Insulin-like growth factor (IGF)-binding protein-3 induces apoptosis and mediates the effects of transforming growth factor-beta1 on programmed cell death through a p53- and IGF-independent mechanism.

Insulin-like growth factor (IGF) binding protein-3 (IGFBP-3) is known to block IGF action and inhibit cell growth. IGFBP-3 is thought to act by sequestering free IGFs or, possibly, act via a novel IGF-independent mechanism. Supporting its role as a primary growth inhibitor, IGFBP-3 production has been shown to be increased by cell growth-inhibitory agents, such as transforming growth factor-beta (TGF-beta), and the tumor suppressor gene p53. In this paper, we demonstrate, for the first time, a novel function of IGFBP-3 as an apoptosis-inducing agent and show that this action is mediated through an IGF.IGF receptor-independent pathway. In the p53 negative prostate cancer cell line, PC-3, the addition of recombinant IGFBP-3 resulted in a dose-dependent induction of apoptosis. 125I-IGFBP-3 bound with high affinity to specific proteins in PC-3 cell lysates and plasma membrane preparations. These membrane-associated molecules may serve as receptors that mediate the direct effect of IGFBP-3 on apoptosis. In addition, in an IGF receptor-negative mouse fibroblast cell line, treatment with recombinant IGFBP-3 as well as transfection of the IGFBP-3 gene induced apoptosis, suggesting that neither IGFs nor IGF receptors are required for this action. Furthermore, treatment with TGF-beta1, a known apoptosis-inducing agent, resulted in the induction of IGFBP-3 expression 6-12 h before the onset of apoptosis. This effect of TGF-beta1 was prevented by co-treatment with IGFBP-3-neutralizing antibodies or IGFBP-3-specific antisense thiolated oligonucleotides. These findings suggest that IGFBP-3 induces apoptosis through a novel pathway independent of either p53 or the IGF.IGF receptor-mediated cell survival pathway and that IGFBP-3 mediates TGF-beta1 induced apoptosis in PC-3 cells.


Insulin-like growth factor (IGF) binding protein-3 (IGFBP-3) is known to block IGF action and inhibit cell growth. IGFBP-3 is thought to act by sequestering free
IGFs or, possibly, act via a novel IGF-independent mechanism. Supporting its role as a primary growth inhibitor, IGFBP-3 production has been shown to be increased by cell growth-inhibitory agents, such as transforming growth factor-␤ (TGF-␤), and the tumor suppressor gene p53. In this paper, we demonstrate, for the first time, a novel function of IGFBP-3 as an apoptosis-inducing agent and show that this action is mediated through an IGF⅐IGF receptor-independent pathway. In the p53 negative prostate cancer cell line, PC-3, the addition of recombinant IGFBP-3 resulted in a dose-dependent induction of apoptosis. 125 I-IGFBP-3 bound with high affinity to specific proteins in PC-3 cell lysates and plasma membrane preparations. These membrane-associated molecules may serve as receptors that mediate the direct effect of IGFBP-3 on apoptosis. In addition, in an IGF receptor-negative mouse fibroblast cell line, treatment with recombinant IGFBP-3 as well as transfection of the IGFBP-3 gene induced apoptosis, suggesting that neither IGFs nor IGF receptors are required for this action. Furthermore, treatment with TGF-␤1, a known apoptosis-inducing agent, resulted in the induction of IGFBP-3 expression 6 -12 h before the onset of apoptosis. This effect of TGF-␤1 was prevented by co-treatment with IGFBP-3-neutralizing antibodies or IGFBP-3-specific antisense thiolated oligonucleotides. These findings suggest that IGFBP-3 induces apoptosis through a novel pathway independent of either p53 or the IGF⅐IGF receptor-mediated cell survival pathway and that IGFBP-3 mediates TGF-␤1 induced apoptosis in PC-3 cells.
We and others have previously demonstrated the effects of IGFBP-3 as a negative regulator of cell proliferation in prostatic and other tissues (8 -10, 12, 13). This negative growth regulation by IGFBP-3 has been proposed to involve a separate cellular signaling pathway (18,19). Further, in support of its role as a negative regulator of cell growth and proliferation, IGFBP-3 gene expression has also been shown to be induced by other growth-inhibitory (and apoptosis-inducing) agents such as transforming growth factor-␤1 (TGF-␤1) (20 -22), retinoic acid (21), tumor necrosis factor-␣ (TNF-␣) (23), and the tumor suppressor gene, p53 (24). However, the direct apoptosis-inducing ability of IGFBP-3 has not previously been demonstrated. In this study, we have investigated a novel role of IGFBP-3 as an apoptosis-inducing agent.
