Blockade of transforming growth factor-beta signaling does not abrogate antiestrogen-induced growth inhibition of human breast carcinoma cells.

We have studied the role of autocrine transforming growth factor-β (TGF-β) signaling on antiestrogen-mediated growth inhibition of hormone-dependent T47D and MCF-7 human breast carcinoma cells. Tamoxifen treatment increased the secretion of TGF-β activity into serum-free cell medium and the cellular content of affinity cross-linked type I and III TGF-β receptors in both cell lines. Anti-pan-TGF-β antibodies did not block anti-estrogen-induced recruitment in G1 and inhibition of anchorage-dependent and -independent growth of both cell lines. Early passage MCF-7 cells, which exhibit detectable type II TGF-β receptors at the cell surface and exquisite sensitivity to exogenous TGF-β1, were transfected with a tetracycline-controllable dominant-negative TGF-βRII (ΔRII) construct. Although the TGF-β1 response was blocked by removal of tetracycline in MCF-7/ΔRII cells, tamoxifen-mediated suppression of Rb phosphorylation, recruitment in G1, and inhibition of cell proliferation were identical in the presence and absence of tetracycline. TGF-β1 treatment up-regulated the Cdk inhibitor p21 and induced its association with Cdk2 in MCF-7 cells; these responses were blocked by the ΔRII transgene product. In MCF-7 cells with a functional TGF-β signaling pathway, tamoxifen did not up-regulate p21 nor did it induce association of p21 with Cdk2, suggesting alternative mechanisms for antiestrogen-mediated cytostasis. Finally, transfection of late-passage, TGF-β1 unresponsive MCF-7 cells with high levels of TGF-βRII restored TGF-β1-induced growth inhibition but did not enhance tamoxifen response in culture. Taken together these data strongly argue against any role for TGF-β signaling on tamoxifen-mediated growth inhibition of hormone-dependent breast cancer cells.

tors of cellular proliferation, differentiation, morphogenesis as well as extracellular matrix formation, extracellular proteolysis, and inflammation (1)(2)(3). In epithelial cells a major effect of TGF-␤ is its ability to inhibit cell proliferation (4). Three different mammalian TGF-␤ isoforms (TGF-␤1, -␤2, and -␤3) encoded by different genes have been identified, and they exhibit similar effects in a variety of biological assays (5). Three membrane ligand-binding proteins with sizes of 53, 73, and ϳ250 kDa have been reported as TGF-␤ receptors type I, II, and III, respectively. Type I and II receptors are transmembrane serine/threonine kinases directly involved in signal transduction, while the type III receptor functions mainly by presenting the ligand to the signaling type I and II receptors and as a storage protein (6,7). Both TGF-␤ and its receptor molecules are expressed ubiquitously by normal and transformed cells.
Both signaling receptors seem to be needed for TGF-␤ responsiveness (8), and type II receptor expression correlates with the anti-proliferative activity of TGF-␤ (9, 10). TGF-␤ arrests cell growth in the G 1 phase of the cell cycle and most probably affects multiple signaling pathways (11). TGF-␤ has been shown to retain the retinoblastoma susceptibility gene (pRb) in a hypophosphorylated form, which prevents cells from entering the S phase (12). By regulating the formation of Cdkcyclin complexes and their inhibitor levels, TGF-␤ contributes to the accumulation of hypophosphorylated pRb (13). In several cell lines, TGF-␤1 also induces rapid down-regulation of c-myc expression (14,15), suggesting this is an additional mechanism by which these peptides suppress cell growth.
All three TGF-␤ isoforms are expressed in mouse mammary gland, and there are data supporting their role in the development of the mouse mammary gland (16,17). Exogenous TGF-␤ administrated by slow release pellets or tissue-specific expression of active TGF-␤1 in the mammary gland of transgenic mice leads to ductal hypoplasia and suppression of ductal branching (16,18,19).
