Identification of a Novel Cell Death Receptor Mediating IGFBP-3-induced Anti-tumor Effects in Breast and Prostate Cancer*

Insulin-like growth factor-binding protein-3 (IGFBP-3), a major regulator of endocrine actions of IGFs, is a p53-regulated potent apoptotic factor and is significantly suppressed in a variety of cancers. Recent epidemiologic studies suggest that IGFBP-3 contributes to cancer risk protection in a variety of cancers, and a polymorphic variation of IGFBP-3 influences cancer risk, although other studies vary in their conclusions. Some antiproliferative actions of IGFBP-3 have been reported to be independent of IGFs, but the precise biochemical/molecular mechanisms of IGF-independent, antiproliferative actions of IGFBP-3 are largely unknown. Here we report a new cell death receptor, IGFBP-3R, that is a single-span membrane protein and binds specifically to IGFBP-3 but not other IGFBP species. Expression analysis of IGFBP-3 and IGFBP-3R indicates that the IGFBP-3/IGFBP-3R axis is impaired in breast and prostate cancer. We also provide evidence for anti-tumor effect of IGFBP-3R in vivo using prostate and breast cancer xenografts in athymic nude mice. Further in vitro studies demonstrate that IGFBP-3R mediates IGFBP-3-induced caspase-8-dependent apoptosis in various cancer cells. Knockdown of IGFBP-3R attenuated IGFBP-3-induced caspase activities and apoptosis, whereas overexpression of IGFBP-3R enhanced IGFBP-3 biological effects. IGFBP-3R physically interacts and activates caspase-8, and knockdown of caspase-8 expression or activity inhibited IGFBP-3/IGFBP-3R-induced apoptosis. Here, we propose that IGFBP-3R represents a novel cell death receptor and is essential for the IGFBP-3-induced apoptosis and tumor suppression. Thus, the IGFBP-3/IGFBP-3R axis may provide therapeutic and prognostic value for the treatment of cancer.

Insulin-like growth factor-binding proteins (IGFBPs) 3 are integral components of the IGF system and modulate biological actions of IGFs such as cellular proliferation, differentiation, increase in metabolic activity, and cell survival (1). Apart from its ability to inhibit or enhance IGF actions, all the IGFBPs, IGFBP-1 to -6, have been reported to exert distinct biological actions such as cell proliferation, differentiation, migration, angiogenesis, and apoptosis through an IGF/IGF-I receptor (IGF-IR)-independent manner (2)(3)(4)(5)(6)(7). These intrinsic biological activities of IGFBPs appear to be critical to cardiogenesis, vascular development, and pathogenesis of cancer.
IGFBP-3, the most abundant IGFBP species in serum, circulates as a 150-kDa ternary complex with an acid-labile subunit and IGF peptide (1-3). Classically, the principal function of IGFBP-3 has been to transport IGFs, protecting them from rapid clearance and/or degradation, and modulating IGF bioavailability to cell-surface IGF receptors (8 -9). Many studies have reported that IGFBP-3 exhibits distinct biological effects independent of the IGF/IGF-IR axis, in particular cell growth inhibition and induction of apoptosis in a variety of cancer cells (10 -14). IGFBP-3 was shown to exert its IGF/IGF-IR-independent, antiproliferative actions via interaction with specific proteins in the extracellular, cytoplasmic and nuclear compartments of cancer cells, but the precise underlying mechanisms of IGFBP-3 have yet to be elucidated (12)(13)(14)(15)(16).
IGFBP-3 binding to a cell surface protein is indispensable for its antiproliferative action in human breast cancer cells, and the mid-region of IGFBP-3, which is the least conserved region among the IGFBP-1-6 protein, is responsible for cell surface binding (17,18). Moreover, IGFBP-3-induced apoptosis through the activation of specific caspases that are involved in the death receptor-mediated apoptotic pathways in breast cancer cells (19). These findings strongly suggest the existence of an IGFBP-3-specific receptor participating in the direct proapoptotic effect of IGFBP-3 in cancer cells.
We describe herein the cloning and characterization of a novel cDNA encoding a new cell death receptor for IGFBP-3, IGFBP-3R. We found that IGFBP-3R is a bona fide IGFBP-3 receptor; IGFBP-3R is expressed on the cell surface, interacts specifically with IGFBP-3, activates initiator caspase-8, and mediates IGFBP-3-induced apoptosis in vitro and tumor suppression in vivo.

EXPERIMENTAL PROCEDURES
Cell Culture-All cell lines, except M12 and the MCF-7:IG-FBP-3 stable lines, were purchased from ATCC. We cultured breast cancer cells in Dulbecco's modified Eagle's medium/high glucose with 10% fetal bovine serum (FBS), PC-3 in F-12 with 10% FBS, and M12 cells in RPMI 1640 medium supplemented with 10 ng/ml epidermal growth factor, 0.02 mM dexamethasone, 5 g/ml insulin, 5 g/ml transferring, 5 g/ml selenium and gentamicin at 37°C under 5% CO 2 . MCF-7:IGFBP-3 stable lines were generated from wild type MCF-7 cells using the ecdysone-inducible expression system (Invitrogen). IGFBP-3 protein production was induced with the addition of ponasterone A (Invitrogen) to the culture medium.