We hypothesized that the growth-inhibitory effect of IGFBP-3 is mediated not only by regulating the availability of free IGFs and by inducing growth arrest, but also by inducing apoptosis. We further considered that this process may involve an IGF-independent mechanism. To test these hypotheses, we investigated the ability of IGFBP-3 to induce apoptosis in a prostate carcinoma cell line (PC-3) and in an IGF receptornegative (R (Ϫ) ) mouse fibroblast cell line (13,19,25). Furthermore, to determine the importance of IGFBP-3 as a critical cell growth-regulatory factor, we also investigated its role as a mediator of the apoptosis induced by TGF-␤1.
PC-3 Cell Cultures-The human PC-3 cells were purchased from ATCC (Rockville, MD) and were originally initiated from a grade IV prostatic adenocarcinoma from a 62-year-old male Caucasian. The PC-3 cells were grown in 75-cm 2 flasks according to the recommended protocol (FK-12 supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin. For each experiment, cells were dissociated, centrifuged, and resuspended in serum containing FK-12 media with antibiotics and inoculated at a density of 1 ϫ 10 5 cells/cm 2 in 24-well or 6-well tissue culture dishes and grown to confluence in a humidified atmosphere of 5% CO 2 at 37°C before treatment. After a quick wash with serum-free FK-12 media (SFM), the confluent cells were treated with various concentrations of IGFBP-3, TGF-␤1, and/or other specified reagents. SFM with antibiotics was used as the control treatment.
R (Ϫ) Cell Cultures-Fibroblasts from an IGF-I receptor knockout mouse were generated from 18-day embryos as described previously (25) and were designated R (Ϫ) cells. The R (Ϫ) cells were maintained in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum and Geneticin (G418). For the generation of cell lines stably expressing the human IGFBP-3 gene, R (Ϫ) cells were co-transfected with the pKGhBP-3 (19) and pLHL4 vectors. Transfections were done in suspension with 5 g of plasmid DNA for 1 ϫ 10 6 cells using the calcium phosphate suspension method (13,25). The cells were then plated at a concentration of 8 ϫ 10 3 cells/cm 2 . Cells selected in hygromycin were designated R (Ϫ) /BP-3 and were continuously grown in media supplemented with hygromycin (50 g/ml). Several clones were confirmed for IGFBP-3 mRNA and protein expression and were used for further biological experiments (19).
DNA Ladder-DNA fragmentation analyses were performed using cells (1 ϫ 10 6 /ml), which were washed with Tris-buffered saline and resuspended in 1 ml of 0.15 M sodium chloride and 0.015 M sodium citrate (pH 7.0) containing 10 mM EDTA, 1% (w/v) sodium laurylsarkosinate, and 0.5 mg/ml proteinase K. Following digestion with proteinase K, the DNA was precipitated with 2 volumes of cold absolute ethanol, pelleted, resuspended in 10 mM Tris-HCl (pH 8.0) containing 1 mM EDTA, heated to 70°C, and loaded onto a 1.5% agarose gel containing 0.1 mg/ml ethidium bromide. Electrophoresis was carried out in 40 mM Tris acetate, 1 mM EDTA, pH 8.0, until the marker dye had migrated approximately 5 cm. DNA was visualized under UV light and photographed.
Terminal Deoxynucleotidyltransferase-mediated dUTP Nick End Labeling (TUNEL)-In situ detection of apoptosis in cultured cells was performed with the use of direct immunoperoxidase detection of biotinlabeled genomic DNA in monolayer cells. In brief, following treatment with different conditions, the monolayer cultures were fixed in 3.7% paraformaldehyde solution for 10 min at room temperature followed by dehydration in 70% ethanol for 5 min at room temperature. Following this step, the endogenous peroxidase was quenched by treatment with 2% hydrogen peroxide in methanol for 5 min. The cells were incubated in the labeling mixture (Biotin dNTP mix; 50 ϫ MgCl 2 , terminal deoxynucleotidyltransferase, and labeling buffer) for 60 min at 37°C. The free 3Ј-OH DNA in the apoptotic cells was visualized using the streptavidin-horseradish peroxidase-diaminobenzidine detection system. These apoptotic cells appeared as dark brown cells.
