The conventional versus a new integrative CSC model for breast cancer.

<p>(A) In luminal epithelial cell-lines (e.g. in HMLE, HMLER, MCF7) the conventional model describes breast CSCs as M cells and ‘non cancer stem cells’ as E cells. Paradoxically, in more basal mesenchymal cancer cell-lines (MDA-MB231 cells, MCF10ACA1, 4T1) the more E gene-expressing cells are associated with CSC-properties. Thus, the identity of CSCs appears to be context-dependent. (B) Our model for CSCs integrates and explains these paradoxes by the existence of stem-like intermediate hybrid E/M cells independent of the tumor cell line. We propose that compared to more ‘polarized’ differentiated E (capable of plasticity) and M cell-types (capable of self-renewal), undifferentiated hybrid E/M cells can generate more mammospheres and heterogeneous progeny due to their capacity of both, self-renewal and plasticity. Arrows indicate possible state transitions (incomplete EMT and MET) between the instable hybrid E/M state and the extreme stable E and M states. Stemness of the intermediate E/M state is reflected both by presence of stem-like hybrid E/M cells and by co-presence of E and M cell-types due to cooperation. Context-independency of the stemness of the intermediate E/M state explains the paradoxical context-dependent meaning of E genes in basal tumors and M cell lines and stemness of M genes in luminal B tumors and E cell-lines (A).