Yeast Two-hybrid Screening-An Hs578T cDNA library in pGAD10 was generated using the Two-Hybrid cDNA Library Construction kit (Clontech). The bait plasmid was constructed by amplifying an internal sequence of the IGFBP-3 cDNA (encoding amino acids 88 -148) by PCR, then cloning this fragment into the pBTM116 vector, inframe with the LexA DNA binding domain coding sequence. The bait plasmid was then transformed into yeast strain L40. A single transformant colony was selected for successive transformation with the pGAD10:Hs578T cDNA library. 663 transformant colonies that grew on selective media were replica-plated onto duplicate fresh plates, one of which was used for a ␤-galactosidase assay. Positive colonies were then cultured on non-selective media until the bait plasmid was lost and further tested in mating assays with the original bait strain and a false-positive screening strain expressing the pBTM116:lamin plasmid. Three colonies tested positive with the IGFBP-3 bait and negative with lamin. Plasmid DNA was isolated from these and transformed into E. coli TG1.
Cell Monolayer Binding Assay-Breast cancer cells were seeded into 24-well plates and grown to 60 -70% confluency then transfected according to manufacturer's instructions in serum-containing medium for 36 h. Sf9 cells were seeded into 24-well plates and infected with virus harboring the IGFBP-3R FLAG cDNA for 48 h. Cells were washed with cold PBS, then incubated for 3 h at 12°C in binding buffer (1ϫ Hanks' buffered salt solution, 1 mM CaCl 2 , 1 mM MgCl 2 , 0.5% BSA) containing 125 I-IGFBP-3 (Diagnostic Systems Laboratories) at 50,000 cpm per well in triplicate with or without unlabeled IGFBP-3 as indicated in the Fig. 1 legend. The binding solution was then aspirated, and the cells were washed in cold binding buffer (without BSA), lysed in 0.25 N NaOH, and then the radiolabeled IGFBP-3 was detected in a gamma counter.
Generation of Adenoviral IGFBP-3 and IGFBP-3R-We used NotI-XbaI restriction fragments from pcDNA3/wild type IGFBP-3 cDNA and EcoR1-Xho1 restriction fragments from CS2-IGFBP-3R F to ligate into NotI-XbaI-digested pShuttle-CMV and EcoRV-digested pShuttle-CMV, respectively. We used SalI-NotI restriction fragments from pcDNA3.1/LacZ (Invitrogen) to ligate into SalI-NotI-digested pShuttle-CMV. Recombination into the pAdEasy viral backbone was accomplished in E. coli strain BJ5183 according to the manufacturer's instructions. The recombination was verified, and the adenoviral recombinant DNA was transferred to DH5␣. We purified recombinant pAdEasy plasmids containing CMV-cDNA inserts using Qiagen columns (Qiagen) and transfected QBI-293A cells with 5 g of PacI-digested DNA using the calcium phosphate method (Promega). Cells were seeded at 2 ϫ 10 6 cells per 150-mm culture dish 24 h before transfection. Lysis of transfected cells, indicating adenoviral growth, occurred within 4 days. After amplification, lysates containing clonal recombinant Ad were prepared from 150-mm culture dishes and purified by CsCl gradient centrifugation. Recovered virus was separated into aliquots and stored at Ϫ20°C in 5 mM Tris, pH 8.0, buffer containing 50 mM NaCl, 0.05% bovine serum albumin (BSA), and 25% glycerol. We performed viral titrate assay using serial dilution infection of QBI-293A cells and counting plaques under an overlay of 0.3% agarose, 10% FBS, and 1ϫ Dulbecco's modified Eagle's medium.
Northern Blotting-We used Human Multi-tissue Northern Blots I & II (Clontech). Total RNAs from monolayer cultures of human breast and prostate cells were harvested using the RNeasy RNA isolation kit (Qiagen). Five g of total RNA from each cell line were run on 1% formaldehyde gels and transferred onto GeneScreen Plus nylon membranes (PerkinElmer Life Sciences). Membranes were UV-cross-linked and stained in 0.02% methylene blue, 0.3 M NaOAc, pH 5.5. 32 P-Labeled cDNA probes were prepared using the Prime IT II kit (Stratagene). Membranes were hybridized in Rapid-Hyb buffer (Amersham Biosciences), washed in 0.3ϫ SSC (1ϫ SSC ϭ 0.15 M NaCl and 0.015 M sodium citrate), and exposed to film.
Cell Proliferation Assays-We carried out cell proliferation assays using the MTT Cell Proliferation Assay (ATCC). Cells were seeded in triplicate in 96-well microtiter plates, grown to 60 -70% confluency, and treated as indicated. MTT was added to the cell culture and incubated for 4 h at 37°C. The assays were performed according to the manufacture's protocol.
In Vitro Transcription and Translation Techniques (TNT)-Each linear plasmid DNA template of IGFBP-3R and caspase-8 was translated separately in the presence of wheat germ extracts, T7b RNA polymerase, and amino acid mixture according to the manufacturer's protocol (Promega). A mixture of the two proteins was co-immunoprecipitated with IGFBP-3R antibodies, and the resultants were further Western-blotted using caspase-8 and IGFBP-3R antibodies.
