The Role of CXCR7/RDC1 as a Chemokine Receptor for CXCL12/SDF-1 in Prostate Cancer*

Several reports have recently documented that CXCR7/RDC1 functions as a chemokine receptor for SDF-1/CXCL12, which regulates a spectrum of normal and pathological processes. In this study, the role of CXCR7/RDC1 in prostate cancer (PCa) was explored. Staining of high density tissue microarrays demonstrates that the levels of CXCR7/RDC1 expression increase as the tumors become more aggressive. In vitro and in vivo studies with PCa cell lines suggest that alterations in CXCR7/RDC1 expression are associated with enhanced adhesive and invasive activities in addition to a survival advantage. In addition, it was observed that CXCR7/RDC1 levels are regulated by CXCR4. Among the potential downstream targets of CXCR7/RDC1 are CD44 and cadherin-11, which are likely to contribute to the invasiveness of PCa cells. CXCR7/RDC1 also regulates the expression of the proangiogenic factors interleukin-8 or vascular endothelial growth factor, which are likely to participate in the regulation of tumor angiogenesis. Finally, we found that signaling by CXCR7/RDC1 activates AKT pathways. Together, these data demonstrate a role for CXCR7/RDC1 in PCa metastasis and progression and suggest potential targets for therapeutic intervention.

The predominant CXCL12 receptor is CXCR4, a member of the cell surface G protein-coupled seven-span transmembrane receptors. CXCR4 has received considerable notoriety because it serves as a co-receptor for entry of T-tropic human immunodeficiency viruses into CD4 ϩ T cells (18,19). During development, CXCR4 is expressed in a broad range of tissues, including brain, lymph node, and small intestine (20), functioning in a number of normal processes (21,22). Gene knock-out approaches in mice have demonstrated that CXCL12 and CXCR4 play an important role in embryologic development (23). CXCL12-deficient mice die in utero with severe cardiac septum defects and poorly developed marrow. CXCR4 receptor deletion is similarly lethal, resulting from circulatory, central nervous system, immune, and hematopoietic defects (7). The phenotype of the CXCL12 Ϫ/Ϫ mouse closely follows those of CXCR4 Ϫ/Ϫ animals, further suggesting that the CXCL12 and CXCR4 are a receptor-ligand pair.
The metastatic process is very similar to the "homing" behavior of hematopoietic stem cells to the bone marrow. Our group was first to demonstrate that PCa uses CXCL12 and CXCR4 as key elements in metastasis and growth in bone (13,24). Moreover, we have demonstrated that CXCL12 and CXCR4 are key elements in the growth of metastatic PCa (13,25). Yet inhibition of CXCR4 in vivo only partially blocks the metastatic behavior of PCa. This suggests that other factors control the tissue-specific migration of epithelial cancer cells.
Recently, CXCL12 was shown to bind with high affinity to the orphan receptor CXCR7/RDC1. CXCR7/RDC1 was originally cloned on the basis of its homology with conserved domains of G protein-coupled receptors (26,49). Combined phylogenetic and chromosomal location studies suggest that CXCR7/RDC1 represents a chemokine receptor structurally close to CXC receptors (27), pointing to CXC chemokines as potential ligands. CXCR7/RDC1 shares homology with the viral gene ORF74 encoding a chemokine receptor, which suggests that it may signal constitutively in the absence of ligand (28). Like CXCR4, CXCR7/RDC1 serves as a co-receptor for some human and simian immunodeficiency virus strains (29). In the vasculature, the expression of CXCR7/RDC1 is elevated in endothelial cells associated with tumors, and overexpression of CXCR7/RDC1 in NIH 3T3 cells strongly supports a role for the receptor in tumorigenesis (29). More recently, Miao et al. (30) further confirmed a critical role for CXCR7/RDC1 in tumor vascular formation, angiogenesis, and promotion of the growth of breast and lung cancer in vivo.
In this study, the role of CXCR7/RDC1 in PCa metastasis was determined. The results suggest that alterations of CXCR7/ RDC1 expression are associated with survival and adhesive and invasive activities of PCa cells. Determination of the role of this newly identified receptor in tumor progression may point to potentially new therapeutic avenues for the treatment of metastatic PCa.

EXPERIMENTAL PROCEDURES
Cell Culture-The PC-3 cell line was originally isolated from a vertebral metastasis, and the LNCaP and the metastatic subline LNCaP C4-2B (C4-2B) cells were originally isolated from a lymph node of a PCa patient with disseminated bone and lymph node involvement. PCa cell lines were passaged and grown to confluence over 5 days. For the production of PCa cell conditioned medium (CM), the cells were replated at 2.0 ϫ 10 4 cells/ cm 2 into 24 well-tissue culture plates and incubated in growth medium (RPMI 1640 medium with 10% fetal calf serum and 1% antibiotics) (Invitrogen). After confluence, the cells were washed with PBS, and the growth medium was replaced and incubated for an additional 96 h. The CM were collected and frozen after passage through a 0.22-m filter. Human dermal microvascular endothelial cells (HDMECs) were grown as described previously (25,31).
