Insulin-like growth factor-binding protein-5 inhibits the growth of human breast cancer cells in vitro and in vivo.

The role of insulin-like growth factor-binding protein (IGFBP)-5 in human breast cancer cell growth is unclear. We determined the effects of IGFBP-5 expression on the growth of human breast cancer cell lines in vivo and in vitro. Expression of IGFBP-5, both by stable transfection and adenoviral-mediated infection, was inhibitory to the growth of MDA-MB-231 and Hs578T human breast cancer cells over a 13-day period. IGFBP-5 expression resulted in a G2/M cell cycle arrest and the induction of apoptosis in both cell lines, an effect that was abrogated in the presence of the broad-spectrum caspase inhibitor, z-VAD-fmk. IGFBP-5-induced apoptosis was associated with a transcriptional increase in expression of the proapoptotic regulator bax and decrease in the anti-apoptotic bcl-2 compared with vector controls. Secreted IGFBP-5 when added exogenously to breast cancer cells was not internalized and had no effect on cell growth or apoptosis, suggesting that IGFBP-5 may elicit its inhibitory effects via a novel, intracrine mechanism. In athymic nude mice, stable expression of IGFBP-5 significantly inhibited both the formation and growth of tumors derived from MDA-MB-231 cells. IGFBP-5-expressing tumors also had a significantly elevated level of bax mRNA and decreased levels of bcl-2 mRNA compared with vector tumors. These data suggest that IGFBP-5 is a potent growth inhibitor and proapoptotic agent in human breast cancer cells via modulation of cell cycle regulation and apoptotic mediators.

The role of insulin-like growth factor-binding protein (IGFBP)-5 in human breast cancer cell growth is unclear. We determined the effects of IGFBP-5 expression on the growth of human breast cancer cell lines in vivo and in vitro. Expression of IGFBP-5, both by stable transfection and adenoviral-mediated infection, was inhibitory to the growth of MDA-MB-231 and Hs578T human breast cancer cells over a 13-day period. IGFBP-5 expression resulted in a G 2 /M cell cycle arrest and the induction of apoptosis in both cell lines, an effect that was abrogated in the presence of the broad-spectrum caspase inhibitor, z-VAD-fmk. IGFBP-5-induced apoptosis was associated with a transcriptional increase in expression of the proapoptotic regulator bax and decrease in the anti-apoptotic bcl-2 compared with vector controls. Secreted IGFBP-5 when added exogenously to breast cancer cells was not internalized and had no effect on cell growth or apoptosis, suggesting that IGFBP-5 may elicit its inhibitory effects via a novel, intracrine mechanism. In athymic nude mice, stable expression of IGFBP-5 significantly inhibited both the formation and growth of tumors derived from MDA-MB-231 cells. IGFBP-5-expressing tumors also had a significantly elevated level of bax mRNA and decreased levels of bcl-2 mRNA compared with vector tumors. These data suggest that IGFBP-5 is a potent growth inhibitor and proapoptotic agent in human breast cancer cells via modulation of cell cycle regulation and apoptotic mediators.
The insulin-like growth factor-binding proteins (IGFBPs) 1 are a family of proteins that bind with high affinity to IGFs and modulate their mitogenic actions by regulating their ability to interact with their signaling receptor, the type I IGF receptor (IGFRI, reviewed in Ref. 1). However, it is now becoming clear that many IGFBPs have direct roles in the regulation of cell growth and cell death. For example, we and others have demonstrated the antiproliferative and proapoptotic effects of IGFBP-3 in breast (2)(3)(4) and prostate (4) cancer cell lines.
The role of IGFBP-5 in cell growth is complex, with reports that it can either stimulate or inhibit cell proliferation in various experimental systems (5)(6)(7). There is also evidence that IGFBP-5 expression is up-regulated by antiproliferative agents such as retinoic acid (8), vitamin D-related compounds (9), and antiestrogen ICI 182780 (10), with some evidence that it may mediate their growth inhibitory effects (8,10). Similarly, growth stimulation of human breast cancer cells by estradiol is associated with a down-regulation of IGFBP-5 expression (10), although exogenous IGFBP-5 had no effect on IGF-I-stimulated DNA synthesis in the breast cancer cell line, MCF-7 (11). However, there are some conflicting data in vivo, with Chan et al. (12) reporting a decrease in IGFBP-5 gene expression following growth inhibition of the rat mammary gland with ICI 182780, whereas Huynh et al. (10) observed an ICI-induced increase in IGFBP-5 mRNA in 9,10-dimethyl-1,2-benzanthracene-induced mammary tumors. These results suggest that IGFBP-5 has cell type-specific effects on cellular proliferation.
