Smad4 dependency defines two classes of transforming growth factor {beta} (TGF-{beta}) target genes and distinguishes TGF-{beta}-induced epithelial-mesenchymal transition from its antiproliferative and migratory responses

Abstract

In response to transforming growth factor beta (TGF-beta), Smad4 forms complexes with activated Smad2 and Smad3, which accumulate in the nucleus, where they both positively and negatively regulate TGF-beta target genes. Mutation or deletion of Smad4 is found in about 50% of pancreatic tumors and in about 15% of colorectal tumors. As Smad4 is a central component of the TGF-beta/Smad pathway, we have determined whether Smad4 is absolutely required for all TGF-beta responses, to evaluate the effect of its loss during human tumor development. We have generated cell lines from the immortalized human keratinocyte cell line HaCaT or the pancreatic tumor cell line Colo-357, which stably express a tetracyline-inducible small interfering RNA targeted against Smad4. In response to tetracycline, Smad4 expression is effectively silenced. Large-scale microarray analysis identifies two populations of TGF-beta target genes that are distinguished by their dependency on Smad4. Some genes absolutely require Smad4 for their regulation, while others do not. Functional analysis also indicates a differential Smad4 requirement for TGF-beta-induced functions; TGF-beta-induced cell cycle arrest and migration, but not epithelial-mesenchymal transition, are abolished after silencing of Smad4. Altogether our results suggest that loss of Smad4 might promote TGF-beta-mediated tumorigenesis by abolishing tumor-suppressive functions of TGF-beta while maintaining some tumor-promoting TGF-beta responses.

FIG. 1.

Characterization of a stable HaCaT cell line expressing a Tet-inducible siRNA targeted against Smad4. (A and B) Knockdown of Smad4 is strongly induced by Tet in HaCaT-TRS4 cells. HaCaT-TR parental cells and HaCaT-TRS4 cells were treated with Tet for 48 h, and equal amounts of protein were analyzed by Western blotting using antibodies against Smad4 and Grb2 as a loading control (A). Smad4 levels were quantified using the Odyssey detection system (B). (C) Knockdown of Smad4 has no effect on Smad2/3 phosphorylation after TGF-β stimulation. HaCaT-TR parental cells and HaCaT-TRS4 cells were treated with Tet for 48 h and then with TGF-β for 3 h. Equal amounts of protein were analyzed by Western blotting using antibodies against Smad4, Smad2/3, phosphorylated Smad2 (P-Smad2), phosphorylated Smad3 (P-Smad3), and Grb2 as a loading control. (D) Knockdown of Smad4 has differential effects on TGF-β-induced transcription from Smad3-dependent and Smad2-dependent synthetic reporters. Cells were transfected with the Smad3-dependent reporters CAGA₁₂-Luc or c-junSBR₆-Luc and the Smad2-dependent reporters ARE-Luc and a plasmid expressing XFoxH1a/XFast-1 or DE-Luc and a plasmid expressing Mixer. Where appropriate, Tet was added at the time of the transfection. Cells were then incubated for 40 h, and TGF-β was added 8 h before assaying luciferase activity. The TGF-β fold inductions are given at the bottom of each graph.

FIG. 2.

Quantitative analysis of the microarray results. Comparison of the fold induction (A) or fold repression (B) ratios of the early (1 h) and late (6 h) TGF-β target genes reveals two different categories of TGF-β-regulated genes dependent on or independent of Smad4. Fold inductions or repressions of one of the replicates for HaCaT-TR (+Tet), HaCaT-TRS4 (−Tet), and HaCaT-TRS4 (+Tet) cells are shown. A significant dependency on the level of Smad4 is observed for TGF-β target genes scored as Smad4 dependent (A, upper panel; B, left panel), whereas the TGF-β target genes scored as Smad4 independent (A, lower panel, B, right panel) display no significant differences in activation/repression under the three conditions. Known TGF-β target genes are shown in bold.

FIG. 3.

Validation of the microarray data. (A) Differences in transcriptional activation upon TGF-β stimulation were confirmed by RT-PCR. HaCaT-TR or HaCaT-TRS4 cells were incubated with or without Tet for 48 h and treated with TGF-β for 0 h, 1 h, or 6 h. Representative candidates of the two different categories of TGF-β-induced genes were analyzed. PAI-1, ITGB6, and AKAP12 were confirmed as Smad4-dependent target genes, and CDKN1A (p21), c-JUN, SMAD7, SNAI2 (Slug), TMEPAI, and CDKN2B (p15) were confirmed as Smad4-independent TGF-β target genes. GAPDH was used as a control. (B) Gene inductions assayed at the protein level. HaCaT-TR or HaCaT-TRS4 cells were incubated with or without Tet for 48 h and treated with TGF-β for 7 h. Whole-cell extracts were prepared, and equal amounts of protein were analyzed by Western blotting using antibodies against p21 and p15 as representatives of Smad4-independent TGF-β targets, against PAI-1 and Smurf1 as representative products of Smad4-dependent TGF-β target genes, and against Smad4 to confirm the Smad4 protein levels in the different conditions. Grb2 serves as a loading control. In all cases except p21, the protein expression in response to TGF-β in the different cell lines/conditions mirrored the RNA expression. (C) The loss of induction of the Smad4-dependent genes is rescued when an siRNA-resistant HA-Smad4 is expressed in the HaCaT-TRS4 cells. HaCaT-TRS4-rescue cells were incubated with or without Tet for 48 h and then treated with TGF-β for 7 h. Whole-cell extracts were prepared, and equal amounts of protein were analyzed by Western blotting. Tet induction of the HA-tagged siRNA-resistant Smad4 in the rescue clone is observed in the HA blot and in the Smad4 blot, where it is detected as a band running with slightly lower mobility compared to endogenous Smad4 (see arrows). Induction of PAI-1, Smurf1, and p21 by TGF-β is rescued when the Tet-induced silencing of the endogenous Smad4 is rescued by the induction of the siRNA-resistant form of Smad4.

