The oncogenic protein Ski associates with Smad proteins and counteracts their activation of gene expression and growth inhibition in response to transforming growth factor β (TGF-β). Here we show that Ski protein levels are increased in all 44 human melanoma tumor tissues analyzed in vivo. In addition, Ski subcellular localization changes from nuclear, in preinvasive melanomas (melanomas in situ), to nuclear and cytoplasmic in primary invasive and metastatic melanomas. Furthermore, Ski/Smad association in the cytoplasm seems to prevent Smad3 nuclear translocation in response to TGF-β. The biological significance of Ski overexpression in melanomas was established by showing that down-regulation of Ski levels, by antisense Ski vectors, restored TGF-β-mediated growth inhibition. Such inhibition is apparently mediated by up-regulation of the cyclin-dependent kinase-I p21Waf-1 and inhibition of cyclin-dependent kinase 2 activity. Our results suggest that high levels of Ski in human melanomas produce a disruption of TGF-β signaling phenotypically similar to that in cells harboring mutations in TGF-β receptors or Smad proteins, and this may represent a significant event in the progression of melanomas in vivo.
The oncogenic protein Ski is a nuclear protein that either activates or represses transcription, depending on the cellular and promoter context (1). The consensus GTCTAGAC sequence previously was found to bind Ski-containing protein complexes and subsequently identified as a SBE3 that is bound by Smad family members (1). Ski suppresses TGF-β signaling by binding Smad2 and Smad3 and functioning as a corepressor of promoters bound by these proteins and the common mediator, Smad4 (2, 3, 4, 5). There is evidence that oncogenic activation and/or loss-of-function mutations of members of the TGF-β signaling pathway contribute to resistance to TGF-β growth inhibition (reviewed by Liu et al. in Ref. 6). Inactivation of the TGF-β pathway has been observed in a variety of human cancers including melanomas (7). Here we describe that primary invasive melanomas in vivo exhibit nuclear and cytoplasmic localization of Ski, whereas in melanoma metastasis, Ski is mostly localized in the cytoplasm and expressed at very high levels. We demonstrate that Ski cytoplasmic localization prevents its associated protein Smad3 to translocate to the nucleus in response to TGF-β. We propose that repression of TGF-β-mediated growth inhibition by Ski may contribute to the progression of human malignant melanomas.
Materials and Methods
Formalin-fixed, paraffin-embedded tissues were obtained from the archives of the Department of Pathology, Baylor College of Medicine, Houston, TX. Forty-four lesions representing histologically recognizable stages of tumor progression in cutaneous malignant melanoma in vivo were included in this study. The tissues included preinvasive MIS (n = 18), primary invasive melanoma (n = 15), and MET (n = 11). All tissues were cut 5 μm thick, mounted on positively charged microscope slides (Fisher Scientific, Pittsburgh, PA), deparaffinized, and rehydrated through a graded series of alcohols. Before immunohistochemistry, tissue sections were subjected to heat-induced epitope retrieval (8) in urea buffer [2.5%, w/v (pH 9.0)].
Immunohistochemistry and Immunocytochemistry.
Immunohistochemistry was performed as previously described (9). Briefly, tissue sections were incubated with the monoclonal G8 anti-Ski Ab (15 μg/ml) for 16 h. at 4°C and then incubated sequentially with biotinylated horse antimouse IgG secondary Ab (Vector Laboratories, Burlingame, CA), streptavidin alkaline phosphatase detection system (Boehringer-Mannheim, Indianapolis, IN), and Vector Red chromogen (Vector Laboratories). All tissues were counterstained with hematoxylin, permanently mounted, and viewed with a standard light microscope. To examine the subcellular distribution of Smad3, IIB-Mel-J and UCD-Mel-N cells were treated with 5 μm TGF-β1 for 0 and 1.5 h, fixed with paraformaldehyde, and stained with a polyclonal anti-Smad3 Ab (SC-8332; Santa Cruz Biotechnology Biotech).
The human melanoma cell lines IIB-Mel-J, UCD-Mel-N (described in text), A375, and derivative lines expressing a human AS-ski (pcDNA3.1-AS-Ski) or EV (pcDNa3.1) were grown and subcultured as described previously (5).
Colony Formation Assay in Agar Plates.