The role of IGFBP-5 in the apoptotic process is also somewhat ambiguous, with conflicting data from in vivo and in vitro studies. In vivo, increased expression of IGFBP-5 has been observed in tissues undergoing apoptosis such as the involuting prostate (13) and mammary (14,15) glands, atretic ovarian follicles (16), and in the rat brain following hypoxic-ischemic injury (17). Allan et al. (18) have also reported that IGFBP-5 expression is highly restricted to regions of cell death in the developing mouse limb bud. In vitro, expression of IGFBP-5 mRNA increases following induction of apoptosis in ovine granulosa cells (19). However, Perks et al. (20) have reported opposing effects of IGFBP-5 in human breast cancer cells in vitro. They demonstrated that addition of exogenous IGFBP-5 to Hs578T cells protects these cells from ceramide-induced apoptosis, suggesting IGFBP-5 may have a survival function in response to apoptotic stimuli (20,21). A similar conclusion was reached by Roschier et al. (22) following their demonstration that induction of apoptosis in cerebellar granule cells is associated with a down-regulation of IGFBP-5 mRNA and protein expression, an effect that is reversed by IGF-I (22). Furthermore, castration-induced IGFBP-5 expression enhances the mitogenic and antiapoptotic effects of IGFs and accelerates progression to androgen independence in prostate cancer models (23).
IGFBP-5 is expressed in breast cancer tissue (24) and cell lines (25) and is positively correlated with estrogen receptor status. Because of the contradictory data on the role of IGFBP-5 in cancer cell growth, we directly examined the effect of stable expression of IGFBP-5 on the growth of human breast cancer cells in vitro and in vivo. Furthermore, using a transient overexpression system, we have characterized the proapoptotic effects of IGFBP-5 in human breast cancer cells. We show that IGFBP-5 significantly inhibits the growth of human breast cancer cells both in vitro and in vivo, and induces a caspasedependent, intrinsic apoptotic pathway. Interestingly, these growth effects appear to be mediated by intracellular IGFBP-5, independent of signaling through a cell-surface receptor.

EXPERIMENTAL PROCEDURES
Cell Culture of Breast Cancer Cells-The human breast cancer cell lines MDA-MB-231 and Hs578T were routinely maintained in RPMI 1640 supplemented with 10% FCS, 10 g/ml insulin, and 2.92 mg/ml glutamine under standard conditions. Stable Transfection-A 0.9-kb IGFBP-5 cDNA, cloned from U2-OS osteosarcoma cells, was kindly provided by Dr. Sue Firth, Kolling Institute (26). MDA-MB-231 cells were stably transfected with the IGFBP-5 cDNA in the expression vector pOP13 (Invitrogen), using LipofectAMINE (Invitrogen) according to the manufacturer's protocol. IGFBP-5 transfectants (MDA/BP-5) and vector controls (MDA/VEC) were selected in media containing 800 g/ml geneticin for 21 days post-transfection, then mixed populations of transfectants were grown up for subsequent experiments.
Adenoviral-mediated Expression of IGFBP-5-The human IGFBP-5 cDNA was subcloned into the adenoviral shuttle vector, pAd-Track-CMV, and replication-deficient IGFBP-5-expressing adenoviruses were produced essentially as previously described (27). Proliferating cultures of MDA-MB-231 and Hs578T were incubated with varying concentrations of adenovirus stock (approximately 4.5 ϫ 10 12 plaque forming units/l) for 6 h in serum-free, insulin-free (SF) media. Virus was then removed and replaced with fresh growth media. As the pAd-Track-CMV vector contains a green fluorescent protein marker, levels of infection were determined by assessing the percentage of green fluorescent protein-positive cells using flow cytometry. For the cell proliferation assays, cells were infected with 0.1 l of adenovirus stock. For analysis of apoptosis induction, Hs578T cells were infected with 2 l of adenovirus stock, and MDA-MB-231 cells were infected with 1 l of adenovirus stock. These concentrations were shown to give Ͼ90% infection efficiency and low toxicity as assessed by flow cytometry (data not shown).
Measurement of Cell Proliferation-Cells were seeded at 1 ϫ 10 4 cells/well in 12-well plates in the presence of 10% FCS media. On days 3, 6, 9, and 13 post-seeding, cells were trypsinized and the viable cell numbers were assessed by trypan blue exclusion.
Analysis of Cell Cycle Distribution-Cells were plated at 5 ϫ 10 5 per well in 6-well plates for 24 h, then incubated in SF media for 48 h. Cells were trypsinized and aliquots of 1 ϫ 10 6 cells were fixed in 70% ethanol at Ϫ20°C for at least 24 h. Fixed cells were then washed and suspended in 1 ml of fluorochrome solution (50 g/ml propidium iodide, 1 mg/ml RNase A) for at least 1 h in the dark at 4°C. Cell cycle analysis was performed using a Coulter ELITE flow cytometer (Coulter, Hialeah, FL). 20,000 cells were analyzed for each sample, and quantitation of cell cycle distribution was performed using Multicycle software (Phoenix Flow Systems, San Diego, CA).