IGFBP-3R Binding Assay-Cos-7 cells were infected with Ad:IGFBP-3R for 2 days. The cells were lysed in HBSST (HBSS with 0.035% sodium bicarbonate, 0.5% Triton X-100, 1 mM MgSO 4 , and 1 mM CaCl 2 , pH 7.4) supplemented with Halt TM Protease Inhibitor (Roche Applied Sciences). A 96-well plate was coated with 50 l of anti-FLAG antibody diluted in PBS at 10 ng/l per well overnight at 4°C. The wells were blocked with 300 l of HBSST containing 5% BSA (HBSST-BSA) for 1 h at room temperature. The wells were washed 3 times with 300 l of PBS containing 0.2% Tween 20. Cell lysates (10 g/50 l/well) were added and incubated for 1 h at room temperature. After washing, 16 nM biotinylated IGFBP-3 was incubated in the presence of 0 -100 nM IGFBP-3 (Celtrix) for 1 h at room temperature. The wells were incubated with HRP-conjugated streptavidin (Thermo Scientific) diluted in 50 l of HBSST-BSA for 1 h at room temperature and further incubated with 50 l of 3,3Ј,5-5Ј-tetramethyl benzidine substrate (Sigma). The reaction was terminated by adding 50 l of 1 N H 2 SO 4 , and absorbance was measured at 450 nm.
Immunofluorescence Staining-We seeded cells in a chamber slide (Lab-Tek II Chamber Slide System, Nalge Nunc International) and infected cells with Ad:IGFBP-3R for 2 days. Adherent cells were washed with PBS and fixed in 3% paraformaldehyde for 10 min at room temperature. After washing twice with PBS, cells were permeabilized with 0.1% Triton X-100 in PBS for 5 min at room temperature, washed twice with PBS, and blocked in 1% BSA with 1% normal goat serum in PBS for 30 min at room temperature. We incubated slides with ␣-IGFBP-3R or ␣-FLAG antibodies at a 1:100 dilution in blocking buffer for 30 min at room temperature and washed slides twice with PBS. We incubated slides with secondary goat anti-rabbit IgG Alexa Fluor 568 or goat anti-mouse IgG fluorescein isothiocyanate at a 1:200 dilution for 45 min followed by 3 washes with PBS. Next we counterstained the nuclei with Hoechst (1 g/ml) for 5 min. Slides were washed three times with PBS, and stained slides were mounted with VECTASHIELD Mounting Medium (Vector Laboratories) and analyzed under a confocal microscope (Leica TCS-SP2 AOBS).
Cell Lysates, Subcellular Fraction Preparations-Cell lysates were harvested by washing monolayer cultures in ice-cold PBS then lysing in Triton lysis buffer (20 mM Tris, pH 7.6, 150 mM NaCl, 1% Triton X-100). Plates were rocked at 4°C for 15 min, and lysates were collected and centrifuged to remove cellular debris. For subcellular fraction preparations, plasma membrane and cytosolic and nuclear fractions were isolated by homogenizing cells using the membrane protein extraction kit (Bio Vision) and centrifuging successively at 700 and 100,000 ϫ g for 10 and 60 min, respectively. All pellets were solubilized in 50 mM HEPES, pH 7.4, 150 mM NaCl, 2 mM MgSO 4 , 1% Triton X-100. Equal amounts of protein from each fraction were analyzed by SDS-PAGE followed by Western blotting for fractionation specific markers.
Immunoprecipitations and Western Blotting-For immunoprecipitations, cleared cell lysates were diluted in 20 mM Tris, pH 7.6, 150 mM NaCl and adjusted to a final concentration of 0.5% Triton X-100 (immunoprecipitation buffer). Antibodies and Protein A or Protein G-Sepharose (Amersham Biosciences) were added, and reactions were incubated at 4°C for 4 h. Beads were washed twice in immunoprecipitation buffer and suspended in 1ϫ reducing sample buffer. Samples were run on 12.5% reducing SDS-PAGE and electrotransferred onto Hybond-ECL nitrocellulose (Amersham Biosciences). Membranes were blocked in 5% nonfat milk, Tris-buffered saline, 0.1% Tween 20 (TBS-T). Primary antibodies were diluted in blocking solution and incubated on the membranes for 2 h at room temperature. Membranes were washed in TBS-T then incubated with HRP-conjugated secondary antibody, diluted 1:3000 for 1 h at room temperature, washed, and detected using the Renaissance Western blot Chemiluminescence reagents (PerkinElmer Life Sciences).
Thymidine Incorporation Assay-Cells were seeded in triplicate into 24-well plates, grown to 60 -70% confluency, and transfected and treated as indicated in the Fig. 4 legend. At 48 h post-transfection, the medium was replaced with serum-free medium, and cells were further incubated for 24 h. To each well 0.1 Ci of [ 3 H]thymidine was added, and the acid-precipitable radioactivity in each well was assayed 24 h later.
Cell Death ELISA Assay-Apoptotic cell death was measured using the cell death detection ELISA (Roche Applied Sciences) according to the manufacturer's instructions. Equivalent cell populations were measured, and absorbance was measured at 405 nm (reference wavelength, 490 nm).