EV and AS-ski cells were seeded in six-well dishes in Dulbecco/F-12 medium supplemented with 10% FBS. Twenty-four h later, the medium was removed and replaced by Dulbecco/F-12 supplemented with 0.2% FBS for 3 days to deplete the cells from serum-derived growth factors. At day 4, the cells were fed with 0 or 400 pm TGF-β1. After 3 days, cells were trypsinized and seeded at a density of 2 × 104 (EV) and 4 × 104 (AS-Ski) in six-well dishes containing 0.33% top low-melt agarose-0.7% bottom low-melt agarose and fed with medium containing 0 or 400 pm TGF-β. Every 3 days, 0.1 ml of fresh medium (containing 0 or 400 pm TGF-β) was added to the dishes. After 15 days of incubation, the colonies were stained overnight with a solution containing 1mg/ml p-iodinitrotetrazolium violet (Sigma Chemical Co.).
An antisense ski (AS-Ski) expression vector was constructed by excising h-ski with the enzymes BamHI and XhoI from pcDNA3.1-ski (5) and then cloning into pcDNA3.1 vector carrying hygromycin as a resistant marker. The plasmid WWP-luc containing the human p21Waf-1 promoter between positions −2300 to +8 was a gift from B. Vogelstein (Johns Hopkins Medical Center, Baltimore, MD).
Transfection and Luciferase Assay.
UCD-Mel-N cells were transfected using the FuGENE 6 reagent (Boehringer Mannheim, Inc.) as described previously (5). Transfections used 0.750 μg of the WWP-Luc and CAGA reporter plasmids and combinations of 0.5 μg each of Smad3 and Smad4 plasmids and 0.250(+), 0.500(++), and 1(++++) μg of Ski expression plasmid together with a β-galactosidase expression plasmid for normalization of transfection efficiencies. Luciferase and β-galactosidase activities were measured 22 h after transfection by using the appropriate reporter assay kits according to manufacturer’s instructions (Promega).
Immunoprecipitation and Immunoblotting.
IIB-Mel-N melanoma cells, treated with 400 pm TGF-β1 for 0 or 24 h, were scraped; solubilized in 0.1% NP40, 50 mm Tris-HCl (pH 7.5), and 100 mm NaCl containing protease inhibitors; centrifuged at 10,000 × g for 15 min at 4°C; and immunoprecipitated as described previously. Cytosolic and nuclear extracts of IIB-Mel-J cells were prepared using the NC EB kit, as described by the manufacturer. Nuclear and cytosolic fractions were incubated on ice for 3 h with 3 μg of Smad3 polyclonal Ab (N-19; Santa Cruz Biotechnology) or normal goat or rabbit serum (Sigma Chemical Co.). Immunocomplexes adsorbed to protein G-agarose beads (Boehringer Mannheim) were collected by centrifugation and analyzed by SDS-PAGE and immunoblotting with either G8 monoclonal Ab or Smad3 (H-2) Ab. The supernatants remaining after centrifugation were depleted of Smad3 by three rounds of immunoprecipitation and then incubation with anti-Ski Ab (G8) for 3 h on ice, immunoabsorbed, collected, and analyzed as described in the previous paragraph.
Three hundred μg of total cell extracts were used for kinase assays as described previously (10). For immunodepletion experiments, extracts were depleted of p21Waf-1 by three rounds of immunoprecipitation with an anti-p21Waf-1 Ab and then immunoprecipitation with an anti-CDK2 Ab.
Results and Discussion
Early studies had demonstrated that ski induces transformation of pigmented avian melanocytes.4 Therefore, we were interested in determining whether Ski was overexpressed in human melanoma tissues, and whether its levels, cellular localization, and/or activity correlated with melanoma tumor progression. All of the melanomas studied here (n = 44) expressed the Ski protein. In preinvasive MIS, the Ski protein was observed predominantly in the nucleus of intraepidermal melanoma cells (Fig. 1,A). However, most of the primary invasive melanomas demonstrated both nuclear and cytoplasmic localization of Ski (Fig. 1,B), although nuclear labeling was greater in the intraepidermal melanoma cells. In MET we observed nuclear and cytoplasmic labeling (Fig. 1,C) or predominantly cytoplasmic Ski labeling (Fig. 1 D).
Two human melanoma cell lines partially recapitulated such changes in Ski localization; UCD-Mel-N cells, which were derived from a malignant melanoma of the skin (to be described elsewhere) exhibited predominantly nuclear localization (Fig. 1,E), whereas IIB-Mel-J cells, which were derived from a cutaneous melanoma metastatic to the lung (11), showed similar nuclear and cytoplasmic Ski distribution in small cells and more prominent cytoplasmic localization in very large cells (Fig. 1 F).