Fluorescence-activated Cell Sorting (FACS)-Cells were harvested from confluent monolayer cultures including the floating cells in the conditioned media and were fixed with 1% paraformaldehyde on ice and treated with 70% ethanol for 10 min before use. Apoptotic cells were detected using the Apoptag system, which utilizes terminal deoxynucleotidyltransferase and digoxigenin-labeled D-UTP (Oncore, Gaithersburg, MD). Following incubation in 70% ethanol, the cells were washed in cold PBS and incubated with fluorescein-labeled anti-digoxigenin. After a 30-min incubation, the cells were rinsed in PBS with 0.05 Triton X-100 and resuspended in 0.5 ml of PBS containing PI (5 g/ml) and RNase (1 mg/ml). The cells were incubated in the dark on ice for 30 min prior to flow cytometric analysis. Stained cells were evaluated for fluorescence, employing a Coulter EPICS Elite flow cytometer operated at 488 nm and 300 milliwatts output. Individual cells were electronically gated to exclude aggregates from evaluation. Single parameter fluores-cence histogram data were generated for both isotypic control antibodytreated cells and apoptosis-specific antibody-treated cells. Saved fluorescence histograms were evaluated by Overton's histogram subtraction statistical model (Immuno 4 software, Coulter Immunology, Hialeah, FL) to determine both the percentage of positive cells and relative fluorescence intensity in mean channel fluorescence units. The K ϩ -selective ionophore valinomycin (1 ϫ 10 5 M) was used as a positive control (27). SFM-treated conditions were used as negative controls. The percentage of apoptotic cells in both control and experimental conditions were compared.
Comparative Analyses of Apoptosis and Necrosis-To confirm that the population of apoptotic cells observed in the IGFBP-3-treated conditions are truly undergoing programmed cell death and not necrosis, we utilized the previously published novel method and detected both apoptosis and necrosis from the same samples using quantitative FACS analysis. This method utilizes the binding of FITC-labeled annexin V to phosphatidylserine in the cell membrane that surfaces only during the early phase of apoptosis, indicating the loss of cell membrane phospholipid asymmetry (28 -30). However, these apoptotic cells with intact cell membranes do not stain with the PI. Utilizing the morphological changes that occur in both apoptotic and necrotic cells, we simultaneously stained the samples with annexin-FITC and PI and subjected the samples to flow cytometric analyses to detect the percentage of apoptotic (FITC-stained cells) and necrotic cells (PI-stained cells) in a given population. A minimum of 6,000 cells was maintained for all samples.
Apoptosis ELISA Assay-Photometric cell death detection ELISA (Boehringer Mannheim) was performed to quantitate the apoptotic index by detecting the histone-associated DNA fragments (mono-and oligonucleosomes) generated by the apoptotic cells. The assay is based on the quantitative sandwich enzyme immunoassay principle using mouse monoclonal antibodies directed against DNA and histones, respectively, for the specific determination of these nucleosomes in the cytoplasmic fraction of cell lysates. In brief, equal numbers of cells were plated in 24-well culture plates (1 ϫ 10 4 /cm 2 ) in serum-supplemented FK-12 medium and grown to confluency for 72 h. At the time of sample collection, the confluent cells were washed with PBS and treated with various concentrations of IGFBP-3, TGF-␤1, or other required agents for the designated time period. The cells were dissociated gently (PBS with 0.1 M EDTA) and pelleted along with the floating cells (mostly apoptotic cells) collected from the conditioned media. The cell pellets were used to prepare the cytosol fractions that contained the smaller fragments of DNA. Equal volumes of these cytosolic fractions were incubated in anti-histone antibody-coated wells (96-well plates), and the histones of the DNA fragments were allowed to bind to the antihistone antibodies. The peroxidase-labeled mouse monoclonal DNA antibodies were used to localize and detect the bound fragmented DNA using photometric detection with 2,2Ј-azino-di-(3-ethylbenzathiazoline sulfonate) as the substrate. Valinomycin (1 ϫ 10 5 M) was used as a positive control (27). SFM-treated conditions were used as negative controls. Each experimental condition was carried out with at least three samples and was repeated at least three times. The reaction products in each 96-well plate were read using a Bio-Rad microplate reader (model 3550-UV). Averages of the values Ϯ S.E. from double absorbance measurements of the samples were plotted.