Intratumoral Delivery of Ad:IGFBP-3R in Vivo-Mouse experiments were performed under a protocol approved by the Virginia Commonwealth University Institutional Animal Care and Use Committee (IACUC). M12 prostate cancer cells or MDA231 breast cancer cells suspended in 100 l of sterile PBS were injected into the dorsal flank of 4-week-old male or female athymic nude mice, respectively. Once tumor volume reached ϳ75 mm 3 , mice were injected intratumorally 2 times at day 0 and day 6 with 5 ϫ 10 9 plaque-forming units of Ad:IGFBP-3R or Ad:Lac-Z in 50 l of PBS or 50 l of PBS alone as a control. Tumor size and volume were measured every 2 days. We harvested tumor tissues for immunohistochemistry of ␤-galactosidase, IGFBP-3R, and CD34 16 days after first injection.
Human Samples-Breast and prostate tumor paraffin blocks were obtained from the archives of the Division of Surgical Pathology with approval from the Institutional Review Board of Virginia Commonwealth University. To perform semiquantitative RT-PCR, we obtained frozen-fresh prostate tumor tissues from the Tissue/Tumor Procurement Facility of the Oregon Cancer Center with approval from the Institutional Review Board of Oregon Health Sciences University. Total RNA was purified from tumor tissues using the TRIZOL (Invitrogen) extraction technique as recommended by the manufacturer. One g of purified total RNA was used for RT-PCR analysis using the ThermoScript RT-PCR System (Invitrogen) and specific primers.
Immunohistochemistry of Human and Mouse Tissue-Slides with the human and mouse tissue sections were dried in a 65°C oven for 1 h before staining. The tissue sections were deparaffinized in xylene and rehydrated in decreasing grades of alcohol (100, 90, and 80%) and rinsed in PBS. We quenched endogenous peroxidase activity by incubating tissue sections in 3% hydrogen peroxide for 5 min. The slides are placed in the steamer for heat target antigen retrieval at the optimal determined pH of 9. We incubated slides with the primary antibodies for IGFBP-3 (1:100 dilution), IGFBP-3R (1:100 dilution), or CD34 (1:100 dilution) for 1 h at room temperature and then rinsed twice in Dako wash buffer. We further incubated slides with appropriate secondary antibody for 30 min then rinsed and applied a HRP-labeled Envisionϩ TM polymer. The specific antibody and conjugated polymer enzyme complex was visualized using the Envisionϩ TM Dual Link System-HRP (DABϩ) system after thorough washing. We counterstained tissue sections with Gill's III hematoxylin for 30 s.
Statistical Analyses-We conducted statistical analysis by analysis of variance with Student's t test. We considered a p value of 0.05 or less as significantly different.

Identification of a Novel IGFBP-3R and Its Endogenous Expression on Cell
Membrane-The yeast two-hybrid screening was performed to identify a potential IGFBP-3 receptor using a cDNA construct encoding the mid region (amino acids 88 -148) of the IGFBP-3 protein as bait against an Hs578T, human breast cancer cell cDNA library. This screen resulted in the identification of a cDNA, later designated IGFBP-3R, representing a novel gene composed of 915 base pairs and encoding a 240-amino acid polypeptide (GenBank TM accession #FJ748884). Genomic data base analysis identified 4 exons composing the cDNA sequence on chromosome 16q13.
Northern blot analysis indicated a wide distribution of IGFBP-3R mRNA in human tissues. The 0.9-kb mRNA was detected to varying degrees in normal human tissues (Fig. 1A). A second roughly 2-kb transcript was detected in testis, the significance of which, if any, is not known. The deduced amino acid sequence indicates that IGFBP-3R consists of an N-terminal putative signal peptide sequence, three potential N-glycosylation sites, a putative single-span transmembrane domain, and a short C-terminal cytoplasmic domain (Fig. 1B). Interestingly, its transmembrane domain contains a leucine zipper-like heptad repeat pattern of amino acids similar to the single-span transmembrane leucine zipper sequences identified in erythropoietin receptor and discoidin domain receptor (Fig. 1B, inset, and Refs. 21 and 22).
To examine the endogenous protein, a rabbit polyclonal antibody was generated against GST-fused IGFBP-3R purified from E. coli. The IGFBP-3R antibody recognized a species of roughly 32 kDa from human breast cancer cell lysates (Fig. 1B, right  panel). We further examined whether IGFBP-3R is a cell membrane protein and whether IGFBP-3 interacts with IGFBP-3R on the cell surface using monolayer binding assays with 125 Ilabeled IGFBP-3, cell fractionation preparations by differential centrifugation, and confocal microscopy. Overexpression of IGFBP-3R resulted in a 1.5-and 3.5-fold increase in IGFBP-3 cell surface binding to MCF-7 and Sf9 cell monolayers relative to cells expressing endogenous levels of IGFBP-3R, respectively (Fig. 1C, left panel). Total binding was competed in a dose-dependent manner with the addition of unlabeled IGFBP-3, demonstrating IGFBP-3R specificity for binding IGFBP-3. The presence of the FLAG tag at the C terminus of the IGFBP-3R protein had no effect on IGFBP-3 binding (Fig. 1C, right panel). In addition, the IGFBP-3R antibody recognized a species of roughly 32 kDa in plasma membrane and cytoplasmic fractions but not in nuclear fraction in human breast cancer cells (Fig.  1D). In accordance with these findings, confocal microscopy confirms that both endogenously produced and overexpressed IGFBP-3R exists on the cell membrane of human breast cancer cells (Fig. 1D).