Clonogenic Survival Assays-The long term survival of MDA-MB-231 transfectants following doses of irradiation was assessed by a clonogenic survival assay. 2 ϫ 10 3 cells were seeded into 6-well plates in triplicate for 24 h. Cells were then washed with fresh media and irradiated with various doses of x-rays (2.5, 5, or 7.5 gray), then incubated for 14 days in 10% FCS. At this time, cells were counted and the percentage survival determined by the proportion of attached cells surviving (assessed by trypan blue exclusion) relative to unirradiated controls.
Measurement of DNA Fragmentation by Flow Cytometry-Both floating and adherent cells were analyzed for induction of apoptosis by flow cytometry. 24 h post-adenoviral infection, cells were rinsed with fresh SF media, then incubated for 48 h with or without the caspase-inhibitor z-VAD-fmk (100 M; Bachem AG, Switzerland). Floating and attached cell populations were combined and 1 ϫ 10 6 cells were prepared and analyzed as described above. Labeled nuclei were gated on light scatter to remove debris and the percentage of nuclei with a sub-G 1 content was determined.
Apoptosis Assay-Cells were plated at 5 ϫ 10 5 per well on sterile glass coverslips in 6-well plates for 24 h then infected with IGFBP-5-or vector-adenovirus. 24 h post-adenoviral infection, cells were washed with SF media and incubated for 48 h. Cell monolayers were rinsed with phosphate-buffered saline, fixed in ice-cold methanol for 10 min, and then stained with 0.8 g/ml 4,6-diamidino-2-phenylindole (Sigma). The percentage of apoptotic cells was determined microscopically as cells with visible nuclear fragmentation.
[ 3 H]Thymidine Incorporation-For analysis of DNA synthesis, MDA-MB-231 and Hs578T cells were plated in 48-well plates at 5 ϫ 10 4 /well and infected with adenovirus as described above. DNA synthesis was determined using incorporation of [ 3 H]thymidine essentially as previously described (28) 24 h post-infection, or following addition of 20 or 100 ng/ml recombinant human (rh) IGF-II or 10 g/ml IGFRI neutral-izing antibody (␣IR-3, Calbiochem-Novabiochem, Alexandria, New South Wales, Australia) as indicated for 24 h in SF media.
Immunoblot Analysis-Total cell lysates were prepared from cells 48 h post-seeding for stable transfectants or post-infection with adenovirus following an incubation in SF media and 50 g were resolved under reducing conditions on 12% SDS-polyacrylamide gels using standard methods. For analysis of Bax expression post-irradiation, cells were irradiated with various doses of x-rays (2.5, 5, or 7.5 gray), then incubated for 4 days in 10% FCS before preparation of total cell lysates as described above. For IGFBP-5 immunoblots, 50 g of conditioned media (CM) samples were electrophoresed under non-reducing conditions. Resolved proteins were transferred to nitrocellulose membranes and probed with anti-Bax polyclonal (1:1000 dilution; BD Pharmingen), anti-Bcl-2 monoclonal (0.8 g/ml; Dako, Carpinteria, CA), or chicken anti-human IGFBP-5 (1:100; Ref. 29) antibodies overnight at 4°C. This was followed by incubation with anti-rabbit (for Bax, 1:10,000 dilution) or anti-mouse (for Bcl-2, 1:2000 dilution) IgG conjugated with horseradish peroxidase. Immunoreactive protein bands were visualized with ECL (Pierce). Blots were checked for equal loading by reprobing with anti-␣-tubulin antibody (1:10,000, Sigma). Protein bands were quantitated by densitometry (Video Densitometer model 620, Bio-Rad). For IGFBP-5 immunoblots, membranes were incubated with anti-chicken IgY secondary antibody (1:5000 dilution; Promega, Madison, WI) conjugated with alkaline phosphatase and visualized with colorimetric detection. Purity of cytoplasmic and nuclear extracts ( Analysis of IGFBP-5 in Conditioned Media-MDA-MB-231 and Hs578T cells were infected with adenovirus as described above, then incubated in SF media for 48 h. CM was collected and IGFBP-5 was purified by binding to, and elution from, an affinity column of IGF-Iagarose, followed by reverse-phase high performance liquid chromatography on a Jupiter 5, 300-Å C18, 250 ϫ 4.6-mm column (Phenomenex, Torrance, CA). IGFBP-5 was eluted with a gradient of acetonitrile in 0.1% trifluoroacetic acid, 15-24% over 10 min, then 24 -33% over 20 min. IGFBP-5 fragments eluted at 26 -27 min, and intact IGFBP-5 at 30 -32 min. Concentrations of IGFBP-5 in CM from IGFBP-5 expressing cells were quantitated by IGFBP-5-specific radioimmunoassay as previously described (29).