Normal and tumorigenic human breast epithelial cells in culture express TGF-␤1 mRNA and secrete TGF-␤ receptor binding activity into their medium (20,21). Published data support the notion that endogenous TGF-␤s function as autocrine growth regulators of breast cancer cell proliferation (22,23). Antibodies that neutralize mature TGF-␤s stimulate the proliferation of estrogen-independent breast cancer cell lines (23). Growth stimulation of estrogen-dependent breast cancer cells with estradiol or the testosterone derivative norethindrone is associated with down-regulation of TGF-␤2 and -␤3 mRNAs (24 -26). Growth inhibition of these cell lines by the antiestrogens tamoxifen or toremifene and the progestin analogue gestodene is associated with enhanced TGF-␤1 mRNA expression or increased secretion of TGF-␤ bioactivity or protein synthesis without associated mRNA changes (22,27,28), thus leading to the hypothesis that autocrine TGF-␤ signaling contributed to antiestrogen's actions. Some reports, however, argue against TGF-␤s' role in the growth inhibitory response to antiestrogens. MCF-7 and T47D breast cancer cells can exhibit resistance to TGF-␤1-mediated growth inhibition despite retaining sensitivity to tamoxifen (29,30). In addition, T47D and CAMA-1 breast cancer lines, which lack mRNA for TGF-␤RII and hence response to exogenous TGF-␤1, remain sensitive to the cytostatic effect of tamoxifen (31,32). By using anti-TGF-␤ neutralizing antibodies as well as dominant negative TGF-␤RII constructs in estrogen-dependent, tamoxifen-sensitive human breast cancer cells, we have formally tested in this study the role of endogenous TGF-␤ signaling on the cellular response to antiestrogens in breast carcinoma.
Collection of Cell-conditioned Medium (CM) and TGF-␤ Radioreceptor Assay-Secreted TGF-␤ bioactivity was measured in serum-free IMEM conditioned for 24 h by adherent breast cancer cells as described previously (23). When indicated, the CM was acidified with 1 N HCl to pH 1.5 for 1 h at 4°C and reneutralized with 1 N NaOH before testing. CM was then tested in a TGF-␤ radioreceptor assay (23) utilizing AKR-2B mouse fibroblasts as indicator cells. Binding was performed in six-well plates with 1 ml/well binding buffer (128 mM NaCl, 5 mM KCl, 5 mM MgSO 4 , 1.2 mM CaCl 2 , 50 mM Hepes, pH 7.5, 0.2% BSA) containing 0.25 ng/ml 125 I-TGF-␤1 (specific activity 173 Ci/g; DuPont NEN) with or without variable volumes of CM for 4 h at 4°C. Human recombinant TGF-␤1 (Genentech) was used to generate a standard curve from which the receptor binding activity in CM was calculated. In this assay, TGF-␤1 and TGF-␤2 are equipotent in displacing 125 I-TGF-␤1 binding (35). Addition of 1 M tamoxifen directly to the radioreceptor assay has no effect on TGF-␤1 binding by AKR-2B cells. 3 125 I-TGF-␤ Binding and Affinity Cross-linking-Binding was performed on intact adherent cells in 100-mm tissue culture dishes. Cells were incubated in binding buffer (see above) containing 1 ng/ml 125 I-TGF-␤1 with or without 100-fold excess unlabeled TGF-␤1 for 4 h at 4°C with gentle rocking. After two washes with ice-cold binding buffer without BSA on ice, the bound 125 I-TGF-␤1 was cross-linked to cell surface receptors with 50 M disuccinimidyl suberate (Pierce) for 15 min at 4°C in 10 ml of binding buffer without BSA. Cells were then scraped and solubilized as described previously (23). The samples were subjected to 5-10% gradient SDS-PAGE and labeled receptors visualized by autoradiography. In some cases, to enhance the sensitivity of the assay, approximately 5 ϫ 10 7 cells were labeled in 1 ml of binding buffer with 5 ng/ml (0.2 nM) 125 I-TGF-␤1 and cross-linked under the same conditions as above. After solubilization, cross-linked cell lysates were precipitated overnight at 4°C with the C-16 TGF-␤RII polyclonal antibody (Santa Cruz Biotechnology) followed by protein A-Sepharose for 1 h. Immune complexes were then resolved by 5-10% gradient SDS-PAGE and visualized by autoradiography.