Specific Interaction of IGFBP-3R with IGFBP-3-To characterize the specific interaction between IGFBP-3 and IGFBP-3R in a mammalian system, coimmunoprecipitation and IGFBP-3R binding assays were performed. A coimmunoprecipitation experiment was initially carried out using the EGFP: IGFBP-3R fusion protein and FLAG-tagged IGFBP-3 produced in a baculovirus system. A specific interaction of IGFBP-3 and IGFBP-3R was observed ( Fig. 2A). Similar results were obtained using MDA231 cells infected with adenoviral constructs containing the FLAG-tagged IGFBP-3R cDNA sequence. When cell lysates from Ad:EV-or Ad:IGFBP-3R F -infected cells were immunoprecipitated with FLAG M2 antibodies and proteins were detected on Western immunoblots with IGFBP-3R antibodies. Both endogenous IGFBP-3R and overexpressed IGFBP-3R F were detected in both cell lysates, whereas FLAG-tagged IGFBP-3R was detected in cell lysates from Ad:IGFBP-3R F -infected cells after immunoprecipitation with FLAG M2 antibodies (Fig. 2B). When the FLAG antibody was used for immuno-precipitation, recombinant IGFBP-3 was precipitated with IGFBP-3R F (Fig. 2, C and D). Specificity of coimmunoprecipitation was demonstrated by showing that IGFBP-3 is only coprecipitated with the Ad:IGFBP-3R F -infected cell lysates but not with the Ad:EV-infected cell lysates (Fig. 2D). To further demonstrate the interaction between endogenously expressed IGFBP-3 and IGFBP-3R, MDA231-conditioned media was immunoprecipitated with IGFBP-3R antibodies in the presence/absence of MCF-7 cell lysates. As shown in Fig. 2E, endogenous IGFBP-3 in MDA231-conditioned media was coprecipitated with endogenously produced IGFBP-3R in MCF-7 cells when IGFBP-3R antibodies were used for immunoprecipitation. Additional coimmunoprecipitation assays with IGFBPs 1-6 reveal that the interaction of IGFBP-3 and IGFBP-3R was evident, but no interaction was detected with other IGFBPs (Fig. 2F). Furthermore, the IGF-I and IGFBP-3 complex was unable to bind IGFBP-3R, indicating that binding of IGFBP-3 to IGF and IGFBP-3R is mutually exclusive (Fig. 2D), and IGF-I may regulate the IGFBP-3/IGFBP-3R-induced biological function by preventing IGFBP-3 binding to IGFBP-3R (10). In vitro IGFBP-3R binding assays were further performed using biotinlabeled IGFBP-3 and purified IGFBP-3R. The interaction between IGFBP-3 and IGFBP-3R was specific and saturable (Fig. 2G, left panel). A competitive binding assay revealed a dose-dependent competition with unlabeled IGFBP-3 with IC 50 ϭ 0.5 nM but not with unlabeled IGFBP-5 (Fig. 2G, right  panel).
Significance of IGFBP-3/IGFBP-3R in Human Cancer-To assess the significance of the IGFBP-3/IGFBP-3R axis in tumorigenesis, the distribution of IGFBP-3 and IGFBP-3R was examined in normal and malignant human breast and prostate specimens. Although IGFBP-3 and IGFBP-3R were readily detected in benign breast lobule and benign prostate tissues, heterogeneous but significantly reduced expression of IGFBP-3 and IGFBP-3R was observed in invasive breast ductal carcinoma and prostate adenocarcinoma cells (Fig. 3A). Heterogeneous but significantly reduced mRNA expression of IGFBP-3 and IGFBP-3R were also observed in prostate tissue (Fig. 3B). These data indicate that the IGFBP-3/IGFBP-3R axis is impaired during the progression of breast and prostate tumor, and restoration of the IGFBP-3/IGFBP-3R axis might play a role in suppression of tumor growth.
