**4.1 The cytokine cocktail secreted by MSCs**

MSCs secrete a plethora of cytokines and chemokines. In addition to the immuno-regulatory proteins, such as TGF1, IL-6 and prostaglandin E2, MSCs produce many other interleukins, including IL-7, IL-8 and IL-9, CC-type chemokines (CCL1, 2, 5, 8, 11, 15, 16, 20, 22, 26, and 27), CXC-type chemokines (CXCL1, 5, 6, 10, 11, 12, 13, and 16) and other factors, such as TIMP (tissue inhibitor of metalloproteases) -1 and -2, TNF and , PDGF A and B, G-CSF (granulocyte colony-stimulating factor), HGF (hepatocyte growth factor), VEGF and angiopoietin (Parekkadan et al., 2007). The syntheses of these factors can be further stimulated. E.g., IL-6 induces the expression of CXCL7, which further enhances the

Involvement of Mesenchymal Stem Cells in Breast Cancer Progression 255

Fig. 3. Paracrine actions of MSCs on breast cancer cells (BCCs). The effects of MSCs on luminal A and basal/mesenchymal subtype BCCs, on murine BCCs and on cancer stem cells (CSCs) are separately displayed. IL-6/-17B(R) = interleukin-6/-17B (receptor), VEGF = vascular endothelial growth factor, SDF-1 = stromal-derived factor-1, DKK = dickkopf, INF-1 = interferon 1, TGF = transforming growth factor , TNF = tumor necrosis factor , EGFR = epidermal growth factor receptor, NF-B = nuclear factor kappa-light-chain-enhancer of activated B-cells, Stat3 = signal transducer and activator of transcription 3, ADAM10 = a disintegrin and metalloprotease 10, EMT = epithelial-to-mesenchymal transition.

MCF-7 cells. In mouse xenografts, MCF-7 tumor formation and growth were fostered by MSCs (Klopp et al., 2010). Though ER-positive MCF-7 cells are dependent on estrogen for growth, MSCs may even trigger estrogen-independent proliferation of MCF-7 cells (Rhodes et al., 2009). The estrogen-independent growth may nevertheless be dependent on ER, as the proliferation-promoting effect of MSCs on MCF-7 cells was found to be blocked by ERspecific inhibitor ICI 182780 (Rhodes et al., 2010). It is thought that, by a yet unknown mechanism, MSCs activate ER which, in turn, stimulates the expression of SDF-1/CXCL12, a chemokine shown to trigger the proliferation of MCF-7 cells. Growth-stimulating effects of MSCs were also found on other ER-positive breast cancer cell lines, including T47D, BT474 and ZR-75-1 (Sasser et al., 2007a) and may be dependent on similar mechanisms as those on MCF-7 cells. There are also two reports that show that MSCs are able to inhibit the proliferation of MCF-7 cells (Goldstein et al., 2010; Qiao et al., 2008a). Dickkopf-1 (DKK-1) secreted by MSCs and known to block differentiation and to promote proliferation of MSCs

expression levels of IL-6, IL-8, CXCL5 and CXCL6 (Liu et al., 2011). TGF (transforming growth factor ) was found to stimulate MSCs to secrete more IL-6, IL-8, angiopoietin-2, G-CSF, HGF, VEGF and PDGF-BB (De Luca et al., 2010). TNF forced MSCs to increase their expression of CXCL9, CXCL10 and CXCL11 (Shin et al., 2010). Exposure of MSCs to conditioned medium from tumor cells also stimulate expression of chemokines, such as CXCL2 and CXCL12 (Menon et al., 2007). Direct interactions of cancer cells with MSCs may as well contribute to the rise of chemokine secretion by MSCs. Direct contacts of MSCs with MDA-MB-231 breast cancer cells were found to strongly upregulate the production of RANTES/CCL5 (Karnoub et al., 2007).