Western Blot Analysis-PCa cells were cultured to confluence, washed, and then serum-starved in RPMI 1640 medium with 0.1% bovine serum albumin for 24 h. Stimulation of the cells was performed with 200 ng/ml CXCL12 (R & D Systems, Minneapolis, MN). At selected time points, the cells were lysed in ice-cold RIPA buffer. Protein concentrations were deter-mined from cell lysates clarified by centrifugation at 14,000 rpm for 10 min (Bio-Rad). Normalized lysates (30 g) were resuspended in loading buffer and were electrophoresed on 10% polyacrylamide gels under reducing conditions and transferred to polyvinylidene difluoride membranes. For detection of CD44 and CDH11 (Santa Cruz Biotechnology, Santa Cruz, CA), the membranes were either blocked in 3% bovine serum albumin in PBS, 0.1% Tween 20 and a rabbit anti-human monoclonal antibody (1 g/ml) (1:1000, Abcam Inc., Cambridge, MA) or a mouse anti-human monoclonal antibody (1 g/ml) (1:1000, Sigma) and were used in conjunction with anti-species conjugated horseradish peroxidase (1:1000, Upstate Biotechnology, Inc., Lake Placid, NY) and detected by chemiluminescence (Amersham Biosciences). AKT detection was similarly performed in 5% dry milk in PBS, 0.1% Tween 20 with a rabbit monoclonal reactive to dual phospho-AKT (Ser-473) and total AKT (Cell Signaling Technology, Beverly, MA).
Tissue Microarray Development, Immunohistochemistry, Digital Image Capture, and Analysis-High density tissue microarrays were constructed from clinical samples obtained from a cohort of over 120 patients, who underwent radical retro pubic prostatectomy at the University of Michigan as a primary therapy (i.e. no previous hormonal or radiation therapy) for PCa from 1994 to 1998. These samples were provided by the University of Michigan Comprehensive Cancer Center Histology and Immunoperoxidase Core as detailed previously (24). Tumors were graded using the Gleason grading system and examined to identify areas of benign prostate, prostatic intraepithelial neoplasia (PIN), localized prostate cancer, and bone metastasis. The formalin-fixed, paraffin-embedded tissues were dewaxed and placed in a pressure cooker containing 0.01 M buffered sodium citrate solution (pH 6.0), boiled, and then chilled to room temperature for antigen retrieval. The slides were then incubated overnight at room temperature with antihuman CXCR7/RDC1 antibody diluted 1:100 (Abcam Inc.). A standard streptavidin/biotin detection method with 3,3-diaminobenzidine tetrahydrochloride was employed for signal detection, and Harris hematoxylin was used as a counter-stain. CXCR7/RDC1 expression was blindly scored by a genitourinary pathologist as negative (1), weak (2), moderate (3), or strong (4) on the basis of staining intensity and the percentage of stained tumors cells using a telepathology system without knowledge of overall Gleason score (e.g. tumor grade), tumor size, or clinical outcome. Breast cancer arrays were purchased from Folio Biosciences (Columbus, OH) (ARY-HH0058), which were generated from clinical samples derived from a cohort of 80 patients. In this case an alkaline phosphatase chromogen substrate (ABC-AP substrate kit; Vector Laboratories, Burlingame, CA) was used for signal detection.
Cytokines ELISA Analysis-Antibody sandwich ELISAs were used to evaluate IL-8 and VEGF levels in the PCa cell CM (R & D Systems) as described previously (25,31).
Cell Invasion-Cell invasion was examined using a reconstituted extracellular matrix membrane (BD Biosciences). Test cells were placed in the upper chamber (1 ϫ 10 5 cells/well) in serum-free medium, and 200 ng/ml CXCL12 was added to the bottom chambers. Invasion into the matrix after 48 h was determined by removal of the invasion chambers, and 40 l of MTT FIGURE 1. CXCR7/RDC1 expression is correlated with human PCa development. A, expression of CXCR7/RDC1 in human prostate cancers. Representative elements of a tissue microarray stained with an anti-CXCR7/RDC1 antibody demonstrate weak staining in benign and PIN (panels a and b) but strong in staining in PCa (panels c and d) and metastatic tissue lesions (panels e-i). Original magnification ϫ20, where the black bars represent 100 m. B, quantitative histologic evaluation of CXCR7/RDC1 expression of PCa progression. CXCR7/RDC1 expression intensity was scored on a four point scale as either negative (1), weak (2), moderate (3), or strong (4). Mean expression scores multiplied by percent positive cells in the field (Quick's combined score system) are presented for benign, PIN, localized PCa, and metastatic PCa (PCa Met) cases in a graphic format using error bars with 95% confidence intervals (CI). Statistically significant differences were noted between benign hyperplasia and localized cancer or metastatic cancer (PCa Met). *, p Ͻ 0.001 and between PCa and metastatic PCa; #, p Ͻ 0.001 (ANOVA). C, FACS analysis of CXCR7/RDC1 expression on PCa cell lines. CXCR7/RDC1 protein expression was evaluated by FACS analysis using rabbit anti-human CXCR7/RDC1 antibody, and rabbit anti-human IgG served as isotype control. Cells expressing CXCR7/RDC1 are labeled as CXCR7. FEBRUARY 15, 2008 • VOLUME 