Analysis of Cell Cycle Distribution-Cells were trypsinized, suspended in cold PBS, and fixed in absolute ethanol (final concentration 67%). After overnight refrigeration, DNA was stained in the dark with 50 g/ml of propidium iodide (PI) containing 125 units/ml protease-free RNase (Calbiochem), both in PBS. DNA histograms were analyzed in a FACScan flow cytometer (Beckton Dickinson, Mansfield, MA) and modeled off-line using Modfit software (Verity, Topsham, ME).
Cell Proliferation Assays-Cells (2 ϫ 10 4 /well) were plated in 24-well dishes in regular growth medium. The following day, 1 M tamoxifen or TGF-␤1 was added. After 4 -5 days, cells from triplicate samples were assessed in a Coulter Z1 counter (Coulter Electronics Limited, Beds., United Kingdom). For testing of anchorage-independent growth, a 1-ml top layer containing a single-cell suspension of 3 ϫ 10 4 cells, 0.8% agarose (Sea-Plaque, FMC Corp. BioProducts, Rockland, ME), IMEM, 10% FCS, and 10 mM Hepes, with or without different concentrations of TGF-␤1 or 1 M tamoxifen was added to a 1-ml bottom layer of 0.8% agarose, 10% FCS in 35-mm dishes. Dishes were incubated in a humidified 5% CO 2 incubator at 37°C, and colonies measuring Ն50 m counted after 10 days using an Omnicon Stem Model II Image Analyzer (Bausch & Lomb, Rochester, NY).

Tamoxifen Increases Secreted TGF-␤ Bioactivity and TGF-␤
Binding in MCF-7 and T47D Cells-We first examined the modulation of TGF-␤ secretion by tamoxifen in the ER-positive MCF-7 and T47D human breast carcinoma cell lines. Exponentially growing cells were treated with 0.1% ethanol (controls) or 1 M tamoxifen for 24 h in serum-free medium. TGF-␤ activity was measured in a 125 I-TGF-␤1 radioreceptor assay. In the absence of acid activation in vitro, TGF-␤ activity was below the detection limit of the binding assay, indicating that the majority of the secreted TGF-␤ was in a latent form. Tamoxifen induced increases of approximately 30-and 4-fold in the secretion of acid-activable TGF-␤ activity in MCF-7 and T47D cultures, respectively (Table I).
To test whether endogenous TGF-␤ ligands in response to the antiestrogen were masking endogenous TGF-␤ receptors, we examined the effect of tamoxifen on 125 I-TGF-␤1 binding and the cellular content of affinity cross-linked TGF-␤ receptors. Both MCF-7 and T47D cells exhibit type I and type III TGF-␤ receptors at the cell surface, whereas TGF-␤RII was undetectable (Fig. 1). To enhance the sensitivity of the assay, we labeled 5 ϫ 10 7 cells with 10 ng/ml 125 I-TGF-␤1 in a 1-ml suspension followed by cross-linking and immunoprecipitation with TGF-␤RII antibodies, but were still unable to detect any binding by type II receptors at 4°C (data not shown). Treatment of cells with 1 M tamoxifen for 3 days up-regulated type I and III TGF-␤ receptors in both cell lines (Fig. 1). This up-regulation was first obvious at 48 h and was maximal at 72-96 h. Bound cpm corrected for protein content in crosslinked cell lysates were Ն5-fold higher in tamoxifen-treated than in control lysates. A mild acid wash (pH 2.8) modestly increased 125 I-TGF-␤1 binding in control and tamoxifentreated cells, 3 arguing against significant masking of TGF-␤ binding sites by autocrine ligands in the presence or absence of the antiestrogen.
