Independent Regulation of Growth and SMAD-mediated Transcription by Transforming Growth Factor β in Human Melanoma Cells¹

TGF³-β superfamily members (activin, bone morphogenic proteins, TGF-βs, decapentaplegic) are multifunctional cytokines that affect cell proliferation, interaction with the extracellular matrix, differentiation, and/or survival. All TGF-β family members signal through multimeric serine/threonine kinase receptor complexes on the cell surface, which phosphorylate cytoplasmic mediators called SMADs. Specifically, upon phosphorylation by activated TGF-β receptors, ligand/receptor-specific SMADs (SMAD2 and/or SMAD3 in the case of the three mammalian TGF-β isoforms) associate with SMAD4 and translocate as a complex to the nucleus to transactivate promoters containing SBSs (reviewed in Ref. 1). Many, if not all, of the biological effects of TGF-β are considered to be SMAD dependent, through transcriptional regulation of extracellular matrix, adhesion, and growth regulatory genes.

Cell cycle progression of normal epithelial cells is inhibited by exogenous TGF-β, whereas malignant epithelial cells are often resistant to the growth-inhibitory effects of TGF-β. Acquired resistance to the growth inhibitory effects of TGF-β is generally considered as a mechanism by which malignant tumor cells subvert normal growth controls. In some epithelial malignancies, TGF-β resistance is associated with functional inactivation of either the TGF-β receptors (2, 3) or of signal transducers of the SMAD family (4), suggesting that resistance to TGF-β-induced growth inhibition is due to disrupted TGF-β/SMAD-dependent transcriptional regulation. In particular, SMAD4, also known as DPC4 (deleted in pancreatic carcinoma), has frequently been found to be nonfunctional, either by chromosomal deletion or by mutation in pancreatic and, to a lesser degree, other epithelial malignancies (5). TGF-β receptor and/or SMAD activation have been shown to up-regulate mRNA expression of the cell cycle inhibitors p21 (6) and p27 (7), suggesting that TGF-β-mediated cell growth inhibition is due, at least in part, to up-regulation of cell cycle inhibitory genes. Taken together, these results show that disruption of TGF-β receptor/SMAD signaling provides one mechanism by which tumor cells may escape TGF-β-induced growth inhibition. On the basis of its role in mediating the growth-inhibitory effects of TGF-β in normal cells and its loss of function in some tumor types, SMAD4 is considered to represent a tumor suppressor protein (5).

Melanocytes are derived from the neuroectoderm and, like epithelial cells, are highly sensitive to growth inhibition by TGF-β (8). By contrast and similar to carcinomas, many malignant melanomas exhibit various degrees of resistance to the growth-inhibitory effects of TGF-β (8, 9). The molecular basis of TGF-β resistance in melanomas and its relationship to TGF-β receptor/SMAD signaling is not understood. This led us to investigate whether, in human melanoma cells, TGF-β resistance was associated with functional inactivation of TGF-β receptor/SMAD-dependent signal transduction and transcription. We describe that, in contrast to pancreatic carcinoma cells, melanoma cells could be induced to efficiently activate SMAD3/4-mediated transcription in a TGF-β-dependent manner. SMAD-induced transcriptional activity in melanoma cells did not correlate with effects of exogenous TGF-β on proliferation of these cells. In addition, tumor-derived TGF-β contributed to comparatively high constitutive activity of SMAD-dependent transcription in some melanoma cell lines. These results indicate that, in contrast to some epithelial cancers, transcriptional regulation of gene expression via SMAD signaling pathways was preserved in human melanoma cells.

TGF-β-induced effects on proliferation of the primary (WM902-B, WM983-B, and WM793) and metastatic (WM239-A, WM164, and WM852) melanoma cell lines used in the present study have been described before (10). The cell lines 451-LU and 1205-LU were derived from WM164 and WM793, respectively, by serial passage through athymic mice and selection of cells metastatic to the lungs (9). FM516SV3/3 are postcrisis, immortalized, nontumorigenic melanocytes transformed by transfection with the SV40T antigen (11). Unless otherwise noted, melanoma and FM516SV3/3 cells were grown in a composite medium (W489) consisting of three parts MCDB153 and one part L15 supplemented with 2% (by volume) FCS. Normal melanocytes (FM1085 and FM1094), kindly provided by Dr. M. Herlyn, were propagated in W489 medium supplemented with 10 μg/ml insulin, 10 ng/ml EGF, ∼100 μg/ml bovine pituitary extract, 10 ng/ml phorbol 12-myristate 13-acetate, and 2% fetal calf serum and used between passages 15 and 20. Human dermal fibroblasts were isolated and propagated from neonatal foreskin using standard procedures.

