Transforming growth factor beta regulates parathyroid hormone-related protein expression in MDA-MB-231 breast cancer cells through a novel Smad/Ets synergism.

The majority of breast cancers metastasizing to bone secrete parathyroid hormone-related protein (PTHrP). PTHrP induces local osteolysis that leads to activation of bone matrix-borne transforming growth factor beta (TGF beta). In turn, TGF beta stimulates PTHrP expression and, thereby, accelerates bone destruction. We studied the mechanism by which TGF beta activates PTHrP in invasive MDA-MB-231 breast cancer cells. We demonstrate that TGF beta 1 up-regulates specifically the level of PTHrP P3 promoter-derived RNA in an actinomycin D-sensitive fashion. Transient transfection studies revealed that TGF beta 1 and its effector Smad3 are able to activate the P3 promoter. This effect depended upon an AGAC box and a previously described Ets binding site. Addition of Ets1 greatly enhanced the Smad3/TGF beta-mediated activation. Ets2 had also some effect, whereas other Ets proteins, Elf-1, Ese-1, and Erf-1, failed to cooperate with Smad3. In comparison, Ets1 did not increase Smad3/TGF beta-induced stimulation of the TGF beta-responsive plasminogen activator inhibitor 1 (PAI-1) promoter. Smad3 and Smad4 were able to specifically interact with the PTHrP P3-AGAC box and to bind to the P3 promoter together with Ets1. Inhibition of endogenous Ets1 expression by calphostin C abrogated TGF beta-induced up-regulation of the P3 transcript, whereas it did not affect the TGF beta effect on PAI expression. In TGF beta receptor II- and Ets1-deficient, noninvasive MCF-7 breast cancer cells, TGF beta 1 neither influenced endogenous PTHrP expression nor stimulated the PTHrP P3 promoter. These data suggest that TGF beta activates PTHrP expression by specifically up-regulating transcription from the PTHrP P3 promoter through a novel Smad3/Ets1 synergism.

Parathyroid hormone-related protein (PTHrP) 1 is a pleiotropic secretory protein that plays a role in a number of biological processes by acting primarily in an autocrine or paracrine fashion (1)(2)(3)(4). PTHrP has been discovered as a tumor-derived humoral agent that causes hypercalcemia of malignancy, a common metabolic disorder in patients with neoplastic disease (5,6). Its calcium-mobilizing activity is based on its ability to interact, like parathyroid hormone, with the parathyroid hormone/PTHrP receptor in bone and kidney (7), thereby stimulating osteoclastic bone resorption and calcium resorption from the kidney, respectively.
PTHrP is expressed by a number of different tumors, including breast carcinoma (8 -10). PTHrP may promote tumor metastasis. In particular, PTHrP facilitates the growth of breast cancer cells metastasized to bone by inducing destruction of the bone because of its bone resorptive activity (11,12). As a consequence of this destruction, transforming growth factor ␤ (TGF␤) gets released from the bone matrix to further stimulate PTHrP production by the breast cancer cells (13). Such a TGF␤/ PTHrP feedback loop is thought to significantly contribute to the progression of breast metastases in bone (14). TGF␤ stimulates PTHrP expression in a variety of cell lines by increasing the PTHrP mRNA level (15)(16)(17)(18). Of the three promoters (P1-P3) that can drive PTHrP transcription in humans, the proximal P3 promoter (formerly called P2 promoter) is always active in PTHrP-expressing breast tumors (19). The P3 promoter is primarily responsible for PTHrP expression in breast cancer cells metastasized to bone (20). We have shown previously that the human PTHrP P3 promoter contains a composite Ets and Sp1 binding element that confers responsiveness of the PTHrP P3 promoter to human T cell lymphotrophic virus type I Tax, Ets1, and Sp1 (21)(22)(23). This element is conserved and is also found in the corresponding murine PTHrP promoter, where it mediates activation by retinoic acid, Ets1, Ets2, and the adenoviral protein E1A (24,25).