Neutralizing Anti-TGF-␤ Antibodies Do Not Block Tamoxifen Growth Inhibitory Action-To determine whether the enhanced TGF-␤ secreted activity was mediating tamoxifen actions, we performed studies with anti-TGF-␤ antibodies. These antibodies have been shown to stimulate the proliferation of breast tumor cells with an operative autocrine TGF-␤ pathway by neutralizing endogenous mature TGF-␤s (23). Cells were treated with 1 M tamoxifen or 0.1% ethanol (control) in the presence of IgGs (100 g/ml) for 3 days and cell cycle distribution analyzed by flow cytometry of PI-labeled DNA. In both lines, tamoxifen induced a marked decrease of the percentage of cells in S phase and an accumulation in G 1 . Neither the anti-pan-TGF-␤ 2G7 or the anti-TGF-␤1 4A11 antibodies nor their respective controls, the 12H5 IgG 2 and an irrelevant IgG 1 , altered tamoxifen-mediated cytostasis ( Fig. 2A). Identical results were obtained in a colony-forming soft agarose assay. In this assay, both MCF-7 and T47D cells were markedly inhibited by the antiestrogen; this inhibition was not altered by any of the TGF-␤ antibodies (Fig. 2B). By themselves, the IgGs utilized had no growth effects on either cell line (data not shown).
High Expression of Functional TGF-␤RII Does Not Enhance Tamoxifen Action in MCF-7 Cells-Transfection with high levels of a tetracycline-repressible TGF-␤RII expression vector into MCF-7 cells (MCF-7/RII cells, Ref. 9) restores sensitivity to growth inhibition by exogenous TGF-␤. Therefore, in these cells, we tested whether up-regulation of an operative TGF-␤ signaling pathway would enhance antiestrogen-induced growth inhibition. MCF-7/RII cells were plated with or without 0.1 M tetracycline. The following day, 1 M tamoxifen or 0.1%  EtOH were added to the monolayers and cell proliferation assessed 4 days later. There was a similar 60% inhibition of growth in tamoxifen-treated monolayers relative to controls in the absence or presence of transfected TGF-␤RII. Similar to its effect on endogenous TGF-␤RI and TGF-␤RIII in MCF-7 and T47D cells (Fig. 1), tamoxifen also increased the cellular content of TGF-␤RII in MCF-7/RII cells with a simultaneous upregulation of type I receptor sites (Fig. 3B). Similar to the parental MCF-7 cells (Table I), incubation of MCF-7/RII cells with 1 M tamoxifen increased the secretion of TGF-␤ activity from 0.1 to 1.5 ng/10 6 cells/24 h as measured by radioreceptor assay of conditioned medium. Since these cells have an impaired ability for anchorage-independent growth, soft agarose colony-forming assays to test the modulation of tamoxifen sensitivity by TGF-␤RII were not useful.

