By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Wiley Online Library will be unavailable on Saturday 7th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 08.00 EDT / 13.00 BST / 17:30 IST / 20.00 SGT and Sunday 8th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 06.00 EDT / 11.00 BST / 15:30 IST / 18.00 SGT for essential maintenance. Apologies for the inconvenience.
To delineate the role of SDF-1 and CXCR4 in metastatic prostate cancer (CaP), positive correlations were established between SDF-1 levels and tumor metastasis. Neutralization of CXCR4 limited the number and the growth of intraosseous metastasis in vivo. Together, these in vivo metastasis data provide critical support that SDF-1/CXCR4 plays a role in skeletal metastasis.
Introduction: Previously we determined that the stromal-derived factor-1 (SDF-1)/CXCR4 chemokine axis is activated in prostate cancer (CaP) metastasis to bone. To delineate the role of SDF-1/CXCR4 in CaP, we evaluated SDF-1 levels in a variety of tissues and whether neutralization of SDF-1 prevented metastasis and/or intraosseous growth of CaPs.
Materials and Methods: SDF-1 levels were established in various mouse tissues by ELISA, immunohistochemistry, and in situ hybridization. To assess the role of SDF-1/CXCR4 in metastasis, bone metastases were established by administering CaP cells into the left cardiac ventricle of nude animals in the presence or absence of neutralizing CXCR4 antibody. The effect of SDF-1 on intraosseous growth of CaP cells was determined using intratibial injections and anti-CXCR4 antibodies and peptides.
Results: There was a positive correlation between the levels of SDF-1 and tissues in which metastatic CaP lesions were observed. SDF-1 levels were highest in the pelvis, tibia, femur, liver, and adrenal/kidneys compared with the lungs, tongue, and eye, suggesting a selective effect. SDF-1 staining was generally low or undetectable in the center of the marrow and in the diaphysis. SDF-1 mRNA was localized to the metaphysis of the long bones nearest to the growth plate where intense expression was observed near the endosteal surfaces covered by osteoblastic and lining cells. Antibody to CXCR4 significantly reduced the total metastatic load compared with IgG control-treated animals. Direct intratibial injection of tumor cells followed by neutralizing CXCR4 antibody or a specific peptide that blocks CXCR4 also decreased the size of the tumors compared with controls.
Conclusions: These data provide critical support for a role of SDF-1/CXCR4 in skeletal metastasis. Importantly, these data show that SDF-1/CXCR4 participate in localizing tumors to the bone marrow for prostate cancer.
PROSTATE CANCER (CaP) frequently metastasizes to the bone marrow, resulting in significant morbidity and mortality. Nearly 90% of patients with advanced prostate cancer are at high risk for pathologic fractures, spinal cord compression, and pain due in part to dysregulated cycles of osteoblastic and osteolytic resorption/formation driven by the growing tumor mass.(1) What defines the molecular and cellular predilection for prostate cancers to metastasize to bone is not well known. Clearly, multiple factors, including the acquisition of metastatic abilities by the cancer cell, chemotactic responses to bone-derived factors, preferential adhesion to bone marrow endothelium, and the interaction of CaP cells with the bone microenvironment leading to tumor in the bone marrow, play considerable roles.(2)
Chemokines are a group of molecules known to play significant roles as activators and chemoattractants for many blood cell types as well as an increasing array of non-hematopoietic cell types (e.g., epithelial cells, fibroblasts, and endothelial cells). Chemokines are 8- to 10-kDa proteins subdivided into four structural groups on the basis of the relative position of the first two cysteine residues in the mature protein (CXC, ML, C3XC, and C). At present, >50 chemokines have been identified, most of which are thought to activate cell migration by binding to G-protein-coupled cell surface receptors. Currently, there are at least 18 known chemokine receptors.(3) Activation of chemokine receptors triggers activation of many downstream pathways including nonreceptor tyrosine kinases, inositol trisphosphate, MAPK, protein kinase C, and calcium mobilization.
