Placental transforming growth factor-beta is a downstream mediator of the growth arrest and apoptotic response of tumor cells to DNA damage and p53 overexpression.

The p53 tumor suppressor gene and members of the transforming growth factor-beta (TGF-beta) superfamily play central roles in signaling cell cycle arrest and apoptosis (programmed cell death) in normal development and differentiation, as well as in carcinogenesis. Here we describe a distantly related member of the TGF-beta superfamily, designated placental TGF-beta (PTGF-beta), that is up-regulated in response to both p53-dependent and -independent apoptotic signaling events arising from DNA damage in human breast cancer cells. PTGF-beta is normally expressed in placenta and at lower levels in kidney, lung, pancreas, and muscle but could not be detected in any tumor cell line studied. The PTGF-beta promoter is activated by p53 and contains two p53 binding site motifs. Functional studies demonstrated that one of these p53 binding sites is essential for p53-mediated PTGF-beta promoter induction and specifically binds recombinant p53 in gel mobility shift assays. PTGF-beta overexpression from a recombinant adenoviral vector (AdPTGF-beta) led to an 80% reduction in MDA-MB-468 breast cancer cell viability and a 50-60% reduction in other human breast cancer cell lines studied, including MCF-7 cells, which are resistant to growth inhibition by recombinant wild-type p53. Like p53, PTGF-beta overexpression was seen to induce both G(1) cell cycle arrest and apoptosis in breast tumor cells. These results provide the first evidence for a direct functional link between p53 and the TGF-beta superfamily and implicate PTGF-beta as an important intercellular mediator of p53 function and the cytostatic effects of radiation and chemotherapeutic cancer agents.

DNA damage arising from exposure to radiation or genotoxic drugs results in the recruitment of DNA repair enzymes and changes in gene expression leading to cell cycle arrest and/or apoptosis (1). The p53 tumor suppressor gene product is an important mediator of the cellular response to DNA damage (2)(3)(4) and functions primarily as a transcription factor that binds to a specific consensus sequence in the promoter or intronic regions of target genes to activate their expression (5,6). The growth arrest function of p53 can be attributed to its ability to directly transactivate the cyclin-dependent kinase inhibitor p21 waf1/cip1 (5,7), leading to G 1 cell cycle arrest (8), as well as genes such as GADD45 (9), 14-3-3 (10), and B99 (11) involved in G 2 checkpoint control. The p53-dependent apoptotic response (12) is not well understood but appears to also involve direct transactivation of genes such as bax (13), Fas/APO1 (14), KILLER/DR5 (15), the PIG genes (16), IGF-BP3 (17), and PAG608 (18). Increased levels of bax (6), for example, promote the release of cytochrome c from mitochondria, leading to activation of initiator caspase 9 and the apoptotic caspase cascade (19,20). However, neither bax nor any other apoptotic regulatory gene identified to date has been shown to be indispensable to apoptotic signaling by p53 (21)(22)(23)(24). Rather, current evidence points to the existence of multiple p53-dependent apoptotic signaling pathways, some of which function in a cell typespecific manner.
The ability of p53 to directly transactivate genes that regulate cell cycle arrest and apoptosis, the high rate of mutation of the p53 gene in tumor cells (25), and the association between loss of p53 function and radiation resistance (26) together provide strong evidence for an important role for p53 in the response of tumor cells to radiation and chemotherapeutic drugs. However, evidence has also been presented for the existence of p53-independent growth arrest and apoptotic pathways that are responsive to DNA damage in cells deficient for wild-type p53 function. For example, loss of p53 function does not always correlate with increased resistance of tumor cells to ionizing radiation (27), and studies in our laboratory have shown that the response of human breast cancer cell lines to DNA damage arising from incorporation of the purine analog gancyclovir (9-[(1,3-dihydroxy-2-propoxy)methyl]guanine) triphosphate can be highly variable and does not correlate with endogenous p53 gene status (28). This issue is further complicated by evidence for cooperation between p53-dependent and p53-independent apoptotic pathways, leading to accelerated cell death when both pathways are activated simultaneously (29). Evidence for cross-talk between p53-dependent and p53-independent apoptotic pathways suggests that they intersect with each other and that these points of convergence represent important regulatory intermediates in these pathways.