Members of the TGF␤ family are cytokines that regulate a broad range of cellular function, including proliferation, differentiation, and invasion. TGF␤ binds to and activates a heterodimeric TGF␤ receptor, composed of type I and II receptor units (26,27). Activation of the type I receptor leads to the phosphorylation/activation of the R-Smad (receptor-activated Smad) proteins Smad2 and/or Smad3, two transcriptional activators of the Smad protein family (28,29). In turn, activated Smad2 or Smad3 associates with Smad4, a common mediator Smad (co-Smad), and translocates into the nucleus. Here they act as transcriptional activators that specifically recognize AGACcontaining sequences (AGAC box) (30). Smad proteins often synergize with other transcription factors, e.g. Smad3 functionally interacts with TFE3 (31), AP1 (32,33), AML-1 (34), and Sp1 (35).
The Ets family of transcription factors comprises proteins that share a unique DNA binding domain, the Ets domain, that specifically recognizes a 5Ј-GGA(A/T)-3Ј-based DNA sequence (36). Ets proteins have been shown to activate many genes (37) and to be involved in a number of physiological and pathophysiological processes (38). As with Smad proteins, activation of genes by Ets proteins often involves synergisms with other transcription factors (36). Interestingly, Ets1 shares with Smad3 the ability to cooperate with AP1 (39), AML-1 (40,41), TFE3 (42), and Sp1 (22).
Here we describe the identification of a new functional element, an AGAC box, in the PTHrP P3 promoter. We show that this sequence is a TGF␤-responsive DNA binding site that allows Smad3 to activate the P3 promoter. We further observed that, for this effect, Smad3 had to cooperate with an Ets transcription factor. By testing several Ets proteins for their ability to act in synergy with Smad3, we found Ets1 to be one potential partner for Smad3.

MATERIALS AND METHODS
Cell Lines and Plasmids-MDA-MB-231 and MCF-7 breast cancer cell lines were maintained in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% fetal calf serum (PAA) in the absence of antibiotics. For PTHrP P3 promoter analysis by transient transfection assay, a Ϫ328/ϩ20 PTHrP P3 promoter fragment (numbering relative to the start site of the P3 promoter) was used. PTHrP P3-luciferase constructs (P3-luc) were produced by digesting wild type, Ets, and Sp1 mutant Ϫ328/ϩ20 PTHrP P3 promoter-chloramphenicol acetyltransferase plasmids (22) with HindIII and SalI and by cloning of the resulting promoter fragments into pIL5P.luc (43) that had been cut by XhoI and HindIII. The AGAC box mutant version of the P3 promoter (see Fig. 2B) was synthesized by polymerase chain reaction and cloned into the pCRII vector (Invitrogen). An internal AccI/PstI promoter fragment containing the AGAC box mutation was excised and inserted into the Ϫ328/ϩ20 wild type PTHrP P3 promoter-luciferase replacing the corresponding wild type sequence. All ets genes as used here were cloned into the pcDNA3 vector (Invitrogen). The human ese-1 and erf cDNAs were kindly provided by T. Libermann and G. J. Mavrothalassitis, respectively. The plasmids pEXL-Flag-Smad3 and pEXL-Smad4 as well as 3TP-Luc were generous gifts from R. Weinberg and Y. Sun, respectively.
Inhibitors-Actinomycin D (Calbiochem) was dissolved in water and used at a final concentration of 5 g/ml. Calphostin C (Calbiochem), dissolved in dimethyl sulfoxide, was added to the medium at a final concentration of 1 M. Control cells were incubated with the equivalent amount of dimethyl sulfoxide. To activate calphostin C, treatment of cells with calphostin C were performed under exposure to light.
Transient Transfection and Luciferase Assay-For luciferase assays, cells grown to 70 -80% confluence were transiently transfected with 3 g of P3-luc (or 3TP-luc) reporter plasmid and 1 g of an expression plasmid or vector (pcDNA3) (Fig. 3D) or 3 g of P3-luc (or 3TP-luc) reporter plasmid and a total of 2 g of expression plasmids and/or vector (pcDNA3) (Figs. 4A, 5, and 6) by using LipofectAMINE (Life Technologies, Inc.). Each transfection mix also contained 0.5 g of a ␤-galactosidase expression plasmid. After incubation overnight, TGF␤1 (R&D Systems) or the equivalent amount of TGF␤1 buffer (1 mg/ml bovine serum album in 4 mM HCl) was added and cells incubated for another 7 or 24 h. Cells were then analyzed for luciferase activity as described previously (43). Relative promoter activity was calculated by normalizing luciferase activity against ␤-galactoside activity. For electromobility shift assays (EMSA), cells were transfected by electroporation. For expression of Flag-Smad3, Smad4, or Ets1, 6 g of pEXL-Flag-Smad3, 2 g of pEXL-Smad4, or 6 g of pcDNA3-Ets1, respectively, were used. The amount of transfected DNA was kept constant by the addition of pcDNA3. Two hours after electroporation, medium and debris were removed and cells were treated with fresh TGF␤1 (5 ng/ml) or TGF␤1 buffer containing medium for another 3 h. Cells were harvested and lysed for nuclear extraction.