Preparation of Cell Lysates-Confluent PC-3 cells were briefly washed with cold PBS and allowed to dissociate in dispersion buffer (1 mM EDTA in PBS, pH 7.4). Free floating cells were collected and centrifuged (2000 rpm for 5 min), resuspended in cold lysis buffer containing 10 mM HEPES, 1.5 mM EDTA, 0.5 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 1 M aprotinin, and 1 mM pepstatin in PBS (pH 7.4), vortexed, and boiled at 100°C for 5 min. Aliquots were stored at Ϫ70°C until further use.
Membrane Preparation-Samples were collected using the method described above for the preparation of cell lysates. The pellets were resuspended in homogenization buffer (10 mM HEPES, 1.5 mM EDTA, 0.5 mM dithiothreitol, and 1 mM phenylmethylsulfonyl fluoride, pH 7.4). Cells were homogenized using a Dounce homogenizer, maintaining the uniformity of the homogenization process. Homogenized cells were centrifuged at 3000 rpm at 4°C for 5 min, and the nuclear pellet was discarded. The supernatant was recentrifuged at 12,000 rpm at 4°C for 30 min, and the pellet containing the debris was discarded. The supernatant containing the plasma membrane was then resuspended in solubilization buffer (50 mM HEPES, 0.15 M NaCl 2 , 2 mM MgSo 4 , 1 mM phenylmethylsulfonyl fluoride, pH 7.4) and further homogenized using the same technique as mentioned above. Following centrifugation at 33,000 rpm at 4°C for 1 h using a Sorvall ultracentrifuge (DuPont, OTD B60), the resulting pellet was resuspended in isolation buffer and used for separation on SDS-PAGE.
Western Immunoblots-The Western immunoblot analysis was performed as described previously (31). Serum-free conditioned media from PC-3 cultures treated with various concentrations of TGF-␤1 for different time periods were used. SFM incubated with similar culture conditions were used as controls. Samples of 100 l (from 1 ϫ 10 6 cells) were electrophoresed through 12.5% nonreducing SDS-PAGE overnight at constant voltage, electroblotted onto nitrocellulose, blocked with 5% nonfat dry milk in Tris-buffered saline, probed with specific IGFBP-3 antibodies, and detected using a peroxidase-linked enhanced chemiluminescence detection system (Pierce).
Western Ligand Blot with 125 I-IGFBP-3 (Reverse Western Ligand Blot)-PC-3 cell lysates and plasma membrane fractions were electrophoresed through 12.5% nonreducing SDS-PAGE overnight at constant voltage and electroblotted onto nitrocellulose, blocked with 1% bovine serum albumin in Tris-buffered saline, and incubated with 5 ϫ 10 4 cpm of 125 I-IGFBP-3 (DSL Webster, TX) for 12 h. The membranes were exposed to film for 3 days and visualized by autoradiography.
Densitometric and Statistical Analysis-Densitometric measurement of immunoblots were performed using a Bio-Rad GS-670 Imaging densitometer (Bio-Rad, Melville, NY). Protein levels were estimated by comparing the optical density of each specific protein band from control (SFM) conditions with that from the TGF-␤1-treated conditions. All experiments were repeated at least three times. When applicable, means Ϯ S.E. are shown. Student's t tests were used for statistical analysis. qualitative (DNA laddering and TUNEL) and quantitative (FACS and ELISA) methods. The apoptotic DNA cleavage in cells treated with IGFBP-3 (500 ng/ml for 72 h) was visualized as DNA laddering (Fig. 1A). The DNA extracted from cells treated with SFM showed limited fragmentation (lane 4). However, DNA extracted from IGFBP-3-treated cells demonstrated significant fragmentation, with bands varying in size primarily from 100 to 300 base pairs (lane 3). Cells treated with 10% fetal bovine serum were used as a negative control to demonstrate the absence of any fragmented DNA (lane 2).

Induction of Apoptosis in
As an alternative method of detection, and to localize the apoptotic cells in situ, we detected the fragmented DNA in monolayer cell cultures treated with SFM or IGFBP-3 using TUNEL (Fig. 1B). The DNA fragments bound to the peroxidase-diaminobenzidine reaction product in apoptotic cells were visualized as dark brown cells. Cells in SFM displayed an insignificant number of apoptotic cells (Fig. 1B, i); however, IGFBP-3 treatment revealed numerous apoptotic cells (Fig. 1B, ii). This method was not used to quantitate the number of apoptotic cells in the SFM-and IGFBP-3-treated conditions, since many of the apoptotic cells, after 72 h of incubation, were found floating in the conditioned media. Loss of cells from the culture plate due to the increased apoptotic index was seen as empty spaces in the IGFBP-3-treated condition. However, the control condition showed confluent cells.