Previously, we determined that the stromal-derived factor-1 (SDF-1)/CXCR4 chemokine axis is activated in CaP metastasis to bone.(4, 5) Specifically, CXCR4 expression is related to increasing tumor grade.(5) Moreover, we have shown that SDF-1 signaling through CXCR4 triggers the adhesion of CaPs to bone marrow endothelial cells. Similar demonstrations have also been made suggesting that the SDF-1/CXCR4 axis may play parallel roles in other tumors that also metastasize to the marrow.(6, 7) For example, Muller et al.(7) reported that CXCR4 and SDF-1 are central players in regulating metastasis by showing that normal breast tissues express little CXCR4, whereas breast neoplasms express high levels of CXCR4. Furthermore, antibody to CXCR4 blocks the metastatic spread of the tumors to the lung and lymph nodes. Similarly, mRNA levels for CXCR4 are elevated in regions of angiogenesis and degeneration but decreased in areas of rapid cell proliferation in glioblastoma multiforme that may prime the cells for metastasis.(8) Results consistent with these have also been reported for human melanoma cell lines, melanoma cells that had macroscopically infiltrated draining lymph nodes,(9) and for pancreatic, neuroblastoma, and renal carcinomas.(10, 11)
Common to most of these reports is the analogy with hematopoietic stem cell homing where gene knockout approaches in mice have shown that SDF-1 and CXCR4 play an important role in embryologic development.(12–14) Specifically, SDF-1-deficient mice die in utero with severe cardiac septum defects and poorly developed marrow. CXCR4 receptor deletion is similarly lethal, resulting from circulatory, CNS, immune, and hematopoietic defects.(13, 15) The phenotype of the SDF-1−/− mouse so closely followed the CXCR4−/− that these investigations showed that the receptor-ligand pair play important roles in cardiac, CNS, and hematopoietic stem cell homing.(16) In addition, CXCR4 expression levels correlate with the ability of human progenitors to engraft into the marrow in nude mice, and antibody to CXCR4 prevented the engraftment of progenitors in the bone marrow.(14) Finally, osteoblasts and marrow endothelial cells express SDF-1, which may thereby localize hematopoietic progenitor cells in the marrow.(14, 17, 18)
In this report, the goal was to delineate the role of SDF-1/CXCR4 receptor in CaP disease. Two independent studies were performed. First, to establish a positive correlation between the SDF-1 gene and protein expression with tumor metastasis, SDF-1 levels were characterized for a variety of murine tissues by ELISA and in situ hybridization. In an in vivo metastasis model, the role of CXCR4 in metastasis of prostate cancer to the bone was examined. In this model, neutralizing antibody to CXCR4 limited the extent of bone metastases. Finally, antibody and blocking peptide to CXCR4 limited the growth of intraosseous CaPs after CaP cell intratibial injections. Together, these in vivo metastasis data provide critical support that SDF-1/CXCR4 plays a role in skeletal metastasis. Most importantly, these novel data show that SDF-1/CXCR4 participates in localizing tumors to the bone marrow in CaP and may help in design of therapeutic treatments to prevent the spread and growth of metastasis.
MATERIALS AND METHODS
CaP cell line
PC3 prostate cancer cells originally isolated from a vertebral metastasis of a human prostate cancer patient were obtained from American Type Culture Collection (Rockville, MD, USA) and were maintained in RPMI containing 10% FBS, 100 U/ml penicillin, and 100 mg/ml streptomycin. PC3 cells infected with the pLazarus retroviral construct expressing luciferase were selected for stable transfectants in G418.(19, 20)
Bone marrow and tissue extracts were derived from C57BL6 mice in PBS with protease inhibitors (Sigma). In some cases, the femur and tibia were dissected free of soft tissues, and 2-mm sections of the bones were generated on ice using a 0.5-mm diamond mandrill set at low speed in a dental drill. Harvested tissues were subjected to sonication and spun, and the resulting supernatants were stored at −80°C until assayed for cytokine SDF-1 levels by double-antibody sandwich method assembled with commercially available components according to the directions of the manufacturer (sensitivity, 31.25 pg/ml; range, 62.5-5000 pg/ml; R&D Systems). SDF-1 levels are presented as mean ± SE picograms per milliliter for duplicate determinations. Conditioned media from human osteoblasts (HOBs) and the human osteosarcoma cell line MG-63 (ATCC CRL1424) were used as positive controls for SDF-1.
SDF-1 bioactivity in tissues
Bone and liver tissue extracts were generated from 20-day-old C57BL6 mice in RPMI. In vitro invasion assays were performed using a reconstituted extracellular matrix membrane (BD BioCoat Matrigel Invasion Chamber; BD Biosciences, San Jose, CA, USA) using PC3 cells after determining the levels of SDF-1 in the extract by ELISA. For these investigations, test cells were placed in the upper chamber (1 × 105 cells/well) in RPMI medium with 1% serum and 0–200 ng/ml SDF-1 or 50% tissue extract: RPMI (vol/vol) was added to the lower chamber. Spontaneous invasion was compared with invasion supported by a SDF-1 gradient. Invasion into the matrix was assayed after 24 h by quantification with MTT. The effect of 30 μg/ml CXCR4 blocking antibody (MAB173; R&D Systems), IgG control, or peptides designed to block SDF-1 binding to the CXCR4 receptor (TC14012 or TN14003(21, 22)) synthesized by the Emory Microchemical Facility served as positive controls. They were added to the top chamber of the Transwell to provide additional proof that observed responses are dependent on CXCR4 receptor binding.