The role of IGFBP-3R in prostate and breast tumor in vivo was further examined with an athymic nude mouse model using M12 human prostate cancer cells and MDA231 breast cancer cells. Initial tests revealed that the adenoviral titer, 5 ϫ 10 9 plaque-forming units, was optimal in terms of in vivo expression efficiency and cytotoxic effect on M12 prostate tumor in the nude mice (Fig. 3C). Intratumoral injection of Ad:IGFBP-3R F into M12 cell tumors in nude mice resulted in a significant shrinkage of tumor mass without any obvious toxic effect on mice 16 days after viral injection. Compared with Ad:Lac-Z-infected tumor, Ad:IGFBP-3R tumor showed a 40% reduction in tumor mass (Fig. 3D). A similar anti-tumor effect of IGFBP-3R was obtained in a MDA231 breast tumor nude mouse model (43% reduction in tumor mass). Immunohistochemistry with CD34 antibody revealed that IGFBP-3R-overexpressed tumor also exhibit significant suppression  (right panel). The presence of the FLAG-tag at the C terminus of the IGFBP-3R sequence had no effect. Error bars show S.D. n ϭ 3 in duplicate. *, p Ͻ 0.1; **, p Ͻ 0.05. D, subcellular localization of IGFBP-3R by cell fractionation and confocal microscopy is shown. Left panel, endogenous IGFBP-3 was detected in plasma membrane and cytoplasmic fractions in human breast cancer cells. Subcellular fractions were prepared by differential centrifugation. 100 g of proteins were subjected to Western immunoblot with ␣-IGFBP-3R antibodies. ␤3 integrin, ␣-tubulin, and histone deacetylase (HDAC) were used as plasma membrane (Memb), cytosol (Cyto), and nuclear (Nuc) markers, respectively. Right panel, both endogenous and overexpressed IGFBP-3R is readily detectable on the cell membrane as well as in perinuclear and cytoplasmic area using IGFBP-3R and FLAG M2 antibodies. Hoechst stain (blue) was used to identify nuclei. of tumor angiogenesis (Fig. 3C). These findings indicate that overexpression of IGFBP-3R may enhance paracrine/autocrine-derived IGFBP-3-induced tumor suppression, and thus, IGFBP-3/IGFBP-3R axis represents a novel signaling pathway that inhibits tumor growth.

IGFBP-3R Mediates IGFBP-3induced Growth Inhibition and
Apoptosis-To determine whether IGFBP-3R mediates antiproliferative effect of IGFBP-3, the growth inhibitory actions of endogenous IGFBP-3 via IGFBP-3R were initially tested using IGFBP-3-expressing Hs578T and MDA231 human breast cancer cells, PC-3 prostate cancer cells, and NCI-H157 nonsmall cell lung carcinoma cells. Significant cell growth inhibition was observed in Ad:IGFBP-3R-infected in a m.o.i.-dependent manner but not Ad:EGFP-infected Hs578T cells (Fig. 4A). IGFBP-3R overexpression resulted in growth inhibition up to 90% in IGFBP-3-expressing cancer cells, whereas adenoviral vector itself shows no growth inhibition (Table 1). It was further tested whether IGFBP-3R-induced growth inhibition results from induction of apoptosis. Treatment with the pan-caspase inhibitor z-VAD-fmk resulted in a complete abrogation of both etoposide-and IGFBP-3R-induced cell growth inhibition in Hs578T cells, suggesting that the IGFBP-3R might play a role in IGFBP-3-induced apoptosis (Fig.  4B). Furthermore, IGFBP-3R overexpression effectively increases caspase-3/-7 activities in IGFBP-3expressing cancer cells, reaching maximum activity 48 h after infection (p Ͻ 0.01 in Hs578T and MDA231 cells and p Ͻ 0.05 in PC-3 and H157 cells). This suggests that IGFBP-3R mediates endogenous IGFBP-3-induced growth inhibition via activation of caspasedependent cell death signaling (Fig. 4C).

JOURNAL OF BIOLOGICAL CHEMISTRY 30239
synthesis by an average of 45% compared with controls in MCF-7 cells (Fig. 4D, left panel). With additional overexpression of IGFBP-3R in these cells, DNA synthesis was further inhibited by an average of 65%. Overexpression of IGFBP-3R had no significant effect on DNA synthesis in the absence of IGFBP-3 (either wild type MCF-7 or uninduced MCF-7: IGFBP-3 #3 cells). In IGFBP-3-expressing MDA231 cells, overexpression of IGFBP-3R resulted in a 2.5-fold increase of the cell death compared with controls measured by cell death ELISA assays (Fig. 4D, right panel). When endogenous IGFBP-3 was suppressed by treatment of IGFBP-3 shRNA (23) in these cells, IGFBP-3R overexpression-induced cell death was significantly inhibited (2.5versus 1.6-fold, p Ͻ 0.05). Suppression of endogenous IGFBP-3 without IGFBP-3R overexpression shows a slight increase of the cell death up to 1.3-fold.
It was further examined whether IGFBP-3R is required for IGFBP-3-induced apoptosis using IGFBP-3R RNAi techniques in non-IGFBP-3-expressing M12 prostate cancer cells. After verifying a significant knockdown of endogenous IGFBP-3R by IGFBP-3R siRNAs at the mRNA and protein levels, IGFBP-3-induced caspase activity was measured. Although overexpression of IGFBP-3 resulted in a 500% increase of caspase activity, specific knockdown of IGFBP-3R completely inhibited IGFBP-3-induced caspase activities (Fig. 4E, top panel). No inhibitory effect of IGFBP-3R siRNA on caspase activities was observed in control cells or control adenoviral vector-infected cells. Two target sequences corresponding to siRNAs #1 and #3 were further selected to construct shRNA plasmids. As shown in Fig. 4E, bottom panel, transfection of two IGFBP-3R shRNAs resulted in complete inhibition of IGFBP-3-induced apoptosis with concomitant suppression of endogenous IGFBP-3R expression in M12 cells. These results indicate that IGFBP-3 induces caspase-dependent apoptosis through IGFBP-3R.