FACS analysis of fragmented DNA antibody staining (Fig.  1C) further demonstrated and quantitated the apoptotic index in SFM-and IGFBP-3-treated conditions. The basal apoptotic index due to serum withdrawal was 1.4% of the total number of cells analyzed. The addition of IGFBP-3 to SFM resulted in the significant increase in the apoptotic index (94.7% of the total) as measured by fragmented DNA antibody staining. Using related FACS approaches (Fig. 1D)  der the treated condition and the ratio of apoptotic cells to necrotic cells were calculated. Treatment with IGFBP-3 resulted in a 3-fold increase in apoptotic cells (Fig. 1E) when compared with SFM treatment. The percentage of necrotic cells was low in both treatment conditions and was not increased by IGFBP-3.
Effects of IGFs on IGFBP-3-induced Apoptosis-Similar to the observations obtained using FACS, quantitative analysis with photometric ELISA (Fig. 2) revealed a basal level of apoptosis in SFM-treated cells. In contrast, the serum-treated cultures were completely devoid of apoptotic cells (Fig. 2A). In addition, the addition of IGF-I (200 ng/ml) also prevented the effect of serum starvation on apoptosis. On the other hand, the addition of IGFBP-3 to the SFM induced a further significant increase (p Ͻ 0.001 compared with SFM) in the apoptotic index above the basal level caused by serum deprivation (Fig. 2A). This induction of apoptosis by IGFBP-3 was as potent as the apoptosis induced by the ionophore valinomycin, which has previously been demonstrated to be a potent apoptosis-inducing agent (27). A dose response study revealed that IGFBP-3 induced apoptosis at concentrations as low as 50 ng/ml and demonstrated a dose response up to 500 ng/ml (Fig. 2B). This effect was only partially inhibited by exogenous IGF (Fig. 2C) (p Ͻ 0.05 compared with IGFBP-3 treatment) and was not inhibited by the IGF analogue (long R3-IGF-I) that does not bind to IGFBPs (Fig. 2C), suggestive of an IGF-independent mechanism for this IGFBP-3 effect. All of these methods confirmed that treatment with IGFBP-3 for 72 h resulted in a significant increase in the apoptotic index in PC-3 cells.
Activation of the ICE Pathway by IGFBP-3-Analyses using the apoptosis ELISA demonstrated that IGFBP-3-induced apoptosis was inhibited by the reversible ICE-I (Ac-Tyr-Val-Ala-Asp-aldehyde) (Fig. 3A). The ICE-I completely suppressed IGFBP-3-induced apoptosis in a dose-dependent fashion at concentrations ranging from 0.4 to 5 mol/liter (Fig. 3B). These concentrations are known to inhibit ICE and ICE-like proteases and to block apoptosis induced by a variety of stimuli in other cell types (32)(33)(34). These data demonstrate the involvement of ICE or ICE-like proteases in the IGFBP-3-induced pathway.
Demonstration of IGFBP-3 Association Proteins/Receptors in PC-3 Cells-Detection of IGFBP-3-binding molecules using reverse Western ligand blots revealed a number of bands varying in size from 18 to 150 kDa that represent proteins with high affinity to 125 I-IGFBP-3 (Fig. 4). These molecules were detected both in whole cell lysates (lane 1) and in the purified plasma membrane fraction (lane 2). However, the 150-, 68-, and 18-kDa bands were strongly enriched in the membrane fraction, while some bands (44 and 35 kDa) were seen more prominently in the cell lysates, suggesting a cytoplasmic or nuclear origin. The selective localization of some of these molecules in the membrane fraction suggests the possibility of these proteins serving as IGFBP-3 cell surface receptors that may mediate IGFBP-3 action.
