![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Department of Molecular and Cell Biology, University of California, Berkeley,1 Life Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, California 94720,2 Department of Cellular and Structural Biology, University of Texas Health Science Center, San Antonio, Texas 78229-39003
Received 28 July 2006/ Returned for modification 19 September 2006/ Accepted 16 October 2006
 |
ABSTRACT |
|---|
 Top ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
SnoN has been shown to be an important negative regulator of transforming growth factor ß (TGF-ß) signaling (56, 58). TGF-ß plays a complex and multifaceted role in the regulation of tumorigenesis (2, 54, 61). On the one hand, TGF-ß induces cell cycle arrest or apoptosis in many cell types, and therefore acts as a potent inhibitor of cell proliferation. These antimitogenic responses to TGF-ß are frequently lost early in malignant transformation, and the remaining effects of TGF-ß on tumor cells and their surrounding microenvironment promote malignant progression. The ability of TGF-ß to induce epithelial-to-mesenchymal transition (EMT) plays a critical role in the acquisition of a migratory, invasive phenotype that is correlated with enhanced metastatic potential in tumor cells. Whereas the signal transduction pathways leading to TGF-ß-induced EMT have not been fully characterized, the mechanism by which TGF-ß elicits growth arrest has been elucidated. The Smad proteins are critical mediators of cytostatic responses to TGF-ß. Upon ligand binding, the activated TGF-ß receptor kinase complex phosphorylates the receptor-associated Smads (R-Smads), Smad2 and Smad3, leading to their hetero-oligomerization with Smad4 and the accumulation of Smad complexes in the nucleus, where they interact with other transcription factors to regulate expression of TGF-ß-responsive genes (4, 23, 40, 41).
We and others have shown that SnoN interacts directly with Smad2, Smad3, and Smad4 and represses their ability to activate expression of TGF-ß target genes by disrupting the formation of an active heteromeric Smad complex, recruiting a transcriptional corepressor complex, and by blocking the interaction of transcriptional coactivators with Smad2 and Smad3 (3, 38, 56, 63). When localized to the cytoplasm, SnoN antagonizes TGF-ß signaling by sequestering Smad proteins in the cytoplasm (36). The transforming activity of SnoN may be dependent on its ability to bind to Smad proteins, since Smad binding was shown to be required for repression of TGF-ß-induced growth arrest as well as oncogenic transformation of chicken embryo fibroblasts (29). In addition to the Smad proteins, SnoN has also been shown to interact with other transcription factors, including nuclear hormone receptor corepressor N-CoR/SMRT, mSin3A, methyl-CpG-binding protein MeCP2, and TAF(II)110 (16, 35, 53), to mediate transcriptional repression. However, it is not clear whether these interactions involve Smad proteins or occur independently of them. It is also not clear whether any of these interactions is required for the transforming activity of SnoN.
In this study, we investigated the role of SnoN in tumorigenesis by ablating its expression in two different human cancer cell lines that exhibit elevated expression of SnoN. We show that reducing SnoN expression inhibits the proliferative potential of cancer cells both in vitro and in vivo but promotes EMT and tumor metastasis. Thus, SnoN exerts opposite effects at different stages of tumorigenesis. In addition, we have also attempted to explore the signaling molecules and downstream effectors associated with these SnoN activities. We find that both Smad-dependent and Smad-independent pathways are involved in the activities of SnoN during tumorigenesis.
|
MATERIALS AND METHODS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
A small hairpin RNA (shRNA) vector targeting human SnoN was generated using the pSUPER vector system as described previously (14). Oligonucleotide pairs encoding shRNA against human SnoN were designed according to established guidelines using Oligoengine software. The 19-nucleotide sequence within SnoN targeted by the shRNA oligonucleotide pair was 5'-GACAGTCAGAGAAGGCTCA-3' at nucleotide position number 1307 within the human SnoN cDNA. Forward and reverse primers were synthesized containing this sequence in sense and antisense orientations with an intervening linker. Primer pairs were designed to generate single-strand overhangs upon annealing that would allow the annealed duplex oligonucleotide to be cloned into BglII and HindIII sites in the pSUPER.retro.puro vector. Forward and reverse primers were annealed and ligated into the pSUPER.retro.puro vector.
Smad-binding mutant SnoN construct was generated as described previously (29, 63). pGEX-2T-TRBD was obtained from M. Schwartz. pGEX-2T-PBD was obtained from G. Bokoch.
Transfections and generation of stable cell lines. All transfections were performed using the Lipofectamine Plus transfection system (Invitrogen) according to the manufacturer's instructions. MDA-MB-231 breast cancer and A549 lung cancer cell lines that stably express shRNA constructs targeting human SnoN were generated by cotransfecting each cell line with the pSUPER-shSnoN vector along with the pBABE puro vector to permit selection using puromycin resistance. At 48 h posttransfection, stably transfected cells were selected by culture in medium containing 1.5 µg/ml puromycin. For "rescue" experiments, wild-type or Smad-binding mutant SnoN constructs were constructed containing nonsense mutations within the sequence targeted by shSnoN and therefore were not recognized by the shRNA. These SnoN constructs were cotransfected into shSnoN A549 cells along with the pLXSN neomycin resistance vector. Transfected cells were selected by growth in medium containing 0.8 mg/ml neomycin (Invitrogen).
