Identification of two cancer-associated fibroblast markers revealing stromal heterogeneity in sustaining cancer progression and chemoresistance

Intra-tumor cellular heterogeneity is a key player of tumor aggressiveness and resistance to therapies. Genetic diversity in tumor cells enables, under selective pressures (e.g., during the metastatic process or under therapies), the rise of certain subclones proficient with survival and metastatic advantages (1). Yet, phenotypic tumor cell heterogeneity also critically dictates such adaptative behaviours, this plasticity relying on stromal signals that reversibly change signaling pathways or gene expression programs in tumor cells, and reciprocally (2,3). An illustration of the importance of this phenotypic heterogeneity in tumor cells and crosstalks with the stroma has been revealed in pancreatic cancer, the aggressiveness of which is closely linked to transcriptomic and epigenetic signatures (4,5). Tumor stroma is represented by acellular (extracellular matrices, vesicles, soluble proteins and lipids) and cellular components [cancer-associated fibroblasts (CAF), endothelial cells, immune cells and nerve cells], the variety of which underscores the complexity of tumor biology. Tumor stroma heterogeneity also ensues from its constant evolution during tumor progression or remodeling under treatments, and from inconsistent spatial distribution into the tumor of stromal components (6-8). Consequently, tumor cells will not, at a specific stage of tumor progression, be exposed to the same signals whether they localize within the primary tumor or at metastatic site, in a small or a large lesion, and in the central area or at the invasive fronts, underlying the great advances that should in a near future bring multi-spatial single cell phenotyping approaches. Another step of diversity within the stroma concerns the immune system which comprises, even within a similar immune cell class, not only anti-tumoral, but also pro-tumoral, cell subsets (e.g., M1 vs. M2 macrophages populations), these populations being evolving to functionally divergent subsets owing to tumor- or stromal (particularly, CAF)-derived polarizing factors (9-11). This implies that, depending on the factors released by the whole tumor, an anti- or pro-tumoral immunity will be built up, this signature impacting on patient prognosis and response to therapy (12). For example, it is now recognized that a stromal anti-tumoral lymphocyte T CD8 signature is of good prognosis in many cancers (13,14), albeit not sufficient since CD8 cytolytic activity is modulated by other microenvironment factors (myeloid cells, cytokines, hypoxia) (15)