283 • NUMBER 7 FIGURE 2. CXCR7/RDC1 expression in PCa cell regulates proliferation, apoptosis, and mediates cell adhesion and invasion. A, FACS analysis of CXCR7/ RDC1 expression in transfected PCa cell lines. A rabbit anti-human CXCR7/RDC1 antibody was used to detect the expression of CXCR7/RDC1 in PC3-and C4-2B-transfected cells. For overexpression studies, CXCR7/RDC1 was overexpressed (OE) using a pcDNA3.1 vector (labeled as PC3 CXCR7 /PC3 OE Control or C4-2B CXCR7 /C4-2B OE Control ). Inhibition of CXCR7/RDC1 was achieved using siRNA to target expression of the receptor. A scrambled CXCR7/RDC1 sequence was incorporated into the siRNA as a vector control. The cells are denoted as PC3 siCXCR7 /PC3 siControl or C4-2B siCXCR7 /C4-2B siControl . Rabbit anti-human IgG served as isotype control. The rabbit primary antibodies were detected by incubating the cells at 25°C for 45 min with FITC-conjugated goat (FabЈ) 2 anti-rabbit IgG. The cells were washed twice with phosphate-buffered saline, resuspended, and fixed in 1% (w/v) paraformaldehyde for analysis. B, proliferation of PCa cell lines in response to altered CXCR7/RDC1 levels. After a 24-h serum withdrawal, PCa cells were digested and washed three times in PBS, and 1 ϫ 10 4 cells were plated in into 96-well flat-bottomed tissue culture plates in 0.1 ml in growth medium. Proliferation was evaluated by XTT assay over a 5-day period. * denotes significant difference from controls (p Ͻ 0.05, ANOVA) for means Ϯ S.E. of n ϭ 5 samples per condition. C, changing apoptosis profile of PCa cell lines in response to altered CXCR7/RDC1 levels. The role of CXCR7/RDC1 in preventing apoptosis in PCa cell lines was evaluated culturing equal numbers of cells grown in complete medium for 24 h. Annexin-V-fluorescein staining and propidium iodide were used to identify apoptotic and necrotic cells by FACS analysis. D, CXCR7/RDC1 prevents apoptosis in PCa cells. Levels of cytoplasmic histone-associated DNA associated with apoptosis were detected by cell death detection ELISA kit. 2 ϫ 10 4 cells were counted to obtain equivalent numbers of cells per assay condition and then treated as described under "Experimental Procedures." * denotes significant difference from controls (p Ͻ 0.05, ANOVA) for means Ϯ S.E. for n ϭ 3. E, CXCR7/RDC1 regulates adhesion of PCa cells to HBMECs and CXCR7/RDC1 in Prostate Cancer (5 mg/ml,Sigma) was added to the top well and 80 l of MTT to the bottom well and further incubated for 4 h at 37°C. After complete removal of the residual cells or medium, the purple residues attached to the bottom or top chambers were released with 1 ml of isopropyl alcohol (Sigma). The invasion chambers were rocked for 30 min at a medium speed, and then 100 l from each well was transferred into 96 wells and read on a multiwell scanning spectrophotometer (Molecular Devices Corp., Sunnyvale, CA) at A 450 .
Flow Cytometry Analysis-For surface chemokine receptor detection, 2 ϫ 10 5 cells were incubated at 2-8°C for 40 min with nonspecific isotype-matched controls, mouse to human IgG (BD Biosciences), and rabbit polyclonal to human IgG (Abcam Inc.) (ab2410), and 20 g/ml of each of the following murine monoclonal antibodies: anti-human CXCR4 (R & D Systems) and rabbit monoclonal antibodies: anti-human CXCR7/RDC1(Abcam Inc.) (ab12870). The mouse or rabbit primary antibodies were then detected by incubating the cells at 25°C for 45 min with fluorescein isothiocyanate (FITC)-conjugated goat (FabЈ)2 anti-mouse or rabbit IgG (Upstate). The cells were washed twice with phosphate-buffered saline, resuspended, and fixed in 1% (w/v) paraformaldehyde for analysis. Ten thousands cells from each sample were evaluated for fluorescence detection using FACScan (BD Biosciences), and the data were analyzed with CellQuest software (BD Biosciences).
Apoptosis Assays-PCa cells were normally cultured with growth medium. After 24 h, annexin-V staining and a cell death ELISA kit were used to detect apoptosis. Annexin-V staining was performed by incubating 1 ϫ 10 6 cells in 24-well plates resuspended in 50 l of binding buffer (10 mmol/liter HEPES/ NaOH (pH 7.4), 140 mmol/liter NaCl, and 2.5 mmol/liter CaCl 2 buffer) (Roche Applied Science). Annexin-V/FITC (Roche Applied Science) and PI (Roche Applied Science) were added to the cell suspension for 15 min at room temperature in the dark. The samples were then washed once in PBS, and flow cytometric analysis was performed immediately thereafter. Propidium iodine staining was also performed using cells suspended in 1 ml of PBS, and 3 ml of absolute ethanol was added during vortexing. The cells were fixed with 4% paraformaldehyde for at least 1 h at 4°C. Following fixation cells were washed in PBS and resuspended in 1 ml of staining buffer (50 g/ml PI, 0.5 g/ml RNase A, PBS). Samples were incubated at 4°C for 2 h, washed once in PBS, resuspended in PBS, and analyzed by flow cytometry. Ten thousand cells were analyzed for annexin-V/PI staining, and 5,000 cells were analyzed for PI staining by fluorescence detection using FACScan (BD Biosciences), and the data were analyzed with Cell-Quest software (BD Biosciences).