Preparation of Nuclear and Whole Cell Extracts-Nuclear extracts were prepared essentially as described (43). Briefly, cells were washed with phosphate-buffered saline and harvested by using a cell scraper. Cells were resuspended in buffer A (10 mM Hepes, pH 7.9, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride), lysed by addition of Nonidet P-40 followed by vortexing for 10 s. After centrifugation at 13,000 rpm for 10 min, nuclei were extracted by addition of buffer C (20 mM Hepes, pH 7.9, 0.4 M NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride). Total protein amount in the extracts was measured using the Bio-Rad Bradford reagent. For whole cell extraction, cells were lysed in 250 mM Tris-Cl, pH 7.5, by three cycles of freezing and thawing, followed by clearing the lysate by centrifugation for 5 min at 13,000 rpm at 4°C.
For quantitative PCR, 1 l of 1:10 diluted cDNA was mixed with 2 l of each primer and 5 l of SYBR Green PCR reaction buffer (PerkinElmer Life Sciences). PCR reactions were performed according to the manufacturer's instructions on an ABI Prism 7700 sequence detector, which allows real-time detection of the PCR product by measuring the increase in SYBR Green fluorescence caused by binding of SYBR Green to double-stranded DNA. Quantification was performed as suggested by the manufacturer using the gapdh gene as endogenous reference gene.

TGF␤1 Induces PTHrP Expression Primarily through the P3
Promoter-The effect of TGF␤1 on PTHrP expression in TGF␤responsive MDA-MB-231 cells (13) were analyzed by Northern blot hybridization (Fig. 1B), conventional RT-PCR (Fig. 1C) and real-time quantitative RT-PCR (Fig. 1D). By targeting exon III (Northern blot) or exon IV-specific sequences (RT-PCR) to detect all transcripts (Fig. 1A), we found a significant increase in the level of PTHrP RNA in response to TGF␤1 (Fig. 1, B-D). Specific amplification of P3, P1, or P1/P2 transcripts by quantitative RT-PCR (Fig. 1A) revealed a substantial effect of TGF␤1 on the P3 transcript level, but an only moderate effect on the levels of the P1 and P1/2 transcripts (Fig. 1, C and D). Based on the data presented in Fig. 1D, we calculated that ϳ25% of the PTHrP RNA in MDA-MB-231 cells was derived from P3-induced transcription in the absence of TGF␤1, whereas, after treatment with TGF␤1 for 24 h, the majority of PTHrP RNA originated from P3-dependent transcription (55-75%). MDA-MB-231 cells also supported TGF␤-dependent activation of the classic TGF␤-responsive PAI-1 gene (44). In contrast, in TGF␤ receptor II-deficient MCF-7 breast cancer cells (45), neither PTHrP nor PAI-1 expression was increased by TGF␤1 (Fig. 1E). This suggests that TGF␤1 acts through its receptor to activate PTHrP expression.
TGF␤ has the potential to affect both PTHrP transcription and stability of PTHrP RNA (15)(16)(17)(18). To assess the contribution of RNA stabilization to the effect of TGF␤ on PTHrP expression in MDA-MB-231 cells, we inhibited transcription by actinomycin D. We found that actinomycin D substantially inhibited up-regulation of the PTHrP P3-specific transcript by TGF␤ (Fig. 1F) and completely abolished TGF␤-dependent stimulation of PAI-1 expression (Fig. 1G). These data suggest that TGF␤ regulates PTHrP synthesis in MDA-MB-231 cells by specifically affecting promoter P3-dependent PTHrP expression, at least in part, by directly modulating transcription from this promoter.