To test this hypothesis, recent efforts in our laboratory have focused on the identification of genes positioned at points of convergence of p53-dependent and p53-independent growth ar-rest and apoptotic signaling pathways that are responsive to DNA damage. One of the genes identified encodes a 1.2-kb 1 mRNA that is strongly up-regulated by wild-type p53 and to a lesser extent by radiation treatment. Sequence analysis indicated that this gene corresponds to placental transforming growth factor-␤ (PTGF-␤) (30,31), a novel member of the transforming growth factor superfamily of proteins involved in the regulation of cell proliferation, differentiation, and apoptosis (32,33).

EXPERIMENTAL PROCEDURES
Cell Growth and Apoptosis-Conditions for growth of MDA-MB-468 and MCF-7 human breast cancer cell lines have been described previously (34). HS574.Mg normal human mammary cells and the T47D and SK-BR-3 human breast cancer cell lines were obtained from the ATCC and cultured under conditions recommended by the supplier. Procedures used to treat cells with radiation or recombinant adenoviral vectors and the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay (Sigma) used to monitor cell growth at various times post-treatment have been described previously (34). Western blot analysis of p21 waf1/cip1 protein expression was carried out essentially as described previously (28,34). For flow cytometry, cells were infected with AdPTGF-␤ or Ad␤gal at 100 pfu/cell, harvested 4 days posttreatment, stained with propidium iodide, and analyzed using an EP-ICS flow cytometer (Coulter) as described previously (34). Procedures used to identify characteristic morphological (acridine orange and ethidium bromide staining) and biochemical (DNA laddering) features of apoptosis have been described elsewhere (34). Western blot analysis of ␤-catenin cleavage products was carried out essentially as described previously (35).
mRNA Differential Display-RNA Image kits (GenHunter, Brookline, MA) were used for mRNA differential display analysis of total RNA isolated from control and treated cells using the RNeasy total RNA kit (Qiagen, Chatsworth, CA) essentially as described elsewhere (36). In these experiments, duplicate total RNA samples isolated from untreated, Adp53 (100 pfu/cell)-, and radiation (10 Gy)-treated MDA-MB-468 cells were analyzed using 24 primer combinations (three downstream H-T11(G,A,C) and eight upstream random primers (H-AP1-8) supplied with the RNA Image kit.
Northern Blot Analysis-Northern blots of total RNA isolated from control and treated cells using the RNeasy total RNA kit (Qiagen, Chatsworth, CA) were prepared essentially as described previously (28). Candidate cDNA fragments that were up-regulated in response to both Adp53 and radiation treatment on duplicate mRNA differential display gels were excised from the dried gel, eluted, reamplified by PCR, gel-purified, radiolabeled, and hybridized to Northern blots as described elsewhere (36). Tissue-specific expression was examined using a commercially available Northern blot prepared from poly(A) ϩ mRNA (2 g/lane) isolated from adult tissues (CLONTECH).
Sequence Analysis-cDNA fragments differentially expressed on Northern blots were subcloned into the TA vector system (In Vitrogen) and sequenced using vector-specific primers and a LI-COR model 400 automated sequencer (ACGT Corp., Toronto, Canada).
PTGF-␤ Promoter Constructs and Reporter Gene Assays-A 997-bp fragment containing 31 bp within the 5Ј-untranslated region in exon 1 and 966 bp upstream of the predicted transcription start site of the PTGF-␤ gene in MDA-MB-468 cells was cloned upstream of the luciferase gene in the pGL3Basic vector (Promega) and transiently transfected into MDA-MB-468 cells using a calcium phosphate precipitation protocol (37). A CMV-␤gal reporter plasmid was included in all experiments as a control for transfection efficiency. Luciferase and ␤-galactosidase reporter gene assays were performed on cell extracts prepared 24 h post-treatment with Adp53 (48 h post-transfection) using the Dual-Light Chemiluminescent Reporter Gene Assay System (Tropix, Inc.) and a Lumat LB9507 luminometer (EG & G Berthold). Cell extract protein concentrations were determined using the BCA protein assay (Pierce). PTGF-␤ promoter activities were normalized for total protein concentrations as well as transfection efficiencies based on control CMV-␤gal promoter activities.