Femurs and tibia were trimmed of musculature, fixed in 10% formalin at 4°C, decalcified in 10% EDTA (pH 7.4) for 10 days, and embedded in paraffin. Longitudinal sections of tibias were cut and stained with H&E for histological evaluation. Where indicated, histomorphometric analyses of the femurs/tibias were performed using a computer-assisted bone histomorphometric analyzing system (Image-Pro Plus version 4.0; Media Cybernetics, Silver Spring, MD, USA).
Digoxigenin in situ hybridization and immunohistochemistry
Normal murine bones were harvested and fixed in 10% buffered formalin and decalcified in EDTA, and 2- to 3-μM paraffin embedded slides were prepared. Human bone was obtained from patients undergoing orthopedic surgery in accordance with the University of Michigan's Investigational Review Board. Each slide was dewaxed and rehydrated and subsequently treated with proteinase K, 10% formalin, and washed in SSC. Hybridization was performed with a digoxin-labeled SDF-1 sense or antisense SDF-1 riboprobe at 55°C overnight. After incubation, the slides were washed in 4× SSC, RNase A, 2× SSC-10 mM DTT, 1× SSC-10 mM DTT, and 0.5× SSC-10 mM DTT.(23) The slides were washed, blocked in 2% blocking buffer for 1 h, 1:500 antibody (anti-digoxigenin), and color developed using nitro-blue tetrazolium chloride (NBT) and 5-bromo-4-chloro-3′-indolyphosphate p-toluidine salt (BCIP). In some cases, antibody to SDF-1 (MAb Clone 79018; R&D Systems) or IgG control (Sigma) were used in conjunction with a DAB Chromogen kit (LSAB + Peroxidase Kit; Dako, Carpinterina, CA, USA) to identify the protein in tissues.
SDF-1 binding assay
To show the validity of the antibody and peptide reagents, PC3 cells were plated in 96-well plates at an initial density of 2 × 104 cells/well. After 24 h, antibody to SDF-1 (polyclonal and monoclonal), unlabeled SDF-1 (0-100 μg), anti-CXCR4 (clone 12G4 or MAB 171 [clone 44708.111]), and anti-CXCR4 peptide or IgG control (0-100 μg/ml) were added to select wells (all reagents from R&D Systems). Similarly, peptides designed to block SDF-1 binding to the CXCR4 receptor (TC14012 or TN14003(21, 22)) were synthesized by the Emory Microchemical Facility and added to several wells. Biotinylated SDF-1 was added to the wells, and binding was allowed to proceed for 1 h at 4°C. Thereafter, the cells were washed, streptavidin horseradish peroxidase (HRP) was added to wells for 0.5 h at 4°C, and O-phenylenediamine dihydrochloride (OPD; Sigma) in citrate buffer was added for development.
In vivo metastasis assay
For these studies, 2 × 105 PC3luc cells that were stably transfected with luciferase before the experiment were used. Two experimental groups were established; sodium azide-free monoclonal antibodies, anti-CXCR4 (clones MAB171 from R&D and 12G5 from BD PharMingen, San Diego, CA, USA) and polyclonal anti-SDF-1 (R&D), were used to neutralize the CXCR4-CXCL12/SDF-1 circuitry. Appropriate irrelevant antibodies (sodium azide-free 44708.111 clone [MBA171] monoclonal antibody [IgG2a]; R&D Systems) or IgG matched controls were used. Before inoculation, the tumor cells were preincubated with antibody to CXCR4 or control at 5 μg per 1 × 104 injected cells. Our choice of reagents and dosing was based on the reported use in blocking human hematopoietic CD34+ cell engraftment into the bone marrows of NOD/SCID animals,(14) the ability to block invasion of LNCaP, C4-2B cells and PC3 invasion in vitro,(4) and the ability to block SDF-1 binding to PC3 cells. Moreover, these reagents were used by Muller et al.(7) to block metastasis of breast cancer cell lines mediated by SDF-1 in vivo. Inbred HSD:athymic male mice were purchased from Harlan Bioscience (Indianapolis, IN, USA) and were housed under constant humidity and temperature, with 12-h light and 12-h dark cycles. The animals were anesthetized with a mixture of xylazine (100-200 mg/kg) and ketamine HCL (5-16 mg/kg; Ketaset; Fort Dodge Laboratories, Fort Dodge, IA, USA), and the left cardiac ventricle was punctured percutaneously and injected using a 25-gauge needle attached to a 1-ml syringe as previously described.(20) The animals were subsequently injected 2 and 24 h afterward with equal doses of the same antibody.
