Cancer stem cells (CSCs) are a subset of tumor cells that are responsible for initiating and maintaining the disease. However, the genetic changes and cellular context determining those characteristics of CSCs remain unclear. In addition, what differentiation properties are suitable or refractory to the generation of CSCs is still unknown. To address these questions, we need to obtain a large number of CSCs and attempted to induce CSC population from mouse normal tissues.
We recently developed novel mouse osteosarcoma (OS) model by transplantation to syngeneic mice with c-MYC overexpressing bone marrow stromal cells derived from Ink4a/Arf(−/−) mice. Single cell cloning revealed that two distinctly different cells were candidates for the origin of OS: bipotent cells (AX) with bilineage (osteogenesis and chondrogenesis) differentiation potential and tripotent cells (AO) having trilineage (adipogenic, osteogenic, and chondrogenic) differentiation potential. Their differentiation potentials and gene expression profiling suggested that AX cells were derived form osteo-chondro-committed progenitor cells, while AO cells were originated from MSCs.
Bipotent AX cells were highly tumorigenic and possessed high propensity for distant metastasis that mimics human disease. In addition, they showed both terminal differentiation and self-renewal capacity in vivo, which are properties ascribed to CSC.
Modulating adipogenic potential by depletion and overexpression of PPARγ in AO or AX, respectively, affected cell proliferation and tumorigenic activity. Thus cellular differentiation potential is a critical factor for inducing OS CSC and regulating their functions.
Interestingly, although tripotent AO showed low tumorigenic activity, they exhibited drug resistance and could be transformed to bipotent highly tumorigenic cells like AX in vivo. These findings suggest that CSCs consist of heterogeneous fractions and that our OS model might provide clues to clarify ‘bona fide’ CSC characteristics.
Moreover, secreted soluble factors provided by normal environmental tissues such as Fgf2 and Lif maintained immature state and contributed to aggressiveness of OS in vivo, suggesting that CSC characteristics are not only regulated by the cell intrinsic factors but also heavily regulated by environmental factors.
Finally, we would like to mention a newly identified mechanism which critically regulates tumorigenic activity of CSC in vivo. Our induced OS CSC mouse model is useful for both basic and clinical research to develop novel therapeutic approaches.