3TP Promoter Assay-The 3TP-lux promoter-luciferase construct (kindly provided by J. Wrana) containing three consecutive 12-O-tetradecanoylphorbol-13-acetate response elements and a portion of the plasminogen activator inhibitor 1 promoter region has been shown to be induced by various members of the TGF-␤ superfamily (38). The 3TPlux plasmid was transfected into (Smad4-null) MDA-MB-468 cells in the presence or absence of pCMV5BMADR4, a eukaryotic expression vector encoding the Smad4 gene under the control of a cytomegalovirus (CMV) promoter (also kindly provided by J. Wrana). After 24 h, cells were washed extensively in PBS and treated with either AdPTGF-␤ (50 pfu/cell), Ad␤gal (50 pfu/cell; negative control), or recombinant TGF-␤1 (Sigma; 10 ng/ml; positive control) in serum-free medium. After an additional 24 h of incubation, cells were harvested, and luciferase activities were determined using the Luciferase Assay system (Promega).
Electrophoretic Mobility Shift Assays (EMSAs)-A histidine-tagged, truncated (aa 82-360) p53 protein was expressed from the pET-15b plasmid (Novagen) in Escherichia coli (strain BL21 (DE3)) and purified using a one-step Ni 2ϩ affinity column yielding a preparation of p53 that was 70 -80% pure as determined by SDS-polyacrylamide gel electrophoresis. Complementary, single-stranded oligomers corresponding to the p53 consensus binding site (TTTGGACATGCCCGGGCATGTC-CTTT) (39) or p53 binding site 1 (CACAGCCATGCCCGGGCAA-GAACTCA) within the PTGF-␤ gene promoter were combined at 50 pmol each and end-labeled with [␥-32 P]dATP. EMSA binding assays (40) contained annealed oligonucleotide probes and either purified, recombinant p53 or nuclear extracts (40) prepared from MDA-MB-468 cells either untreated or treated 24 h postinfection with a recombinant adenovirus expressing full-length, wild-type p53 protein (Adp53) at a multiplicity of infection (m.o.i.) of 100 pfu/cell. EMSA competition assays were performed over a range of 30 -300-fold molar excess of the unlabeled, double-stranded competitor oligonucleotide. Binding reactions were incubated for 15 min at 4°C and then for 15 min at 25°C and analyzed by electrophoresis on 5% polyacrylamide gels in 1ϫ TBE buffer at 20 mA for 1 h (41). Gels were dried and exposed to x-ray film for 24 h.
Recombinant Adenoviral Vectors-A PTGF-␤ minigene consisting of the 1.2-kb PTGF-␤ cDNA cloned from MDA-MB-468 cells and CMV promoter and bovine growth hormone polyadenylation sequences cassette derived from the pRc/CMV vector (Invitrogen) was cloned into the p⌬E1Sp1B adenoviral shuttle vector (42) to generate p⌬E1Sp1B-CM-VPTGF-␤. A recombinant adenoviral (Ad) vector expressing the PTGF-␤ minigene was generated by co-transfection of p⌬E1Sp1B-CM-VPTGF-␤ and pJM17 into 293 cells essentially as described elsewhere (42). Recombinant adenoviral plaques were clonally expanded and tested for integration of the PTGF-␤ minigene by PCR using genespecific primers, as well as by restriction fragment analysis of recombinant adenoviral DNA. Sequence analysis verified the integrity of the PTGF-␤ minigene. A recombinant adenoviral vector expressing ␤-galactosidase from a CMV promoter (Ad␤gal) was kindly provided by Dr. F. Graham (McMaster University, Hamilton, Canada). Procedures used in the growth and maintenance of recombinant adenoviral stocks have been described elsewhere (34).

