In human carcinomas, the acquisition of an invasive phenotype requires a breakdown of intercellular junctions with neighboring cells, a process termed the epithelial-to-mesenchymal transition (E-MT). Upon arriving at secondary sites, the mesenchymal cells revert to an epithelial phenotype via a mesenchymal-to-epithelial transition (M-ET). Human carcinoma tissues and cells typically show extensive heterogeneity in both phenotype and genotype, suggesting a role for genetic instability in cell type determination. To test this possibility, we have developed methods to continually isolate phenotypic variants from epithelial or mesenchymal subclones of carcinoma cell lines. We have explored the signal pathway underlying E-MT/M-ET phenotypic switching by gene expression analysis, spectral karyotyping (SKY), and fluorescent in situ hybridization (FISH). We found that changes in chromosome content are associated with phenotypic switching. We further showed that these changes dictated the expression of specific genes, which in E-MT events are mesenchymal and in M-ET events are epithelial. Our results suggest that chromosome instability can provide the diversity of gene expression needed for tumor cells to switch phenotype.
Anti-cancer therapy based on blocking the HGF-Met signaling pathway has emerged as an important goal of pharmaceutical research. One of the limitations of studying the altered Met HGF/SF signaling of human cancers grafted in mouse models has been that the murine HGF/SF protein has a low affinity for human MET. To overcome this, our lab developed the transgenic human HGF-SCID mouse model (hHGFtg-SCID), which generates a human-compatible HGF/SF protein and thus allows for the propagation of human tumors. This model has proven to be a valuable preclinical tool for in vivo study of MET-dependent cancers and is used to evaluate treatment strategies that aim to target this pathway.