Reciprocal repression between Sox3 and snail transcription factors defines embryonic territories at gastrulation

Abstract

In developing amniote embryos, the first epithelial-to-mesenchymal transition (EMT) occurs at gastrulation, when a subset of epiblast cells moves to the primitive streak and undergoes EMT to internalize and generate the mesoderm and the endoderm. We show that in the chick embryo this decision to internalize is mediated by reciprocal transcriptional repression of Snail2 and Sox3 factors. We also show that the relationship between Sox3 and Snail is conserved in the mouse embryo and in human cancer cells. In the embryo, Snail-expressing cells ingress at the primitive streak, whereas Sox3-positive cells, which are unable to ingress, ensure the formation of ectodermal derivatives. Thus, the subdivision of the early embryo into the two main territories, ectodermal and mesendodermal, is regulated by changes in cell behavior mediated by the antagonistic relationship between Sox3 and Snail transcription factors.

Graphical abstract

Figure 1

Snail2 Expression in the Epiblast Induces Ectopic EMT and Cell Delamination (A–C) Chick embryos were coelectroporated with a vector encoding GFP and another containing the coding region of Snail2. The embryos were subsequently allowed to develop for 15 hr. Endogenous Snail2 expression at the primitive streak and ectopic expression induced by electroporation could be observed in a dorsal view (B) and in a transverse section through the epiblast (C). Observe the delamination of cells from the epiblast. Dotted line in (B) indicates the level of the section shown in (C). Electroporation of a control empty vector (pCX) or that containing GFP failed to induce cell delamination (not shown). (D–H) Snail2 electroporation was accompanied by the induction of RhoB expression, a known Snail2 downstream target (D and E), cell delamination, and disruption of the basement membrane as assessed by the absence of laminin staining (F and G) and by the repression of E-cadherin expression in the electroporated cells (H).

Figure 2

Snail2 and Sox3 Genes Are Complementarily Expressed in the Chick Gastrula (A–C) Snail2 expression starts at the posterior part of the blastula at stage EGXIV (A), and its expression is associated with the primitive streak as it progresses in the gastrulating embryo (B and C). (D–F) Sox3 expression is detected in the epiblast and it gets progressively downregulated at the primitive streak as it forms. (G and H) Double labeling for Snail2 (green) and Sox3 (red) transcripts at the full primitive streak stage (HH4) shows the complementary expression pattern between Snail2 and Sox3. A transverse vibratome section taken at the level shown by the dotted line in (G) allows a better assessment of the mutually exclusive expression pattern and the continued Snail2 expression in migratory mesendodermal cells emanating from the streak (H).

Figure 3

Snail2 and Sox3 Behave as Mutual Transcriptional Repressors in the Chick Embryo (A–E) Control embryo electroporated with a vector encoding GFP (A) shows the normal pattern of Sox3 expression in a dorsal view (B) and in a transverse section (C). (D) and (E) are high power images of the areas demarcated in (C). (F–K) Similar images taken from an embryo electroporated with Snail2 in the anterior epiblast shows the repression of Sox3 expression in the electroporated region (arrow in H and sections in I). High power images comparing the electroporated with the control side confirm Sox3 downregulation (J and K). (L–Q) Electroporation of a dominant-negative form of Snail2 (DN-Snail2) in the primitive streak extended Sox3 expression to the embryonic midline (arrow in N), as better assessed in the images shown in (O–Q). (R–X) Ectopic Sox3 expression by electroporation at the primitive streak inhibited Snail2 expression (black arrow in T and a section at different magnifications in (U–X).

Figure 4

Sox3 Overexpression or Snail2 Inhibition Block Ingression at the Primitive Streak (A–X) Time-lapse confocal analysis of cell movements in cultured embryos electroporated with GFP alone (A–D), GFP plus Snail2 (E–H), GFP plus Sox3 (I–L), GFP plus a dominant-negative form of Snail2 (M–P), GFP plus a dominant-negative form of Sox3 (Q–T), and GFP, Sox3, and Snail2 (U–X). BF (t = 0) and GFP (t = 0) show transmitted light and fluorescent images obtained 6 hr after electroporation, just before the start of time lapse analyses. Colored dots in (B, F, J, N, R, and V) represent the initial position of cells that were subsequently followed. Dots in (C, G, K, O, S, and W) represent the final position of those cells and the lines their tracks. (D), (H), (L), (P), (T), and (X) show the percentages of ingressing and noningressing cells in each condition. The white dotted lines indicate the relative position of the midline; black arrows and ps indicate primitive streak. Control GFP-expressing embryos showed the normal movements of epiblasts cells toward the primitive streak and subsequent ingression (C). Snail2 overexpression increases the percentage of ingressing cells (E–H). Ectopic Sox3 expression blocked cell ingression through the primitive streak without affecting the convergence movement toward the midline (compare C and D with K and L). A dominant-negative form of Snail2 also blocked cell ingression (compare C and D with O and P), whereas a dominant-negative form of Sox3 increased ingression (compare T with D). Overexpression of both Snail2 and Sox3 could rescue the ingression behavior of epiblast cells (compare C and D with W and X).