Ligand Blot-To examine intracellular IGFBP-5, cells were infected with adenovirus or treated with exogenous IGFBP-5, then incubated for 48 h in SF media. Cytoplasmic and nuclear extracts were prepared using NE-PER Nuclear and Cytoplasmic Extraction Reagents (Pierce) following the manufacturer's instructions. Protein content was estimated and samples were electrophoresed (30 g for nuclear samples from stable transfectants, 50 g for all other samples) and transferred to nitrocellulose membranes, then probed with 125 I-IGF-II (1 ϫ 10 6 cpm/50 ml) overnight.
Northern Analysis-RNA extraction and Northern analysis were performed essentially as previously described (32). The IGFBP-5 cDNA probe was kindly provided by Dr. S. Shimasaki (University of California San Diego, La Jolla, CA) and the bax cDNA probe was kindly provided by Dr. G. Packham (Kolling Institute of Medical Research, St. Leonards, NSW, Australia). Probes were labeled using a HexaLabel DNA labeling kit (MBI Fermentas, Vilnius, Lithuania) and [␣-32 P]deoxy-CTP. Filters were hybridized at 42°C overnight, then washed in 1ϫ SSC (standard saline citrate) at 42°C, followed by a wash in 0.1ϫ SSC heated to 42°C. The ribosomal phosphoprotein, 36B4, was used as a loading control (33). Filters were quantitated using a phosphorimaging FLA3000 (Fuji, Japan). Data were normalized by expressing the ratio of target mRNA to 36B4 mRNA.
Real Time Quantitative PCR-Total RNA was isolated and reverse transcribed using oligo(dT) 17 and SuperScript II reverse transcriptase polymerase as previously described (2). Real time quantitative PCR was performed using the Bio-Rad iCycler in combination with SYBRgreen dye and a final concentration of 400 nM gene-specific primers: bcl-2 F, 5Ј-GTTCGGTGGGGTCATGTGTGTGGAGAGCG; bcl-2 R, 5Ј-TAGCTG-ATTCGACGTTTTGCCTGA; bax F, 5Ј-GGGGACGAACTGGACAGTAA; bax R, 5Ј-CAGTTGAAGTTGCCGTCAGA. Human actin primers were used as a control for quantitative analysis.
Tumor Growth in Vivo-MDA/VEC and MDA/BP-5 cells were trypsinized and resuspended at 1 ϫ 10 7 viable cells/100 l in SF media. Cell suspensions were mixed with an equal volume of Matrigel (BD Biosciences, Bedford, MA) as a support matrix for initial cell growth. 1 ϫ 10 7 cells were injected subcutaneously at the dorsal neck into two groups (MDA/VEC and MDA/BP-5) of six mice (in the repeat experi-ment, seven mice were injected with MDA/BP-5 cells). Tumor growth was monitored weekly by measuring tumor length and width using calipers, and tumor volume (cm 3 ) was calculated using the formula: length ϫ width 2 ϫ 0.5. Experiments were terminated when tumors reached ϳ1 cm 3 . Blood samples were taken from all animals at termination and tumors were frozen in liquid nitrogen for RNA analysis. Results were pooled from two separate experiments and all experiments were carried out with the approval of the Institutional Animal Care and Ethics Committee.
Statistical Analysis-Statistical analysis was carried out using Stat-View 4.02 (Abacus Concepts, Inc., Berkeley, CA). Differences between groups was evaluated by Fisher's protected least significant difference test after analysis of variance using repeated measures or factorial analysis where appropriate.

Stable and Adenoviral-mediated Expression of IGFBP-5 Is
Growth Inhibitory to Human Breast Cancer Cells-The effects of IGFBP-5 on the growth of human breast cancer cells were examined using both stable transfection of human IGFBP-5 cDNA as well as transient adenoviral-mediated expression. Cell growth was assessed by cell counting over a 13-day period, and concentrations of IGFBP-5 secreted into the media were also determined at each time point. Secretion of IGFBP-5 remained high throughout the 13-day growth period and was low or undetectable in vector control cells (data not shown). Fig. 1A shows the effects of stable expression of IGFBP-5 on the growth of MDA-MB-231. MDA/BP-5 cells were significantly growth inhibited over the 13-day growth period compared with vector controls (p Ͻ 0.03 by repeated measures analysis of variance). Interestingly, despite much higher levels of IGFBP-5 secretion, a similar inhibitory effect was observed following adenoviralmediated expression of IGFBP-5 in MDA-MB-231 (Fig. 1B) and Hs578T (Fig. 1C). In both cell lines, transient overexpression of IGFBP-5 significantly inhibited cell growth compared with vector controls (p Ͻ 0.0005 and p Ͻ 0.005, respectively, by repeated measures analysis of variance). The effects of IGFBP-5 on cell cycle progression were examined by flow cytometric analysis after a 48-h incubation in SF media. Fig. 2B demonstrates that stable expression of IGFBP-5 in MDA-MB-231 cells resulted in a significant accumulation in G 2 /M phase of the cell cycle compared with vector controls (p Ͻ 0.05). Adenoviral-mediated expression of IGFBP-5 also resulted in a significant increase in the percentage of MDA-MB-231 and Hs578T cells in the G 2 /M phase compared with vector controls (Fig. 2C, p Ͻ 0.005 and p Ͻ 0.02, respectively). A significant decrease in the percentage of cells in G 0 /G 1 was also observed when compared with vector controls (p Ͻ 0.03 and p Ͻ 0.01, respectively). Interestingly, no significant differences in the S phase fraction were observed between IGFBP-5-expressing cells and vector controls, which may reflect the relative insensitivity of this method compared with analysis of active DNA synthesis.