To determine the biological consequences of Ski overexpression in melanoma tumors, we isolated nuclear and cytoplasmic fractions from ∼3 × 107 IIB-Mel-J cells, treated with 0 or 400 pm TGF-β for 20 h. In the absence of TGF-β, Ski was evenly distributed in nuclear and cytoplasmic fractions, whereas treatment with TGF-β for 20 h caused partial down-regulation of the endogenous levels of both nuclear and cytoplasmic Ski (Fig. 2,A). In support of these results, no changes were observed previously in ectopic Ski levels associated to Smad2 and Smad3 in a UCD-Mel-N-derivative line (UCDSki+) overexpressing the human ski gene (5). We previously demonstrated that TGF-β increases Ski-Smad3 association in melanoma cells expressing high levels of the nuclear ski protein (5). We therefore asked whether cytoplasmic Ski also associated with Smad3 and whether the association prevented Smad3 nuclear translocation. In the absence of TGF-β, the bulk of Smad3 in IIB-Mel-J melanoma cells seemed to be associated with cytoplasmic Ski, whereas low levels of nuclear Smad3 were found associated with high levels of nuclear Ski (Fig. 2,B, Lanes 3 and 4). No Ski was immunoprecipitated from the 20-h TGF-β lysates by an irrelevant (negative control) antiserum (Fig. 2,B, Lanes 1 and 2). Surprisingly, treatment with TGF-β did not substantially change the nuclear levels of Smad3 or levels of the cytoplasmic Ski associated with Smad3 (Fig. 2,B, Lanes 5 and 6). These results suggest that the association of Smad3 with Ski in the cytoplasm prevents Smad3 nuclear translocation in response to TGF-β and raise the possibility that the association of Ski with Smad3 may impair binding to importin β and nuclear translocation (12). We also investigated whether in addition to forming Smad3 complexes, a free fraction of Ski existed that was not associated with Smad3. Smad3-immunodepleted extracts were immunoprecipitated and immunoblotted with an anti-Ski Ab (Fig. 2,C). In the absence of TGF-β, a significant amount of “free” Ski remained in cytoplasmic and nuclear fractions (compare Fig. 2,C, Lanes 3–4 with Fig. 2,B, Lanes 3–4), whereas in the presence of TGF-β, the majority of Ski protein was coprecipitated with Smad3 (compare Fig. 2,C, Lanes 5–6, with Fig. 2 B, Lanes 5–6).
Immunocytochemistry analyses confirmed Smad3 localization and demonstrated that only a small number of IIB-Mel-J cells exhibited Smad3 nuclear translocation after treatment with TGF-β (Fig. 2, D and E). However, in UCD-Mel-N cells, which exhibited predominant nuclear localization of Ski (Fig. 1,E), Smad3 was able to translocate to the nucleus after exposure to TGF-β (Fig. 2, F and G). Thus, the biological consequence of Ski/Smad3 interaction in the cytoplasm seems to be similar to nuclear localization signal mutations in Smad3, because this mutant remains in the cytoplasm and functions as dominant-negative inhibitor of TGF-β signaling (13).
Loss of TGF-β sensitivity is frequently observed in tumors derived from cells that are otherwise sensitive to inhibition by this protein, and the extent of TGF-β resistance often correlates with metastatic progression (7). Melanoma cells are highly resistant to the inhibitory activity of TGF-β, but no measurable defects in the TGF-β pathway have been found to date (14, 15). To determine whether high levels of Ski down-regulate the growth inhibitory response to TGF-β in melanoma cells, we constructed an antisense Ski vector (AS-ski) that spanned the entire Ski coding region, and we used it to transfect the melanoma cell lines UCD-Mel-N and A375N. Several stable transfected clones with antisense (AS-Ski) or empty vectors (EV) resistant to hygromycin were recovered. Two clones derived from UCD-Mel-N cells demonstrated reduced Ski levels by antisense Ski mRNA, whereas only partial Ski reduction was observed in one A375 AS-ski clone (Fig. 3,A). We chose a clone of AS-ski from each cell line to test their response to TGF-β and compared these with control clones (EV). All growth curves and clonogenic assays were performed in the presence of 0.2% FBS and in the absence of other exogenously added growth factors. UCD-AS-ski cells showed reduced growth ability in 0.2% serum and complete growth inhibition by 400 pm TGF-β, compared with UCD-EV control cells (Fig. 3,B). Conversely, A-375 AS-ski cells were more resistant to inhibition by low serum and were inhibited by TGF-β after 3 days of treatment. Clonogenic assays of UCD AS-ski (seeded at a density of ∼4 × 104 cells/plate) showed an almost complete inhibition of colony formation in the presence of TGF-β, whereas UCD-EV cells (seeded at a density of ∼2 × 104 cells/plate) exhibited a minimal reduction in colony formation (Fig. 3 C). The clonogenicity of A375AS-Ski was only slightly reduced compared with controls, which possibly reflects the delayed growth inhibitory activity of TGF-β in these cells.
