Twist2 could accumulate in nuclei during initial invasion and metastasis, and functions as a transcriptional factor to regulate EMT. Twist2 in nuclei could remarkably repress Ecadherin in the invasion edge to R428 promote EMT, thus increase cell motility and invasiveness to enter the new adjacent tissue. Recent findings suggest that cells undergone EMT were responsible for degrading the surrounding matrix to enable invasion and intravasation of both EMT and non-EMT cells. Only those non-EMT cells that had entered the blood stream were able to re-establish colonies in the secondary sites. Similarly, high nuclear b-catenin expression at the invasion front and less nuclear b-catenin in central tumor regions exist in colorectal carcinoma tissues. Thus, carcinoma cells may experience EMT in invasive front area, then the MET process in metastasis. When cancer cells move to their new homing sites, Twist2 redistributes to the cytoplasm with E-cadherin re-expression, thus carcinoma cells revert into a noninvasive state in the absence of ongoing exposure to the microenvironmental signals. This plasticity might result in the formation of new tumor colonies of carcinoma cells exhibiting a histopathology similar to those of carcinoma cells in the primary tumor that did not undergo an EMT. It is likely that EMT is triggered by genetic and epigenetic alterations of the tumor cells and their interaction with the surrounding microenvironment including stromal cells and matrix components. Little is known on the mechanisms controlling the release of these EMT signals within a tumor. In part, the understanding of these mechanisms is complicated by the fact that the EMT signals controlling cell number and position within tissues are thought to be transmitted in a temporally and spatially regulated fashion from one cell to its neighbors. Such paracrine signaling is difficult to access experimentally. The sequences necessary for dimerization in other T-box factors are not conserved in Mid, which is also consistent with Mid binding as a monomer. Xbra homodimerizes through a relatively small interface of 250 A˚ 2 found near the centre of the T-box domain. The small polar N129 residue in Xbra is replaced with a large hydrophobic F281 in Mid and F130 in Xbra is replaced by S282 in Mid. Likewise Xbra M85 is substituted with R235, and Xbra V173 corresponds to L326 in Mid. Overall, 4 of the 8 dimerization residues are not conserved in Drosophila Mid. Furthermore, Tbx20 also differs from both Mid and Xbra at these same 4 positions. The crystal structure of Tbx3 bound to a palindromic T-site shows that the two monomers are rotated with respect to one another on the DNA strand and use different residues to contact one another. These residues fall within a poorly conserved region of the T-box domain. Comparison to the corresponding Mid residues shows that none of these amino acids are conserved. Similarly, only Tbx3 D239 is identical to the corresponding Tbx20 residue. The small monomer-monomer contacts defined in the Tbx3 crystal structure are thought to be insufficient to facilitate dimerization and as such, Tbx3 is believed to bind as a monomer. Finally, the crystal structure of Tbx5 bound to a half-site unlikely to be in involved in dimerization. Taken together, the site selection data and the comparison of the Mid amino acid sequence with evidence from the crystal structures of the Xbra, Tbx3 and Tbx5 suggest that Mid binds DNA as a monomer. We have also found that Mid is able to directly regulate the transcription of the wingless gene, in vivo, by binding to sequences within the wg enhancer. The sequences Mid binds in order to regulate wg resemble the motif we present in this study.