Human recombinant TGF-β1 and pan-TGF-β neutralizing antibody were from R&D Systems (Minneapolis, MN). Recombinant TGF-β2 was a gift from Dr. David Olsen (Celtrix Corp., Palo Alto, CA). The two TGF-β isoforms exert comparable effects on melanoma cell growth (10) and on the activity of the (SBS)₂-TK/CAT promoter construct.⁴ They were thus used interchangeably and are referred to as TGF-β throughout the text.

Cells were seeded in their respective growth medium. In the case of melanoma cells, medium was changed after 12 h to W489 supplemented with insulin and 0.2% BSA (fatty acid-free; Boehringer Mannheim, Indianapolis, IN) and TGF-β at a final concentration of 10 ng/ml. Melanocytes were incubated in their fully complemented growth medium (see above) in the presence and absence of TGF-β. Cell numbers were determined after 6 days of incubation, and inhibition of cell growth was calculated relative to the cell number obtained in cultures maintained in the absence of exogenous TGF-β.

Transient transfections using a TGF-β-responsive reporter construct [(SBS)₂TK/CAT] were performed using the calcium phosphate/DNA coprecipitation method. The reporter construct consisted of the CAT gene driven by a tandem repeat of the SBS binding site of the human COL7A1 promoter cloned upstream of a minimal thymidine kinase promoter (12). After incubation of transfected cells with 10 ng/ml TGF-β in W489 medium containing 0.2% BSA-FAF for 6 h, cells were rinsed once with PBS, harvested by scraping, and lysed in 200 μl of Reporter Lysis Buffer (Promega Corp., Madison, WI). The protein concentrations in the extracts were determined using a commercial assay (Bio-Rad, Hercules, CA). To allow quantitative comparison of basal and induced CAT activities between different cell lines, 2 μg of pRSV-β-galactosidase were cotransfected with (SBS)-₂TK/CAT, and β-galactosidase activity was measured. Aliquots of cell extracts corresponding to identical β-galactosidase activity were used for CAT assays with [¹⁴C]chloramphenicol as substrate, using TLC. After autoradiography, the plates were cut, and the amount of acetylated[¹⁴C]chloramphenicol isoforms were determined using a liquid scintillation counter.

A fragment spanning the region-525/−444 of the COL7A1 promoter, corresponding to the SMAD binding TGF-β response element, was used as a probe as described (13). Nuclear extracts were isolated from cells treated with 10 ng/ml TGF-β for 30 min, using a small-scale preparation, aliquoted in small fractions to avoid repetitive freeze-thawing, and stored at-80°C until use. Nuclear extracts (5 μg) were incubated for 20 min on ice in binding reaction buffer (10 mm HEPES-KOH, pH 7.9 at 4°C, 4% glycerol, 40 mm KCl, 0.4 mm EDTA, and 0.4 mm DTT), in the presence of 1 μg of poly(deoxyinosinic-deoxycytidylic acid), prior to the addition of [³²P]5′-end-labeled oligomers (0.05–0.1 pmol, 2–6x 10⁴ cpm) for another 20-min incubation at 4°C. For supershift experiments, nuclear extracts were preincubated overnight with the pan-SMAD antibody 367 (kindly provided by Drs. R. J. Lechleider and A. B. Roberts, National Cancer Institute, NIH, Bethesda, MD).

Previously, we (8, 10) and others (9) have reported that the melanoma cell lines used in this study were less responsive to TGF-β-induced growth inhibition than normal melanocytes. Except for minor differences, we confirmed these earlier results (Table 1). Complete resistance to TGF-β-induced growth inhibition was observed in WM 852 metastatic melanoma cells, the metastatic variant 451-LU line, and the SV40T transformed melanocyte line FM516. Consistent with earlier results, proliferation of the 1205-LU metastatic variant line was moderately stimulated by exogenous TGF-β.