A cell death ELISA kit (Roche Applied Science) was used to detect cytoplasmic histone-associated DNA fragments. The assay was performed by the manufacturer's directions with the following modifications. Both adherent and nonadherent cells were collected from serum-starved cultures, and the resulting cell counts were normalized. 2 ϫ 10 4 cells per condition were lysed, and 20 l of lysate were utilized for each reaction and evaluated at A 405 using a plate reader from Molecular Devices Corp equipped with Softmax software (Sunnydale, CA).
Endothelial Sprout Formation Assay-Growth factor-reduced basement membranes were placed into 4-chamber slides (Matrigel TM 125 l/chamber; BD Biosciences) and 0.8 ϫ 10 4 endothelial cells were added on top. The chambers were incubated at 37°C for 24 h. After incubation, the slides were fixed with methanol and stained with Diff-Quick solution II (Sigma). The slides were examined, and the sprouts were counted from five random fields under a microscope (ϫ200). For co-culture assays, equal numbers of PCa cells and endothelial cells were plated together, or 500 l of the PCa CM or control was added daily.
Adhesion Assays-HDMECs were cultured to confluence in microvascular endothelial cell growth medium (EGM-MV; Clonetics, Walkersville, MD). PCa cells were labeled with 2.5 g/ml of the lipophilic dye carboxyfluorescein diacetate succinimidyl ester (Molecular Probes, Eugene, OR) for 30 min at 37°C and washed in PBS. Thereafter, the PCa cells were rested for 30 min to reduce nonspecific background, and subsequently the cells were resuspended in PBS to deliver 10 5 cells/well in the adhesion assays. The adhesion assays were performed in PBS containing Ca 2ϩ /Mg 2ϩ where the PCa cells were added to a final reaction volume of 100 l. The plates were then spun at 500 rpm for 5 min at 4°C and allowed to incubate for an additional 10 min at 4°C. The nonadherent cells were removed in three subsequent washes, and the remaining fluorescence was quantified in a 96-well fluorescent plate reader (Molecular Devices Corp.).
RT 2 Profiler TM PCR Array-Total RNA was extracted from stable transfectants of PC3 cells overexpressing CXCR7/RDC1, C4-2B cells with CXCR7/RDC1 knocked down, and their respective controls. For PCR array experiments, an RT 2 Profiler custom PCR array was used to simultaneously examine the mRNA levels of 89 genes closely associated with tumor metastasis, including five "housekeeping genes" in 96-well plates following the manufacturer's protocol (catalog number APH-028X, SuperArray Bioscience, Frederick, MD). Briefly, first-strand cDNAs were synthesized from 1 g of total RNA using TaqMan RT reagent kit (Applied Biosystems, Inc., Foster invasion. Top, adhesion of labeled PCa cells to HBMECs monolayers was evaluated after a 15-min co-culture. The data are presented as % binding Ϯ S.D. for n ϭ 5 replicates. * denotes significant difference from controls; Krusal-Wallis test and Dunn's multiple comparisons tests were utilized with the level of significance set at p Ͻ 0.05. Bottom, PCa cell lines were placed in the top chamber of cell invasion plates containing a reconstituted extracellular matrix in serum-free RPMI medium, and CXCL12 (200 ng/ml) was added to the lower chambers. Invasion was determined at 48 h by MTT staining, and the data are presented as % invasion binding Ϯ S.D. for n ϭ 5. * denotes significant difference from nontreated controls; #, significant difference from invasion mediated by CXCL12 (200 ng/ml) (p Ͻ 0.05, ANOVA). FEBRUARY 15, 2008 • VOLUME 283 • NUMBER 7

JOURNAL OF BIOLOGICAL CHEMISTRY 4287
City, CA) according to the manufacturer's protocol. The reaction mixtures (50 l) were incubated at 25°C for 10 min, followed by incubation at 48°C for 30 min and 95°C for 5 min, and then cooled on ice. Arrays were performed independently at least twice for each cell line; values were obtained for the threshold cycle (C t ) for each gene and normalized using the average of four housekeeping genes on the same array (HPRT1, RPL13A, GAPD, and ACTB). C t values for housekeeping genes and a dilution series of ACTB were monitored for consistency between the arrays. Change (⌬C t ) between CXCR7/RDC1-siRNA, CXCR7/RDC1 overexpression vector, and control siRNA control vector were found by the following: ⌬C t ϭ C t(CXCR7/RDC1siRNA) Ϫ C t(Scramble siRNA) ; ⌬C t ϭ C t(RDC1) Ϫ C t(control) , and fold change by fold change ϭ 2 (Ϫ⌬Ct) . The resulting values were reported as fold change; only genes showing 2-fold or greater change were considered. The negative controls ensured a lack of DNA contamination and set the threshold for the absent/present calls.