Cell lysis, immunoprecipitation, and immunoblotting. Cells were lysed in high-salt lysis buffer (420 mM NaCl, 50 mM HEPES-KOH, pH 7.8, 5 mM EDTA, 0.1% NP-40, 3 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride [PMSF], 10 µg/ml aprotinin). Prior to immunoprecipitation, cell lysates were precleared on 30 µl of protein A-Sepharose beads for 30 min at 4°C. Precleared lysates were then transferred to beads complexed with the appropriate antibody for immunoprecipitation. Endogenous SnoN was immunoprecipitated using protein A-Sepharose beads coupled to a polyclonal antibody recognizing a C-terminal SnoN peptide (KELKLQILKSSKTAKE). Endogenous Ski was immunoprecipitated using a polyclonal anti-Ski antibody raised against a glutathione S-transferase (GST)-fusion protein containing amino acid residues 1 to 605 of human c-Ski.
Antibodies used for immunoblotting were anti-Ski (G8; Cascade Bioscience), anti-SnoN (H-317; Santa Cruz Biotechnology), anti-Smad2 (BD Transduction Labs), anti-Smad3 (FL-425; Santa Cruz Biotechnology), anti-cofilin (Sigma), anti-phospho-cofilin (Cell Signaling), anti-E-cadherin (BD Transduction Laboratories), anti-fibronectin (BD Transduction Laboratories), anti-RhoA (Santa Cruz Biotechnology), anti-Rac1 (Upstate), and anti-Cdc42 (Santa Cruz Biotechnology). Anti-phospho-Smad2 and anti-phospho-Smad3 antibodies were kind gifts of A. Moustakis (Ludwig Institute for Cancer Research, Uppsala, Sweden).
Growth inhibition assay. A total of 3 x 104 MDA-MB-231 cells or 5 x 104 A549 cells were cultured with the indicated concentrations of TGF-ß1 for 4 days. Relative cell growth was determined by counting cells and calculating the number of TGF-ß-treated cells relative to that of unstimulated cells.
Immunofluorescence. For actin stress fiber staining, cells were cultured in medium containing 1% serum on glass coverslips for 20 h prior to treatment with 50 pM TGF-ß1 for another 3 days before staining. Cells were fixed in 4% paraformaldehyde in phosphate-buffered saline (PBS) for 20 min, and the paraformaldehyde was quenched with three washes in 100 mM glycine in PBS. Cells were permeabilized in 0.1% Triton X-100, and actin filaments were stained with rhodamine-phalloidin (Molecular Probes) for 30 min at room temperature. Coverslips were then washed three times for 5 min each in PBS before being mounted on slides in one drop of mounting medium, consisting of 80% glycerol, 20 mM Tris, pH 8.0, and 1:1,000 DABCO [1,4-diazabicyclo-(2,2,2)-octane; Sigma].
To visualize transfected Myc-RhoA T19N, coverslips were blocked for 1 h following the permeabilization step in blocking buffer (PBS containing 10% newborn calf serum, 1% bovine serum albumin, and 0.02% Triton X-100). After blocking, coverslips were incubated with anti-Myc antibody (Sigma) in staining buffer (blocking buffer lacking bovine serum albumin) overnight at 4°C. Coverslips were washed three times in staining buffer and then incubated with Alexa 488-conjugated anti-mouse antibody (Molecular Probes) for 1 h at room temperature. Coverslips were then washed three times in PBS before being mounted on slides in one drop of mounting medium. For E-cadherin staining, cells were grown until reaching 90% confluence and then fixed in 50/50 methanol-acetone for 10 min at –20°C. Anti-E-cadherin antibody was purchased from Transduction Laboratories.
Immunofluorescence was observed with a Zeiss Axiophot epifluorescence microscope or a Zeiss confocal LSM 510 microscope.
Semiquantitative reverse transcription-PCR (RT-PCR). Total RNA from parental A549 and MDA-MB-231 cells and their shSnoN-expressing derivatives was extracted using the RNeasy mini kit (QIAGEN) and treated with RNase-free DNase (QIAGEN). cDNA was produced using reverse transcriptase (SuperScript II; Invitrogen) and was amplified with the following sets of primers: GAPDH, 5'-CGTCTTCACCACCATGGAGA-3' (forward) and 5'-CGGCCATCACGCCACAGTTT-3' (reverse); EMP1, 5'-GGTTAGAGAGATTGGCCAGC-3' (forward) and 5'-CAGTACTAGAACAATCCACC-3' (reverse); PLAU, 5'-AGCAGCTGAGGRCTCTTGAG-3' (forward) and 5'-AAACTGAGACAGTGCTGGTC-3' (reverse); Twist1, 5'-GAAAGCGAGACAGGCCCGTG-3' (forward) and 5'-GATTGGCACGACCTCTTGAG-3' (reverse); EGFR, 5'-TGTCTCTGCCTTGAGRCATC-3' (forward) and 5'-ACTGCTGTTAACCAGTCAGG-3' (reverse); VEGF, 5'-GGGCAACTTGTATTTGTGTG-3' (forward) and 5'-CTGCACTAGAGACAAAGACG-3' (reverse); JunB, 5'-ACGTCAGCAACGGCTGTCAG-3' (forward) and 5'-GAATCGAGTCTGTTTCCAGC-3' (reverse); GADD45A, 5'-CTGAACGGTGATGGCATCTG-3' (forward) and 5'-GCAAAGTCATCTATCTCCGG-3' (reverse).