IGFBP-3/IGFBP-3R Activates Caspase-8-dependent Cell Death
Signaling-To explore the mechanisms by which the IGFBP-3/IGFBP-3R axis induces apoptosis, the hypothesis was tested that IGFBP-3R might be a novel cell death receptor and activate initiator caspases such as caspase-8 and -10 and the subsequent downstream caspases such as caspase-3 and -7. Because IGFBP-3 has been reported to activate caspase-8 but not caspase-10 in breast cancer cells (19), it was tested whether IGFBP-3R attributes to activation of caspase-8 in IGFBP-3-expressing MDA231 breast cancer cells. IGFBP-3R overexpression resulted in activation of caspase-8 and -3 in a time-dependent manner, as shown by a decrease of procaspase-8 and -3 and concomitant appearance of cleaved active forms of both caspases at 36 h after viral infection (Fig. 5A). Co-treatment with pan-caspase inhibitor z-VAD-fmk completely prevented the IGFBP-3Rinduced caspase-8 and caspase-3/-7 activities and apoptosis (Fig. 5B). At the same time, the caspase-8 specific inhibitor z-IETD-fmk completely abolished the changes in not only caspase-8 but also caspase-3/-7 activities and apoptotic index. We further tested the specific role of caspase-8 in the IGFBP-3/IGFBP-3R axis-induced apoptosis using caspase-8 shRNAs (24). Caspase-8 shRNA treatment resulted in more than 80% suppression of endogenous caspase-8 expression and inhibited IGFBP-3/IGFBP-3R-induced apoptosis up to 60% (Fig. 5C). These data clearly demonstrate that caspase-8 is a key initiator caspase involved in the induction of apoptosis by the IGFBP-3/IGFBP-3R axis.
Specific Interaction between IGFBP-3R and Caspase-8-Because IGFBP-3R contains no death domain (DD) sequence, which is a critical sequence to recruit adapter proteins such as FADD and caspase-8 to the activated death receptors, the underlying mechanisms for IGFBP-3R-induced activation of caspase-8 were further investigated. Coimmunoprecipitation studies using IGFBP-3R-overexpressed Cos-7 cell lysates clearly demonstrated that IGFBP-3R binds to caspase-8 without FADD (Fig. 5D, left  panel, third lane). Further coimmunoprecipitation experiments were performed to verify interaction between endogenous IGFBP-3R and caspase-8 using Cos-7 cells that produce both proteins. Endogenous caspase-8 but not FADD was co-precipitated with endogenously produced IGFBP-3R in Cos-7 cells when IGFBP-3R antibodies were used for immunoprecipitation (Fig. 5D, right panel, fourth lane), whereas neither caspase-8 nor FADD was detected in the immunoprecipitated samples in the absence of ␣-IGFBP-3R antibodies (fifth lane). Because recent studies demonstrated that membrane proteins containing no DD directly interact with caspase-8 and induce apoptosis (25,26), the direct binding of IGFBP-3R to caspase-8 was investigated using a eukaryotic in vitro TNT. IGFBP-3R and caspase-8 were successfully produced by TNT and were comparable with endogenous proteins expressed in human breast cancer cells. When TNT-derived IGFBP-3R and caspase-8 were incubated and immunoprecipitated with IGFBP-3R antibodies, caspase-8 was coprecipitated with IGFBP-3R (Fig. 5E). These data collectively suggest that IGFBP-3R exists as a complex with caspase-8 at the resting stage and IGFBP-3 binding to IGFBP-3R activates procaspase-8 and release from the IGFBP-3R, subsequently activating executioner caspases such as caspase-3 and -7, thereby inducing apoptosis.