RESULTS
The PTGF-␤ Gene Is Up-regulated upon Activation of Either p53-dependent or p53-independent DNA Damage-signaling Pathways-The molecular events associated with the early stages of p53-dependent and p53-independent apoptotic signaling have not been well defined, but they appear to involve distinct pathways that can intersect and cooperate with each other (29,43). This led us to postulate that key regulators of these apoptotic pathways might be coordinately regulated in response to death-promoting agents that specifically activate these signaling pathways. To explore this possibility, the mRNA differential display technique (45) was used to examine changes in gene expression in human MDA-MB-468 breast cancer cells associated with either wild-type p53 overexpression or radiation treatment. MDA-MB-468 cells express high levels of mutant p53 (273 Arg-His ) (46) and undergo rapid and complete p53-dependent growth arrest and apoptosis following transduction with an adenoviral vector expressing wild-type p53 (Adp53; 100 pfu/cell) or p53-independent growth arrest and apoptosis following treatment with ionizing radiation (10 Gy) in the absence of functional p53 (Fig. 1a). Differential display analysis led to the identification of several candidate genes that are similarly up-or down-regulated in response to each of these apoptotic stimuli (Fig. 1b). One of these genes was shown by Northern blot analysis to correspond to a 1.2-kb mRNA that is strongly up-regulated in response to wild-type p53 overexpression and to a lesser extent by radiation treatment (Fig. 1c). Further analysis demonstrated that this gene is also induced by cisplatin and doxorubicin but not by serum withdrawal or treatment with retinoic acid (Fig. 1c). These results suggested that this gene is uniquely positioned at a convergence point of p53-dependent and -independent apoptotic pathways that are specifically responsive to DNA-damaging agents (47).
Sequence analysis of a 250-bp cDNA fragment isolated from the differential display gel indicated that this gene is highly homologous to six independent entries in the GenBank TM data base corresponding to novel members of the TGF-␤ superfamily: Homo sapiens mRNA for TGF-␤ superfamily protein (Gen-Bank TM accession no. AB000584) (48); human placental bone morphogenic protein (GenBank TM accession no. U88323) (31); H. sapiens macrophage-inhibitory cytokine-1 (GenBank TM accession no. AF019770) (49); human trophoblast mRNA (Gen-Bank TM accession no. U51731); H. sapiens prostate differentiation factor mRNA (GenBank TM accession no. AF003934) (50); and H. sapiens prepro-placental TGF-␤ gene (GenBank TM accession no. AF008303) (30). Based on the high sequence homology between the gene identified in our study and each of these recently described TGF-␤ superfamily members, we concluded that all of these cDNAs probably correspond to a single gene. This gene encodes a 296-amino acid protein containing an amino-terminal hydrophobic domain corresponding to a putative signal sequence, a potential protease cleavage site between aa 177 and 183 consisting of two overlapping dibasic amino acid motifs (RXXRXXR), and a 113-aa carboxyl-terminal domain containing seven positionally conserved cysteine residues that are nearly invariant within the mature regions of all members of the TGF-␤ superfamily (32). Cloning and sequence analysis of reverse transcriptase-PCR products generated by primers corresponding to the full-length 1.2-kb H. sapiens mRNA for TGF-␤ superfamily protein (GenBank TM accession no. AB000584) confirmed that this gene is induced in MDA-MB-468 cells in response to both Adp53 and radiation treatment (Fig. 2a). Northern blot analysis indicated that this gene is expressed at high levels in placenta, at moderate levels in adult kidney, lung, pancreas, heart, and skeletal muscle, and at low levels in adult brain and liver (Fig. 2b). This tissue-specific profile of expression is almost identical to that reported for the prepro-placental TGF-␤ gene (30), providing further evidence that these cDNAs probably represent the same gene. The existence of larger, cross-hybridizing bands in several of these tissues suggests that this gene is alternatively spliced or that closely related genes are co-expressed in these tissues. Interestingly, no expression was seen in transformed counterparts of some of these normal tissues (e.g. H661 lung cancer cells) or in any human breast cancer cell line examined (Fig. 2c). This may suggest that expression of this novel TGF-␤ superfamily member is not compatible with malignant transformation and/or tumor progression. On the basis of its high levels of expression in placental tissue and its nearly complete homology to the prepro-placental TGF-␤ gene, we subsequently refer to this gene as placental TGF-␤ (PTGF-␤).