EMP1, PLAU, Twist1, EGFR, VEGF, JunB, and GADD45A mRNA levels were determined by semiquantitative PCR. GAPDH served as an internal control. The following PCR program was performed: 94°C for 5 min (initial denaturation) and 94°C for 30 s, 55°C for 30 s, and 72°C for 45 s. Within the linear range of amplification, all of the PCR products were prepared under appropriate cycling conditions and separated on a 1% agarose gel. The band densities were compared between samples from parental and shSnoN-expressing cells.
Northern blotting. Prior to RNA extraction, MDA-MB-231 cells or A549 cells were starved in medium containing 0.1% FBS overnight, and parallel dishes were treated with 100 pM TGF-ß for the times indicated. Total RNA was extracted using the RNeasy mini kit (QIAGEN). Twenty micrograms of RNA was resolved on a 1% formaldehyde gel, transferred to a nylon membrane, and analyzed by Northern blotting. DNA probes for p21 and PAI-1 were radiolabeled by random priming (Stratagene) and hybridized with the nylon membrane in QuikHyb (Stratagene).
Microarray hybridization, data collection, and analysis.
Affymetrix human genome U133 A+B GeneChips were used in this study. Fifteen micrograms of total RNA collected from parental and shSnoN-expressing A549 cells was used for microarray probe synthesis as described in the Affymetrix GeneChip manual. Microarray hybridization, data collection, and analysis were performed by the Stanford PAN facility. The gene expression profiles were compared between parental and shSnoN-expressing samples. The parental samples were used as the "baseline." The expression pattern of genes was considered changed in shSnoN cells only when two statistical analysis criteria were satisfied: (i) a probe set with a detection P value that was less than 0.4, which is considered PRESENT by the Affymetrix GeneChip program, and (ii) the change in gene expression between the experimental (shSnoN sample) and baseline (parental sample) conditions was greater than 1.75-fold. The signal log ratio is related to the change (n-fold) by the following formulas: change (n-fold increase) = 2signal log ratio for signal log ratios of 
0; change (n-fold decrease) = (–1) x 2signal log ratio for signal log ratios of <0.
Selected microarray results were verified by Northern blotting analysis, semiquantitative RT-PCR, and immunoblotting.
Wound healing assay. A549, MDA-MB 231, or their derivative cell lines were seeded into six-well dishes and grew until 90% confluence. One-milliliter plastic tips were used to generate wounding across the cell monolayer. Phase contrast pictures were taken under an inverted microscope at the time of wounding (0 h) and 48 h later. The wound closure was estimated as the percentage of the closure area to the initial wounded area. Experiments were carried out in triplicate at least three times.
Rho/Rac/Cdc42 GTPase assays.
The protocol for measuring GTP loading on Rho family GTPases was as described previously (8). The GST fusion construct used to bind Rho-GTP was pGEX-2T-TRBD, which contains the Rho-binding domain of rhotekin corresponding to amino acids 7 to 89. The GST fusion construct used to bind Rac-GTP or Cdc42-GTP was pGEX-2T-PBD. GST fusion proteins were expressed in DH5
Escherichia coli bacteria and purified using glutathione-Sepharose beads (Amersham). After washing to remove unbound protein, the beads were resuspended in wash buffer containing 10% glycerol and assayed for protein concentration before being stored at –80°C.
To prepare cell lysates, A549 cells or MDA-MB-231 cells were washed in ice-cold Tris-buffered saline and lysed in RIPA buffer (50 mM Tris, pH 7.2, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate, 500 mM NaCl, 10 mM MgCl2, 5 µg/ml each of leupeptin and aprotinin, and 1 mM PMSF). Lysates were clarified by centrifugation at 13,000 x g at 4°C for 5 min, and protein concentration was measured for all samples. For determination of RhoA-GTP loading, 700 µg of whole-cell lysate in a 500-µl total volume was incubated with 60 µg of GST-TRBD beads. For determination of Rac-GTP or Cdc42-GTP binding, 500 µg of whole-cell lysate in a 400-µl total volume was incubated with 30 µg of GST-PBD beads. Lysates were incubated with beads for 1 h at 4°C with constant rotation and subsequently washed three times with wash buffer (50 mM Tris, pH 7.2, 1% Triton X-100, 150 mM NaCl, 10 mM MgCl2, 5 µg/ml each of leupeptin and aprotinin, and 0.1 mM PMSF). The beads were resuspended in 20 ml of 2x sample buffer containing dithiothreitol and heated at 95°C for 10 min. The samples were run on a 12% polyacrylamide gel electrophoresis gel and transferred to a nitrocellulose membrane. Bound Rho proteins were detected by immunoblotting using either anti-RhoA (Santa Cruz Biotechnology), anti-Rac1 (Upstate), or anti-Cdc42 (Santa Cruz Biotechnology) antibody. A small sample of each cleared lysate was reserved for analysis of total Rho protein present in the original sample.