TGF␤1 Stimulates PTHrP P3 Promoter Activity through a Smad3 Recognition Site-A potential TGF␤-responsive element, an AGACAGAC motif, is located immediately downstream of the Ets/Sp1 composite element within the PTHrP P3 promoter (Fig. 2B). This sequence shows a high homology to a Smad3/TGF␤-responsive AGAC box identified in the c-jun gene (32) (Fig. 2A). To check first whether MDA-MB-231 supports nuclear translocation of Smad3 by TGF␤, we performed Western blot analyses. Smad3 could only be detected in nuclear extracts when MDA-MB-231 cells had been treated with TGF␤1 (Fig. 3A). TGF␤1 also increased nuclear translocation of ectopically expressed Flag-Smad3 (Fig. 3B). In this case, however, a fair amount of Smad3 was also detectable in nuclear extracts in the absence of TGF␤1, a finding also reported by others (46).
For promoter studies we used a luciferase construct containing a Ϫ328/ϩ20 P3 promoter fragment that harbors the AGAC box (Fig. 2B). TGF␤1 stimulated P3 promoter activity by 1.7fold, which compares with a 2.4-fold induction by TGF␤1 of the PAI-1 promoter construct 3TP (Fig. 3C). A similar weak response of the PAI-1 promoter to TGF␤ in MDA-MB-231 cells has been reported by others (47). Flag-Smad3 activated the PTHrP P3 promoter by 1.9-fold in the absence and by 4.1-fold in the presence of TGF␤1 (Fig. 3D). The Smad3/TGF␤1 effect depended upon the AGAC box, the Ets binding site, and the Sp1 DNA binding motif (Figs. 2B and 3D). These data suggest that the PTHrP P3 promoter is responsive to TGF␤1 and its effector Smad3.
Smad3 Cooperates with Ets1 to Activate the PTHrP Promoter in a TGF␤1-dependent Manner-The importance of the Ets binding site for the Smad3/TGF␤1-mediated stimulation of the PTHrP P3 promoter prompted us to study the contribution of Ets1 to this effect. Ets1 is a potent transcriptional activator of the P3 promoter (23) and is expressed in MDA-MB-231 cells ( Figs. 4B and 6C). Alone, Ets1 increased P3 promoter activity by 3.5-fold (Fig. 3D). A mutation in either the Ets or the Sp1 binding site abrogated this effect. This is consistent with earlier findings that Ets1 cooperates with Sp1 to regulate the P3 promoter (22). In contrast, a mutation in the AGAC box did not affect Ets1-dependent activation. The presence of Flag-Smad3 increased the Ets1 effect slightly (Fig. 4A). However, when TGF␤1 was also added, Ets1 cooperated with Smad3 to increase promoter activity 14-fold. Importantly, co-expression of Ets1 and Smad3 did not affect the expression levels of these proteins (Fig. 4B). A mutation in either the AGAC box or the Ets or Sp1 binding site abrogated the TGF␤1-dependent Smad3/Ets1 synergistic effect (Fig. 4A). In comparison, the 3TP promoter did not respond to Ets1 alone, nor did it support a Smad3/Ets1 synergism (Fig. 4A). Rather, Ets1 reduced the Smad3 effect on the 3TP promoter by ϳ2.5-fold. Collectively, these data suggest that TGF␤1-mediated activation of the PTHrP P3 promoter involves a novel promoter-specific Smad3/ Ets1 synergistic interaction, which depends not only on the Smad3 and Ets1 binding sites of the promoter, but also on the Sp1 binding motif.
Next we tested other Ets transcriptional activators, Ets2, Ese-1, and Elf-1, that are also endogenously expressed in MDA-MB-231 cells (data not shown), for their ability to cooperate with Smad3. Ets2 supported Smad3-dependent activation to some extent, but failed to induce transcription from the P3 promoter on its own (Fig. 5). Ese-1 stimulated the promoter as strongly as Ets1, but was unable to support promoter induction by Smad3. It seems that activations of the P3 promoter by Ese-1 and Smad3 are mutually exclusive. Elf-1 as well as the Ets protein ERF, a transcriptional repressor (48), had no effect on the promoter at all. Collectively, these data suggest that only certain Ets factors, such as Ets1, are capable of synergizing with Smad3.
In TGF␤ receptor II-deficient MCF-7 cell line, Smad3 alone or in conjunction with Ets1 was unable to efficiently activate the PTHrP P3 promoter (Fig. 6A). Because Ets1 and Smad3 were ectopically expressed in these cells at levels comparable with those in MDA-MB-231 cells (Fig. 6B), these results suggest that MCF-7 cells do not support the TGF␤-dependent Smad3/Ets1 synergism. It is noteworthy that, in these cells, even Ets1 alone failed to activate the P3 promoter (Fig. 6A). However, MCF-7 cells supported Ets2-dependent P3 promoter activation (data not shown), which was not observed with MDA-MB-231 cells.
