A SNAIL1-SMAD3/4 transcriptional repressor complex promotes TGF-beta mediated epithelial-mesenchymal transition

Abstract

Epithelial-mesenchymal transition (EMT) is essential for organogenesis and is triggered during carcinoma progression to an invasive state. Transforming growth factor-beta (TGF-beta) cooperates with signalling pathways, such as Ras and Wnt, to induce EMT, but the molecular mechanisms are not clear. Here, we report that SMAD3 and SMAD4 interact and form a complex with SNAIL1, a transcriptional repressor and promoter of EMT. The SNAIL1-SMAD3/4 complex was targeted to the gene promoters of CAR, a tight-junction protein, and E-cadherin during TGF-beta-driven EMT in breast epithelial cells. SNAIL1 and SMAD3/4 acted as co-repressors of CAR, occludin, claudin-3 and E-cadherin promoters in transfected cells. Conversely, co-silencing of SNAIL1 and SMAD4 by siRNA inhibited repression of CAR and occludin during EMT. Moreover, loss of CAR and E-cadherin correlated with nuclear co-expression of SNAIL1 and SMAD3/4 in a mouse model of breast carcinoma and at the invasive fronts of human breast cancer. We propose that activation of a SNAIL1-SMAD3/4 transcriptional complex represents a mechanism of gene repression during EMT.

Figure 1. Loss of junction proteins is associated with nuclear accumulation of SNAIL1 and SMAD3/4 during EMT in breast epithelial cells

Analysis of the expression of CAR, occludin, claudin-3, E-cadherin, vimentin, SNAIL1 and SMAD3/4 during EMT in EpH4/EpXT (a, b, e, g) and NMuMG (c, d, f, h) cells. (a, c) Immunoblot analysis of protein expression during EMT. A time course experiment showing differences in expression in NMuMG cells treated with TGF-β (10 ng/ml) (c). Calnexin was used as a loading control. Full scans are shown in Supplementary Information, Fig. S5. (b, d) Semi-quantitative analysis of mRNA levels during EMT. A time course experiment showing differences in NMuMG cells treated with TGF-β. Data in b and d represent mean ± SEM of three independent experiments. (e, f) Immunofluorescence analysis of the expression and localization of junction proteins (green) during EMT. Nuclei were visualized with DAPI staining (blue). Morphological evidence of EMT is shown in bright-field images. Bars indicate 40 μm. (g, h) Immunofluorescence analysis of SNAIL1 and SMAD3/4 during EMT. NMuMG cells were treated with TGF-β for 12 hours. Bars indicate 40 μm.

Figure 2. SNAIL1 and SMAD3/4 interact and are recruited to CAR and E-cadherin promoters during EMT

(a) Diagram indicating putative SNAIL1 sites (E-box) and SMAD-binding elements (SBE) in different regions (I-V) of the 1000 bp genomic DNA sequence upstream of the ATG in the CAR gene (upper panel). Sequences of E-box and SBE sites in regions I, II and IV are shown. (b) Chromatin immunoprecipitation (ChIP) analysis of regions I-V was performed with antibodies against SNAIL1 and SMAD4. (c) ChIP analysis of the interaction of SNAIL1 and SMAD3/4 with the CAR (regions I and IV) and E-cadherin promoters in EpH4 versus EpXT cells. The interaction of phosporylated SMAD3 (p-SMAD3) with region I of the CAR promoter was analyzed. (d) Interaction of SNAIL1, SMAD3/4 with the CAR promoter (regions I and IV) in untreated or TGF-β treated NMuMG cells (12 h). (e, f) Co-immunoprecipitation (Co-IP) analysis of the interaction of SNAIL1 with SMAD4 and SMAD3 in EpXT cells (e) and in untreated or TGF-β treated NMuMG cells (12 h) (f). Full scans are shown in Supplementary Information, Fig. S5.