The PTGF-␤ Gene Is Directly Trans-activated by p53-To begin to examine the transcriptional mechanisms that govern the PTGF-␤ gene response to p53-dependent and p53-independent DNA damage signaling events, a 997-bp genomic fragment corresponding to the promoter region and exon 1 of the PTGF-␤ gene (30) in MDA-MB-468 cells was PCR-amplified, cloned, and sequenced (Fig. 3a). Only minor sequence varia-tions were observed relative to the previously published PTGF-␤ promoter sequence (30), and analysis of this sequence against the TFMATRIX transcription factor binding site profile data base identified several candidate regulatory motifs, many of which have been implicated in regulating gene expression during development (51)(52)(53), hematopoiesis (54 -56), and in cellular responses to cytokines and growth factors (57-59) (Fig.  3a). Also identified were two potential p53 binding sites (39) located at ϩ21 bp (p53 binding site 1) and at Ϫ455 bp (p53 binding site 2) relative to the predicted transcription start site for the gene (Fig. 3a). When cloned upstream of the luciferase reporter gene in the pGL3Basic vector and transiently transfected into MDA-MB-468 cells, PTGF-␤ gene promoter activity increased 20-fold (relative to untreated controls) in response to WT p53 overexpression (Fig. 3b). The level of PTGF-␤ promoter induction by WT p53 was equal to or greater than that of the p21 waf1/cip1 promoter (5) included as a positive control in all experiments. PTGF-␤ promoter activity was also seen to be induced in response to radiation treatment, although to levels that were significantly lower (4-fold induction) than those seen following Adp53 infection (Fig. 3b). This pattern is consistent, however, with the relative levels of induction of endogenous PTGF-␤ transcripts following radiation treatment. Together, these observations support the notion that transcription of the PTGF-␤ gene increases in response to Adp53 or radiation treatment of MDA-MB-468 cells and that binding sites for transcription factors that mediate these p53-dependent and p53-independent signaling events are positioned within 966 bp upstream of the first exon of the gene.
To examine whether p53 directly transactivates the PTGF-␤ gene promoter, constructs containing mutations in each of the two putative p53 binding sites were tested for p53 induction relative to the wild-type promoter. As shown in Fig. 3b, mutation of p53 binding site 1 resulted in a 3-4-fold decline in PTGF-␤ promoter activity following Adp53 treatment, although promoter activity was still 5-6-fold higher than untreated controls (Fig. 3b). Mutation of p53 binding site 2, on the other hand, had no effect on the level of induction of PTGF-␤ promoter activity by WT p53 (Fig. 3b). These results suggest that p53 activates PTGF-␤ gene expression in two ways: directly, by binding to p53 binding site 1, and indirectly, perhaps by influencing the binding of one or more as yet unidentified transcription factors to other regions of the PTGF-␤ gene promoter.
Confirmation that site 1 functions as a p53 binding site was provided by EMSAs in which recombinant wild-type p53 protein and MDA-MB-468 cell nuclear extracts were incubated with radiolabeled, double-stranded oligonucleotides corresponding to either PTGF-␤ p53 binding site 1 or a consensus p53 binding site control (Fig. 3c). Binding reactions containing PTGF-␤ p53 binding site 1 generated EMSA bands with mobilities and intensities identical to the consensus p53 binding site control. This was observed with nuclear extracts prepared from MDA-MB-468 cells treated with Adp53 (468 ϩ Adp53; Fig. 3) as well as with purified, recombinant wild type p53 protein (Rec p53). No bands were seen in binding reactions that contained either no protein extract (probe alone; P) or nuclear extracts prepared from untreated MDA-MB-468 cells (468). The increased mobility of bands formed in the presence of recombinant p53 reflects the smaller size of the recombinant p53 protein (aa 82-360) used in these binding reactions as compared with the full-length p53 protein (393 aa) expressed from the adenoviral vector. Further evidence for the specificity of this binding reaction was provided by the observation that band formation could be successfully competed by the addition of a 100-fold molar excess of an unlabeled, double-stranded to untreated controls by 7 days post-treatment, despite the absence of pRb and Smad4 function (Fig. 4a). Treatment with Adp53 resulted in a 94% decline in MDA-MB-468 cell viability under these conditions, consistent with the notion that PTGF-␤ is not the sole downstream mediator of the growth inhibition function of p53. Cell viability declined less than 20% following treatment with Ad␤gal under the same conditions.