In situ zymography. Glass coverslips were coated with a thin layer of Alexa 488-conjugated gelatin (0.2 mg/ml in 2% sucrose-PBS). The coated coverslips were fixed with cold 0.5% glutaraldehyde in PBS for 15 min and incubated for 3 min in 5 mg/ml NaBH4 at room temperature. The coverslips were then sterilized with 70% ethanol, washed three times in PBS, and quenched with serum-free DMEM for 1 h at 37°C. Cells were plated on coated coverslips in complete medium. After the cells had attached, the medium was aspirated, and the cells were incubated in DMEM containing 0.1% FBS with or without either 100 pM TGF-ß1 alone or in the combination with 10 µM matrix metalloprotease (MMP) inhibitor GM6001 for 24 h (in breast cancer cells) or 48 h (in lung cancer cells) before being processed for immunostaining.
Soft agar assay. A concentrated bottom layer comprised of 2 ml of growth medium containing 0.66% Bacto-agar (Difco) was poured into a six-well dish. A total of 5 x 103 A549 cells, MDA-MB 231 cells, or their shSnoN-expressing derivatives were resuspended in 2 ml of medium containing 0.4% agar and overlaid on the hardened bottom layer. One milliliter of fresh medium containing 0.4% agar was added to the dish every week. After 4 weeks of incubation, colonies were visualized by staining with 0.5 mg/ml thiazolyl blue tetrazolium bromide (MTT; Sigma) and scanned on a Hewlett-Packard ScanJet 5500C to visualize the colonies.
Tumorigenicity in nude mice. Female athymic BALB/c mice were supplied by Charles River Laboratories at 4 to 6 weeks of age. A549 cells, MDA-MB-231 cells, or their shSnoN-expressing derivatives were injected subcutaneously into the left and right flanks of nude mice. To prepare cells for injection, cells were trypsinized and resuspended in medium containing 10% FBS to neutralize the trypsin. The cell pellets were then washed twice in PBS and resuspended in PBS at a concentration of 5 x 106 cells/0.1 ml PBS (A549) or 2 x 106 cells/0.1 ml PBS (MDA-MB 231). A 0.1-ml aliquot of the cell suspension was injected into each injection site using a 26-gauge needle with a 1-ml sterile syringe. Tumor growth was monitored once a week. Mice were sacrificed at 8 weeks after injection, and tumors were surgically isolated. Tumor volume (V) was calculated by using the formula: V = length x width x thickness.
Metastasis assays. MDA-MB-231 cells (1 x 105 cells in 0.1 ml PBS/mouse) were injected into the left cardiac ventricle of 4-week-old female nude mice. The body weight of each mouse was measured weekly. Animals were monitored for paraplegia and were terminated at the end of 4 weeks after the injection at the sign of paraplegia of some animals. After the mice were sacrificed, whole lungs were excised and fixed with Bouin's solution. Metastatic nodules were identified by color and appearance and counted under a dissecting microscope. The right tibia and femur bones were fixed in buffered formalin and used for histomorphometry analyses.
To detect paraplegia as a measure of metastatic body burden, a wire hang test was employed (28). Each mouse was placed on a wire cage lid. The lid was inverted, and the latency to fall was recorded, with a 60-second cutoff time. The results were plotted as mean latency to fall in seconds ± the standard error of the mean (SEM).
To examine metastasis of A549 lung cancer cells, 2 x 106 parental or shSnoN-expressing cells in 0.1 ml PBS were injected into the lateral tail veins of 4-week-old female nude mice. The body weight of each mouse was taken every week. After 3 months, the mice were sacrificed. Lungs were excised and fixed in Bouin's solution. Metastasis nodules showed a different color from that of lung tissue and were counted.
|
RESULTS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
Downregulation of SnoN expression restores TGF-ß responses. We chose two model systems, the A549 human lung adenocarcinoma cell line and the MDA-MB-231 human breast cancer cell line, to investigate the role of SnoN in malignant progression. A549 cells express high levels of SnoN but have very little if any Ski (Fig. 1A). MDA-MB-231 cells are derived from invasive and metastatic breast adenocarcinoma and express both SnoN and Ski at high levels. These cell lines are weakly or nonresponsive to TGF-ß-induced growth inhibition and, in addition to various aspects of morphological and mitogenic transformation, are also capable of undergoing or have undergone at least some aspects of EMT, an important process in later stages of tumorigenesis that may be necessary for tumor invasiveness and metastasis. The multiple features of tumorigenesis exhibited by these cell lines provide an opportunity to probe the contribution of SnoN to various aspects of epithelial transformation. In addition, the different tissue origins of the two cell lines allow us to determine whether the processes affected by SnoN are tissue specific or are common to epithelial tumors.
We generated stable A549 and MDA-MB-231 cell lines in which expression of SnoN was reduced using shRNA specific for human snoN. pSUPER vector expressing snoN shRNA was introduced into A549 and MDA-MB-231 cells by transfection together with a plasmid expressing a puromycin resistance gene. For each cell line, multiple stable clones were generated that exhibited reduced expression of SnoN (Fig. 2A). Stable reduction of SnoN expression had no effect on the intensity or duration of TGF-ß-induced phosphorylation of Smad2 or Smad3, as expected given that SnoN plays no direct role in the regulation of Smad phosphorylation (Fig. 2B).