After 4 weeks, bioluminescence imaging was used to follow prostate-derived bone metastases as a primary outcome. The mice were injected intraperitoneally with luciferin (100 μl at 40 mg/ml in PBS) before imaging. This dose and route of administration have been shown to be optimal for rodent studies with a 10- to 20-minute after luciferin injection.(24) Mice were anesthetized with 1.5% isoflurane/air, and the Xenogen IVIS cryogenically cooled imaging system was used as described.(20) Selected mice were imaged weekly after tumor injection to monitor tumor development. Bioluminescence generated by the luciferin/luciferase reaction served as a locator for cancer growth and was used for quantification using the LivingImage software on a red (high intensity/cell number) to blue (low intensity/cell number) visual scale. A digital grayscale animal image was acquired followed by acquisition and overlay of a pseudocolor image representing the spatial distribution of detected photon counts emerging from active luciferase within the animal. Signal intensity was quantified as the sum of all detected photons within the region of interest during a 1-minute luminescent integration time.
PC3Luc cells were inoculated intratibially to measure the effect of blocking CXCR4 on tumor growth.(25) Intratibial injection was performed as previously described.(25) Briefly, animals were anesthetized, and both legs were cleaned with betadine and 70% ethanol. PC3luc cells were injected through the cortex of the anterior tuberosity of the tibia with a drill-like motion to prevent cortical fracture using a 25-μl syringe fitted with a 25-gauge needle. Ten microliters of cell suspension was injected (107 cells/ml) to the tibia. Growth of the intraosseous tumor was examined using the a CXCR4 antagonist synthesized by the Microchemical Core Facility at Emory University. A control peptide was produced by randomly scrambling the amino acid sequence of CXCR4 antagonist while maintaining the disulfide bond to maintain the U-type structure of the antagonist (NH2-KY-Nal-YR-DK-Cit-RCRRP-Cit-C-amide).(26)
Calcium and pyridinoline cross-links levels in serum
Serum calcium and pyridinoline (PYD) levels were used as measures of bone turnover. Blood samples were obtained from the experimental mice after death, serum was separated by centrifugation, and total calcium was determined by colorimetric assay with the cresolphthalein complexone method (Sigma). Serum PYD levels were assayed using an EIA kit (Quidel Corp., San Diego, CA, USA) after filtration by adding ∼200 μL of each serum to a 30-k molecular weight cut-off (MWCO) Spinfilter and centrifugation at 10,000g for 30 minutes (Quidel Corp.).
Numerical data are expressed as means ± SD. Statistical differences between the means for the different groups were evaluated with Instat 4.0 (GraphPAD software) using one-way ANOVA, with the level of significance at p < 0.05.
There is now substantial evidence to suggest that chemokines and their receptors are involved in the pathogenesis of many diseases, ranging from infectious to autoimmune disorders. Recently, chemokine receptors have been implicated in the homing of tumors to the marrow microenvironment, and our own data suggest that SDF-1 and CXCR4 may play a critical role in these phenomena based on the expression of CXCR4 by CaPs and SDF-1 secretion by human osteoblasts in vitro. To further characterize the role that SDF-1 may play in CaP metastasis, a variety of murine tissues were examined for the level of expression of SDF-1 protein by ELISA. As shown in Fig. 1, high SDF-1 levels were identified in tissues to which CaPs frequently metastasize to including the femur, tibia, and pelvis. Surprisingly SDF-1 levels were also elevated in the optic nerve region and in cardiac muscle. In contrast, tissues that are rarely sites of CaP metastasis in humans had lower levels of SDF-1 (eye, tongue, maxilla, mandible). To localize the sites of SDF-1 production in bone, bone marrow extracts derived from the tibia were examined by ELISA after sectioning the bones into 2- to 3-mm lengths. The highest levels of SDF-1 in the medium fraction of the marrow were localized to the distal and proximal extremities of the long bones (data not shown).