To explore this question further, the effects of PTGF-␤ overexpression were examined in a number of other human breast cancer cell lines, including MCF-7 cells, which we and others had previously shown to be resistant to WT p53 overexpression (34,63). As shown in Fig. 4b, treatment of T47D and SK-BR-3 breast cancer cells with AdPTGF-␤ (50 pfu/cell) resulted in 64 and 68% reductions in viability, respectively, relative to Ad␤gal-treated controls. As with MDA-MB-468 cells, Adp53 treatment resulted in significantly greater decreases in T47D and SK-BR-3 cell viability (10 and 29% of Ad␤gal-treated controls, respectively), consistent with the notion that PTGF-␤ is a downstream mediator of p53 function but is not solely responsible for the growth-inhibitory effects of p53. Interestingly, AdPTGF-␤ treatment was seen to reduce MCF-7 cell viability to 44% of Ad␤gal-treated controls, whereas no significant decline in MCF-7 cell viability was seen following treatment with Adp53 under these conditions. These results suggest that, like p53 and other members of the TGF-␤ superfamily, PTGF-␤mediated growth effects can vary significantly from one tumor cell type to another. Furthermore, the absence of a significant decrease in viability in either HS574.Mg normal mammary cell or normal human skin fibroblasts suggests that PTGF-␤ overexpression may be particularly detrimental to tumor cell growth.
PTGF-␤ Overexpression Can Lead to the Induction of p21 waf1/cip1 , G 1 Cell Cycle Arrest, and Apoptosis-To begin to explore the mechanism by which PTGF-␤ overexpression can lead to decreased tumor cell viability, MDA-MB-468 and MCF-7 breast cancer cells were transduced with AdPTGF-␤ and examined for established indicators of cell cycle arrest and apoptosis. Northern blot analysis verified that high levels of PTGF-␤ mRNA persist for up to 4 days postinfection in MDA-MB-468 and MCF-7 cells treated with AdPTGF-␤ (Fig. 5a). To investigate whether PTGF-␤ can activate signal transduction pathways common to other members of the TGF-␤ superfamily, MDA-MB-468 cells were transfected with a 3TP-lux reporter plasmid in the presence or absence of a second plasmid encoding wild-type Smad4 under the control of a constitutive viral (CMV) promoter. The 3TP-lux reporter has been previously shown to be responsive to Smad signaling initiated by members of the TGF-␤ superfamily (38). Treatment of MDA-MB-468 cells with Ad␤gal (50 pfu/cell), AdPTGF-␤ (50 pfu/cell), or recombinant TGF-␤1 (10 ng/ml) resulted in no significant increases in 3TP-lux reporter activity in the absence of the Smad4 expression plasmid. However, cells co-transfected with the Smad4 expression plasmid exhibited 15-and 12-fold increases in 3TP-lux reporter activities following treatment with AdPTGF-␤ and recombinant TGF-␤1, respectively, relative to Smad4-deficient cells treated under the same conditions (Fig.  5b). This represented a 5-6-fold increase over the levels of 3TP-lux induction in untreated cells expressing Smad4, demonstrating that, like TGF-␤1, PTGF-␤ can induce 3TP-lux reporter activity in a Smad4-dependent manner.
TGF-␤ inhibits epithelial cell growth by causing growth arrest at the G 1 phase of the cell cycle, at least in part by up-regulating cyclin-dependent kinase inhibitors like p21 waf1/cip1 (64). TGF-␤ has also been shown to induce apoptosis in a variety of cancer cell lines including MCF-7 cells (65,66). Flow cytometric analysis demonstrated that PTGF-␤ does signal G 1 cell cycle arrest in MCF-7 cells, as evidenced by an increase in the G 1 /S ratio from 1.1 in untreated cells to 21.8 following AdPTGF-␤ treatment (Fig. 5c). G 1 arrest was not evident in MDA-MB-468 cells following AdPTGF-␤ treatment (data not shown), a finding consistent with the absence of pRb and/or Smad4 function in these cells. To determine whether increased p21 waf1/cip1 levels are associated with PTGF-␤-mediated G 1 arrest, Western blots of MDA-MB-468 and MCF-7 cell extracts prepared at 0, 24, and 48 h post-treatment with AdPTGF-␤ were probed with a p21 waf1/cip1 -specific antibody. As shown in Fig. 5d, AdPTGF-␤ treatment was seen to increase p21 waf1/cip1 protein levels in both MDA-MB-468 and MCF-7 cells by 48 h postinfection, consistent with studies showing that TGF-␤ can induce p21 waf1/cip1 gene expression in a p53-independent manner (67). No increase in p21 waf1/cip1 protein was seen following infection of either cell line with a control (Ad␤gal) adenovirus under the same conditions. MDA-MB-468 and MCF-7 cells were also examined for changes characteristic of apoptosis following AdPTGF-␤ treatment. 