To assess functionality of the TGF-β receptor/SMAD signaling cascade, we first tested the effects of exogenous TGF-β on the activity of the SMAD-responsive (SBS)₂-TK/CAT reporter construct transfected into normal melanocytes and melanoma cells. This plasmid contains two copies of the SBS corresponding to the TGF-β response element of the human type VII collagen gene (COL7A1) promoter, cloned upstream of the heterologous TK promoter, and driving the expression of the CAT reporter gene in a SMAD-dependent manner (13). With the exception of metastatic melanoma cell line WM239-A, (SBS)₂-TK-driven CAT activity was induced to variable degrees by exogenous TGF-β in all cells used including normal melanocytes. A functional mutation in the SBS sequence that abolishes SMAD binding (12) prevented TGF-β-inducible transcription of the CAT construct in 1205-LU melanoma cells (not shown), consistent with the notion that TGF-β-dependent transcriptional activation of the (SBS)₂-TK/CAT reporter construct in melanoma cells depends on SMAD binding to SBS. Importantly, induction of SMAD-dependent transcription by exogenous TGF-β did not correlate with the degree of growth inhibition induced by TGF-β in the different cell lines (Table 1). In fact, relatively low levels of transcriptional activation by exogenous TGF-β as observed in normal melanocytes were associated with strong inhibitory effects of exogenous TGF-β on cell growth. Conversely, in WM793 and 1205-LU melanoma cells, very high levels of SBS-driven transcription were associated with partial and complete resistance to TGF-β–induced growth effects, respectively.

A comparison of (SBS)₂-TK/CAT activity between melanoma cells and normal dermal fibroblasts, measured in the absence of exogenous TGF-β, revealed higher levels of transactivation of the reporter construct in some melanoma cells (Table 1). The results shown were normalized to protein content and transfection efficiency by cotransfecting pRSV-β-galactosidase. As compared with normal dermal fibroblasts, particularly high basal levels of SMAD-mediated transgene expression were observed in WM793 primary melanoma cells and the metastatic variant lines 1205-LU and 451-LU (Table 1). We (8) and others (9) have reported earlier that these melanoma cell lines produce and secrete TGF-β. To determine whether endogenous SMAD-mediated transactivation of the (SBS)₂-TK/CAT construct was contributed to by endogenously produced TGF-β, 1205-LU melanoma cells were incubated for 48 h in the presence or absence of a neutralizing antibody recognizing all known isoforms of human TGF-β. These cells were then transfected with the (SBS)₂-TK/CAT construct and incubated further for another 40 h, always in the presence or absence of the antibody, followed by determination of CAT activity. As shown in Fig. 1, a significant reduction in CAT activity was observed in response to the TGF-β neutralizing antibody. As a control, a (NF-κB)₅-TK/CAT construct was transfected into the same cells using identical experimental conditions. As expected, the TGF-β antibody had no effect on the activity of the NF-κB-responsive construct. Similar results (30% reduction of specific CAT activity in the presence of the neutralizing TGF-β antibody) were obtained when 451-LU melanoma cells were preincubated with the TGF-β antibody (not shown). These experiments indicate that the high basal activity of the (SBS)₂-TK/CAT reporter construct in 1205-LU and 451-LU melanoma cells was due, at least in part, to endogenously produced TGF-β.

To verify that SBS-driven transcription in melanoma cells was accompanied by actual SMAD/DNA binding activity, EMSA experiments were performed using the COL7A1 SBS as a probe (13). Several melanoma cell lines were treated with 10 ng/ml TGF-β for 30 min prior to nuclear extract preparation. As shown in Fig. 2,A, TGF-β treatment resulted in the appearance of a specific DNA/protein complex that comigrated with the previously described SMAD-containing complex induced by TGF-β in fibroblasts (13). In all melanocytic cells, a weaker TGF-β-specific band was present in the same position, even in the absence of exogenous TGF-β and consistent with constitutive SMAD/DNA binding activity in these cells. Formation of the TGF-β-induced complex in melanocytic cell nuclear extracts was abrogated by a pan-SMAD antibody, attesting to the presence of a SMAD member in the complex (Fig. 2 B). These findings indicate the presence of functionally active SMADs in melanoma cells, capable of appropriately binding the SBS.