IGF-independent Effects of IGFBP-3-The possibility that IGFBP-3 acts to induce apoptosis independently of IGFs and IGF receptors was investigated by testing the ability of IGFBP-3 to induce apoptosis in the IGF receptor-negative (R (Ϫ) ) fibroblast cells derived from an IGF-1R knockout mouse (25). These cells have been shown previously to neither bind nor respond to IGFs. To test the effect of IGFBP-3 on these cells, we used both treatment with exogenous IGFBP-3 protein and transfection with the IGFBP-3 gene (Fig. 5). The R (Ϫ) cell line demonstrated a basal level of apoptosis when cultured in 30% serum. The DNA extracted from R (Ϫ) cells and R (Ϫ) cells transfected with IGFBP-3 (R (Ϫ) /BP-3) grown in 30% serum for 72 h (Fig. 5A) reveals that the DNA fragmentation was far more prevalent in R (Ϫ) /BP-3 (Fig. 5A, lane 3). This observation was also quantitated using photometric ELISA (Fig. 5B). The transfection of the IGFBP-3 gene resulted in a substantial increase in the degree of apoptosis (p Ͻ 0.001). The addition of exogenous IGFBP-3 (500 ng/ml) also significantly increased the

Demonstration of the Role of IGFBP-3 in TGF-␤1-induced
Apoptosis-Since TGF-␤1 is known to induce apoptosis in some cells and also to up-regulate IGFBP-3 expression in similar cells, we examined its relation to IGFBP-3-induced apoptosis. At a concentration of 1 ng/ml, TGF-␤1 induces apoptosis in PC-3 cells. We recorded the changes in the apoptotic index after treatment with TGF-␤1 both qualitatively (TUNEL; Fig. 6A) and quantitatively (ELISA; Fig. 6B). The in situ localization of apoptotic PC-3 cells using TUNEL revealed an increase in the number of cells with fragmented DNA (Fig. 6A, i, arrow) in TGF-␤1-treated cells compared with those grown in SFM (Fig.  5A, ii). Quantitative analyses by ELISA (Fig. 6B) revealed a basal level of apoptosis in SFM, the suppression of this basal level by addition of IGF-I, and a significant level of apoptosis induced by TGF-␤1. This induction of apoptosis by TGF-␤1 was 95% as potent as the apoptosis induced by the ionophore, valinomycin. In addition, using the photometric ELISA, we compared and quantitated the apoptotic index induced by 1 ng/ml TGF-␤1 and 500 ng/ml IGFBP-3 under similar conditions. When compared with serum-free conditions, both IGFBP-3 and TGF-␤1 demonstrated a significant increase in the apoptotic index (p Ͻ 0.001).
As shown in the inset of Fig. 7, immunoblotting the conditioned media from TGF-␤1-treated PC-3 cells using IGFBP-3specific antibodies revealed a dramatic (Ͼ10-fold) elevation (p Ͻ 0.0001 compared with SFM) of the 40 -44-kDa IGFBP-3 protein by 12 h after treatment. However, the increase in the apoptotic index after TGF-␤1 treatment was observed later, at 18 -24 h (data not shown).
To test whether TGF-␤1-induced apoptosis is mediated through IGFBP-3, we treated PC-3 cells with TGF-␤1 concomitantly with IGFBP-3 sense or antisense oligonucleotides or with IGFBP-3-specific, affinity-purified, neutralizing antibod-ies (Fig. 8). IGFBP-3 and TGF-␤1 induced apoptosis as shown above. The IGFBP-3 antisense oligomer effectively blocked the TGF-␤1-induced apoptosis in PC-3 cells (p Ͻ 0.001 compared with TGF-␤1 treatment), suggesting that TGF-␤1 induces apoptosis by increasing IGFBP-3 expression. The sense IGFBP-3 oligomer had no such effect. In addition, specific neutralizing antibodies to IGFBP-3 also blocked (p Ͻ 0.001) TGF-␤1induced apoptosis in these cells. No blocking effect was observed with the addition of control IgG. These results suggest that the IGFBP-3 has to be secreted and presumably bind to its receptor before it induces apoptosis.

DISCUSSION
The role of IGFBP-3 as a growth-inhibitory protein has been previously demonstrated by us and others in various cell types (9,(12)(13)(14)(15)(16)(17). Initially, IGFBP-3 was thought to inhibit growth by binding to IGFs and sequestering them from their receptor. Later, the cell growth-inhibitory effect of IGFBP-3 was suggested to also be IGF-independent and to involve cell growth arrest (14,15). Recently, this inhibitory effect of IGFBP-3 was suggested to be mediated by interaction with a putative IGFBP-3 receptor. Although the IGF-independent growth-inhibitory role of IGFBP-3 has been recently investigated, an apoptosis-inducing role for IGFBP-3 has not been previously determined. This is the first demonstration of IGFBP-3 as a cell death-promoting agent.