60-kDa ␤-catenin cleavage products associated with caspase 3 activation were evident in both cell lines as early as 2 days post-treatment with AdPTGF-␤ and appeared to increase over the 5-day period studied (Fig. 5e). Acridine orange/ethidium bromide staining confirmed that treatment of MDA-MB-468 cells with AdPTGF-␤ results in morphological changes (e.g. nuclear condensation and fragmentation) characteristic of apoptotic cell death (data not shown). Taken together, these results are consistent with a role for PTGF-␤ as a downstream mediator of the p53-dependent and p53-independent growth arrest and apoptotic response of cells to DNA damage. DISCUSSION Mutations in genes that mediate growth arrest and apoptosis in response to DNA damage contribute to carcinogenesis and the development of resistance to genotoxic anticancer agents (68). The p53 tumor suppressor gene plays a central role in signaling growth arrest and apoptosis in response to DNA damage, primarily through its ability to activate the expression of downstream target genes such as p21 waf1/cip1 and bax. However, cells deficient in p53 function can still invoke a growth arrest and apoptotic response following DNA damage, pointing to the existence of one or more parallel, p53-independent signaling pathways. Cell-specific variations in the activating thresholds for different death-promoting agents, along with evidence for cross-talk between different apoptotic signaling pathways (29,69), have led to the notion that apoptotic thresholds are determined, at least in part, by the expression of key regulatory proteins positioned at points of intersection or convergence of different signaling pathways (70,71). In a search for genes that are transcriptionally responsive to disparate death-promoting stimuli, we have identified the gene encoding placental transforming growth factor-␤ (PTGF-␤) as a downstream target of both p53-dependent and p53-independent DNA damage pathways. Nucleotide and amino acid sequence alignments suggest that PTGF-␤ is a novel, distantly related member of the TGF-␤ superfamily of cytokines and appears to be identical to several genes cloned independently in other laboratories on the basis of their high levels of expression in placental or prostate tissues (30,31,48,50), the presence of a secretory signal sequence (48), or an association with macrophage activation (49). The apparent functional diversity and tissue-specific expression of this gene point to an important role for PTGF-␤ in normal tissue development and homeostasis.
In the present study, Northern blot analysis demonstrated that PTGF-␤ gene expression is strongly induced by wild-type p53 and to a lesser extent by radiation and genotoxic chemotherapeutic agents in the absence of wild-type p53. Sequence analysis of the PTGF-␤ gene promoter identified two putative p53 binding site motifs, and mutation of one of these p53 binding motifs was seen to abolish promoter responsiveness to WT p53. Direct binding of WT p53 to this p53 binding site motif was confirmed by EMSA, pointing to a direct role for WT p53 in the activation of PTGF-␤ gene expression. Thus, PTGF-␤ appears to be one of a growing list of p53 target genes, many of which have been shown to play a role in the growth arrest and apoptotic functions of p53.
The absence of detectable levels of PTGF-␤ gene transcripts in a series of seven human breast tumor cell lines as well as in several transformed counterparts of tissues that normally express the gene suggests that PTGF-␤ expression may be incompatible with malignant transformation and/or tumor progression. Support for this view was provided by functional studies that demonstrated that overexpression of PTGF-␤ alone can signal growth arrest and apoptosis of human breast cancer cell lines in vitro. These results are consistent with a role for PTGF-␤ as an important downstream mediator of the cellular response to DNA damage, although the precise contribution of PTGF-␤ to the overall growth arrest and apoptotic response remains to be determined. Previous studies in our laboratory (28,34) and in others (63) have shown that recombinant WT p53 overexpression can result in differential growth arrest and apoptotic responses in different tumor cell types (e.g. those expressing mutant p53 versus wild type p53). The differential effects of Adp53 on MDA-MB-468 and MCF-7 cell viability illustrate this point. The 80% reduction in MDA-MB-468 cell viability following adenovirus-mediated PTGF-␤ gene transfer is consistent with a role for PTGF-␤ as a downstream mediator of WT p53 function. Furthermore, the greater sensitivity of MCF-7 cells to PTGF-␤ overexpression (Ͼ50% decrease in viability) points to a defect in p53-dependent signaling upstream of the PTGF-␤ gene and may account, at least in part, for the resistance of MCF-7 cells to WT p53 overexpression. Analysis of PTGF-␤ effects in several different human breast cancer cell lines indicated that, like p53, the response of individual tumor cell lines to PTGF-␤ can be highly variable. These results are similar to those in studies of other known downstream mediators of p53 function and support the view that activating thresholds for individual death-promoting stimuli can vary significantly from one tumor cell type to another.
