The Mesenchymal Stem Cell (MSC) is a self-renewing multipotent progenitor originally found in the bone marrow. It has drawn strong interest from translational research because of the multipotency of MSCs, their immunosuppression, their intrinsic homing and their promotion of wound healing in human patients. Currently available methods rely on in vitro culture for the isolation and expansion of MSCs and most studies used these cells for their investigation. Fibroblasts constitute the connective tissue and support epithelial or hematopoietic cells. Although they have already been extensively investigated, it is not clear whether stem cells exist in vivo which specifically regenerate fibroblasts. MSCs are a potential candidate, because of their fibroblastic shape in vitro, their differentiation towards several mesodermal lineages and their supportive function. We aim to characterize the in vivo differentiation potential of MSCs toward fibroblasts in normal and tumor tissues. A new transplantation approach was developed in order to transfer labeled MSCs in the bone marrow and assess their recruitment and lineage contribution in vivo. Long-term engraftment was accomplished by isolating and transplanting these cells within one day. To this end, strategies for their purification, preconditioning of the host and transplantation methodologies were established during this thesis. We were able to enrich MSCs over 1000 fold by combining a new extraction protocol, a negative selection by MACS and a positive selection by FACS. Cloning of sorted MSCs was performed to confirm their multipotency at single cell level. GFP-expressing MSCs were able to engraft in host mice tolerant for GFP and their expansion in vivo was stimulated by block of GSK3 activity. Altogether, the chimeric MSC model was used to track the recruitment of bone-marrow derived MSCs into spontaneous breast tumors. The tumor microenvironment encompasses several non-tumorigenic cells interacting with cancer cells. Among them, cancer-associated fibroblasts are known to participate in carcinogenesis. Whilst this heterogeneous stromal population has been extensively investigated, their source(s) remain elusive. I now propose tumoral MSCs as one source of cancer-associated fibroblast, based on their ability to self-renew and reconstitute stromal heterogeneity upon serial transplantation. Finally, I introduced the new concept that the multipotency of MSCs is subjugated by cancer cells to meet their requirements for enhanced tumor progression and prepare for metastasis. I found MSCs present in the tumor microenvironment to differentiate into cancer-associated osteoblasts and produce hydroxyapatite-mineralization in vivo. Microcalcification is an accepted poor prognostic feature in breast cancer patients but its function and origin is not documented. I showed that hydroxyapatite was able to promote proliferation of tumor cells and increased intracellular calcium concentration. Lastly, cancer-associated osteoblast and hydroxyapatite were able to promote tumor growth and metastasis in vivo. To conclude, bone marrow derived MSCs are circulating stem cells that participate in the establishment of the pathogenic stroma in breast cancer. Owing to their multipotency, MSCs give rise to cancer-associated osteoblasts that induce pro-malignant mineralization in situ. Subsequently, cancer cells harbor increased intracellular calcium concentration, enhanced proliferation and metastasis. We foresee the development of new stroma-targeted therapies that prevent the mineralization of tumor tissue and reduce metastasis in breast cancer patients.