The epithelial cell default-phenotype hypothesis and its implications for cancer.

The expression of epithelial cell adhesion and cytoskeletal genes is orchestrated by an apparently unique set of rules. No tissue-specific transactivator proteins have been found to drive them; only ubiquitous factors are utilized. In non-epithelial cells, they are actively repressed. Moreover, it was recently found that a single protein (adenovirus E1a) coordinately represses non-epithelial genes while inducing epithelial genes. A simple model is offered to explain how epithelial gene expression is coordinated. Under this model, the epithelial cell gene expression program is a transcriptional 'default'; that is, it occurs in the absence of tissue-specific transactivation. Conversion to this default requires only that mesenchymal transactivators are not expressed, or that central 'integrator' proteins are inactive. In their absence, mesenchymal gene expression cannot occur. Moreover, because the repressors cease to be expressed, the epithelial genes are induced. Oncogenes generally cause the breakdown of the epithelial phenotype--generating carcinomas--so genes such as E1a that cause epithelial conversion may prove useful for both understanding and controlling cancer.