IGFBP-5 Expression Induces Caspase-dependent Apoptosis via Modulation of Bcl-2 Proteins-
The effect of IGFBP-5 on the induction of apoptosis was examined in MDA-MB-231 and Hs578T cells using transient adenoviral-mediated expression. The extent of DNA fragmentation characteristic of apoptotic cell death was assessed using both scoring of 4,6-diamidino-2phenylindole-stained nuclei for fragmentation and flow cytometric analysis of the hypodiploid (pre-G 1 ) fraction. Analysis of 4,6-diamidino-2-phenylindole-stained nuclei showed a significant increase in nuclear fragmentation in IGFBP-5-infected MDA-MB-231 (from 4.7 Ϯ 0.5 to 7.6% Ϯ 1.1, p Ͻ 0.0001) and Hs578T (from 4.8 Ϯ 0.7 to 11.2% Ϯ 1.6, p Ͻ 0.0001) cells compared with vector controls (Fig. 3A). Fig. 3B illustrates that expression of IGFBP-5 resulted in a significant increase in the percentage of the population in the pre-G 1 fraction compared with vector controls in both MDA-MB-231 (from 2.7 Ϯ 0.5 to 7.3% Ϯ 1.1, p Ͻ 0.02) and Hs578T (from 2.9 Ϯ 0.8 to 9.0% Ϯ 3.3, p Ͻ 0.006) cells. Incubation of IGFBP-5-expressing MDA-MB-231 and Hs578T cells with the caspase inhibitor z-VAD-fmk resulted in a significant decrease in the level of apoptosis (p Ͻ 0.01 and p Ͻ 0.02, respectively) to levels that were not significantly different from vector controls (Fig. 3B). Levels of apoptosis in vector-infected cells were not significantly affected by z-VAD-fmk (p ϭ 0.2 for MDA, p ϭ 0.9 for Hs578T). The effect of stable IGFBP-5 expression on the long term survival of MDA-MB-231 transfectants following exposure to IR was examined. Cells were irradiated with various doses of x-rays, then radiosensitivity was evaluated using clonogenic survival assays. At doses of 2.5 and 5 gray x-rays, MDA/BP-5 cells showed significantly reduced long term (up to 14 days) survival (p Ͻ 0.0001 and p Ͻ 0.002, respectively) compared with vector controls (Fig. 3C). At the higher dose of 7.5 gray, the surviving fraction was low in both MDA/VEC and MDA/ BP-5 cells.
We have previously shown that IGFBP-3-induced apoptosis and radiosensitivity is associated with modulation of members of the Bcl-2 family of apoptotic regulators (2, 3). We investigated whether the apoptotic effects of IGFBP-5 were mediated via a similar pathway. Fig. 3D shows that IGFBP-5 expression in stable transfectants was associated with elevated levels of Bax protein compared with vector controls, both basally and 4 days after exposure to increasing doses of radiation. This suggests that the increased radiosensitivity of the IGFBP-5-expressing cells might result from Bax-mediated apoptosis.
To further investigate these IGFBP-5-mediated effects, both stable transfectants and adenoviral-infected cells were analyzed for expression of the proapoptotic Bax and antiapoptotic Bcl-2 proteins. Levels of bax and bcl-2 mRNA were also determined in these cell populations. sponsive to IGF-II. Furthermore, IGF-I and IGF-II levels are undetectable in these cells (data not shown), confirming the results of Sciacca et al. (35). To confirm a lack of autocrine IGF action, we examined the levels of [ 3 H]thymidine incorporation in cells incubated with an IGFRI-blocking antibody (␣IR-3) compared with untreated cells. Fig. 5A demonstrates that 10 g/ml ␣IR-3 had no effect on basal DNA synthesis, suggesting that endogenous IGFs are not significantly mitogenic in MDA-MB-231. We have previously demonstrated in IGF-responsive T47D human breast cancer cells, that 10 g/ml ␣IR-3 can reverse the mitogenic effects of 20 ng/ml IGF-II (3). Fig. 5B illustrates that treatment with 10 g/ml ␣IR-3 also had no significant effect on the basal level of apoptosis as determined by flow cytometry (p ϭ 0.17), indicating that IGF activation of IGFRI does not elicit an autocrine survival signal in MDA-MB-231 cells. Hs578T cells have been previously demonstrated to be unresponsive to the mitogenic and anti-apoptotic effects of IGFs (36). These data suggest that the inhibitory and proapoptotic effects of IGFBP-5 in MDA-MB-231 and Hs578T cells are unlikely to be mediated via sequestration of IGFs and are independent of IGFRI signaling.