|
We next examined whether reducing SnoN expression augmented activation of endogenous TGF-ß target gene expression (Fig. 2D). In parental A549 and MDA-MB-231 cells, TGF-ß induced expression of PAI-1 mRNA after 3 h of treatment (Fig. 2D, left panel). This induction of PAI-1 by TGF-ß was enhanced in both A549 and MDA-MB-231 cells expressing snoN shRNA (Fig. 2D, left panel). Likewise, the expression of p21CIP1, another TGF-ß-inducible gene, was elevated in shSnoN A549 cells even in the absence of TGF-ß treatment and further enhanced by TGF-ß stimulation (Fig. 2D, right panel). This elevation in basal-level p21 expression in shSnoN A549 cells is not due to the increased autocrine TGF-ß activity in these cells, because inhibition of TßRI activity by an inhibitor, SB431542, had no effect on the basal expression of p21, even though this treatment readily blocked TGF-ß-induced p21 expression (data not shown). This observation is also consistent with the proposed role of SnoN in maintaining the basal states of some TGF-ß-responsive genes. Taken together, these results suggest that reducing SnoN expression in human lung and breast cancer cells enhances cellular responses to TGF-ß.
Reducing SnoN expression suppresses tumor growth. TGF-ß signaling suppresses tumor growth at early stages of tumorigenesis through its ability to elicit growth arrest (22, 54). Since SnoN represses TGF-ß signaling, we reasoned that reducing SnoN expression in cancer cells might diminish or reverse their transformed phenotype. To test this, we first examined the ability of shSnoN-expressing lung and breast cancer cells to undergo anchorage-independent growth in a soft agar assay. Cells were embedded in soft agar and allowed to form colonies for approximately 3 weeks. Under these conditions, parental A549 and MDA-MB-231 cell lines formed colonies readily, whereas cells with reduced expression of SnoN were severely impaired in their growth in soft agar (Fig. 3A). Interestingly, Ski does not appear to play a major role in anchorage-independent growth, since shSki-expressing MDA-MB-231 cells have significantly reduced expression levels of Ski protein yet are still able to form as many soft agar colonies as parental cells do (Fig. 3A, bottom panel). Furthermore, reducing Ski expression in shSnoN-expressing cells did not result in further reduction in soft agar colony formation (data not shown). These results suggest that SnoN but not Ski functions to promote mitogenic transformation.
|
|
Downregulation of SnoN enhances TGF-ß-induced EMT. The EMT is a process by which tumor cells lose epithelial characteristics and acquire the features of mesenchymal cells. It is believed to promote the ability to invade surrounding tissues and blood vessels and undergo metastasis and is thought to be important for malignant progression in vivo. EMT is characterized by a number of morphological and biochemical changes, including increased cell motility and stress fiber formation, downregulation of adherens junctions and their affiliated proteins, including E-cadherin, induction of extracellular matrix (ECM) proteins, and increased MMP activity (66). Since SnoN potentiates oncogenic transformation and tumor growth, we next asked whether and how SnoN affects the EMT using the A549 and MDA-MB-231 cell lines expressing shSnoN.
We first examined whether expression of SnoN affects the motility of A549 and MDA-MB-231 cells in a wound healing assay. A wound was created by scratching a confluent monolayer of cells with a pipette tip, and relative rates of cell motility were assessed by measuring percent closure of the wound after 48 h of cell migration. Parental A549 cells showed only 18% wound closure within this time period, whereas migration of shSnoN A549 cells resulted in 74% wound closure (Fig. 4A), suggesting that reducing SnoN expression markedly increased cell motility. Similar results were obtained in a transwell migration assay (data not shown). Increased cell motility was also observed in SnoN-deficient MDA-MB-231 breast cancer cells (Fig. 4A, right panel). Since the MDA-MB-231 cells already exhibit a high rate of cell migration, reducing SnoN expression only resulted in a moderate, but reproducible, increase in cell motility. Taken together, these data suggest that SnoN functions to repress cell motility.
|
Upon treatment with TGF-ß, A549 cells exhibited morphological changes characteristic of EMT, becoming more scattered and elongated, and these TGF-ß-induced changes in cell morphology were more pronounced in shSnoN cells (data not shown). Morphological changes occurring during EMT are thought to be due to increased actin stress fibers as well as loss of adherens junctions resulting from downregulation or mislocalization of E-cadherin (27, 49). We therefore examined whether the reduction of SnoN expression affected stress fiber formation by staining cells with fluorescently labeled phalloidin. In parental A549 lung cancer cells and MDA-MB-231 breast cancer cells, cellular actin was arranged cortically, with few or no stress fibers present (Fig. 4D). As has been demonstrated previously in several cell types (9, 10, 51), TGF-ß treatment resulted in increased stress fiber formation (Fig. 4D). In shSnoN-expressing lung and breast cancer cells, actin stress fibers were observed even in the absence of TGF-ß, and stress fiber formation was further enhanced upon stimulation with TGF-ß (Fig. 4D). Thus, downregulation of SnoN in both lung and breast cancer cells augmented actin stress fiber formation.
Loss of adherens junctions due to downregulation or mislocalization of E-cadherin is frequently observed as tumor cells progress to later, more invasive stages of carcinogenesis (49, 60). MDA-MB-231 cells have already lost E-cadherin expression, since they are quite advanced in malignant progression. We therefore examined expression of E-cadherin in parental and shSnoN-expressing A549 cells. TGF-ß induced a very slight reduction in E-cadherin levels in parental A549 cells (Fig. 4E). Basal levels of E-cadherin were reduced in shSnoN cells compared with parental A549 cells, and this reduction was more pronounced upon treatment with TGF-ß (Fig. 4E), suggesting that downregulation of SnoN promotes disruption of cell-cell contacts. These data are consistent with the morphological observation of enhanced EMT in shSnoN cells.