Subcutaneous Injection in Vivo Assay of CXCR7/RDC1 Effect of Tumor Development-All experimental animal procedures were performed in compliance with the institutional ethical requirements and approved by the University of Michigan Committee for the Use and Care of Animals. To evaluate tumor growth, subcutaneous tumors were established from the various transfectants by resuspending 2 ϫ 10 6 PCa cells in growth factor-reduced Matrigel. 5-7-Week-old male SCID mice (CB.17.SCID; Taconic, Germantown, NY) were anesthetized with isofluorane inhalation. After shaving and cleaning the skin, subcutaneous injections using a 27-gauge needle was used to establish the tumors. The animals were monitored daily, and tumor volumes were evaluated every 6 days starting at day 10. Tumor volumes were calculated using the formula V ϭ (the shortest diameter) 2 ϫ (the longest diameter). After 36 days the animals were sacrificed; the tumors were weighed, measured, and prepared for histology.
Immunohistochemistry-For immunostaining with von Willebrand factor (factor VIII-related antigen), tissue sections were blocked with Sniper for 5 min and incubated overnight at 4°C with 28 mg/ml rabbit anti-human von Willebrand factor antibody (Dako North America Inc., Carpinteria, CA) diluted in PBS. The sections were incubated with appropriate secondary antibodies for 30 min, followed by processing with a Lincoln Label 41 detection system (Biocare Medical, Concord, CA). The HRP-AEC chromogen system (R & D Systems) or a solution of HSS-HRP and AEC chromogen in chromogen buffer were used to visualize bound antibodies. The numbers of stained microvessels were blindly counted in 10 random fields per implant at ϫ200 magnifications. Four or five implants were analyzed per condition and for each time point. For immunostaining with AKT-2 and phosphorylated AKT (Ser(P)-473) (Cell Signaling Technology), the tissue sections were boiled in 10 mM sodium citrate buffer (pH 6.0) and maintained at a subboiling temperature for 10 min and cooled for 30 min. An HRP-AEC chromogen system (R & D Systems) and a solution of 3,3-diaminobenzidine tetrahydrochloride buffer were used to the visualize the bound antibodies.
Statistical Analysis-Numerical data are expressed as means Ϯ S.D. Statistical differences between the means for the different groups were evaluated with Instat 4.0 (GraphPAD software, San Diego, CA) using one-way analysis of variance (ANOVA) with the level of significance at p Ͻ 0.05. Where indicated, a Krusal-Wallis test and Dunn's multiple comparisons tests were utilized with the level of significance set at p Ͻ 0.05.

RESULTS
High density tissue microarrays were stained with an antihuman CXCR7/RDC1 antibody to explore whether the receptor plays a role in PCa development. Representative images are shown in Fig. 1A, demonstrating that CXCR7/RDC1 expression increases as the tumor becomes more aggressive, whereas normal epithelium demonstrates weak cytoplasmic staining (Fig. 1A). High grade (Gleason score 4 -5) and metastatic lesions of the bladder, bone, lymph node, liver, and dura also demonstrate strong cytoplasmic staining (Fig. 1A). Quantitative analysis confirmed that CXCR7/RDC1 expression increases with increasing tumor grade (Fig. 1B). Tissues obtained from the Rapid Autopsy Program at the University of Michigan were also evaluated. These examinations resulted in similar findings in which CXCR7/RDC1 was found to be strongly expressed in metastatic PCa lesions derived from lymph nodes, ribs, humerus, dura, and liver (supplemental Fig.  1A) (32). Similarly, tissue microarrays established from breast cancer tissues demonstrated that the expression of CXCR7/ RDC1 is elevated in metastatic cancer compared with normal breast tissues (supplemental Fig. 1B). FACS analysis of CXCR7/ RDC1 also showed significant expression in the PC3, C4-2B, and LNCaP PCa cell lines (Fig. 1C).
Next CXCR7/RDC1 expression was modulated by overexpressing the receptor in PC3 and C4-2B cells (referred to as PC3 CXCR7 , C4-2B CXCR7 ) or by reducing its expression by siRNA (PC3 siCXCR7 , C4-2B siCXCR7 ) ( Fig. 2A). Overexpression of CXCR7/RDC1 increased the basal proliferation rates of the PC3 and C4-2B cell lines, whereas reducing the receptor expression decreased the effects (Fig. 2B). In the presence of CXCL12 (200 ng/ml), the CXCR7/RDC1-transfected PC3 cells showed a substantial increase in cell numbers compared with the control transfected cells (supplemental Fig. 2). Likewise, a significant inhibition of growth in the C4-2B siCXCR7 cells was observed compared with cells transfected with a scrambled CXCR7/RDC1 sequence even in the presence of CXCL12 (200 ng/ml) (supplemental Fig. 2).