Partial blocking of IGFBP-3-induced apoptosis by IGF suggests two possibilities. First, the presence of IGF may prevent the cells from undergoing apoptotic changes through the IGF receptor-mediated cell survival pathway. Second, some of the IGFBP-3 would not be available to induce apoptosis through its own receptors, since it formed IGF⅐IGFBP-3 complexes. Furthermore, the inability of IGF to fully block IGFBP-3-induced apoptosis even at a 5-fold higher molar concentrations supports the notion that the pathway of IGFBP-3-induced apoptosis may not always involve IGF and IGF receptor. In addition, since IGF analogues that do not bind IGFBP-3 did not reverse the IGFBP-3 effect at all, this further suggests that IGFBP-3 induces apoptosis via an IGF-independent pathway through an IGFBP-3 receptor. This IGFBP-3 cell surface receptor has been first proposed in Hs578T breast cancer cells by affinity crosslinking of 125 I-IGFBP-3 to cell membrane and cell lysate extracts (18). In this study, we have shown that PC-3 cells also bind IGFBP-3 and that several potential IGFBP-3 receptors exists in PC-3 cells.
IGFs have been shown to protect cells from undergoing apoptosis through an IGF receptor-mediated cell survival pathway (35)(36)(37)(38). Both the effects of decreases in the number of IGF receptors causing massive apoptosis and the overexpression of IGF receptors protecting cells from apoptosis have been demonstrated in vivo (35). The roles of IGFs and the IGF receptors as autocrine survival factors (36) and as protective agents that prevent apoptosis induced by other agents such as etoposide have been shown extensively (37). Mutant versions of p53 protein, commonly associated with malignant states, have been shown to derepress the IGF receptor promoter, with ensuing mitogenic activation by locally produced or circulating IGFs (38). All of the above mentioned studies indicate the important role of IGFs and IGF receptors in preventing cells from undergoing apoptosis through a cell survival pathway. We demonstrated here an alternate pathway for the induction of apoptosis that is independent of these apoptosis-protecting agents. By demonstrating IGFBP-3-induced apoptosis in the IGF receptornegative (R (Ϫ) ) murine fibroblast cell line, we proved our hypothesis that IGFBP-3 may induce apoptosis independently of the IGF receptor-mediated survival pathway. Therefore, the ratio of free IGFs and IGFBP-3 will regulate cell growth not only by balancing the rate of cell proliferation and cell growth arrest, but also by regulating the rate at which the cells might be induced to undergo apoptosis.
The apoptosis-inducing effect of IGFBP-3 in R (Ϫ) cells provides ample evidence to suggest that similar IGF receptorindependent pathways are present in PC-3 cells and possibly in other cell lines. Treatment with IGF-I partially decreased the incidence of apoptosis in IGFBP-3-overexpressing cells but did not have any effect on R (Ϫ) cells, suggesting that the partial suppression of apoptosis by IGF is through the formation of IGF⅐IGFBP-3 complexes. Similar to the results found in PC-3 cells, IGFBP-3-neutralizing antibodies partially decreased the degree of apoptosis in IGFBP-3-overexpressing R (Ϫ) cells. In PC-3 cells, IGF-I partially blocked IGFBP-3-induced apoptosis, but the IGF analogue, which binds to the IGF receptor and not to IGFBP-3, was unable to block IGFBP-3-induced apoptosis. These observations not only suggest the involvement of an IGF-independent pathway, but they also demonstrate that IGFBP-3 must be free of IGF to be able to bind to its receptor and initiate its effect on cell death and that the activation of the IGF receptor does not protect cells from IGFBP-3-induced apoptosis.
The expression of the cell growth-inhibitory IGFBP-3 has been shown to be induced by various apoptosis-inducing agents, such as TGF-␤1 (20 -22), retinoids (21), TNF-␣ (23), and the tumor suppressor gene p53 (24). IGFBP-3 has been previously shown to mediate the growth-inhibitory effect of both retinoic acid and TGF-␤1 (20,21). However, the mechanism by which the IGFBP-3 reduces the cell number, under these conditions, is not known. In this work, we have demonstrated that IGFBP-3 mediates the growth-inhibitory effect of TGF-␤1 by inducing apoptosis. This may apply to other agents that have not yet been investigated.