Further studies were performed to localize the sites of SDF-1 production in the marrow. SDF-1 mRNA expression as determined in situ hybridization was predominantly localized to the endosteal surfaces, particularly near the metaphysis in both the human and murine bone samples examined. SDF-1 staining was generally low or undetectable in the center of the marrow and in the diaphysis. Importantly, SDF-1 mRNA was localized to the metaphysis of the long bones examined nearest to the growth plate. At higher magnification, intense SDF-1 expression was frequently observed near the endosteal surfaces covered by osteoblastic and lining cells (Figs. 2A-2F). Immunohistochemistry for SDF-1 protein further verified that SDF-1 expression was generally low or undetectable in the diaphysis regions of long bones (Figs. 3A-3F). Peak SDF-1 protein expression was observed at the growth plates (Figs. 3E and 3F).
SDF-1 bioactivity in tissues
The preceding ELISA, in situ hybridization, and immunohistochemistry results show that SDF-1 is expressed in tissues that are frequent sites of metastasis. To determine if SDF-1 is biologically active, bone and liver tissue extracts were generated from 20-day-old C57BL6 mice. In vitro invasion assays were performed using the PC3 prostate cancer cell line to evaluate bioactivity. In the presence of SDF-1, significantly more PC3 cells invaded into the invasion chambers (Fig. 4). Function blocking antibody and specific peptides that block SDF-1 binding to CXCR4 significantly reduced invasion of the PC3 cells mediated by tissue extracts (Fig. 4). As SDF-1 is the only known ligand for CXCR4, these data show that SDF-1 in tissue extracts is biologically active.
Inhibition of SDF-1 binding by antibodies to CXCR4
Previous reports suggest that metastasis is frequently found at sites of high bone turnover, including the region near the growth plate. Our findings that SDF-1 levels were elevated in the growth plate region and that it is bioactive suggest that sites of elevated SDF-1 production may be particularly prone to CaP metastasis. To test this hypothesis, agents that blocked the binding of SDF-1 to CXCR4 on CaP cells are desired. To validate the employment of several commercial reagents for use in our investigations, we first evaluated whether anti-CXCR4 or anti-SDF-1 antibodies would block the binding of biotinylated SDF-1 to PC3 prostate cancer cells. For these investigations PC3 cultures were established in 96-well plates, where unlabeled SDF-1 (0-100 μg), antibody to SDF-1 (polyclonal and monoclonal), anti-CXCR4 antibodies (clone 12G4 or clone 44708.111), antibody IgG controls (0-100 μg/ml), or an anti-CXCR4 peptide designed to block SDF-1 (TC14012) binding to the CXCR4 receptor were added to the wells. Subsequently biotinylated SDF-1 was added to the wells, and binding was allowed to proceed for 1 h at 4°C. Thereafter, the cells were washed, and streptavidin HRP and OPD were added for detection.
The data showed that either unlabeled SDF-1, which competes for binding of the CXCR4 receptor with biotinylated SDF-1, antibody to CXCR4, which binds to the ligand-binding domain of CXCR4, neutralizing polyclonal or monoclonal antibody to SDF-1, or the peptide TC14012 reduced binding of labeled SDF-1 to PC3 cells (Fig. 5). These data confirmed that these reagents were appropriate for further investigation.
In vivo metastasis investigations
To test our hypothesis that the SDF-1/CXCR4 pathway is critical for the development of bone metastases, in vivo we established two experimental groups (Fig. 6A). In each case, PC3 cells tagged with luciferase were preincubated with antibody to CXCR4 or an isotype-matched nonspecific antibody control group on ice immediately before intracardiac injection. The antibody that was chosen to block CXCR4 (MBA171 monoclonal antibody; R&D Systems) was used at 5 μg/1 × 104 injected cells. The animals were subsequently injected intraperitoneally twice at 2 and 24 h. After 4 weeks, bioluminescence imaging to follow the CaP cell-derived metastases was used as our primary outcome.(24) As shown in Fig. 6A, we observed that all of the animals that received the IgG control antibody established osseous metastases. Antibody to CXCR4 significantly reduced the total metastatic load of the animals compared with IgG control, regardless of whether the animal was imaged from the dorsal or ventral surface (Fig. 6C). Examination of individual sites of bone metastasis also showed that administration of antibody to CXCR4 significantly reduced the total luminescent signal (i.e., total tumor burden), further showing that the SDF-1/CXCR4 chemokine axis in part regulates osseous metastasis, including femur/tibia regions, spine, and maxilla/mandible areas (Fig. 6D). Radiographic analysis confirmed these findings, suggesting that the antibody to CXCR4 had a significant effect on reducing osseous metastases.