One possible explanation of the increase in PCa proliferation observed after transfection with CXCR7/RDC1 is that the receptor may protect the cell lines from apoptosis. Two related assays were utilized to evaluate this possibility. The loss of cellular membrane integrity as a reflection of cells undergoing apoptosis was determined by staining the cells for annexin-V. CXCR7/RDC1 overexpression reduced the apoptotic fraction of cells in culture for both PC3 and C4-2B cells compared with the controls (25.2 versus 35.1% and 41.9 versus 71%) (Fig. 2C). The percent viable cells (double negative) after equal treatment was 34.7 versus 21.3% for the CXCR7/RDC1 overexpressing C4-2B cells. The total number of dead or dying cells (63.6% C4-2B CXCR7 versus 78.39% C4-2B OE Control ) was less as well. Why the overall percentage of necrotic cells (double-positive for PI and annexin-V) increased was not clear. However, the level of cytoplasmic histone-associated DNA levels associated with apoptosis also demonstrated that overexpression of CXCR7/RDC1 protected the cells from apoptosis (Fig. 2D), but inhibition of CXCR7/RDC1 expression below basal levels had no observed effects (supplemental Fig. 3).
To evaluate the role of CXCR7/RDC1 in adhesion, the ability of PCa cells to bind to HDMECs monolayers was determined. Cells overexpressing CXCR7/RDC1 adhered to the HDMECs more vigorously than cells in the control groups (Fig. 2E, top). In contrast, reducing the expression of CXCR7/RDC1 decreased the PCa cell adhesiveness to HDMECs (Fig. 2E, top).
Once tumor cells are bound to the endothelium, they must invade through the extracellular matrix to establish a metastasis. The ability of CXCR7/RDC1 to regulate invasion was studied using reconstituted extracellular matrices in porous culture chambers (Matrigel, Beckman Coulter Labware, Franklin Lakes, NJ). As reported previously (31), CXCL12 supported the invasion of PC3 and C4-2B cells (Fig. 2E, bottom). Overexpression of CXCR7/RDC1 increased the invasive abilities of PC3 or C4-2B cells in the presence CXCL12 (Fig. 2E, bottom). As expected, PC3 or C4-2B cells with reduced expression of CXCR7/RDC1 were less invasive in response to CXCL12 stimulation compared with control cells (Fig. 2E, bottom).
To determine whether expression of CXCR4 directly regulates CXCR7/RDC1, CXCR4 levels were altered by transfection in PCa cell lines. Cells overexpressing CXCR4 expressed less CXCR7/RDC1 protein than controls (Fig. 3A). Conversely, reducing the expression of CXCR4 resulted in enhanced levels of CXCR7/RDC1 by PC3 and C4-2B cells (Fig. 3A). To determine whether CXCR4 and CXCR7/RDC1 reciprocally regulate each other, similar studies were performed in cells with altered CXCR7/RDC1 levels. Surprisingly, alterations of CXCR7/ RDC1 expression did not significantly regulate CXCR4 levels in either cell type (Fig. 3B). Together these findings suggest that signaling through or expression of CXCR4 alters CXCR7/ RDC1 levels, whereas CXCR7/RDC1 expression is not directly linked to CXCR4 expression.
Previously, we and others reported that the changes in CXCR4 levels altered the secretion of IL-6, IL-8, TIMP-2, and VEGF by PCa cells (27,30,31). To determine whether CXCR7/ RDC1 regulates angiogenesis, CM derived from cells overexpressing CXCR7/RDC1 or controls were evaluated for IL-8 and VEGF levels. C4-2B cells overexpressing CXCR7/RDC1 pro-duced more IL-8 and VEGF relative to the parental or control cells (Fig. 4B). CXCL12 stimulation dramatically increased both IL-8 and VEGF levels in C4-2B cells overexpressing CXCR7/ RDC1. It was also observed that by reducing the expression of . CXCR4 expression alters CXCR7/RDC1 levels. A, expression of CXCR4 alters CXCR7/RDC1 levels in PC3 or C4-2B cells. CXCR7/RDC1 expression was determined from FACS analysis in cells expressing altered levels of CXCR4 using an anti-CXCR7/RDC1 antibody or an IgG isotype-matched control. B, alteration of CXCR7/RDC1 expression does not alter CXCR4 levels in PC3 or C4-2B cells. PCa cells with altered expression of CXCR7/RDC1 were evaluated by FACS analysis for the expression of CXCR4 using an anti-CXCR4 antibody and isotype controls. FEBRUARY 15, 2008 • VOLUME 283 • NUMBER 7

JOURNAL OF BIOLOGICAL CHEMISTRY 4289
CXCR7/RDC1 in C4-2B cells, the secreted levels of IL-8 and VEGF were decreased (Fig. 4B). However, it was not clear why the basal levels of IL-8 increased following control transfection of the C4-2B cell line. In PC3 cells, similar IL-8 production in cells overexpressing CXCR7/ RDC1 was enhanced and decreased in cells with reduced CXCR7/RDC1 expression (Fig. 4B). VEGF secretion followed a similar pattern, but the trend failed to reach a level of significance (Fig. 4B).
To explore whether CXCR7/ RDC1 expression is biologically relevant to vascular recruitment by PCa cells, CM derived from the transfected PCa cell lines were evaluated. As shown in Fig. 4C, little or no endothelial sprout formation occurred in the absence of external stimuli applied to endothelial cells alone in vitro. Co-culture of the endothelial cells with either the CM derived from PC3 or C4-2B cells or the PCa cells themselves robustly stimulated vascular sprout formation (Fig. 4, C and D). Overexpression of CXCR7/RDC1 in both cells dramatically increased blood vessel sprout formation, although decreased expression of CXCR7/ RDC1 in either C4-2B or PC3 cells inhibited vessel formation relative to controls (Fig. 4, C and D).