Expression of a Dominant Negative TGF-␤RII Does Not Block
Cellular Responses to Anti-estrogens in MCF-7 Cells-Since the neutralizing TGF-␤ antibodies may not be effective in blocking a potential ligand/receptor intracellular coupling or a direct ligand-independent effect of tamoxifen on TGF-␤RII, we used a dominant negative approach to block TGF-␤RII function in a cell-autonomous experimental system. For this purpose we used early passage MCF-7 breast cancer cells, which, different than the late passage cells used above, exhibit detectable TGF-␤RII protein at the cell surface and are markedly growth inhibited by exogenous TGF-␤s (36). Expression of a tetracyclinerepressible dominant negative type II TGF-␤ receptor in these cells (MCF-7/⌬RII) blocks cellular responses to exogenous TGF-␤1. 2 We first studied the proliferation effects of prolonged tamoxifen treatment in cells preincubated (for 24 h) or not with 0.1 M tetracycline, concentration known to maximally induce the ⌬RII mutant protein. 2 Both anchorage-dependent and -independent proliferation were inhibited by 1 M tamoxifen in the presence or absence of tetracycline (Fig. 4). Similar results were obtained with 5-10 M amounts of the antiestrogen toremifene (27) in monolayer culture (data not shown). Both TGF-␤1 (4, 12) and antiestrogens (38) can recruit sensitive cells in the G 1 phase of the cell cycle while suppressing Rb phosphorylation (12,39). Therefore, we studied the impact of the ⌬RII mutant on tamoxifen-mediated cell cycle arrest and Rb phosphorylation in MCF-7/⌬RII cells. A 72-h incubation with 1 M tamoxifen markedly increased the proportion of cells in G 1 , while reducing those in G 2 M and S phases of the cycle. These changes in response to antiestrogen were almost identical in the presence or absence of tetracycline (Fig. 5). Consistent with this result, a 24-h incubation with 1 M tamoxifen reduced Rb hyperphosphorylation in MCF-7/⌬RII cells under conditions in which the ⌬RII mutant type II receptor was expressed or not (Fig. 6) further arguing against any role for endogenous TGF-␤ signaling on antiestrogen-mediated G 1 arrest.
TGF-␤1 but Not Tamoxifen Induces p21 WAF1/CIP1 in MCF-7 Cells-TGF-␤1 has been shown to inhibit the kinase activity of cyclin E-Cdk2 complexes (7) and hence suppress Rb phosphorylation. One reported mechanism for such inhibition is induction of the Cdk inhibitor p21, which then associates with the cyclin E-Cdk2 complex and inhibits its kinase (40,41). A similar induction of p21 has been reported recently by the antiestrogen ICI182780 in MCF-7 breast cancer cells (39). Therefore, we examined whether TGF-␤1 and tamoxifen suppressed Rb phosphorylation by inducing p21 in MCF-7/⌬RII cells. An overnight incubation with 1 ng/ml TGF-␤1 in the presence of tetracycline markedly up-regulated p21 protein levels and induced its association with Cdk2, as supported by coprecipitation of p21 with Cdk2 antibodies. This induction and association with Cdk2 were abrogated by removal of tetracycline, which eliminates endogenous TGF-␤RII signaling (Fig. 7). Interestingly, 1 M tamoxifen did not induce p21 nor did it induce its association with Cdk2, suggesting its suppressive action on Rb phosphorylation is mediated by a mechanism(s) other than up-regulation of autocrine growth inhibitory TGF-␤s. Prolongation of tamoxifen treatment to 72 h still did not induce p21 protein. A 24-h treatment of MCF-7/⌬RII cells with 1 M tamoxifen or 1 ng/ml TGF-␤1 in the presence or absence of tetracycline, did not induce the Cdk inhibitor p27 as measured by immunoblot analysis (data not shown). DISCUSSION It is proposed that antiestrogens induce growth inhibition of human breast tumor cells by up-regulating expression and/or secretion of TGF-␤s. We have directly tested this hypothesis in MCF-7 and T47D breast carcinoma cells, which exhibit enhanced secretion of TGF-␤ bioactivity upon treatment with the antiestrogen tamoxifen. All this secreted TGF-␤ activity was in a latent form requiring acid