Because antibody to CXCR4 decreased the metastatic load, we tested if this also affected tumor-induced bone resorption. For these evaluations, both serum calcium and serum pyridinoline cross-links levels were evaluated as markers of bone turnover. There was no difference in total serum calcium concentration in the experimental group injected with the IgG antibody compared with those animals receiving the CXCR4 antibody (Fig. 7A). The PYD levels were lower in the CXCR4 antibody group compared with those of the control group, indicating that there was an overall decrease in osteoclastic activity in the CXCR4 experimental groups (Fig. 7A). In animals with bone lesions, however, there were no statistical differences in the tumor size/bone area in anti-CXCR4 antibody treated versus IgG control-treated animals (41.8 ± 9.3 versus 45.5 ± 5.6 mm), in number of osteoclasts/mm of tumor adjacent to the trabecular or cortical bone surfaces (1.8 ± 1.0 versus 1.0 ± 0.8 mm), or for the percentage of the growth plate that was associated with tumor (38.7 ± 13.1% versus 57.69 ± 19.3%; Figs. 7B and 7C). We interpret these data to suggest that, whereas the tibia/femur markers were not significant on a per animal basis with many metastatic lesions in the IgG control group, the systemic PYD levels were elevated, reflecting the total metastatic burden. These data suggest that if CaP cells localize to the skeleton, the early treatment to block CXCR4 had no influence on tumor growth under these conditions.
Growth of intraosseous tumors in vivo
The observation that antibody to CXCR4 blocked metastasis induced by PC3 and our prior observation that antibody to SDF-1 decreased PC3 proliferation in vitro(5) prompted us to determine what effect blocking CXCR4 would have on the growth of prostate cancers in bone. Accordingly, PC3 cells were injected intratibially into nude mice. One day later, anti-CXCR4 antibody, anti-CXCR4 peptide (TC14012), or control antibody was administered daily for 4 weeks intraperitoneally. Skeletal lesions identified by bioluminescence imaging were used as our primary outcome. Tumors were identified in all of the control groups. The majority of the animals treated with either the anti-CXCR4 antibody or TC14012 also had tumors; however, the tumors were significantly smaller (Fig. 8A). Histological analyses revealed that tumor infiltration in all the control-treated animals was greater than the anti-CXCR4 antibody or TC14012-treated mice (Fig. 8B). The levels of pyridinoline cross-links were lower in the CXCR4 antibody and TC14012 peptide groups compared with those of the control group, indicating that there was an overall decrease in osteoclastic activity; however, serum calcium levels were not altered (Fig. 8C). Radiographic analysis of tumor burden of the two groups also showed smaller tumors in the treatment groups versus control groups (data not shown). Similarly, histologic analysis of osteoclast number, percent tumor at the growth plate, and tumor area revealed a reduction in all parameters for the anti-CXCR4 antibody and specific peptide-treated groups relative to the control-treated animals (Fig. 8D). Taken together, these data show that blockage of the SDF-1 receptor, CXCR4, inhibits the development of PC3-derived tumors. Furthermore, the data strongly suggest that the development of bone metastases is dependent on SDF-1 for growth.
In this study, we examined the hypothesis that SDF-1/CXCR4 plays a significant role in the skeletal metastasis of CaP. In a first set of investigations, SDF-1 levels in several murine tissues were examined. High levels of SDF-1 were noted in many tissues to which CaP frequently metastasize. In bone, SDF-1 expression was observed near the endosteal surfaces covered by osteoblasts and lining cells in both human and murine tissues. Importantly, SDF-1 message was localized to the metaphysis of the long bones nearest to the growth plate. SDF-1 message and protein levels were generally low in the center of the marrow and in the diaphyseal regions. These findings correlate with the sites of osseous metastasis observed in our in vivo intracardiac model metastasis model. Moreover, similar results have been presented by other investigators and other tumors using these antibodies, suggesting their appropriateness for use in investigations of the SDF-1/CXCR4 axis in metastasis. Peled et al.(14) showed that the engraftment of human hematopoietic CD34+ cells into the bone marrow of NOD/SCID animals was could be blocked by CXCR4 neutralization. Likewise, Muller et al.(7) showed that antibody to CXCR4 or SDF-1 blocked metastasis of breast cancer cell lines in vivo, and neutralization of CXCR4 prevented homing of human NHL cells and lung cancer cell lines to a number of organs.(27, 28) Further support for a participation of SDF-1/CXCR4 in osseous metastasis was provided using antibody to CXCR4 immediately before and at 2–24 h after tumor inoculation. Overall, less tumor was observed and fewer osseous lesions were identified by Xenogen imaging, radiographic, and histologic examinations. These observations were further substantiated in that pyridinoline cross-links in serum were significantly lower in the anti-CXCR4 antibody-treated group relative to control IgG-treated groups. It is therefore interesting to speculate why SDF-1 levels are elevated in the cardiac muscle and the optic nerve area. These are not frequently sites of CaP metastasis in humans; however, CaP cell metastasis in the intracardiac model often are localized to these sites,(20, 29) but neutralization of CXCR4 decreased metastasis. Clearly, additional events must participate in localizing metastatic tumors to bone in addition to the SDF-1/CXCR4 axis in humans.