Taken together, these data indicate that in addition to prooncogenic activity, SnoN also possesses antitumorigenic activity by inhibiting the EMT and possibly tumor metastasis.
Downregulation of SnoN enhances tumor metastasis in vivo. The EMT is thought to play a role in tumor cell metastasis (59). To investigate how SnoN affects tumor cell metastasis, we tested the ability of MDA-MB-231 cells with reduced SnoN expression to form secondary bone and lung metastases in an in vivo metastasis mouse model system. Cells were injected into the left cardiac ventricle of nude mice, and osteolytic bone metastases were quantified after 4 weeks by the histomorphometric measurement of tumor area/burden (in percentage), and metastasis to lung was examined by anatomical analysis of fixed lung tissue at the end of 4 weeks. Interestingly, cells with reduced expression of SnoN appeared to have a moderate but reproducible increase in metastasis to both bone and lung than that of the parental breast cancer cells (Fig. 5A and B). The heightened skeletal metastatic tumor burden in mice inoculated with shSnoN-expressing cells also resulted in severe paraplegia, as evidenced by their diminished latency to fall from a wire hang test (Fig. 5C). Since the formation of metastatic colonies in the in vivo metastasis assay is dependent upon a number of cellular attributes acting in concert, including proliferative potential, migration, and invasion, and in light of the fact that shSnoN cells exhibited significantly reduced proliferative potential, the observation of increased skeletal and lung metastasis may in fact be an under-representation of the migratory and invasive potential of these cells.
|
|
|
|
In contrast, introduction of the mutant SnoN defective in binding to the Smad proteins (mSnoN) back into the shSnoN cells failed to restore the anchorage-independent growth (Fig. 6C), even though the expression level of ectopically expressed mSnoN was higher than that of WTSnoN in rescued cells and was comparable to that of endogenous SnoN in parental A549 cells (Fig. 6A). Surprisingly and interestingly, mSnoN can rescue some of the EMT phenotypes found in shSnoN cells, including E-cadherin localization, fibronectin production, and MMP2 activity, but not others, such as stress fiber formation and cell motility (Fig. 6E to H). These data suggest that SnoN inhibits the EMT through both Smad-dependent and Smad-independent pathways and that both pathways are necessary for its effects on malignant progression.
SnoN inhibits RhoA GTPase activity to repress actin stress fiber formation. Since SnoN can activate both Smad-dependent and Smad-independent pathways to regulate EMT, we began to dissect downstream signaling events that may mediate these activities of SnoN. Two approaches were taken. In the first approach, we examined signaling molecules known to regulate various aspects of cell growth, migration, and morphology. In the second approach, microarray analysis was carried out to compare the patterns of gene expression between parental and shSnoN-expressing A549 cells.
The Rho family of small GTPases, including RhoA, Rac, and Cdc42, play important roles in the regulation of cell growth, motility, and actin stress fiber formation (25). Since reducing SnoN expression markedly enhanced actin stress fiber formation, we examined whether SnoN regulates the Rho family of proteins by comparing the expression and activity of RhoA, Rac, and Cdc42 in parental and shSnoN-expressing A549 cells. While SnoN expression had no effect on the expression levels of RhoA, Rac, or Cdc42, the GTP-binding activity of RhoA, but not that of Rac or Cdc42, was significantly elevated in SnoN-deficient cells (Fig. 7A). As reported before (9), TGF-ß stimulation resulted in a modest increase in RhoA activity, and this increase was further enhanced in shSnoN cells (Fig. 7B). To determine whether this increase in RhoA activity was required for the enhanced stress fiber formation observed in SnoN-deficient cells, a dominant negative RhoA (DNRhoA:RhoA T19N) was introduced into the shSnoN cells. In untransfected shSnoN A549 or shSnoN MDA-MB-231 cells, actin was arranged in elongated stress fibers as observed previously (Fig. 4D and 7C). In contrast, cells expressing DNRhoA exhibited diffused cytoplasmic actin staining with no detectable stress fibers (Fig. 7C), suggesting that RhoA activity is required for the increased stress fiber formation observed in SnoN-deficient cells.
|
Taken together, these results indicate that the RhoA pathway appears to function downstream of SnoN to mediate its effect on EMT.
SnoN modulates multiple signaling pathways involved in the regulation of cell growth, migration, and morphology. To further understand the signaling pathways regulated by SnoN and to confirm the biological observations outlined above, we performed microarray analysis to uncover changes in gene expression resulting from downregulation of SnoN expression. Total RNA was isolated from parental A549 cells and those expressing shRNA for SnoN and hybridized to an Affymetrix human U133 (A+B) array. Among the 44,500 total human genes on the array, the expression of 7.8% of them (3,471 genes) was altered upon downregulation of SnoN expression. Out of these 3,471 genes, 2,086 genes were upregulated and 1,348 were downregulated. An annotated list of a subset of these genes is shown in Table 2, and the expression of some of the relevant genes, including JunB, GADD45A, EGFR, Twist1, VEGF, PLAU, and EMP1 has been confirmed by RT-PCR (Fig. 8).