TGF-β inhibits cellular proliferation by up-regulating the expression of the CDK inhibitors p21Waf-1 (16) and p15INK4b (17) through a mechanism that involves the cooperation of Smad2, Smad3, and Smad4 with Sp1. We sought to determine whether growth arrest induced by TGF-β in AS-Ski cells is mediated by p21Waf-1 induction (the UCD-Mel-N tumor cell line does not express detectable levels of p15INK4b; data not shown). UCD-EV and UCD-AS-ski cells were left untreated or treated with TGF-β for up to 3 days. AS-Ski cells showed an ∼2.3-fold increase in basal and an ∼4.3-fold increase in TGF-β-stimulated p21Waf-1 protein levels compared with ∼1- and 2.5-fold, respectively, in EV cells (Fig. 4,A). We were therefore interested in determining whether increased p21Waf-1 levels in AS-ski cells resulted in CDK2 inhibition, a condition required for growth arrest in most cell types (18). Histone H1 kinase assays demonstrated that EV cells have more than five times more CDK2 activity than AS-ski cells. Consistent with these observations, high p21Waf-1 levels were found to be associated with CDK2 in AS-Ski cells (Fig. 4,B). Furthermore, increased p21Waf-1 levels seemed to be essential for CDK2 inhibition, inasmuch as depletion of p21Waf-1 by three rounds of immunoprecipitation increased CDK2 activity by ∼8-fold in AS-ski cells, whereas it only increased CDK2 by ∼1.8 fold in EV cells (Fig. 4 C).
We and others have demonstrated that Ski, Smad2/Smad3, and Smad4 form a complex with the Smad/Ski binding element (SBE) GTCTAGAC (3, 5). The p21Waf-1 promoter contains a proximal Sp1-binding site that confers Smad cooperativity and a SBE in its distal part (16). We sought to determine whether Ski represses Smad-dependent activation of the p21Waf-1 promoter. In agreement with published data (7, 16), Smad3 in combination with Smad4 also activated the p21Waf-1 promoter (WWP-Luc) in UCD-Mel-N melanoma cells (Fig. 4,D). However, such activation is dramatically repressed by Ski. As an independent control for the specificity of promotor-reporter Ski repressor activity, we used the established 12X(CAGA) Luc, the transactivation of which by TGF-β depends solely on the Smads (16). In agreement with previous data (5), Ski also repressed Smad3 and Smad4 activity (Fig. 4 E). Thus, in addition to the inactivating of this retinoblastoma protein (19) and the repressing of retinoic acid receptor (20), Ski represses induction of genes that are negative regulators of cell cycle progression.
In conclusion, our results suggest that high levels of Ski in human melanomas results in a disruption of TGF-β signaling similar to that caused by mutations in genes encoding TGF-β receptors or Smad proteins, and it may represent a significant event in the progression of melanomas in vivo.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Supported by a Shannon Award and a R0-1 CA84282 grant from the National Cancer Institute (to E. E. M.).
The abbreviations used are: SBE, Smad-binding element; TGF-β, transforming growth factor β; MIS, melanoma(s) in situ; MET, metastatic melanoma; AS-ski, antisense ski gene; EV, empty vector(s); FBS, fetal bovine serum; Ab, antibody; CDK2, cyclin-dependent kinase 2.
A. E. Barkas, (Dissertation). New York: New York University, 1986.
We thank Xin-Hua Feng and Olivia Pereira-Smith for critically reading the manuscript and Denise J. Schwahn for critical suggestions during the course of this work. We are grateful to Ed Stavnezer for the gift of the G8 Ski Ab, Rick Derynck for Smad3 and Smad4 plasmids, Jean-Michel Gauthier for the 12× CAGA plasmid, and Bert Vogelstein for the WWP-luc plasmid.