also been reported to increase the growth of breast cancer cell lines in vitro, allowing the growth of large tumors in the skeleton.40 Furthermore, OPG has been demonstrated to act as a survival factor by binding the TNF receptor TRAIL and blocking TRAIL-mediated tumor cell apoptosis.24 Consequently, the RANK/RANKL/OPG system may not only be an index of the advanced state of the cancer, but may also contribute to the growth and invasive properties of the tumor per se.
Overall, therefore, results in these studies provide compelling evidence for the expression of all components of the RANKL/RANK/OPG signally system in more undifferentiated prostate cancer and for a functional role for RANKL/RANK/OPG in prostate cancer growth and skeletal metastases. Further studies are warranted to examine the mechanism of action of these proteins in prostate cancer growth and progression and possible cross talk with other signaling pathways.