To determine the proapoptotic ability of these secreted forms of IGFBP-5 on MDA-MB-231 and Hs578T cells, we incubated  1 and 4). Expression of ␣-tubulin was used as a loading control. Immunoblots are representative of two independent experiments.  Fig. 6C shows that there was no significant change in the rate of DNA synthesis following treatment with any of the fractions compared with untreated controls.
These results indicate that the cytostatic and cytotoxic effects that we have observed in IGFBP-5-expressing cells are not initiated by IGFBP-5 or its fragments from outside the cell, suggesting that they may be mediated by intracellular IGFBP-5. We examined the expression of IGFBP-5 in the cytoplasmic and nuclear compartments of both stable transfectants and adenoviral-infected MDA-MB-231 and Hs578T cells by IGF-II ligand blot. Purity of each fraction was confirmed by immunoblot analysis of CPP32 (cytoplasmic, Ref. 30) and poly-(ADP-ribose) polymerase (nuclear, Ref. 31) expression. Fig. 6D illustrates that active IGFBP-5 was expressed in both the nuclear and cytoplasmic fractions of IGFBP-5-expressing cells. Levels of IGFBP-5 expression in Hs578T cells were higher than adenoviral-infected MDA-MB-231 cells, despite these cells being infected with similar efficiency as assessed by green fluorescent protein expression (data not shown). This may explain the more potent proapoptotic effects of IGFBP-5 in these cells. Interestingly, an additional, higher molecular weight band of ϳ60,000 was observed in Hs578T lysates and to a lesser extent in MDA-MB-231 cells, consistent with an IGFBP-5 dimer. IGFBP-5 bands visible in stable vector-transfected MDA-MB-231 cells, representing low-level endogenous expression, result from the long exposure of the blot. However, interestingly, there did not appear to be a significant increase in the expression of nuclear IGFBP-5 in MDA/BP-5 compared with MDA/ VEC, despite the fact that IGFBP-5 expression is proapoptotic in the former cell population. This is consistent with our recent observations that nuclear translocation of IGFBP-3 is not required to mediate its proapoptotic effects (3).
To determine whether the active (i.e. able to bind IGFs) IGFBP-5 detected intracellularly in adenoviral-infected cells had first been secreted and then taken up by the cells, we incubated MDA-MB-231 and Hs578T cells with 30-kDa CMderived IGFBP-5 (Fig. 6A, lanes 1 and 4, respectively) for 48 h, then examined the levels of intracellular IGFBP-5 by ligand blot analysis. Fig. 6E demonstrates that there was no detectable uptake of exogenous IGFBP-5 after 48 h incubation. Faint bands visible in some of the lanes represent low levels of endogenous IGFBPs because of the long exposure of the blot. Use of immunoblot to detect IGFBP-5 was unsuccessful because of multiple nonspecific bands (data not shown).
IGFBP-5 Is Inhibitory to the Growth of MDA-MB-231 Breast Cancer Cells in Vivo-MDA/VEC or MDA/BP-5 cells were injected into groups of 12 or 13 female nu/nu mice, respectively, and tumor size was measured weekly up to 4 weeks postinjection when some tumors had reached ϳ1 cm 3 . The percentage of animals that developed visible tumors was significantly reduced in the MDA/BP-5 group (46.15%) compared with the MDA/VEC group (100%, p ϭ 0.002 by 2 test). Tumors derived from MDA/BP-5 had a significantly decreased growth rate (as determined by tumor volume) compared with MDA/VEC tumors ( Fig. 7A; p Ͻ 0.0001). In addition, there was a significant decrease in the wet tumor weight of MDA/BP-5 tumors compared with MDA/VEC tumors (Fig. 7B, p Ͻ 0.0005).
To confirm that tumors derived from MDA/BP-5 cells maintained expression of IGFBP-5 in vivo, we isolated RNA from the tumors and determined the levels of IGFBP-5 mRNA expression by Northern blot analysis. The transfected human IGFBP-5 cDNA codes for a message of 0.9 kb. A band of this size was detected in all the MDA/BP-5 tumors and not in the vector control tumors (Fig. 7C). We also analyzed RNA from MDA/BP-5 cells as a positive control for IGFBP-5 cDNA expression (Fig. 7C, lane C). Serum samples were analyzed for IGFBP-5 expression using an IGFBP-5-specific radioimmunoassay but levels were too low to be detected by this method.