To confirm that CXCR7/RDC1 plays a role in tumor growth, SCID mice were implanted subcutaneously with cells engineered to express altered levels of CXCR7/ RDC1. Tumors generated from C4-2B overexpressing RDC1/CXCR7 were significantly larger than tumors generated by the transfected  Table 1. Whole cell lysates were immunoblotted with antibody to CD44 and CDH11. The blots were stripped and reprobed with anti-␣-tubulin antibody to confirm equal protein loading. B, CXCR7/RDC1 regulates secretion of IL-8 and VEGF by PCa. IL-8 and VEGF levels were evaluated by ELISA for PCa cells with altered expression of CXCR7/RDC1 at 48 h. The data are presented as means Ϯ S.D. for triplicate determinations and normalized against total protein. * denotes significant difference from respective controls; #, significant from nontreatment controls by CXCL12 (200 ng/ml) stimulation (p Ͻ 0.05, ANOVA). C and D, effect of PCa expression of CXCR7/RDC1 on blood vessel formation. HDMECs were plated in growth factor reduced Matrigel TM in the presence or absence of the conditioned medium as a functional in vitro assay of blood vessel formation. After 24 h the cultures were fixed and stained, which indicated PCa cells themselves were co-plated with the HBMECs. As a negative control, only HDMECs were plated. Original magnification was ϫ20, where the black bars represent 100 m. Vessel sprout formation was quantified by direct microscopic counting. The data are presented as mean Ϯ S.D. for triplicate determinations. * indicates significant difference from the controls (p Ͻ 0.05, ANOVA).

CXCR7/RDC1 in Prostate Cancer
control group (Fig. 5, A and B). In contrast, tumor growth was dramatically suppressed in tumors established from cells in which CXCR7/RDC1 expression was decreased by siRNA transfection relative to controls (236.1 Ϯ 57.8 mg versus 59.3 Ϯ 32.9 mg) (Fig. 5, A and B). To address directly whether CXCR7/ RDC1 plays a role in tumor angiogenesis, the tumors were stained with an antibody to factor VIII. Overexpression of CXCR7/RDC1 in C4-2B or PC3 cells resulted in significantly larger and more abundant blood vessels than in tumors generated by control cells (Fig. 5, C and D, and supplemental Fig. 4).
Previously, we demonstrated that CXCR4 signaling results in activation of ERK and AKT pathways in PCa cells (31). This results in alterations in several angiogenic factors known to regulate the angiogenic phenotype of PCa cells (31). Given that CXCR7/RDC1 functions as a second chemokine receptor for CXCL12, it is reasonable to hypothesize that CXCR7/RDC1 signaling will trigger similar activation pathways. As shown in Fig. 6A, cells expressing less CXCR7/RDC1 expressed less phosphorylated AKT (p-AKT) levels relative to controls. Examination of p-AKT levels in tumor tissues substantiated the in vitro findings demonstrating that reducing CXCR7/RDC1 levels in C4-2B cells resulted in less expression of p-AKT (Fig. 6B), whereas no significant alterations in the overall AKT2 expression were noted (Fig.  6B). Taken together, these data suggest that signaling by the CXCR7/ RDC1 in C4-2B cells activates AKT signaling pathways, which may provide survival advantage to the cells.

DISCUSSION
Previous work by our group and others has shown that CXCL12 and its receptor (CXCR4) are critical elements in growth and metastasis of PCa to tissues that produce large quantities of CXCL12 (13,20,31,34). However, blockade of the CXCR4 receptor only partially blocks the metastatic behavior in vivo. This suggests that other factors exist to control the tissue-specific migration of epithelial cancer cells. CXCR7/RDC1, formerly an orphan receptor, has recently been identified as a receptor for CXCL12, where it is expressed by many human cell lines, vascular endothelial cells, and in rodent brain, kidney, lung, heart, spleen, pancreas, small intestine, blood, colon, and blood vessels (28,35). In addition, recent finding suggests that expression of CXCR7/RDC1 is elevated in endothelial cells associated with tumors (36). Recently, Begley et al. (37) reported that PC3 cells express CXCR7/RDC1. However, to our knowledge, there is no report on the expression and functional contribution of CXCR7/RDC1 in PCa development.
In this study, we demonstrate that CXCR7/RDC1 expression is correlated with tumor aggressiveness. Our results suggest that alterations of CXCR7/RDC1 expression are associated with survival advantage for tumors by facilitating adhesive and invasive activity of PCa cells in in vitro and in vivo tumor  FEBRUARY 15, 2008 • VOLUME 283 • NUMBER 7

CXCR7/RDC1 in Prostate Cancer
growth. This correlated with the signaling of CXCR7/RDC1 through activation of the AKT pathways upon CXCL12 stimulation in vitro. Additionally, immunohistochemistry confirmed the relationship between CXCR7/RDC1 expression and activation of AKT pathways in PCa.