Smad3 and Its Co-Smad Smad4 Bind to the PTHrP P3 AGAC Box-Smad3 can bind to its cognate DNA binding site as a homodimer or as a heterodimer with its partner, the co-Smad Smad4. To test whether these Smad proteins are able to interact with the PTHrP P3 AGAC box, we performed EMSA. Using nuclear extracts from MDA-MB-231 cells we found that, upon treatment with TGF␤1, two new complexes (C1 and C2) were formed with a DNA-probe corresponding to the AGAC box containing PTHrP P3 sequence between nucleotides Ϫ51 and Ϫ28 (Fig. 7B, compare lane 3 with lane 2). Both complexes could be partially shifted by an anti-Smad4 antibody (lane 4) but not by an anti-Flag antibody (lane 5). The formation of these complexes increased strongly when nuclear extracts contained overexpressed Flag-Smad3 (lane 7). Under these conditions, C1 and C2 reacted with the anti-Flag antibody (lane 9).
In the presence of Flag-Smad3, more C2 was found than C1 (lane 7), whereas, in its absence, C1 was preferentially formed (lane 2). Adding Flag-Smad3 plus Smad4 reversed the C2 to C1 ratio again, leading to an increase in the level of C1 and to a reduction of the level of C2 (lane 11). Concomitantly, a strong increase in supershifting by the anti-Smad4 antibody was observed (lane 12). Again, both complexes could be shifted by the anti-Flag antibody (lane 13). These data show that Smad3 and Smad4 can bind to the PTHrP P3 AGAC box in a TGF␤1-dependent manner. They further show that, with the Ϫ51/Ϫ28 PTHrP P3 probe, two complexes are formed, one that likely contains Smad3 and Smad4 heterodimers (C1) and another that may preferentially harbor Smad3 homomers (C2). Given the slower migration of C2 relative to C1, the Smad3 proteins in the C2 complex were probably multimerized. The formation of multimeric Smad3 complexes upon TGF␤1 stimulation has also been reported by others (49).
When Flag-Smad3 and Smad4 were challenged for their ability to bind to a longer PTHrP P3 probe (Ϫ79/Ϫ28) (Fig. 7A), only one complex (C3) was observed (Fig. 7C, compare lane 5 with lane 3). This complex migrated slightly slower than the C2 complex and was completely shifted by the anti-Flag antibody (lane 6) and partially shifted by the anti-Smad4 antibody (lane 7). Similar data were observed when nuclear extracts were used that contained Flag-Smad3 alone (Figs. 8 and 9).
Next, we analyzed whether the mutation in the AGAC box that was used to create the AGAC mutant PTHrP P3 promoter (Fig. 2B) affects the interaction of this DNA element with Smad3. Insertion of this AGAC box mutation into the Ϫ51/Ϫ28 oligonucleotide (Fig. 8A) drastically reduced Smad3 binding to the Ϫ51/Ϫ28 DNA probe (Fig. 8B, compare lane 4 with lane 2). The same mutation abrogated the ability of this oligonucleotide to compete with the Ϫ79/Ϫ28 wild type probe for binding to Smad3 when added at 10-or 25-fold molar excess over the probe (Fig. 8C, compare lanes 3-6 with lane 2). These data show that the same mutation in the AGAC box that inhibited the Smad3/Ets1/TGF␤1 effect also strongly interfered with binding of Smad3 to this DNA element.