DISCUSSION
The IGF/IGF-IR-independent actions of IGFBP-3 have been shown to contribute to the pathophysiology of a variety of human diseases such as cancer, diabetes, ischemia, and Alzheimer disease (10 -14, 27-31). There has been intensive investigation toward characterizing molecular and cellular mechanisms for IGF/IGF-IR-independent anti-tumor effects of IGFBP-3 in human cancer in vitro and in vivo. IGFBP-3 appears to exert its IGF/IGF-IR-independent bio-logical actions through interaction with a variety of binding partners on the cell surface and within cells. IGFBP-3 binds to the low density lipoprotein receptor-related protein-1/ ␣ 2 M receptor (15), autocrine motility factor/phosphoglucose isomerase (32), caveolin, and transferrin/transferrin receptor (33,34) on the cancer cell surface. Functional significance of these binding partners on the IGF/IGF-IR-independent actions of IGFBP-3 is largely unknown, although some studies suggested that these interactions may mediate antiproliferative effects of IGFBP-3 or may result in IGFBP-3 internalization and subsequent biological actions in cytoplasmic and nuclear compartments. In addition, IGFBP-3 is also shown to inhibit cell growth and induce apoptosis through interaction with nuclear proteins such as retinoid X receptor (RXR)-␣, retinoic acid receptor (RAR), and Nur77 (35). IGFBP-3 binds RXR-␣ and RAR and subsequently modulates RAR/RXR and RXR/Nur77 signaling, thereby inducing apoptosis (36,37). However, recent studies also showed that IGFBP-3 mutants that failed to translocate to the nucleus and lost binding ability to RXR-␣ still induced apoptosis in breast cancer cells (38,39). Current findings suggest that IGFBP-3 may utilize multiple mechanisms for its antitumor actions depending upon the cellular environment, or a major key IGFBP-3 signaling pathway is not yet identified. This is the first report of an authentic IGFBP-3 receptor that interacts and functions cooperatively with IGFBP-3 in its IGF/ IGF-IR-independent antiproliferative and proapoptotic functions in cancer cells. Current findings provide evidence for IGFBP-3R as a cell surface receptor for IGFBP-3, including subcellular localization of IGFBP-3R using cell surface binding, cell fractionation, and confocal microscopy, specific interaction of IGFBP-3 with IGFBP-3R on the cell surface, and indispensability of IGFBP-3R for IGFBP-3-induced antiproliferative and proapoptotic functions in cancer cells. Apoptosis can be triggered by internal signals (the intrinsic or mitochondrial pathway) or external signals (the extrinsic or death receptor pathway) (40,41). The initiator caspase in the mitochondrial pathway is caspase-9, which subsequently activates the executioner caspases -3, -6, and -7. In contrast, the initiator caspase of the death receptor pathway is caspase-8. Caspase-8 (like caspase-9) further activates the cascade of executioner caspases (-3, -6, and -7). Recent studies reported that the growth inhibitory effect of IGFBP-3 results mainly from its induction of apoptosis via activation of caspases-8 and -7 without release of cytochrome c into the cytosol and caspase-9 activation in breast cancer cells, suggesting that IGFBP-3 induces apoptosis through the activation of caspases via a death receptor-mediated pathway (19). Hence, IGFBP-3R could be responsible for IGFBP-3-induced activation of the death receptor-mediated pathway. It is noted that IGFBP-3 has also been shown to induce the mitochondria-mediated apoptosis by suppression of antiapoptotic proteins Bcl-2 and Bcl-X L (42) or interaction with RXR-␣ and Nur77 (35)(36)(37). However, our IGFBP-3R siRNA data clearly demonstrate that IGFBP-3R is required for IGFBP-3-induced caspase-3/-7 activity, and the preceding activation of caspase-8 is indispensable for IGFBP-3-induced apoptosis in our cell systems.
IGFBP-3R possesses very unique characteristics; its putative transmembrane domain contains a leucine zipper-like heptad repeat pattern of amino acids. Only a few proteins such as erythropoietin receptor and discoidin domain receptor are known to possess transmembrane leucine zipper sequences, which may involve dimerization/oligomerization of membrane proteins (21,22). Interestingly, the transmembrane leucine zipper sequences of IGFBP-3R are similar to those in erythropoietin receptor; in particular, Ser 203 , Leu 206 , Ser 210 , and Leu 213 in IGFBP-3R are identical to Ser 231 , Leu 234 , Ser 238 , and Leu 241 that occupy a and d positions of an (abcdefg) n heptad repeat motif in transmembrane leucine zipper sequences in erythropoietin receptor (Fig. 1B,  inset). Recent studies demonstrated that those sequences are critical to form the dimer/oligomers in erythropoietin receptor (21). Further structure-function analysis using specific mutants would warrant the significance of the transmembrane leucine zipper sequences on the activity and signaling pathway of the IGFBP-3R. Additionally, IGFBP-3R interacts with initiator caspase-8 without a DD sequence in the intracellular portion of the receptor unlike other known death receptors such as tumor necrosis factor-␣ (TNF-␣) receptor, TNF-related apoptosis-inducing ligand receptor 1 (TRAIL-R1/DR4), TRAIL-R2 (APO-2/DR5), and CD95 (Fas, APO-1). Although the DD sequence is critical in forming death-inducing signaling complex (DISC) by recruiting adaptor proteins (FADD) and initiator caspases (caspase-8) after receptor activation, recent studies demonstrated that clusters of unligated integrins containing no DD directly interact with caspase-8 and induce apoptosis (26). In addition, a novel 600-kDa complex has been shown to recruit and activate caspase-8 independent of FADD and induce apoptosis after p53 activation (25). It is of note that the C-terminal truncated IGFBP-3R mutant shows no interaction with caspase-8, thus strongly suggesting that caspase-8 interacts with the cytoplasmic tail of IGFBP-3R (data not shown). Taken together, we speculate that IGFBP-3R exists as a complex with caspase-8 at the resting stage and IGFBP-3 binding to IGFBP-3R may facilitate dimerization/oligomerization of IGFBP-3R, resulting in proximity-induced activation of the initiator caspase-8 (43).
In summary, we cloned IGFBP-3R, a novel IGFBP-3 death receptor, and identified the IGFBP-3/IGFBP-3R axis as a novel caspase-8-dependent cell death signal in human cancer cells. Our results not only show suppression of IGFBP-3 and/or IGFBP-3R expression in breast and prostate cancer but also the anti-tumor effect of IGFBP-3R in vivo. IGFBP-3R represents a new mammalian cell death receptor and promotes the utilization of the IGFBP-3/IGFBP-3R axis for further development of drug screening assays, and diagnostic/prognostic and therapeutic methods for cancer treatment.