Cancer-associated fibroblasts are very active promoters of breast carcinogenesis through functional cross-talks with cancer cells via secretions that contain metabolites, cytokines and other types of molecules. CAFs constitute a large heterogeneous population of cells that express various types of markers, which are not unique for these cells. This vast heterogeneity could be owing to various origins of CAFs or their mode/condition of activation [2, 4]. We have shown in the present study that MDA-MB-231- and IL-6-dependent activation of normal fibroblasts upregulates the epithelial markers (E-cadherin and EpCAM) but downregulates the mesenchymal markers (N-cadherin and vimentin). This suggests that the activation of fibroblasts is accompanied by a transition from the mesenchymal to the epithelial state (MET). Intriguingly, despite this MET, these cells showed higher migration, invasion and proliferation capacities as compared to controls (Fig. 1C and ). This could be explained by the fact that these cells express high levels of the active form of the pro-invasive/migratory and -proliferative protein kinase AKT (Fig. 1D). Interestingly, ectopic expression of p16 or knock-down of STAT3 in breast stromal fibroblasts suppressed the IL-6-related activation of these cells and their transition to epithelial status. This indicates that while p16 inhibits the IL-6 induced MET, STAT3 activation was required for the acquisition of the epithelial traits.
The development of epithelial features and down-regulation of mesenchymal markers were confirmed in 12 patient-derived CAFs and their adjacent counterpart TCFs. The active status of CAFs was confirmed by the higher expression level of a-SMA, which was accompanied in most CAFs with increased level of E-cadherin and reduced level of N-cadherin and vimentin as compared to their corresponding TCFs. This indicates that, like MDA-MB-231 cells and IL-6, breast cancer cells activate fibroblasts and trigger MET within tumors. It has been previously shown that the complete reprogramming of fibroblasts into pluripotent stem cells entails also up-regulation of the epithelial markers and down-regulation of the mesenchymal markers [15–17]. This confirms the importance of the MET process during fibroblast reprogramming.
Importantly, active CAFs expressed high levels of the myoepithelial markers cytokeratin 14, cytokeratin 19, cytokeratin 5/6 as well as c-Kit, in addition to a-SMA, which is highly expressed in both myofibroblasts and myoepithelial cells. However, CAFs did not express cytokeratin 18, the luminal cells biomarker. This indicates that active CAFs acquired myoepithelal cells characteristics. These findings confirm what has been recently shown by Su et al. who have shown the expression of the myoepithelial cells marker CD10 on a subset of active CAF cells . This suggests that active fibroblasts develop myo-fibroblast as well as myo-epithelial traits, with a novel shape different from that of fibroblast and myoepithelial cells. Indeed, while myoepithelial cells are stellate in shape, and fibroblasts have spindle-shape, myofibroblasts are characterized by branching and flat morphology [2, 4].
Since a-SMA-expressing myoepithelial cells were shown to exhibit some features of lineage-restricted stem cells , we checked this possibility in active breast fibroblasts. We have shown that IL-6-activated breast and skin fibroblasts also exhibited the (CD44high/CD24low/ALDHhigh) feature of mammary stem cells, in a p16- and STAT3-dependent manner. In addition, we have shown that the proportions of CD44high as well as ALDHhigh sub-populations were higher in CAF cells compared to their counterpart TCF cells isolated from the same patient. This showed that activated fibroblasts acquired stem/progenitor cell features. This was confirmed in 12 pairs CAF/TCF cells isolated from 12 different patients. Indeed, most of these cells exhibited the (CD44high/CD24low/ALDHhigh) feature of mammary stem cells (Fig. 5C). Together, these findings indicate that CAFs are myofibroblasts that express myoepithelial as well as stem cell markers, with high migratory/invasive and proliferative capacities. This confirms that they do not represent a cell type but rather a cell state, as has been previously proposed by Madar et al. .
Interestingly, we have also shown that this CD44high sub-population of CAF cells has much higher capacity to grow in suspension and form mammospheres compared to CD44low sub-population. Furthermore, these cells have higher capacity to promote cell growth of breast cancer cells in a co-culture setting (Fig. 5). Similarly, it has been previously shown that CD44 was abundantly expressed on immortalized CAF cells isolated from melanoma bearing mice, and that CD44-positive cells supported the stemness and drug resistance of cancer cells . This indicates that a proportion of CAFs exhibit stemness features and have higher paracrine pro-carcinogenic effects than the corresponding CD44-negative cells. The fact that the pluripotency marker Sox-2 was only lowly expressed in most CAF/TCF pairs, indicates that these CAF-stem cells did not reach pluripotency. Thereby, which type of stem cells are formed in active breast fibroblasts ? and what are their origin ? Ishikawa et al. have shown simultaneous expression of cancer stem cell-like and CAF-like properties in a primary culture of breast cancer cells . In another study, it has been reported the presence of multipotent CAF cells in hepatocellular carcinoma tissues . Furthermore, Herrera et al showed that most pro-migratory CAFs from human colon tumors expressed stem cell markers . Together, these findings indicate that active CAFs contain a sub-population of cells with stem features, and that this population possesses the higher pro-carcinogenic ability. In a recent publication, Nair et al. have proposed a cancer stem cell model as the point of origin of CAFs . This suggests that CAF cells in a tumor may have different origins with different gene expression signature and consequently different effects on cancer cells. It’s also possible that CAF heterogeneity arises from different paracrine signaling effects from heterogeneous cancer cells as well as other active stromal cells such as adipocytes and macrophages. Indeed, we have shown here that TNBC cells (MDA-MB-231) as well as IL-6 can promote stemness in breast stromal fibroblasts.