Consistent with previous data suggesting that SnoN has pleiotropic effects in the regulation of tumorigenesis, SnoN-deficient cells exhibited changes in gene expression indicative of cell cycle arrest as well as EMT leading to tumor metastasis. In particular, several genes involved in negative regulation of cell cycle (p21, JunB, and cyclin G2) and apoptosis (GADD34, GADD45, and IGFBP1) were upregulated, while genes that promote cell cycle progression (cyclin E2, A2, D3, and E2F) and survival (survivin-b) were downregulated in shSnoN cells, in agreement with the reduction of tumor growth both in vitro and in vivo. Many genes involved in extracellular matrix remodeling were induced in SnoN-deficient A549 cells, including FN1, ECM2, PLAU, PAI-1, and MMP-16. In addition, expression of Twist1, an important regulator of EMT that has been shown to promote tumor cell metastasis, was elevated in SnoN-deficient cells. Thus, the ability of SnoN to inhibit cellular processes characteristic of EMT was also substantiated by microarray analysis.
As expected from its role as a negative regulator of TGF-ß signaling, many SnoN-regulated genes are also TGF-ß-responsive genes. These include several well-described TGF-ß-inducible genes (p21, JunB, PAI-1, cyclin G2, MMP2, CTGF, VEGF, etc.) that were upregulated in shSnoN cells as well as TGF-ß-repressed genes (cyclin A2, TIAF1, and TGFßR3) that were also downregulated in shSnoN cells (7, 15, 19, 31, 34, 37, 39, 43, 46, 47), suggesting that reduction of SnoN expression can enhance or repress expression of TGF-ß-regulated genes, even in the absence of TGF-ß stimulation (Table 2 and Fig. 2D and 8). This is consistent with a role of SnoN in maintaining the basal states of TGF-ß-responsive genes. Not surprisingly, SnoN also altered the expression of many genes not currently known to be directly involved in TGF-ß signaling. These include genes involved in cell proliferation and apoptosis (IGFBP1, GADD34, E2F, etc.) and angiogenesis, EMT, and tumor metastasis/invasion (VEGF, SEL1L, and autotaxin) as well as adhesion and cell-matrix interaction (FN1, ECM2, 2 integrin, MMP2, MMP16, Decorin, and PLAU) (21, 30, 54). These genes contribute prominently to the complex effects of SnoN on tumor growth and progression.
Taken together, the results of the microarray analysis support cell biological and biochemical observations in SnoN-deficient cells and provide transcriptional data supporting the dual role of SnoN in tumorigenesis.
|
DISCUSSION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The proposed dual role of SnoN in transformation lends itself to the hypothesis that the activity of SnoN might be regulated differently during various stages of tumorigenesis. Elevated expression of SnoN may help to establish the primary tumor colonies at the early stages of tumorigenesis by promoting tumor cell growth, by blunting negative regulatory signals such as those elicited by TGF-ß, and by limiting the motility of these cells. When the primary tumors have progressed to a stage to spread, SnoN activity may be inhibited either through modification of the SnoN protein or through degradation of the protein, permitting EMT and subsequent metastasis of tumor cells. Once the tumor cells migrate to a new tissue location and are ready to establish secondary colonies, a high level of SnoN expression again will promote tumor growth in the new location. Thus, it is possible that SnoN expression and activity may vary during different stages of malignant progression and that epigenetic or other posttranslational modifications of SnoN may play an important role in the regulation of its activity or expression during malignant progression. While we did not investigate whether there was a correlation between tumor cell invasiveness and reduced SnoN expression among the cancer cell lines analyzed here, it is technically difficult to accurately compare expression levels of a particular protein among cell types with highly variable cell size and protein contents. In addition, if reduced activity of SnoN is sufficient to promote progression of tumor cells to a more advanced phenotype, this change in SnoN expression may be a transient event and may not be captured by analyzing cancer cell lines derived from tumors with established phenotypes. The possession of antitumor invasion activity by an oncogene product has also been reported to happen to other oncogenes, including Akt1 (33, 36a, 65). Akt1 has been shown to inhibit breast cancer cell motility while promoting the proliferation of these cells. The inhibitory activity of Akt1 on cell motility and invasion appears to be dependent on its ability to activate the E3 ubiquitin ligase HDM2, leading to ubiquitination and degradation of NFAT, an invasion-promoting protein (65). It is possible that other traditionally defined oncoproteins may also play inhibitory roles in EMT, tumor invasion, and metastasis.