The PC-3 cells are p53-negative (39) and have the machinery to express low levels of IGFBP-3 (8) under serum-free conditions. TGF-␤1 is a potent growth inhibitor of epithelial cells and has been shown to induce apoptosis and down-regulate Bcl-2 expression (40,41). The dramatic elevation of the 44-kDa IGFBP-3 protein within 12 h of TGF-␤1 treatment and the significant effect of TGF-␤1 on apoptosis that was observed about 18 -24 h after treatment suggest that the TGF-␤1-induced elevation of IGFBP-3 protein in the conditioned media is the primary signal that activated apoptosis in this cell line. Blocking TGF-␤1-induced apoptosis at the IGFBP-3 transcriptional level confirmed the role of IGFBP-3 as the mediator of TGF-␤1-induced apoptosis in PC-3 cells. Co-treatment with IGFBP-3 antisense (but not sense) thiolated oligonucleotide and TGF-␤1 verified the role of IGFBP-3 in the TGF-␤1-induced apoptosis. Furthermore, neutralization of IGFBP-3 action in TGF-␤1-treated cells with IGFBP-3-neutralizing antibodies (but not control IgG) confirmed that IGFBP-3 must be secreted and allowed to bind to its receptor to initiate apoptosis. The latter observation also confirms that the TGF-␤1mediated increase in IGFBP-3 transcription must pass through steps such as IGFBP-3 secretion and the binding of this protein to its receptor to initiate apoptosis.
Inappropriate expression of genes involved in cell proliferation has been shown to alter regulation of apoptosis. Both Bcl-2, which promotes cell survival, and Bax, which promotes cell death, have been implicated as major mediators in the control of apoptotic pathways, and it has been suggested that the ratio of Bcl-2 to Bax controls the relative susceptibility of cells to death stimuli. TGF-␤1, retinoic acid, TNF-␣, and p53 are known to induce apoptosis by regulating Bcl-2 and Bax expression (40 -47). Since all of these apoptosis-inducing agents also induce IGFBP-3 expression, we anticipate that IGFBP-3-induced apoptosis may also involve regulation of the Bcl-2:Bax ratio. In addition, the expression of ICE or ICE-like proteases that are final mediators of the apoptosis pathway is involved in the mechanism of action of IGFBP-3 as well as the above agents.
The role of IGFBP-3 in mediating p53 effects was proposed when p53 was demonstrated to activate the IGFBP-3 promotor (24). Recently, it has been shown that mutants of p53 that have lost the ability to activate IGFBP-3 and Bax expression but maintained their activation of the cyclin-dependent kinase inhibitor p21 are able to induce cell cycle arrest but are unable to induce apoptosis (48). Furthermore, a p53 mutant that activates Bax expression but only partially activates the IGFBP-3 promotor is only partially effective in inducing apoptosis (49). Thus, a p53-dependent role of IGFBP-3 has been previously demonstrated. By demonstrating IGFBP-3-induced apoptosis in PC-3 cells that lack the p53 gene, we have demonstrated that IGFBP-3 can also induce apoptosis in a p53-independent fashion.
We present a hypothesis based on the results from this study and other previous reports from this and other groups in the diagrammatic representation shown in Fig. 9. We propose that the independent and interdependent effects of IGFs and IGFBPs on the regulation of cell number involve two pathways that interact at several levels. IGFs mediate survival via the IGF receptor. IGFBP-3 is able to block this pathway by sequestering IGFs away from the IGF receptor. IGFBP-3 mediates apoptosis via its own receptors, while IGFs can prevent this effect by binding to IGFBP-3. Thus, IGFBP-3 can mediate cell death by both IGF-dependent and IGF-independent pathways.
Both normal cell growth (50) and various pathologies associated with neoplastic cell proliferation, such as breast cancer (20,26,51,52), prostate cancer, and benign prostatic hyperplasia (11), are also associated with altered expression of IGFs and IGFBPs. Earlier observations, however, did not directly demonstrate a role for IGFBP-3 in inducing apoptosis but provided ample evidence to suggest that IGFBP-3 is important in regulating cell number in such situations. Our data demonstrate that IGFBP-3 induces apoptosis at physiological concentrations and that IGFBP-3 may act through an IGF⅐IGF receptor-independent pathway. IGFBP-3 mediates the induction of apoptosis by TGF-␤1 and may mediate similar actions of other growth-regulatory factors.