Spatial self-organization of cancer stem cell niches revealed by live single-cell imaging

Phenotypic plasticity is a major factor of tumor heterogeneity and treatment resistance. In particular, cancer stem cells (CSCs) represent a small subpopulation within tumors with self-renewal and tumor-forming capabilities. Understanding reprogramming, maintenance, and lineage properties of CSCs requires dedicated tools to disentangle the respective influences of phenotypic inheritance and cell-cell interactions. Here we set up ultra-wide field microscopy of breast cancer cell lines expressing a stemness fluorescent reporter for several days. The fluorescent reporter distinguishes three phenotypes: cancer stem cells (CSCs), cancer differentiated cells (CDCs) and intermediate/transiting cancer cells (iCCs). Spatial statistics indicate significant zonation, aka phenotypic niches, with CSC clustering near each other but away from CDCs. Surprisingly, single cell time series reveal spontaneous reprogramming events from CDC to CSC even in unperturbed populations. We identify that such transitions are prone to arise during the cell cycle. Moreover, lineage analysis shows that the phenotype is partially inherited from ancestor cells. However, such heredity is not sufficient to explain the spatial properties of the cell population, which also depend on cell-cell interactions. Indeed, we identified that phenotypic transitions of cancer cells are influenced by the phenotypic state of neighboring cells. Reprogramming into CSCs is respectively promoted and inhibited by the presence of CSCs and CDCs in the neighborhood. Altogether, our results disentangle how phenotypic inheritance and intercellular interactions orchestrate the spatio-temporal self-organization of cancer cell heterogeneity, maintaining a subpopulation of CSCs within niches.