FIG. 4.

Smad4 is necessary for TGF-β-induced cell cycle arrest in HaCaT cells. HaCaT-TR or HaCaT-TRS4 cells were incubated with or without Tet for 48 h as indicated and synchronized by serum starvation (0.2% serum for 24 h). Cells were then collected before (0h serum starved) or after treatment with 10% serum for 22 h in the absence (22h serum) or presence of TGF-β (22h serum + TGF-β). Samples were analyzed by fluorescence-activated cell sorting to determine the number of cells in the G₁, S, and G₂/M phases. FL3-H on the x axis corresponds to propidium iodide fluorescence.

FIG. 5.

Smad4 is not required for EMT induced by TGF-β. (A) A Colo-357-derived cell line that expresses a Tet-inducible Smad4 siRNA shows a strong silencing of Smad4. Colo-TR parental cells and Colo-TRS4 cells that express the Smad4 siRNA were treated with or without Tet for 48 h, and equal amounts of protein were analyzed by Western blotting using antibodies against Smad4 and Grb2 as a loading control. (B) The Smad4 dependency of some representative Smad4-dependent and independent TGF-β target genes is validated in the Colo-TRS4 cells. Colo-357 or Colo-TRS4 cells were incubated with or without Tet for 48 h and treated with TGF-β for 0 h, 1 h, or 6 h. PAI-1, ITGB6, and AKAP12 were confirmed as Smad4-dependent target genes, and SMAD7, SNAI2 (Slug), and TMEPAI were confirmed as Smad4-independent TGF-β target genes. GAPDH was used as a loading control. (C and D) Smad4 is not involved in TGF-β-induced EMT in Colo-357 cells. Colo-TR parental cells and Colo-TRS4 cells were grown on collagen type I and treated with Tet for 48 h. TGF-β or SB-431542 was then added for 6 days. In the presence of SB-431542, cells display an epithelial-like phenotype, whereas growth in the presence of TGF-β induces a mesenchymal-like phenotype independently of the presence of Smad4. Cells were processed for immunofluorescence using an anti-E-cadherin (E-Cad) antibody, to analyze adherens junctions (B) or an antivimentin (Vim) antibody to analyze mesenchymal marker expression (C). Actin reorganization was visualized with Texas red-conjugated phalloidin, and nuclei were visualized with DAPI.

FIG. 6.

Smad4 is necessary for TGF-β-dependent migration in a scratch assay. (A, B, and C) The Tet-induced silencing of Smad4 in HaCaT-TRS4 cells abolishes TGF-β-induced migration in a scratch assay. HaCaT-TR, HaCaT-TRS4, or HaCaT-TRS4-rescue cells were grown to confluence with or without Tet for 48 h and were then serum starved for 24 h. A scratch was introduced into the confluent cell monolayer, and TGF-β-induced migration was monitored with time-lapse microscopy for 48 h. A photograph of the cells after 48 h of migration is shown in panel A. Tracking of cells at the leading edge of the scratch was performed on the acquired images using the Tracker software for the HaCaT-TR cells (+Tet) and HaCaT-TRS4 cells (±Tet) (B). Mean speeds were calculated from the cell tracking analysis using the Mathematica software (C). The mean speeds of cells in different condition are represented, and the asterisk indicates conditions that are significantly different from the others with a P value of <0.00001. (D) Smad4 is also necessary for TGF-β-induced migration in pancreatic cells. Colo-TR and Colo-TRS4 cells were grown to confluence with or without Tet for 48 h. A scratch was introduced into the monolayer, and TGF-β-induced “wound” closure was photographed after 48 h of cell migration. (E) EGF-induced migration is still functional in the absence of Smad4. HaCaT-TR or HaCaT-TRS4 cells were grown to confluence with or without Tet for 48 h and then serum starved for 24 h. After scratching, cells were incubated with or without TGF-β or EGF for 48 h, at which time the images were taken.

FIG. 7.

Knockdown of Smad4 affects only a subset of TGF-β-regulated target genes and functions. Knockdown of Smad4 in HaCaT cells and in Colo-357 cells reveals that the TGF-β signaling pathway has a Smad4-dependent component and a Smad4-independent component. TGF-β-induced functions that we have shown to be Smad4 dependent or Smad4 independent are shown. Examples of genes that may be involved in these functions, based on the literature, are indicated.