In the meanwhile, mesenchymal markers such as vimentin, fibronectin and N-cadherin are up regulated. Our data on the morphological and molecular changes between SOX2 knock down and mock control cells reflected the corresponding MET process, suggesting SOX2 is involved in the EMT process. The EMT is also marked by nuclear relocalization of ß-catenin. We also observed that relocalization of ß-catenin from cytoplasmic/nuclear to membrane after SOX2 knock down. Our data suggests a model for SOX2’s role regulating the balance between ß-catenin in the adhesion complex at the plasma membrane and in the signaling complex in the nucleus EMT process: SOX2 could reduce the expression of E-cadherin at plasma membranes, thus reducing the binding capacity of the adhesion complex for ßcatenin. In the meantime, the WNT signaling components in the nucleus recruit more ß-catenin, activates WNT signaling and its downstream targets. The WNT pathway is widely regarded as the crucial pathway for colorectal carcinogenesis. Our data also suggested that SOX2 activates WNT signaling activity as demonstrated by reduced LEF/TCF activity and reduced expression of well-known downstream target gene cyclin-D1 and c-Myc in SOX2-knock down cells. Our data in CRC is consistent with the report by Chen et al. who showed that SOX2 physically interacts with ß-catenin in human breast cancer cells. The relationship between other SOX family proteins and ßcatenin has also been reported. For example, Zorn et al. showed that in Xenopus, two SOX family members physically bind to beta-catenin and inhibit WNT pathway activity. Zhang et al. showed that in Xenopus, XSox3 did not physically interact with beta-catenin, but could regulate Xnr5, a beta-catenin/VegT-regulated early zygotic gene, thereby indirectly involved in the WNT signaling. The existence of a link between EMT and SOX2 is not surprising, as EMT is key developmental GDC-0941 program in embryo development and SOX2 plays a critical role in the early embryonic development and formation of tissues and organs. Furthermore, the EMT process is often activated during cancer invasion and metastasis and SOX2 is also over expressed in many types of cancers. With multiple attempts, we over-express SOX2 in the previously SOX2 knock down CRC cells, and showed that we achieved SOX2 overexpression in the SOX2 knock down cells comparing to mock control. However, we were not able to recue or reverse the MET process, or initiate an EMT process. We also overexpressed the SOX2 in the SOX2-negative SW480 CRC cells and showed that vimentin expression increased dramatically. However, the increase expression of vimentin did not generate the corresponding morphological changes related to EMT. This observation, although to our surprise, might suggest that SOX2 interacts or works with other factors in the process. The discrepancy between the effect of gene knock down and over expression have prior incidences. For example, Burbridge et al. knock down of Dcdc2 resulted in hippocampal malformations and a bimodal migration pattern of the transfected neurons. However, they showed that the treatment of shRNA-transfected neurons with the DCDC2 overexpression construct failed to rescue the migration phenotype. Retinoid X receptor alphanull mutants exhibit hypoplasia of their ventricular myocardium and die at the fetal stage. However, Subbarayan et al. showed that RXRa overexpression in cardiomyocytes causes dilated cardiomyopathy but fails to rescue myocardial hypoplasia in RXRa-null fetuses. Another example involves perlecan, which is a major heparan sulfate proteoglycan of extracellular matrices.