Figure 3. SNAIL1 and SMAD3/4 act as transcriptional co-repressors

(a-d) Reporter assays showing SNAIL1 and SMAD3/4 repression of CAR, E-cadherin, occludin and claudin-3 promoters. Data represent mean ± SEM of four independent experiments. *, P < 0.05, different from negative control. ^§, P < 0.05, different from SNAIL1. (e, f) Mutational analysis of Ebox and SBE sites in region I of the CAR promoter. Reporter constructs containing wild-type CAR (CAR WT), E-box (CAR ΔEbox) or SBE (CAR ΔSBE) mutations, or double mutations (CAR ΔΔ) were generated and used to analyze the importance of these sites in mediating repression by SNAIL1 and SMAD3/4. Data indicate mean ± SEM and are representative of three independent experiments. *, P < 0.05, different from control. ^§, P < 0.05, different from WT. (g, h) Effect of GSK-3β inhibition on the repression of CAR and E-cadherin mRNA (g) and protein (h) during TGF-β induced EMT (24 h) in NMuMG cells. SNAIL1 and SMAD3/4 proteins were also analyzed (h). Data represent mean ± SEM of four independent experiments. *, P < 0.05, different from untreated cells. ^§, P < 0.05, different from TGF-β alone. Full scans are shown in Supplementary Information, Fig. S5.

Figure 4. SNAIL1 and SMAD3/4 play essential and cooperative roles as transcriptional repressors of junction components during EMT

(a, b) RNA analysis of the effect of siRNA mediated silencing of SNAIL1 and SMAD4 in NMuMG cells in untreated and TGF-β treated NMuMG cells (24 h). (c-e) Effect of SNAIL1 and SMAD4 siRNAs on CAR, occludin and E-cadherin mRNA levels during TGF-β induced EMT. Data indicate mean ± SEM and are representative of three independent experiments. *, P < 0.05, different from untreated cells. ^§, P < 0.05, different from TGF-β + control siRNA. ^†, P < 0.05, different from TGF-β + SNAIL1 siRNA. (f) Effect of SMAD3 inhibition on CAR and E-cadherin mRNA levels during TGF-β induced EMT. *, P < 0.05, different from untreated cells. ^§, P < 0.05, different from TGF-β treatment. Data represent mean ± SEM of three independent experiments. (g) Cooperation of SNAIL1 and SMAD3/4 transcription factors during TGF-β-induced EMT. SNAIL1 and SMAD3/4 interact and form a transcriptional repressor complex, which is targeted to adjacent Eboxes (E) and SBE sites in genes encoding junction proteins like CAR, occludin and E-cadherin. As a result, these genes are repressed.

Figure 5. Loss of junction proteins correlates with nuclear co-expression of SNAIL1 and SMAD3/4 in mouse breast carcinomas and at the invasive front in human breast cancer

(a, b). Immunofluorescence analysis of the expression of HER-2, vimentin, CAR, E-cadherin, SNAIL1 and SMAD4 in subcutaneously grown mouse D2F2/E2 breast carcinomas. Tumour compartments with epithelial (E) or mesenchymal (M) morphology (separated by dashed lines) were identified by hematoxylin and eosin (H&E) staining. Boxes in b indicate areas magnified in lower panels. Arrows indicate nuclear co-expression of SNAIL1 and SMAD4. Scale bars = 50 μm or 20 μm (b, lower panels). (c, d) Immunofluorescence analysis of the expression of SNAIL1, SMAD3/4, CAR and E-cadherin in sections of invasive human ductal breast carcinomas. (c) Representative images showing correlation of nuclear co-expression of SNAIL1 and SMAD3 (upper panels) or SMAD4 (lower panels) with loss of CAR expression (outlined regions). Scale bars = 30 μm. (d) Representative images showing correlation between nuclear co-expression of SNAIL1 and SMAD3 and loss of E-cadherin expression (outlined regions). Scale bars = 30 μm. (e) Quantification of the percentage of SMAD3+ cases and SMAD4+ cases of tumour cells expressing CAR, E-cadherin and SNAIL1. *, P = 0.026 (CAR), P = 0.012 (E-cadherin), different from SNAIL1 negative cells in each graph. 100 cells per field in 10 different fields of each tumour (n=8) were quantified.