Epithelial–mesenchymal transition in cancer

The generation of an organized architectural pattern during tissue formation is a process of paramount importance during development and organogenesis. A large proportion of the vertebrate body is made of epithelial tissues; these provide organized barriers between organ compartments and also differentiate into glandular, secretory specializations (Kahata et al., 2017). One feature of epithelial developmental history is their ability to generate mesenchymal cell types via a process best known as epithelial–mesenchymal transition (EMT) (Hay, 1995; Lim and Thiery, 2012). The barrier function of epithelia is also met in the organization of the vascular and lymphatic walls that are built by endothelial and lymphatic endothelial cells, respectively (van Meeteren and ten Dijke, 2012). A similar generation of mesenchymal cells from the endothelium is also relevant in heart morphogenesis. Actually, one of the very early observations of the EMT (or more properly, of the endothelial–mesenchymal transition (van Meeteren and ten Dijke, 2012) stems from studies of endothelial cells and formation of tissue diversifications within the heart, such as the valves and their septa. The EMT does not necessarily generate terminally differentiated mesenchymal cells but rather produces intermediate cell phenotypes that retain the potential to regenerate new epithelial tissue via the process of mesenchymal–epithelial transition (MET) (Nieto et al., 2016). EMT and MET may represent distinct and interdependent biological processes or, alternatively, they may embody reversible phases of one and the same process. As the locations where an EMT and a connected MET take place within the vertebrate body are often distinct, evidence for the reversibility of these two processes is often ambiguous, and the process is of unclear biological relevance. Clear cases of the MET can be understood by studying the dedifferentiation of mesenchymal cells, for example, fibroblasts, to renal epithelium during nephrogenesis (Davies, 1996). Moreover, MET is required for the generation of pluripotent stem cells from fibroblasts using the popular Yamanaka factor protocols (Sanchez Alvarado and Yamanaka, 2014). Sequential EMT and MET processes are needed to differentiate, for example, induced pluripotent cells to hepatocytes (Li et al., 2017). These studies indicate the relevance of these transitions in embryonic development and can provide new hints on their mechanisms. EMT and MET are not only embryonic physiological processes, but become activated during chronic inflammation, wound healing, and cancer metastasis (Lambert et al., 2017; Nieto et al., 2016). In the latter case, a substantial body of literature describes the contribution of EMT to the invasive state of various carcinomas (epithelial tumors), whereas MET is thought to operate once metastatic cells have reached a distant site of secondary growth (Nieto et al., 2016). In this respect, it is difficult to demonstrate and even understand how the two processes connect with each other when they are separated in time and space, and the cells that connect the EMT to MET may undergo multiple pathophysiological and epigenetic changes in the interim period. In addition, specific in vivo murine models of cancer metastasis occasionally dispute the contribution of the EMT to metastatic spread away from the primary oncogenic site (Fischer et al., 2015; Zheng et al., 2015). Further complication to the above concept is the proposal that cancer-associated EMT is not complete, and epithelial tumor cells generate intermediate phenotypes that express mixed epithelial and mesenchymal genes; such ‘hybrid’ cells may exert more malignant properties compared to the more differentiated epithelial or mesenchymal cells (Jolly et al., 2016). Although the EMT field has progressed to generate long lists of proteins and noncoding RNA, whose identity both marks and functionally defines the process, all modern studies focus on phenotypic analyses of key regulatory processes that can be referred to as hallmarks of the EMT (Nieto et al., 2016). These include disruption of cell–cell adhesion complexes, the most characteristic of which involve the adherens, tight, and desmosomal junctions; remodeling of the three classes of cytoskeleton, microfilaments, microtubules, and intermediate filaments; and finally, the