Figure 5

Snail2 and Sox3 Are Direct Mutual Transcriptional Repressors (A) Conserved Snail-binding site in the Sox3 promoter (see also Figure S1). (B) Snail2 overexpression in cultured embryos repressed the activity of the Sox3 promoter, whereas a deletion that removes the Snail binding site significantly restores promoter activity. (C) Chromatin immunoprecipitation (ChIP) analysis confirms that Snail2 can directly bind to the Sox3 promoter. ChIP assays were carried out with anti-myc antibodies on embryo extracts that were previously electroporated with either GFP or myc-Snail2. PCR fragments were obtained after amplification of the indicated promoter region from the input (1), histone H3 positive control (2), IgG negative control (3), and myc immunoprecipitated fractions (4). (D) Conserved sequence for Sox transcription factors binding in the Snail2 promoter (see also Figure S1). (E) Sox3 overexpression in cultured embryos repressed Snail2 promoter activity, and the deletion of the Sox-binding site restored most of the promoter activity. (F) Chromatin immunoprecipitation (ChIP) analyses confirmed that Sox3 can directly bind to the Snail2 promoter. ChIP assays were carried out as described above.

Figure 6

Snail2 Represses Neural and Nonneural Ectodermal Markers without Inducing Mesodermal or Endodermal Fate (A–E) Snail2 or Sox3 were electroporated together with GFP in the epiblast of stage HH2 embryos. Snail2 repressed the expression of Dlx5, a nonneural ectodermal marker (A) and Otx2, an anterior neural marker (B), whereas Sox3 extended Dlx5 expression (A). Snail2 was unable to induce the expression of mesodermal markers, as assessed by examining the expression of the T-box genes Brachyury and Tbx6L (C and D) or the endodermal marker Sox17 (E), whereas Sox3 strongly repressed mesodermal (C and D) but not endodermal markers (E). Overexpression of both Snail2 and Sox3 did not affect the expression of the ectodermal markers Dlx5 and Otx2 (F), suggesting that their repression by Snail2 acts through Sox3 downregulation.

Figure 7

The Antagonistic Relationship between Sox3 and Snail Factors Is Conserved in the Mouse (A) Expression of Snail1 and Sox3 in 7,5 d.p.c. mouse embryos. Snail1 is the family member expressed in the primitive streak and early mesendodermal cells in the mouse (Sefton et al., 1998). Note the similarities in the expression patterns between chick and mouse embryos. A mutually exclusive expression pattern is observed in the transverse sections obtained at the level of the primitive streak. (B) Snail1 expression is increased in embryoid bodies (EB) obtained from Sox3 knockout embryonic stem cells compared to wild-type stem cells (CCE line). Consistent with the increase in Snail1 expression, E-cadherin transcripts are downregulated in Sox3 null EBs (T-test statistical analysis). (C) EBs derived from the CCE line are round and smooth, whereas those derived from Sox3 null ES cells show irregular edges with budding cells delaminating from the EBs, leading to abundant isolated cells in the culture medium (not shown). E-cadherin is strongly downregulated in Sox3^−/− compared to wild-type (WT) EBs. Conversely, Snail1 mRNA and protein are detected only in Sox3^−/− EBs, mainly in cells at the edges. (D) Expression of Snail1 in WT and Sox3^−/−; Sox2^+/− mouse embryos. Ectopic Snail1 expression can be observed in the mutant embryos leading to an extended region of EMT and cell ingression. The dotted lines in the whole mounted embryos indicate the level of the sections shown in the lower panels. Red brackets indicate the areas of cell delamination in the enlarged pictures and in the diagrams, which schematically show the defective phenotype in the gastrulating embryos. Dotted lines in the diagram indicate the midline.

Figure 8

The Antagonistic Relationship between Sox3 and Snail Factors Is Conserved in Human Cancer Cells (A) Expression of Sox3 and Snail1 in different human cancer cell lines. Epithelial MCF7 cells express high levels of Sox3 and low levels of Snail1. Conversely, mesenchymal MDA231, MDA435, and A375P lines are almost devoid of Sox3 transcripts and show high Snail1 expression. (B) Efficient Sox3 downregulation by specific siRNA in MCF7 cells is accompanied by an increase in Snail1 and associated mesenchymal markers Adam12 and Fibronectin and a decrease in the epithelial marker Claudin1. (C) TGF-β mediated Snail1 induction in MCF7 cells is accompanied by a decrease in Sox3 expression. This decrease is Snail1-dependent, as it is prevented by transfection with Snail1 siRNA. (D and E) Sox3 stable overexpression in MDA435 leads to morphological changes compatible with a partial mesenchymal to epithelial transition (D). These morphological changes are associated with a decrease in Snail2 expression and an increase in E-cadherin expression (E). (F) The migratory behavior of MDA435-Sox3 stable transfectants was tested in culture using a wound healing assay. Control MDA435 cells cover the wound in 12 hr, in clear contrast with Sox3-expressing cells. (G) Invasive behavior of MDA435-Sox3 cells. The nuclei of cells that invaded the collagen matrix were stained with DAPI and quantified. MDA435 cells expressing Sox3 significantly decreased their invasive properties. (H) Diagram showing the relationship between Snail and Sox3 factors in gastrulating embryos and in cancer cells. In gastrulating embryos, the mutual repression between Sox3 and Snail1 regulates the decision to ingress at the primitive streak. Sox3 expressing cells do not ingress, ensuring the development of ectodermal derivatives in developing embryos. In cancer cells, Snail induces EMT and its expression correlates with invasive properties. Sox3 represses Snail, thereby preventing EMT to maintain epithelial integrity.