To test whether Ets1 and Smad3 can simultaneously bind to the PTHrP P3 promoter, we analyzed the effect of Ets1 on Smad3 binding to the Ϫ79/Ϫ28 probe that contains the binding sites for both transcription factors (Fig. 7A). In the presence of Ets1, a new complex (C4) was formed (Fig. 9, A (compare lane  5 with lane 3) and B (compare lane 3 with lane 2)). This was accompanied by a reduction in the formation of C3. Like C3, C4 could be recognized by the anti-Flag antibody (Fig. 9A, lane 6) suggesting that the C4 complex contains Flag-Smad3 in addition to Ets1. Competition experiments with an oligonucleotide (Ets cons.) that bears an Ets consensus binding site in 50-fold molar excess over the probe supported the notion that Ets1 was present in the C4 complex. Addition of this sequence prevented the formation of the C4 complex and, at the same time, increased the level of the C3 complex (Fig. 9A, lane 7). The C4 complex also disappeared when an anti-Ets1 antibody was included in the reaction mix. Three new complexes (Ets1-␣Ets1) were formed instead (Fig. 9B, lane 4). Of these, the two faster migrating Ets1-␣Ets1 complexes were also seen in the presence of the Ϫ51/Ϫ28 wild type PTHrP competitor DNA (AGAC box), when added at 50-fold molar excess over the probe (Fig. 9B, lane 6). This oligonucleotide suppressed the formation of the Smad3 complexes (Fig. 9B, lane 5). This suggests that the two faster migrating Ets1-␣Ets1 complexes did not contain Smad3, whereas the slowest Ets1-␣Ets1 complex comprised both Ets1 and Smad3. Collectively, these data imply that Smad3 and Ets1 can simultaneously bind to the PTHrP P3 promoter. Note that the anti-Ets1 antibody that recognizes the C terminus increased Ets1 binding to the probe. The C terminus is part of the inhibitory module that regulates Ets1 DNA binding activity (50). Interestingly, the amount of Ets1 bound to the probe also increased when Smad3 was present (compare intensity of C4 band in lane 3 with intensity of Ets1 band in lane 5). It may suggest that Smad3 stimulates Ets1 binding to the probe.
Protein Kinase C (PKC) Inhibitor Calphostin C Inhibits Endogenous Ets1 Expression and TGF␤1-mediated Up-regulation of PTHrP P3 Transcripts in MDA-MB-231 Cells-In MDA-MB-231 cells, as opposed to MCF-7 cells, PKC activity is constitutively high (51). Activation of PKC by phorbol ester has been shown to stimulate Ets1 expression (52)(53)(54). It is, therefore, possible that, in MDA-MB-231 cells, Ets1 expression is at least in part dependent on PKC activity. To test this hypothesis, we treated MDA-MB-231 cells independently with two inhibitors of PKC, staurosporine and calphostin C. As judged by Western blot analysis, both inhibitors were able to strongly inhibit endogenous Ets1 protein expression in these cells (Fig. 10, A and  B). Treatment of cells with 1 M calphostin for 3 h or with 58 nM staurosporine for 24 h completely abrogated endogenous Ets1 protein expression. Shorter incubation times (1 or 2 h) with calphostin C resulted in the appearance of a second, slower migrating Ets1-specific band (Fig. 10B). This band most likely represents a form of Ets1 that is specifically phosphorylated on  Ϫ28 (B and C) or the Ϫ79/Ϫ28 PTHrP probe (C). Flag-Smad3 and Smad4 were overexpressed by performing transient transfection using electroporation. Cells were transfected with either 4 g of Flag-Smad3 and/or 2 g of Smad4 plasmids or 6 g of control vector as indicated and incubated for 5 h in the presence or absence of TGF␤1 (5 ng/ml). ␣Flag and ␣Smad4 denote complexes that were formed after addition of the anti-Flag or anti-Smad4 antibody, respectively. the exon VII domain. Such phosphorylations, demonstrated to increase the apparent molecular weight of Ets1, lead to the inhibition of the DNA binding activity of Ets1 (55) and to a reduced half-life of the Ets1 protein (56). Calphostin C also inhibited Ets1 RNA expression. After 2 h of treatment, calphostin C completely down-regulated the level of the major 6.8kilobase pair Ets1 transcript (Fig. 10C). Of note, calphostin C also inhibited the endogenous expression of Ets1 in invasive MDA-MB-435 breast cancer cells (data not shown). Strikingly, in contrast to Ets1 expression levels, the level of Ets2 RNA or protein was not affected by the PKC inhibitors (Fig. 10, A-C). The interference of PKC inhibitors with Ets1 synthesis in MDA-MB-231 cells prompted us to study the effect of calphostin C on basal and TGF␤1-dependent PTHrP. As shown in Fig.  10D, treatment of MDA-MB-231 cells with calphostin C for 3 h only weakly inhibited the basal levels of the PTHrP P3-specific transcript or of all PTHrP transcripts combined (Fig. 10D). However, the TGF␤1-mediated up-regulation of the PTHrP P3-specific RNA was completely abrogated by calphostin C. This effect was not