To evaluate what role SDF-1 has on intraosseous growth of CaPs, intratibial injection of PC3 cells was performed. Antibody or specific peptide against CXCR4 decreased the growth of pre-established CaP cell lesions in bone. These data correlated with our previous finding that showed that human CaPs express elevated levels of SDF-1 mRNA relative to benign hyperplasia in situ and that CaP cell lines that secrete biologically active SDF-1 protein regulate their expression of SDF-1 mRNA in response to exogenous SDF-1.(5) Correlating with the in vivo data, neutralizing antibody to SDF-1 decreased the proliferation metastatic tumor cells that were isolated from bone, but not those isolated from other tissues.(5) These findings, together with those from our metastasis study, suggest that neutralization of the SDF-1/CXCR4 axis may prevent osseous metastasis by both inhibiting metastatic spread within the vasculature, and once in bone, preventing tumor growth.
We recently reported that SDF-1 activates binding of prostate cancer cells to marrow endothelial cells.(4) We determined that part of the mechanism that facilitates CaP cell binding to endothelium in response to SDF-1 may be through activation of αvβ3 integrins already expressed on CaP cancer cell lines (Y-X Sun and RS Taichman, unpublished data, 2004). Clearly, binding to endothelium is a major event regulating the establishment of metastasis of blood-borne tumors. Recently, it has been reported that SDF-1 is bound to bone marrow endothelial cells by endothelial heparan sulfate and chondroitin/dermatan proteoglycans, where it would activate adhesion receptors on the tumor cells to facilitate transendothelial migration.(30–32) For hematopoietic cells, it is well established that efficient rolling on endothelial cells requires P-selectin, E-selectin, and the CD44 under physiologic flow conditions. On the other hand, intercellular adhesion molecule-1 (ICAM-1) alone does not seem to facilitate binding of hematopoietic cells under flow conditions. In the presence of surface-bound SDF-1, however, rolling on endothelial cells is rapidly developed into a firm adhesion, mediated by interactions between ICAM-1 and SDF-1.(33) CXCL12/SDF-1 may also regulate the development of pseudopodia in breast cancers that may lead to invasive metastasis.(7) These findings are important in that blockade of CXCR4 signaling may result in decreased metastases. Part of the mechanism whereby CXCR4 inhibition may function is to limit extravasation of circulating tumors by limiting binding or by yet another unidentified mechanism. For example, Bertolini et al.(27) enumerated the number of circulating tumor neutralization in non-Hodgkin's lymphoma (NHL) cells in the peripheral blood of mice injected with an NHL cell line intravenously after CXCR4 neutralization. One day after tumor injection, circulating tumor cells were undetectable. For mice injected with CXCR4-neutralized NHL cells, the frequency of circulating tumor cells was >0.5%. These observations, along with data that show that antibody to CXCR4 impairs transendothelial/stromal cell migration, suggest a pivotal role of the CXCR4-CXCL12/SDF-1 in tumor extra- and intravasation.(27) Whether SDF-1 serves a similar role in CaPs is not known, but seems likely.