Inhibitory Effects of IGFBP-5 on the Growth of MDA-MB-231 Cells in Vivo Are Associated with an Induction of bax mRNA and Decrease in bcl-2 mRNA-We investigated the possible mechanism of IGFBP-5-induced growth inhibition in vivo, by examining the mRNA levels of the apoptotic regulators, bax and bcl-2 in MDA/BP-5-and MDA/VEC-derived tumors using real time PCR analysis (Fig. 7D). Levels of mRNA were expressed as a ratio of the control gene actin and as a percentage of the levels in MDA/VEC-derived tumors. There was a significant induction of bax mRNA expression and a significant decrease in bcl-2 mRNA levels in IGFBP-5-expressing tumors compared with vector controls (p Ͻ 0.02, p Ͻ 0.04, respectively).

DISCUSSION
The role of IGFBP-5 in the growth of human breast cancer cells is not well understood, with some studies suggesting it may be growth inhibitory and associated with induction of apoptosis, whereas others demonstrate a survival role in response to apoptotic stimuli. To address this ambiguity, we have examined and characterized the effects of IGFBP-5 on the growth of human breast cancer cells both in vitro and in vivo.
The human breast cancer cell lines MDA-MB-231 and Hs578T both secrete low levels of IGFBP-5 (Ͻ10 ng/ml). Stable transfection of MDA-MB-231 cells with IGFBP-5 cDNA resulted in an increase in secreted levels to ϳ40 ng/ml. Other breast cancer cell lines secrete similar or higher levels than this (e.g. MCF-7 cells secrete ϳ30 ng/ml) emphasizing the relevance of this model in studying the effects of IGFBP-5 on breast cancer cell growth. Interestingly, T47D breast cancer cells secrete much higher levels of IGFBP-5 (ϳ200 ng/ml), yet are tumorigenic and have low basal levels of apoptosis (2). This suggests that some cancer cells may acquire resistance to the cellular effects of IGFBP-5, a phenomenon that we have previously reported to occur for IGFBP-3 (37), and which may involve ras-dependent signaling pathways (28).
Both stable and adenoviral-mediated expression of IGFBP-5 in MDA-MB-231 and Hs578T resulted in a significant inhibition of in vitro cell growth, an arrest in G 2 /M phase of the cell cycle, and a decrease in DNA synthesis. Interestingly, despite much higher levels of IGFBP-5 expression in adenoviral-infected cells compared with stable transfectants, a similar de-gree of growth inhibition was observed. This suggests that the mechanisms that initiate these inhibitory effects can reach a saturation level beyond which further increases in IGFBP-5 have no additional effect. Indeed, even at the lower levels of IGFBP-5 expression in stable transfectants, breast tumor growth in nude mice was ablated. Antiproliferative effects have been observed in cervical carcinoma cells following addition of up to 200 ng/ml exogenous IGFBP-5 (8) and following stable transfection of IGFBP-5 cDNA in osteosarcoma cells (6). In breast cancer cells, endogenous IGFBP-5 has been shown to mediate the growth inhibitory effects of antiestrogens (10) and . Samples (30 g for nuclear samples from stable transfectants, 50 g for all other samples) were probed with 125 I-IGF-II, then exposed to x-ray film overnight for adenoviral-infected samples and for 3 days for samples from stable transfectants. 50 ng of rhIGFBP-5 expressed by 911 retinoblastoma cells (rhBP5) was analyzed as a positive control. Lower panel, purity of cytoplasmic and nuclear extracts was confirmed by immunoblotting with CPP32 and poly(ADP-ribose) polymerase antibodies, respectively. E, MDA-MB-231 and Hs578T cells were incubated with 1 g/ml 30-kDa CM-derived IGFBP-5 for 48 h in SF media or untreated (cont), then cytoplasmic and nuclear extracts were prepared and samples were probed with 125 I-IGF-II as described above. Blots shown are representative of two independent experiments. vitamin D-related compounds (9,38). However, Perks et al. (20) reported that addition of 100 ng/ml exogenous IGFBP-5 had no direct effects on the growth of Hs578T cells. Interestingly, we have previously reported that the antiproliferative effects of IGFBP-3, which shares much structural and functional homology with IGFBP-5, are associated with a cell cycle arrest in the G 1 /S phase (3). This suggests that IGFBP-3 and IGFBP-5, while both inhibitory to breast cancer growth, may influence cell cycle progression via distinct pathways.
