The current interest in sequencing the genomes of cancer cells has distracted attention away from the role of epigenetic programs, which may in the end be responsible for the lion’s share of distinct cancer cell-associated phenotypes. One type of epigenetic program derives from differentiation programs of the normal cells-of-origin of cancer cells, which continue to influence the phenotypes of derived cancer cells. A second type of program is represented by the epithelial-mesenchymal transition (EMT) program, which operates in normal epithelial tissues that undergo wounding and dominates the behavior of carcinoma cells at various grades of tumor progression. When activated by heterotypic signals that carcinoma cells receive from the adjacent tumor-associated stromal cells, the EMT program confers a variety of malignancy-associated phenotypes, including motility, invasiveness, a resistance to various therapies, an ability to disseminate to distant sites in the body, and tumor-initiating ability, the last being the trait of cancer-stem cells (CSCs). This association between stemness and the EMT program operates in a wide variety of normal and neoplastic epithelial cell types, and enables cancer cells to serve as founders of metastatic colonies. Experimental activation of the EMT program in primary carcinoma cells enables their dissemination and blocking of this program prevents such dissemination. As tumor progression proceeds in the mouse mammary gland, the Slug EMT-inducing transcription factor (EMT-TF) expressed in normal SCs is replaced by its paralogous cousin, Snail, albeit in a different cell layer. Accordingly, the resulting CSCs do not arise from normal tissue SCs. Residence of mammary carcinoma cells in epithelial (E) or mesenchymal (M) cell states holds important implications for their responses to immune attack: activation of anti-checkpoint immunotherapies results in blocking the growth of tumors composed of E carcinoma cells but not of M carcinoma cells.
Hence, the current limitations of checkpoint immunotherapies in the oncology clinic may be due in no small part to the representation of more M carcinoma cells within tumors under treatment. These biological complexities reveal how limited are the insights gleaned from focusing exclusively on the somatically mutated genomes of cancer cells.