The autocrine production of SDF-1 resulting in proliferation is an exciting aspect of the biology of the SDF-1/CXCR4 chemokine axis in CaP cells,(5) more so in that CaPs use SDF-1 to localize to end-organs that supply high levels of ligand (e.g., bone, lung, liver, lymph node(7)). Unfortunately, we have not been able to consistently show SDF-1 at the protein level using tissue microarrays, although weak nuclear staining was observed in localized cancers (data not shown). Recently, Zhou et al.(34) showed that overexpression of CXCR4 in glioblastoma cell lines enhanced their growth in soft agar, and expression of antisense CXCR-4 in glioblastoma cell lines caused neurite outgrowth and cellular differentiation. Moreover, treatment of the glioblastoma cell lines with antibody to CXCR4 or SDF-1 inhibited proliferation, suggesting that the CXCR-4 gene is required to prevent apoptosis after serum withdrawal.(34,35) Similarly systemic administration of CXCR4 antagonist AMD 3100 inhibits growth of intracranial glioblastoma and medulloblastoma xenografts by increasing apoptosis and decreasing the proliferation of tumor cells.(35) Moreover, in some instances, CXCR4 may not be responsible for invasion but rather be critical for the outgrowth of micrometastases.(36) However, not all tumors expressing CXCR4 proliferate in response to SDF-1, including rhabdomyosarcoma cells.(37) Likewise, chemokines may stimulate proliferation of tumor cells directly, as was shown with melanoma growth stimulatory activity (MGSA), causing the growth of melanoma and pancreatic cell lines; interleukin (IL)-8 stimulates growth of non-small-cell lung and prostate carcinomas.(28,38–40) Several tumor-derived chemokines may protect neoplasms from immune attack of tumor-infiltrating lymphocytes including CCR5 ligands (CCL5 [RANTES], CCL3 [MIP-1α], CCL4 [MIP-1β]) and SDF-1 (CXCL12).(41) Conversely, the release of chemotactic chemokines has been shown to recruit leukocyte infiltrates that may limit tumor growth.(38,40,41–44)
A further possibility is that SDF-1 secretion facilitates retention of the CaP cells in the marrow. This mechanism may be envisioned through the continued activation of integrin molecules, or autocrine produced SDF-1 may desensitize CaP cells to further CXCR4 signaling. Recently, it was reported from several groups that the uncoupling of G-coupled receptors including CXCR4 after receptor internalization by endocytosis may persist even after the receptor is recycled to the cell surface.(45,46) This would result in receptor ligand interactions that fail to induce downstream signals such as calcium fluxes and/or MAP kinase signaling or cell migration. Additionally, the production of SDF-1 by CaP cells may help to establish a “repellant migratory” system to localize the tumor cells between sources of SDF-1 in the marrow (endothelium, osteoblasts). A similar system was shown in T-cells, when different concentrations of SDF-1 were immobilized in vitro, and time-lapse video microscopy was used to show a bidirectional effect.(46) T-cells were found to move away from the high-concentration sites of SDF-1 and toward those of low SDF-1 concentrations.(46) Moreover, antigen-induced T-cell recruitment into the peritoneal cavity was reversed by high but not low concentrations of SDF-1. Whether such findings represent mechanisms to explain why high SDF-1-producing organs such as the thymus and bone marrow are not chronically infiltrated with T-cells is not known.
Our hypothesis was among the very first to suggest a role for SDF-1/CXCR4 in prostate tumor metastasis. Recently Muller et al.(7) and Kang et al.(6) reported that CXCR4 and SDF-1 are central players in regulating metastasis by showing that normal breast tissues express little CXCR4, whereas breast neoplasms express high levels of CXCR4, and antibody to CXCR4 blocked the metastatic spread of the tumors to the lung and lymph nodes. Similarly, CXCR4 mRNA levels are elevated in glioblastoma multiforme in regions of angiogenesis and degeneration but decreased in areas of rapid cell proliferation and may be a general characteristic of neuroblastoma cells that supports their preferential metastasis into the bone marrow.(8) Results consistent with these have also been reported for human melanoma cell lines, melanoma cells that had macroscopically infiltrated draining lymph nodes,(9) and pancreatic as well as renal carcinomas.(1147) These results are also consistent with those of Zeelenberg et al.,(48) who transfected SDF-1 fused to a KDEL sequence into mouse T-cell hybridoma TAM2D2. The SDF-KDEL fusion protein was retained in the endoplasmic reticulum by the KDEL receptor and bound to CXCR4, which was therefore also retained, preventing the metastasis of the cell line to many different tissues.(48) However, little is known about the role of SDF-1/CXCR4 in CaP. These in vivo metastasis data provide critical support for a role of SDF-1/CXCR4 in skeletal metastasis. Most importantly, these data show that SDF-1/CXCR4 participate in localizing tumors to the bone marrow for CaP.
Expertise on the Xenogen IVIS imaging system was provided by Dan Hall of the University of Michigan's Small Animal Imaging Resource Core (http://www.med.umich.edu/msair/). These investigations were supported in part by the Tissue Procurement Core of the University of Michigan Comprehensive Cancer Center and National Institutes of Health Grants DK067688 (RST), R01 DE13701 (RST), R01 AR46024 (RST), P01 CA46952 (KJP, ETK, LKM, RST), P50 CA69568 (KJP), CA93900 (ETK, LKM, RST), and the Department of Defense DAMD17-02-1-0100 (RST).