resulting from an interference of calphostin C with TGF␤1 signaling in general, because the TGF␤1-dependent increase in PAI-1-RNA level was nearly unaffected by this agent (Fig. 10D). Neither could this effect be attributed to the possibility that TGF␤1 acts on PTHrP P3 RNA expression through PKC, as phorbol ester failed to mimic TGF␤1 action on PTHrP expression in MDA-MB-231 cells (data not shown). We conclude from these data that Ets1 is involved in TGF␤-dependent regulation of the endogenous PTHrP gene in MDA-MB-231 cells. DISCUSSION The data presented here show that, in MDA-MB-231 cells, TGF␤1 stimulates PTHrP synthesis by specifically up-regulating the level of PTHrP promoter P3-derived RNA. This effect is, at least in part, the result of an increase in P3-dependent transcription. We further demonstrate that TGF␤1 and Smad3 can activate the PTHrP P3 promoter. This depended on the Ets binding site, the Sp1 element, and a newly identified Smad3binding site. We report a Smad3/Ets1 synergism that substantially enhanced the stimulatory effect of TGF␤ on the P3 promoter. The significance of this finding is demonstrated by the observation that abrogation of endogenous Ets1 expression by calphostin C resulted in complete loss of TGF␤'s ability to induce PTHrP expression in vivo, whereas it had no effect on TGF␤-mediated PAI-1 expression, which depends on a Smad3/ TFE-3 synergism (31).
In contrast to MDA-MB-231 cells, MCF-7 cells failed to support an Ets1/Smad3 synergism. In these cells, TGF␤1 had also no effect on the endogenous PTHrP expression. One reason for this is certainly a defect in proper expression of the TGF␤receptor II. In addition, other differences between MCF-7 cells and MDA-MB-231 cells may be important for the TGF␤ response. In contrast to MDA-MB-231 cells, MCF-7 cells do not express Ets1 endogenously ( Fig. 6C and data not shown) and even prevent stimulation of the P3 promoter by ectopically expressed Ets1 (Fig. 6A). On the other hand, MCF-7 cells do support P3 promoter activation by ectopically expressed Ets2 (data not shown), which failed to stimulate the same promoter in MDA-MB-231 cells (Fig. 5). Unlike Ets1, Ets2 is endogenously expressed by both cell lines (data not shown). Thus, additional factors seem to be required for the ability of Ets1 or Ets2 to regulate P3 promoter activity. In support of this notion, Ets2 was found to activate the murine counterpart of the human PTHrP P3 promoter in P19 embryonal carcinoma cells only when these cells had been stimulated by retinoic acid (25). Erk1/2 kinases may be important for Ets1-dependent activation of the P3 promoter. These kinases have been shown to mediate superactivation of Ets1 through Ras (57), known to stimulate PTHrP expression (59,60), and are constitutively active in MDA-MB-231, while inactive in MCF-7 cells (58). We are currently testing the possibility of an involvement of Ras and/or Erk1/2 in Ets1/Smad3-dependent activation of the PTHrP P3 promoter.
The PKC inhibitor calphostin C strongly interfered with endogenous Ets1 expression in MDA-MB-231 cells and, concomitantly, prevented induction of PTHrP expression by TGF␤1. At the same time, endogenous Ets2 synthesis and TGF␤1-mediated Ets-independent expression of PAI-1 were not affected. This suggests that calphostin C inhibited specifically the TGF␤1 effect on P3-derived PTHrP production by inhibiting Ets1 synthesis. This supports the notion that Ets1 is involved in TGF␤-mediated activation of PTHrP expression in MDA-MB-231 cells. Interestingly, calphostin C had no substantial effect on the PTHrP P3-specific RNA level in the absence of TGF␤1, suggesting that Ets1 is dispensable for basal P3 promoter activity in these cells. On the other hand, we show that the Ets binding site is essential for basal P3 promoter activity in MDA-MB-231 cells (Fig. 3D). It is possible that a different Ets protein is responsible for maintaining basal transcription from the P3 promoter. A potential candidate is Ese-1, which FIG. 8. The same mutation in the AGAC box that abrogated Smad3-dependent activation of the PTHrP P3 promoter inhibits Smad3 binding to this element. A, schematic of the Ϫ79/Ϫ28, the Ϫ51/Ϫ28 wild type, and AGAC mutant PTHrP P3-specific probes used for EMSA. B, EMSA of Flag-Smad3 containing MDA-MB-231 nuclear extracts (NE) incubated either with the Ϫ51/Ϫ28 wild type or the AGAC mutant PTHrP probe. Cells were electroporated with 4 g of the Flag-Smad3 plasmid and treated with TGF␤1 (5 ng/ml) for 5 h. C, EMSA of the same extracts as in B but treated with the Ϫ79/Ϫ28 PTHrP probe. As a competitor either the wild type or the mutant version of the Ϫ51/Ϫ28 PTHrP oligonucleotide in 10-or 25-fold molar excess over the probe or a nonspecific oligonucleotide (ns) in 50-fold excess over the probe was used.
