Easily moulded

Credit: Lara Crow/NPG

Previous studies have indicated that some cells that lack cancer stem cell (CSC) properties can transition to CSC-like cells. However, it is unclear how common these types of transitions are or how they are controlled. Chaffer, Marjanovic, Weinberg and colleagues have investigated this further in breast cancer cell lines.

CSCs in breast cancers have been shown to exist exclusively in the CD44^hi compartment. The authors found that CD44^hi and CD44^low populations of cells coexist in several cell lines derived from basal breast cancers, but that cell lines from less aggressive luminal cancers are mostly composed of CD44^low cell populations. Purified CD44^low cells (>99.7% purity) from both basal and luminal cancer cell lines were able to seed tumours in the mammary fat pads of immunocompromised mice. Although tumours formed more frequently from CD44^hi cells, these data suggested that populations that lack CD44^hi cells (and that therefore presumably lack CSCs) can generate tumours. An investigation of the tumours arising from CD44^low cells showed that they contained a small fraction of CD44^hi cells (2–22% of the tumour cells), which indicates that some cells had converted from a CD44^low to a CD44^hi state.

How might these cells convert from a CD44^low to a CD44^hi state? Previous data have shown that stem-like cells have more mesenchymal properties than non-stem cells, and have linked this to the epithelial-to-mesenchymal transition (EMT) programme. Analysis of EMT-controlling transcription factors in HME-flopc cells (non-transformed mammary epithelial cells that frequently undergo spontaneous CD44^low-to-CD44^hi transitions) indicated that zinc finger E-box binding homeobox 1 (ZEB1) was highly expressed in CD44^hi cells compared with CD44^low cells. Doxycycline-inducible short hairpin RNA (shRNA)-mediated knockdown of ZEB1 substantially reduced the ability of CD44^low cells to convert to CD44^hi cells, and this was reversed when doxycycline was withdrawn.

The miR-200 microRNA family and ZEB1 antagonize each other, and the authors showed that the inhibition of miR-200 increased ZEB1 mRNA levels and the CD44^low-to-CD44^hi transition in HME-flopc cells; this was abrogated when ZEB1 was also knocked down, which indicates that ZEB1 is a key miR-200 target in this process. Similar results were observed in transformed HME-flopc cells. Furthermore, ZEB1 and miR-200 expression were inversely correlated in breast cancer cell lines, with higher expression of ZEB1 in cells from basal tumours (which have higher levels of CD44 expression) than in cells from luminal tumours.

populations that ... presumably lack CSCs ... can generate tumours

The induction of ZEB1-targeted shRNAs in basal breast cancer cell lines decreased the ability of the CD44^low population to give rise to tumours in immunocompromised mice, which shows the importance of ZEB1 expression for tumour initiation. Intriguingly, although ZEB1 was required for the CD44^low to CD44^hi conversion and continuous ZEB1 expression was necessary for the maintenance of stem-like properties in vitro, continuous ZEB1 expression was not required for the maintenance of high cell surface expression of CD44, which indicates that stemness and CD44 expression can be uncoupled.

The authors hypothesized that ZEB1 might be rapidly induced as a result of epigenetic regulation; specifically, that ZEB1 might exist in a bivalent state in which the gene promoter contains both permissive and repressive histone H3 modifications so that the gene is repressed but is also ready for rapid activation. This is common in genes that control the cell state in embryonic stem cells. Analysis of histone methylation patterns in CD44^low HME-flopc cells and basal breast cancer cells (which are associated with more aggressive clinical behaviour) showed that the ZEB1 promoter was indeed in a bivalent state in these cells, whereas the ZEB1 promoter in CD44^hi cells had mostly activating marks. By contrast, the more benign CD44^low luminal breast cancer cells carried mostly repressive marks on the ZEB1 promoter, which indicates that these cells are unable to rapidly upregulate ZEB1 expression. Furthermore, transforming growth factor-β (TGFβ), a common inducer of EMT that upregulates ZEB1 expression, was shown to enhance CD44^low-to-CD44^hi transitions in basal but not in luminal breast cancer cells. Therefore, the differing clinical behaviours of these two subtypes of breast cancer cells could partly be explained by the differences in the configuration of the ZEB1 promoter.

These data raise several interesting questions, including whether the plasticity of CSC-like states is present in primary human breast tumours, whether it differs depending on the subtype and the aggressiveness of the tumour, and whether it exists in other tumour types.