IGFBP-5 expression resulted in the induction of a caspasedependent pathway of apoptosis, and reduced the survival of breast cancer cells following the apoptotic stimulus of IR, in association with increased protein levels of the proapoptotic regulator Bax. Furthermore, expression of IGFBP-5 both in vitro and in vivo induced a transcriptional up-regulation of bax and down-regulation of the anti-apoptotic bcl-2, suggesting IGFBP-5 may induce a mitochondrial apoptotic pathway. Surprisingly, changes in Bcl-2 protein were not seen following transient, adenoviral-mediated expression of IGFBP-5, possibly because of the short term nature of this experiment. However, levels of Bax were up-regulated both basally and in re-sponse to radiation, altering the critical ratio of these mediators toward apoptosis (39). We have previously demonstrated that IGFBP-3 also has intrinsic antiproliferative and proapoptotic effects in human breast cancer cells and modulates expression of Bcl-2-like proteins (3). These current results suggest that IGFBP-5 may act along a similar intracellular apoptotic pathway, possibly by direct transcriptional modulation of apoptotic genes. These results are in contrast to studies by Perks et al. (20) in Hs578T cells who showed that, whereas exogenous IGFBP-5 had no direct cytotoxic effects, it inhibited apoptosis induced by ceramide or integrin detachment (21). This may reflect post-translational differences in exogenous and endogenously produced IGFBP-5 (for example, glycosylation state), or indicate that intracellular expression of IGFBP-5 interacts with different, cell surface-independent signaling pathways as has been suggested by our recent studies with IGFBP-3 (3).
The inhibitory and proapoptotic effects of IGFBP-5 on breast cancer cell growth appear to be independent of IGF signaling, as neither MDA-MB-231 nor Hs578T cells (36) are responsive to the mitogenic and anti-apoptotic effects of IGFs. Previous studies demonstrating the growth-promoting effect of IGFBP-5 in prostate cancer cells have suggested this occurs through enhancement of the mitogenic effects of IGFs (23). This led us to hypothesize that, whereas IGFBP-5 can potentiate the stimulatory effects IGFs, its intrinsic, IGF-independent effects are inhibitory in cancer cells. However, the mechanism by which IGFBP-5 mediates its direct effects in these cells is unclear. Whereas cellular receptors for IGFBP-5 have been reported in osteoblasts (40), IGFBP-5 has direct, stimulatory growth effects in these cells (41) suggesting that IGFBP-5 may utilize different signaling pathways in breast cancer cells. Furthermore, our observations that IGFBP-5 secreted by IGFBP-5 transfectants has no growth effects and is not internalized, suggests it is not acting through a cell-surface receptor in these cells, nor are the intracellular levels derived from IGFBP-5 that has been secreted and taken up by the cells. We have previously demonstrated that IGFBP-5 translocates to the nucleus in T47D human breast cancer cells (42), where it may directly or indirectly influence transcription of growth inhibitory and proapoptotic genes. Indeed, in this present study, IGFBP-5 was observed in the nuclear compartment of MDA-MB-231 and Hs578T cells and its expression resulted in an up-regulation of bax and down-regulation of bcl-2 mRNA. However, levels of nuclear IGFBP-5 were not significantly elevated in IGFBP-5 stable transfectants compared with vector controls, suggesting that nuclear translocation of IGFBP-5 is not required to mediate its inhibitory growth effects. We have previously demonstrated that IGFBP-3 also translocates to the nucleus (42) and transcriptionally down-regulates the antiapoptotic bcl-2 (2), although its nuclear translocation is also not necessary for its proapoptotic and antiproliferative effects in breast cancer cells (3). Interestingly, ligand blot analysis revealed the presence of a higher molecular weight form of IG-FBP-5 in cell lysates from adenoviral-infected cells. This suggests the presence of intracellular IGFBP-5 dimers that retain the ability to bind IGFs. However, whether IGFBP-5 monomers or dimers mediate its cytostatic and cytotoxic effects intracellularly remains to be determined.
Both MDA-MB-231 and Hs578T cells express mutant, nonfunctional p53 protein (43). The latter plays a critical role in effecting the cellular response to DNA damage, mediating both a cell cycle arrest and induction of apoptotic cell death. Our studies suggest that IGFBP-5 can induce p53-independent apoptotic and growth-inhibitory pathways. This has particular relevance to the treatment of breast cancers that frequently harbor inactivating mutations in p53 (44) leading to increased resistance to radio-and chemotherapy, and suggests the possibility of therapeutic use of IGFBP-5 to induce apoptosis in resistant cancers.
In conclusion, we have demonstrated that expression of IGFBP-5, both by stable transfection and transiently by adenoviral-mediated infection, has IGF-independent cytostatic and cytotoxic effects on the growth of human breast cancer cells both in vitro and in vivo. Furthermore, IGFBP-5 appears to mediate its growth effects independently of a cell-surface receptor signaling pathway, and transcriptionally modulates intracellular regulators of the apoptotic process, suggesting a novel, intracrine pathway for its actions. These data appear to support in vivo studies that have shown increased IGFBP-5 expression in breast tissue undergoing apoptosis, and emphasize the important role of IGFBP-5 in controlling breast cancer cell growth.