Previous studies of the biological and signaling activities of the PTGF-␤ homologues macrophage-inhibitory cytokine (49) and prostate-derived factor (50) demonstrated that this gene encodes a secreted protein that activates signal transduction along the same pathway as other members of the TGF-␤ superfamily. Our observation of Smad4-dependent 3TP promoter activation by PTGF-␤ in MDA-MB-468 cells provided further evidence that PTGF-␤ is secreted from cells and can activate signal transduction pathways common to other TGF-␤ superfamily members. Members of the TGF-␤ superfamily initiate signaling from the cell surface by interacting with distinct serine/threonine kinase receptors (72), which in turn initiate a complex series of downstream signaling events involving phosphorylation of receptor-activated Smad proteins (Smad1, -2, -3, -5, and -9) (73,74). Smad4 forms heteromeric complexes with receptor-activated Smads to promote their translocation to the nucleus (75,76), where they can associate with transcriptional coactivators and transactivate TGF-␤-regulated genes (77). The growth-inhibitory function of TGF-␤ is mediated, at least in part, through transcriptional activation of the cyclin-dependent kinase inhibitor p21 waf1/cip1 (78). Consistent with this view, PTGF-␤ was seen to increase p21 protein levels and induce G 1 cell cycle arrest in MCF-7 cells. However, while Smads are required for growth inhibition by TGF-␤, they are not by themselves sufficient for p21 waf1/cip1 gene induction. Rather, p21 waf1/cip1 gene activation appears to also involve TGF-␤-mediated activation of the Ras/mitogen-activated protein kinase pathway (79). This issue is further complicated by our observation of increased p21 waf1/cip1 protein levels in MDA-MB-468 cells treated with AdPTGF-␤. Since MDA-MB-468 cells carry a homozygous deletion of the complete Smad4 coding region (80), this suggests that p21 waf1/cip1 induction can occur in the absence of Smad signaling. Whether this involves transcriptional activation through alternative (Smad4-independent) receptor-Smad or Ras/mitogen-activated protein kinase pathways or post-transcriptional mechanisms remains to be determined.
In addition to signaling G 1 cell cycle arrest, TGF-␤ also plays an important role in activating apoptotic signaling pathways in a variety of cell types including cancer cell lines (66). In MCF-7 cells, increased TGF-␤ levels have been associated with the induction of apoptosis in response to several anticancer agents, including tamoxifen (65). Recent evidence suggests that TGF-␤-mediated apoptotic signaling involves up-regulation of connective tissue growth factor, leading to reduced levels of Bcl-2 protein expression (81). Our observation that PTGF-␤ can induce apoptosis in MCF-7 cells is consistent with evidence that PTGF-␤ utilizes signal transduction pathways common to other TGF-␤ superfamily members. However, apoptosis of Smad4-deficient MDA-MB-468 cells treated with AdPTGF-␤ points to the existence of an alternative, Smad4-independent apoptotic signaling pathway. Whether this involves the Ras/ mitogen-activated protein kinase pathway or alternative, PTGF-␤-specific signaling pathways remains to be determined (73,82,83). In this regard, it is interesting to note that c-Jun N-terminal kinase has been implicated in DNA damage-mediated apoptosis (84) as well as TGF-␤-mediated, Smad4-independent activation of fibronectin synthesis in a human fibrosarcoma cell line (82).
In summary, the PTGF-␤ gene is up-regulated in response to both p53-dependent and p53-independent signaling events arising from DNA damage. The PTGF-␤ gene promoter appears to be directly transactivated by WT p53, and overexpression of PTGF-␤ alone was seen to signal growth arrest and apoptosis in human breast tumor cell lines. These results provide the first evidence for a direct functional link between DNA damage, WT p53, and the TGF-␤ superfamily of cytokines and implicate PTGF-␤ as an important downstream mediator and point of convergence of p53-dependent and p53-independent DNA damage pathways. In addition, secretion of PTGF-␤ provides a novel route through which signals arising from DNA damage can be communicated to neighboring cells. This raises the intriguing possibility that PTGF-␤ contributes to the bystander effect associated with WT p53 gene transfer therapy (44,85,86).