Interestingly, we found that CXCR7/RDC1 regulates CD44 and CDH11 levels. CD44 is a multifunctional protein involved in cell adhesion and signaling. The role of CD44 in PCa devel-opment and progression is controversial with studies showing both tumor-promoting and tumor-inhibiting effects (38). CD44 was also found to play a role in metastatic behavior of tumor cells progenitor/PCa stem cell homing (38). Similar observations have been found in head and neck cancer (39), pancreatic cancer (40), and PCa stem cells (33). Our finding that CXCR7/RDC1 expression alters CD44 levels in PCa cell lines suggests that the CXCL12/CXCR7/RDC1 axis is likely to  play a potential role in PCa stem cell niche; however, further studies are required (41). Similarly, expression of CDH11 was also regulated by RDC1. CDH11 may be critical to the invasive activities of PCa. CDH11, initially identified in osteoblasts (OBcadherin), is expressed in mesenchymal cells (fibroblasts) and in a subset of highly invasive cell lines and aggressive breast carcinomas (42). In this study we also observed that at the level of transcription, CXCR7/RDC1 expression also regulates TGF-␤1 expression, a factor intimately associated with tumorigenesis (11). For example, recent studies have demonstrated that TGF-␤1 may participate as a molecular switch resulting in Akt activation by CXCR4 signaling (43). Moreover, TGF-␤1 has been implicated in the induction of an epithelial to mesenchymal transition, or mesenchymal transition, that may be an essential component of malignant progression (44). In addition to the adhesive changes brought about by CXCR7/RDC1, the receptor appears to regulate blood vessel formation by PCa cells. Here, it was noted that both IL-8 and VEGF levels were altered in response to changes in CXCR7/ RDC1 expression. These findings were in keeping with the number of blood vessels formed by the tumors in vivo. IL-8 and VEGF have been shown to be important factors involved in the development of tumor blood supply in the progression of solid tumors (45,46). Numerous published reports and our studies describe the association of microvessel density with IL-8 and VEGF expression and activation of ERK and AKT pathways (45,46). Together, our results suggest that CXCR7/RDC1 enhances tumor growth through activation of AKT pathways to induce angiogenesis.
We also showed that CXCR7/RDC1 expression might be required for tumor survival. By interrupting CXCR7/RDC1 expression using its sequence-specific CXCR7/RDC1 siRNA, we observed that PCa growth was suppressed. These findings are consistent with those reported by Burns et al. (47), in which small molecular antagonists to CXCR7/RDC1 impeded tumor growth in vivo. Thus, CXCR7/RDC1 signaling may be an attractive new therapeutic target for treatment of PCa. As CXCR7/ RDC1 functions as a second chemokine receptor for CXCL12 in PCa progression, it is reasonable to suspect that downstream events activated by both receptors may be similar. Indeed, using a cell line that does not express CXCR4 (SaOS-2, an osteosarcoma cell line), we observed that signaling by the CXCR7/ RDC1 activates ERK and AKT pathways upon CXCL12 stimulation. 3 However, Burns et al. (47) suggested that the CXCR7/ RDC1 signaling pathway may be distinct from the typical G protein-coupled receptor mechanism. The divergent results may also be due to the use of different tumor cell lines, although other possibilities clearly exist. In addition, CXCR7/RDC1 may complex with CXCR4 forming a heterodimer that potentates CXCL12 signaling (48). If true in our systems, the finding that CXCR4 regulates CXCR7/RDC1 expression may be because of a requirement for CXCR7/RDC1 in CXCR4 signaling. Along these lines, the findings that alterations of CXCR4 expression enhance the levels of CXCR7/RDC1 suggest an important consideration for therapy. Specifically, it might be difficult to block CXCR4 alone to prevent metastasis or entry of T-tropic viruses human immunodeficiency viruses, as inhibition of CXCR4 enhanced CXCR7/RDC1 expression. This possibility will require further examination but may provide an opportunity for therapeutic manipulation to prevent metastasis. In summary, these data demonstrate a role for CXCR7/RDC1 in PCa metastasis and progression. Elucidation of newly identified receptor roles in tumorigenesis and progression may indicate new avenues of potentially therapeutic intervention in PCa. . CXCR7/RDC1 mediates activation of AKT signaling pathway upon CXCL12 stimulation. A, evaluation of the effect of CXCR7/RDC1 on AKT signaling. After a 24-h serum withdrawal, PCa cells were stimulated with 200 ng/ml CXCL12 over a 45-min period at 37°C. Whole cell lysates were separated on SDS-PAGE and immunoblotted with antibodies to p-AKT. The blots were stripped and reblotted with antibodies against total AKT, respectively (top). Quantitative densitometry was used for direct comparison. The data are presented as mean Ϯ S.D. for triplicate determinations. * denotes significant difference from the controls (p Ͻ 0.05, Krusal-Wallis test, and Dunn's multiple comparisons test). B, immunohistochemical evaluation of AKT-2, and phosphorylated AKT (Ser-473) in PCa tumors grown in SCID mice. Subcutaneous. tumors expressing various levels of CXCR7/RDC1 were grown in SCID mice and were harvested and processed for immunohistochemistry staining for AKT-2 and p-AKT. Original magnification ϫ40, where the black bars represents 50 m. FEBRUARY 15, 2008 • VOLUME 283 • NUMBER 7

CXCR7/RDC1 in Prostate Cancer
JOURNAL OF BIOLOGICAL CHEMISTRY 4293