was as capable as Ets1 in activating that promoter.
Ets1 and Smad3 share the ability to synergize with Sp1 (22,35). Interestingly, Ets1 and Smad3 recognition elements within the PTHrP P3 promoter are separated by an Sp1 DNA binding site. This DNA sequence has previously been shown to be crucial for Ets1-dependent activation of this promoter (22). Here, we demonstrate that a mutation in the Sp1 binding motif had the same effect on the TGF␤1-dependent Smad3-or Smad3/Ets1-induced activation as a mutation in the AGAC box site. This suggests that these elements are equally important for the Ets1/Smad3 synergism. On the other hand, electromobility shift assays using oligonucleotides containing an Sp1 consensus sequence did not reveal Sp1 or an Sp1-like protein to be present in the Smad3 or Smad3/Ets1 complexes formed with the PTHrP DNA probe (data not shown). It cannot be ruled, however, that the conditions which were required to analyze the interactions of Smad3 and Ets1 with the PTHrP-specific DNA probe were not favorable for Sp1 binding. Further studies are required to clarify the role of Sp1 for the Smad3/Ets1 synergism.
Ets1 was originally described as a T cell-specific Ets transcription factor (61) that is important for T-cell survival (62,63). Now, a new role of Ets1 as a critical factor promoting tumor invasion and metastasis is emerging (38). Consistent with this notion, we could detect the Ets1 protein in invasive, but not in noninvasive, breast cancer cells (data not shown). Likewise, TGF␤ has been shown to play a crucial role for tumor progression at later stages (64). It is interesting that Ets1 and TGF␤ have some features in common and that some of their activities depend on each other, e.g. both Ets1 and TGF␤ can cooperate with Ras to increase invasiveness (65)(66)(67). In addition, both proteins are able to activate the gene coding for urokinase type plasminogen activator (67,68), whose expression is often associated with Ets1 expression (69,70). Regulation of the urokinase type plasminogen activator promoter by either Ets1 or by Smad3/TGF␤ requires AP1 (32,33,39) which, like Sp1, can synergize with both Ets1 and Smad3. Furthermore, TGF␤-dependent activation of the human germline C␣1 gene has been reported to be dependent on Ets1 and AML-1 (71). Additionally, AML-1 has been shown to be able to cooperate with Ets1 as well as with Smad3 (34, 40, 41). Finally, the Ϫ28 PTHrP probe. Cells were electroporated with 4 g of the Flag-Smad3 and/or of the 4 g Ets1 plasmids or with 8 g control vector and incubated in the presence of 5 ng/ml TGF␤1 for 5 h. An oligonucleotide (Ets cons.) bearing the Ets consensus sequence and the Ϫ51/Ϫ28 PTHrP oligonucleotide (AGAC box) were used as competitor DNAs in 50-fold excess over the probe. Ets1/␣Ets1 and ␣Flag denote complexes formed after addition of anti-Ets1 or anti-Flag antibody, respectively. TGF␤ receptor II expression and hence the regulation of Smad3 activity by TGF␤ has been shown to be dependent on Ets proteins (72), whereas, inversely, TGF␤ seems to be able to increase Ets1 expression (data not shown). Collectively, these observations imply that functional interactions between Ets transcription factors and TGF␤, such as those involving a Smad3/Ets1 synergism, may be important for the regulation not only of the PTHrP gene, but also of other cellular genes. Given the growing body of evidence suggesting a correlation of Ets protein and TGF␤ activities with cellular invasiveness, one could speculate that, in particular, TGF␤/Ets interactions may play a role for the regulation of Ets-and TGF␤-responsive genes involved in the acquisition of an invasive phenotype.