This study describes that in TGF-β-resistant human melanoma cells, SMAD-mediated regulation of gene expression was functionally intact. Furthermore, high levels of basal transactivation of a SMAD-induced, TGF-β-responsive reporter construct were observed in some melanoma cells derived from different stages of melanoma progression. This activity was traced in part to endogenous TGF-β produced by the melanoma cells themselves. These results suggest that functional inactivation of either the TGF-β receptors or of SMAD signaling molecules is not the main cause of resistance to the growth-inhibitory effects of TGF-β frequently observed in malignant melanoma cells (8, 9). Of note, even the highly aggressive metastatic variant melanoma lines 451-LU and 1205-LU retained high levels of basal and TGF-β-inducible SMAD activity, underscoring the notion that intact SMAD signaling pathways are maintained throughout tumor progression in the melanocytic cell system. These results are contrasted by the occurrence of inactivating mutations of either one of the TGF-β receptors or of SMADs 2 or 4 in select epithelial cancers, notably colon and pancreatic carcinoma; these mutations are predicted to abrogate all TGF-β-dependent signaling relevant to transcription. Taking into consideration that. in vivo, melanoma cells overexpress TGF-β when compared with normal melanocytes (14), the question arises whether certain autocrine transcriptional effects by tumor-derived TGF-β are beneficial to melanoma development.

All of the melanoma cell lines used in this study resisted growth inhibition by exogenous TGF-β to various degrees when compared with normal melanocytes (8, 9). The extent of growth inhibition was not coupled to their SMAD-specific transcriptional response to either exogenous or endogenous TGF-β. The dissociation of transcriptional effects and effects on cell growth by TGF-β may be explained by the finding that many of the TGF-β-regulated inhibitors of CDKs are functionally inactive in melanoma cells, either through mutational events or through reduced expression. These include p16ink4a (15), p15ink4b (16), and p27KIP1 (17), all of which inhibit phosphorylation of the retinoblastoma protein (pRb), a necessary event for the G₁-S transition of the cell cycle to occur. In addition, a mutant CDK4 protein has been identified in select melanomas that cannot be inhibited by the CDK inhibitor p16ink4a and, thus, can be considered to be constitutively active (18). Our present data indicate that transformed melanocytes in which pRb function has been abrogated by transfection with the SV40T antigen (FM516SV3/3) were resistant to growth inhibition by exogenous TGF-β and exhibited intact SMAD-mediated signaling and target gene transactivation similar to melanoma cells. This finding underscores that uncoupling the TGF-β growth response from the transcriptional response pathway is possible in melanocytic cells provided that inhibition of G₁-S progression is compromised by functional inactivation of the pRb protein. Collectively, these findings suggest that, in human melanoma, TGF-β-induced cell cycle arrest is countered by directly altering expression or function of genes controlling G₁-S cell cycle progression rather than by decreasing TGF-β receptor/SMAD signaling capacity.

The comparatively high transcriptional activity of the (SBS)₂-TK/CAT construct in some melanoma cells measured in the absence of exogenous TGF-β led us to investigate whether TGF-β produced by melanocytic cells themselves contributed to SMAD-dependent transcription. This was indeed the case as shown by lower levels of SMAD-dependent transcription in two melanoma cell lines in the presence of a pan-TGF-β neutralizing antibody. This result was unexpected, because we observed earlier that most of the TGF-β protein released by melanoma cells is in the latent, inactive form requiring physicochemical treatment to become bioactive (8). However, others have identified low levels of active TGF-β in conditioned media of certain melanoma cells (19) consistent with the production of both active and latent TGF-β by these cells in culture. The constitutive activation of SMAD-dependent transcription in melanoma cells supports the notion that low-level production of bioactive TGF-β by melanoma cells is sufficient to exert autocrine effects such as transcriptional activation of TGF-β/SMAD-responsive target genes.

The functional contribution of TGF-β-induced autocrine gene expression to melanoma progression is presently unknown. However, because proliferation of some melanomas is stimulated by exogenous TGF-β (Refs. 9 and 19; this study), induction of TGF-β-responsive gene expression in the absence of negative effects on cell cycle progression may be beneficial to melanoma progression.

We are grateful to Dr. C. Kari for helpful discussions. We thank Drs. R. J. Lechleider and A. B. Roberts for the pan-SMAD antibody 367.