activation in vitro for it to be detected. This may reflect the inability of the radioreceptor assay to detect TGF-␤s already utilized by the cells as well as to estimate in situ activation of TGF-␤s at 37°C over a more prolonged time and in the presence of a potential target cell. Supporting the latter possibility, medium conditioned by MCF-7 cells in the presence of tamoxifen inhibits the growth of cocultured tamoxifen-insensitive MDA-231 breast cancer cells. This inhibition by MCF-7 cell medium was reversed by anti-TGF-␤ antibodies (22). MCF-7 cells but not the T47D line express TGF-␤RII mRNA (32). Although the growth inhibitory response of both cell lines to high concentrations of exogenous TGF-␤1 and TGF-␤2 is minor (36,37), this does not rule out a potential response to lower concentrations of endogenous TGF-␤s. The absence of TGF-␤RII mRNA and protein, presumably indispensable for TGF-␤ cellular responses, in tamoxifen-sensitive T47D cells argues per se against TGF-␤'s involvement in antiestrogen response. However, these cells bind TGF-␤1 (37, Fig. 1) and, in response to the progestin analog gestodene, secrete 90-fold higher levels of TGF-␤1 and -␤2 proteins and become growtharrested (28). This inhibitory effect of gestodene in T47D cells is partially reversed by a polyclonal TGF-␤ antiserum, suggesting these cells are perhaps responsive to autocrine TGF-␤s  Fig. 6 in the presence or absence of tetracycline and the following day treated for 24 h with 0.1% ethanol (ctl), 1 M tamoxifen (Tam), or 1 ng/ml recombinant TGF-␤1. After this incubation, total cell lysates were prepared as indicated under "Materials and Methods" and 150 g of protein/treatment condition subjected to p21 immunoblotting (top). To assess association of p21 with Cdk2, 300 g of protein/treatment condition were precipitated with 2 g of a Cdk2 antiserum (Cdk2ip). Coprecipitated p21 was assessed by immunoblot analysis of immune complexes as indicated under "Materials and Methods" (bottom). despite lacking TGF-␤RII.
We first used antibodies that neutralize mature TGF-␤s in an attempt to block antiestrogen action. This approach has been shown to partially block retinoic acid-mediated inhibition of keratinocytes by counteracting the autocrine action of TGF-␤2 (42). In our study, blockade of all three mammalian TGF-␤ isoforms with different monoclonal antibodies did not alter tamoxifen-induced cell cycle arrest or inhibition of MCF-7 and T47D colony growth. Furthermore, transfection of TGF-␤RII into MCF-7 cells did not enhance their sensitivity to tamoxifen, even though the antiestrogen did enhance the secretion of TGF-␤ activity and binding by these cells. These results also argue that, at pharmacologically achievable concentrations, tamoxifen does not require the contribution of autocrine TGF-␤ to induce growth inhibition.
Treatment with tamoxifen increased ligand binding by all three TGF-␤ receptor types. This may argue against an autocrine interaction between secreted TGF-␤s and endogenous receptors as shown with mouse keratinocytes by Glick et al. (42). Upon terminal differentiation, these cells exhibit a 10 -20-fold increase in TGF-␤2 mRNA and peptide with simultaneous down-regulation of type I and II TGF-␤ receptors available for affinity cross-linking; a mild acid wash significantly increased the number of receptor sites in differentiated keratinocytes suggesting masking of TGF-␤ receptors by endogenous ligand (43). On the other hand, other steroid molecules can up-regulate TGF-␤ receptors at the protein and/or mRNA level despite a simultaneous increase in expression/secretion of TGF-␤ ligands (44). Further work is needed to study the mechanism(s) by which tamoxifen up-regulates TGF-␤ receptors in MCF-7 and T47D cells. Preliminary experiments with MCF-7 cells, however, failed to show an increase in steady-state TGF-␤RII mRNA levels after a 24-h incubation with 1 M tamoxifen, 4 arguing against a transcriptional effect to explain the result with MCF-7/RII cells (Fig. 3B).