The pathways that SnoN activates to regulate oncogenic transformation and EMT are not well understood. One key question related to SnoN function is whether its transforming activity can be fully attributed to its ability to antagonize TGF-ß signaling. TGF-ß has been shown to inhibit tumor growth at the early stages of tumorigenesis through the Smad-dependent growth inhibitory pathway and to promote EMT and tumor metastasis at late stages through both Smad-dependent and Smad-independent pathways (22, 66). Since SnoN is a negative regulator of the Smad proteins, one would predict that SnoN may operate in the opposite direction as TGF-ß during tumorigenesis if these activities are dependent upon its ability to bind to the Smad proteins. Our results suggest that while SnoN indeed has an opposite effect on tumor growth and tumor metastasis as TGF-ß, the ability of SnoN to antagonize TGF-ß signaling is necessary but not sufficient for its complex roles in tumor development. While many of SnoN-regulated genes are TGF-ß-responsive ones and are clearly involved in the regulation of cell proliferation, survival, and EMT, other SnoN target genes that are involved in regulating proliferation and apoptosis, including E2F, GADD34, and survivin-b, as well as a large number of genes involved in EMT, angiogenesis, and metastasis, have not been known to be regulated directly by TGF-ß signaling. Although it is possible that some of these genes may not be direct SnoN targets and are instead activated indirectly by primary SnoN-responsive genes, many of them may be regulated directly by SnoN through its ability to affect signaling pathways other than the TGF-ß/Smad pathway. One of the pathways regulated by SnoN is the RhoA small GTPase. We showed here that SnoN inhibits the GTPase activity of RhoA and a downstream signaling molecule cofilin to regulate stress fiber formation. Introduction of a dominant negative RhoA blocked stress fiber formation, confirming the importance of RhoA in SnoN signaling. At this time, however, we do not know how SnoN is linked to the activation of RhoA. Microarray analysis showed alterations in the expression of several genes, including Rho GDP dissociation inhibitor alpha and Rho GTPase-activating protein, that could contribute to the activation of RhoA in shSnoN cells. It is also possible that SnoN may inhibit RhoA through a direct signaling cascade independent of its activity as a transcriptional factor. Interestingly, the expression of over 1,000 genes is downregulated in cells lacking SnoN, suggesting that SnoN activity is required for the transcriptional activation of these genes. This could imply that SnoN may function in the capacity of a transcription activator, in addition to acting as a transcriptional corepressor of the Smad proteins. Alternatively, it could be due to an indirect effect of the repressive activity of SnoN: SnoN may inhibit the activity of one or more transcriptional repressors that normally prevent the expression of these genes.
In earlier studies using chicken and quail embryo fibroblasts, Ski and SnoN shared many similarities in their biological activities, including promoting transformation and terminal differentiation. Both proteins bind to the Smad proteins and are potent negative regulators of TGF-ß signaling (12, 17, 18, 38, 56, 57, 64). However, it is not clear whether in mammalian cells the two proteins are functionally redundant. Our survey of the various normal and cancerous human cell lines indicated that the two proteins clearly exhibit different patterns of expression. While the expression pattern of SnoN in normal and cancer cells is consistent with its prooncogenic activity, that of Ski is not. Although Ski expression was reported as elevated in malignant melanoma cells in one earlier study (26), a more recent survey of more human melanoma cell lines suggests that most melanoma cell lines may not express detectable level of Ski (48). In addition, we showed that reducing Ski expression had no effect on anchorage-independent growth of breast cancer cells in vitro (Fig. 3A) and tumor growth in vivo (data not shown) but led to increased tumor metastasis (data not shown). Thus, unlike what has been found in chicken embryo fibroblasts that readily undergo anchorage-independent growth in response to Ski overexpression, in mammalian carcinogenesis, Ski may play a role in regulating tumor cell metastasis and invasion rather than tumor growth. Indeed, a recent study by Azuma et al. (5) showed that Ski could inhibit metastasis of a breast cancer cell line to lung and liver. These results raise the important question of whether earlier studies on transformation of chicken embryo fibroblasts by Ski are relevant to human cancer. Future studies hopefully will shed more light on this issue.
|
ACKNOWLEDGMENTS |
|---|
This work was supported by NIH grant CA87940 and Philip Morris grant 019016 to K. Luo and NIH grants CA 17542 to G. S. S. Martin and CA79683 to L.-Z. Sun.
|
FOOTNOTES |
|---|
Published ahead of print on 30 October 2006.
|
REFERENCES |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
2. Akhurst, R. J., and R. Derynck. 2001. TGF-beta signaling in cancer—a double-edged sword. Trends Cell Biol. 11:S44-S51.[Medline]
3. Akiyoshi, S., H. Inoue, J. Hanai, K. Kusanagi, N. Nemoto, K. Miyazono, and M. Kawabata. 1999. c-Ski acts as a transcriptional co-repressor in transforming growth factor-beta signaling through interaction with smads. J. Biol. Chem. 274:35269-35277.
4. Attisano, L., and J. L. Wrana. 2002. Signal transduction by the TGF-beta superfamily. Science 296:1646-1647.
5. Azuma, H., S. Ehata, H. Miyazaki, T. Watabe, O. Maruyama, T. Imamura, T. Sakamoto, S. Kiyama, Y. Kiyama, T. Ubai, T. Inamoto, S. Takahara, Y. Itoh, Y. Otsuki, Y. Katsuoka, K. Miyazono, and S. Horie. 2005. Effect of Smad7 expression on metastasis of mouse mammary carcinoma JygMC(A) cells. J. Natl. Cancer Inst 97:1734-1746.
6. Bamburg, J. R. 1999. Proteins of the ADF/cofilin family: essential regulators of actin dynamics. Annu. Rev. Cell Dev. Biol. 15:185-230.[CrossRef][Medline]
7. Barlat, I., B. Henglein, A. Plet, N. Lamb, A. Fernandez, F. McKenzie, J. Pouyssegur, A. Vie, and J. M. Blanchard. 1995. TGF-beta 1 and cAMP attenuate cyclin A gene transcription via a cAMP responsive element through independent pathways. Oncogene 11:1309-1318.[Medline]