TGF-␤ activation can occur locally within the cell surface of target cells (45,46). Neutralizing anti-TGF-␤ antibodies may not be able to block TGF-␤ activity to a threshold required for the reversal of growth inhibitory signals or alter a ligandindependent direct effect of tamoxifen on TGF-␤RII signaling. Therefore, we examined the effect of a kinase negative truncated TGF-␤RII (⌬RII) on tamoxifen response in early passage MCF-7 cells. These cells are more sensitive than late passage MCF-7 cells to exogenous TGF-␤1 and exhibit detectable levels of TGF-␤RII at the cell surface thus providing an appropriate model to test directly both TGF-␤1 and antiestrogen response. TGF-␤1 responses were abrogated by the dominant negative ⌬RII mutant. 2 However, tamoxifen-mediated cell cycle arrest, suppression of Rb phosphorylation, and antiproliferative effects in these MCF-7 cells were identical with or without endogenous TGF-␤RII signaling, disproving any major role for autocrine TGF-␤s on the response to antiestrogens.
Once a dissociation between TGF-␤sand tamoxifen-mediated growth inhibition was established, we studied whether they independently suppressed Rb phosphorylation by similar mechanisms in the MCF-7/⌬RII cells. The pure antiestrogen ICI182780 and TGF-␤1 induce p21 and p27, which, by complexing with the cyclin E-Cdk2 complex, prevent Rb phosphorylation and hence progression beyond the G 1 phase of the cell cycle (7, 39 -41, 47). Exogenous TGF-␤1 but not tamoxifen induced p21 as well as association of p21 with Cdk2. These responses were abrogated by the ⌬RII mutant receptor, supporting the need of intact TGF-␤RII signaling to elicit ligand-mediated effects on the Cdk2 inhibitor p21. Neither tamoxifen nor TGF-␤1 induced p27 in MCF-⌬RII cells. In addition to the dissociation between both growth inhibitory pathways at a cellular level, these data with p21 further suggest that TGF-␤s and tamoxifen suppress Rb phosphorylation by independent molecular mechanisms.
These data, generated with cell-autonomous experimental systems, do not rule out a possible role for antiestrogen-induced TGF-␤s in the anti-tumor response to tamoxifen in clinical breast carcinoma by a paracrine/endocrine mechanism. Conflicting data have been published on this topic. Butta et al. reported that 3 months of tamoxifen therapy resulted in ERindependent enhanced TGF-␤1 staining around stromal fibroblasts in breast tumor biopsies (48). The correlation between antiestrogen-induced enhancement of peritumoral TGF-␤1 protein and a clinical response was not reported in this study. In two other studies, a rise in the circulating level of TGF-␤2 (49) or in the tumor levels of TGF-␤2 mRNA (50) correlated with a clinical response to antiestrogens, suggesting up-regulation of TGF-␤s is a surrogate marker or epiphenomenon of an antitumor effect. On the other hand, a more recent immunohistochemical study in 19 patients failed to show alterations in TGF-␤1 staining with intervening tamoxifen therapy, despite a Ͼ50% clinical response rate (51). Transfection of MCF-7 cells with a TGF-␤1 expression vector does not alter tamoxifen sensitivity (52). Finally, breast tumors unresponsive to tamoxifen, when rebiopsied, expressed significantly higher levels of TGF-␤1 mRNA than clinically responsive tumors (53). A causal association between ligand overexpression and the antiestrogen-resistant phenotype, if any, would require additional mechanistic studies.
In any event, the data presented strongly argue against a significant involvement of TGF-␤ ligands and receptor signaling on the growth inhibition of human breast carcinoma cells by antiestrogens. Prospective epidemiologic studies will likely address whether treatment-induced up-regulation of TGF-␤s expression in tumors in situ can be used as a marker of response (or lack of response) to antiestrogens. Although it is still possible that autocrine/paracrine TGF-␤s can be involved in antiestrogen response in some mammary carcinomas, the effect of TGF-␤s on the host's immune system and on tumor's stroma, cell adhesion, and angiogenesis (reviewed in Ref. 54) can easily mask the net contribution of this putative autocrine pathway to breast tumor cell viability and progression by indirectly favoring breast tumor maintenance.