### **4.2 Modulatory effects of MSCs on breast cancer cell function**

MCF-7 cells are ER-positive luminal A-type breast cancer cells that show many features of normal breast epithelial cells, including the formation of E-cadherin-based cell-cell interactions and the ability to generate multicellular 3D-aggregates that can mature to lumen-containing spheroids (dit Faute et al., 2002; do Amaral et al., 2010). Decreased expression or complete loss of E-cadherin has been linked to epithelial-mesenchymal transition (EMT) and increased cellular migration of breast epithelial cells as well as to metastasis (Cano et al., 2000; Chua et al., 2007; Mani et al., 2008; Onder et al., 2008). We and others have shown that MSCs negatively interfere with the E-cadherin status of MCF-7 cells either by downregulation of the full length protein (Fierro et al., 2004; Hombauer & Minguell, 2000; Klopp et al., 2010) or by increasing E-cadherin shedding as triggered by the transmembrane protease ADAM10 (a disintegrin and metalloprotease 10) (Dittmer et al., 2009). It is noteworthy that as few as one MSC per 500 MCF-7 cells was sufficient to induce E-cadherin shedding. E-cadherin shedding leads to extracellular E-cadherin fragments that may block E-cadherin-based cell-cell contacts by competing with membrane-bound Ecadherin proteins (Ryniers et al., 2002). Hence, both downregulation of E-cadherin expression and increased E-cadherin shedding may decrease the strength of E-cadherinbased cell-cell interactions. With intercellular adhesions weakened cellular migration may increase. In fact, MSCs have been shown to significantly enhance the migratory activity of MCF-7 cells (Dittmer et al., 2009; Rhodes et al., 2010). Also along with the destabilization of cell-cell contacts, disruption of the architecture of MCF-7 spheroids was observed. It is interesting that, despite these changes in the E-cadherin status, MSCs did not induce EMT of MCF-7 cells, as indicated by the failure of MSCs to stimulate the expression of mesenchymal markers, such as vimentin or snail (Dittmer et al., 2009; Klopp et al., 2010). However, in the luminal A-type T47D breast cancer cell line, MSCs not only downregulated E-cadherin levels, but also increased expression of vimentin, snail, twist and N-cadherin (Martin et al., 2010) suggesting that, under certain conditions, MSCs can induce EMT of breast cancer cells. MSCs were also shown to increase the proliferation of MCF-7 cells in a dose-dependent manner (Fierro et al., 2004; Klopp et al., 2010; Rhodes et al., 2010; Sasser et al., 2007a). These effects may be mediated by IL-6, VEGF and/or SDF-1/CXCL12 as secreted by MSCs (Fig. 3) (Fierro et al., 2004; Sasser et al., 2007b). MSCs or similarly IL-6 induced the phosphorylation of STAT3 (signal transducer and activator of transcription 3) on tyrosine-705 in MCF-7 cells (Sasser et al., 2007b). Incubation of MSCs with TGF, a ligand of the EGFR (epidermal growth factor receptor), further stimulated the secretion of IL-6 and other factors (De Luca et al., 2010). This suggests that TGF-primed MSCs would even be more effective in promoting proliferation of MCF-7 cells. MSCs also enhanced the tumorigenic activity of

expression levels of IL-6, IL-8, CXCL5 and CXCL6 (Liu et al., 2011). TGF (transforming growth factor ) was found to stimulate MSCs to secrete more IL-6, IL-8, angiopoietin-2, G-CSF, HGF, VEGF and PDGF-BB (De Luca et al., 2010). TNF forced MSCs to increase their expression of CXCL9, CXCL10 and CXCL11 (Shin et al., 2010). Exposure of MSCs to conditioned medium from tumor cells also stimulate expression of chemokines, such as CXCL2 and CXCL12 (Menon et al., 2007). Direct interactions of cancer cells with MSCs may as well contribute to the rise of chemokine secretion by MSCs. Direct contacts of MSCs with MDA-MB-231 breast cancer cells were found to strongly upregulate the production of

MCF-7 cells are ER-positive luminal A-type breast cancer cells that show many features of normal breast epithelial cells, including the formation of E-cadherin-based cell-cell interactions and the ability to generate multicellular 3D-aggregates that can mature to lumen-containing spheroids (dit Faute et al., 2002; do Amaral et al., 2010). Decreased expression or complete loss of E-cadherin has been linked to epithelial-mesenchymal transition (EMT) and increased cellular migration of breast epithelial cells as well as to metastasis (Cano et al., 2000; Chua et al., 2007; Mani et al., 2008; Onder et al., 2008). We and others have shown that MSCs negatively interfere with the E-cadherin status of MCF-7 cells either by downregulation of the full length protein (Fierro et al., 2004; Hombauer & Minguell, 2000; Klopp et al., 2010) or by increasing E-cadherin shedding as triggered by the transmembrane protease ADAM10 (a disintegrin and metalloprotease 10) (Dittmer et al., 2009). It is noteworthy that as few as one MSC per 500 MCF-7 cells was sufficient to induce E-cadherin shedding. E-cadherin shedding leads to extracellular E-cadherin fragments that may block E-cadherin-based cell-cell contacts by competing with membrane-bound Ecadherin proteins (Ryniers et al., 2002). Hence, both downregulation of E-cadherin expression and increased E-cadherin shedding may decrease the strength of E-cadherinbased cell-cell interactions. With intercellular adhesions weakened cellular migration may increase. In fact, MSCs have been shown to significantly enhance the migratory activity of MCF-7 cells (Dittmer et al., 2009; Rhodes et al., 2010). Also along with the destabilization of cell-cell contacts, disruption of the architecture of MCF-7 spheroids was observed. It is interesting that, despite these changes in the E-cadherin status, MSCs did not induce EMT of MCF-7 cells, as indicated by the failure of MSCs to stimulate the expression of mesenchymal markers, such as vimentin or snail (Dittmer et al., 2009; Klopp et al., 2010). However, in the luminal A-type T47D breast cancer cell line, MSCs not only downregulated E-cadherin levels, but also increased expression of vimentin, snail, twist and N-cadherin (Martin et al., 2010) suggesting that, under certain conditions, MSCs can induce EMT of breast cancer cells. MSCs were also shown to increase the proliferation of MCF-7 cells in a dose-dependent manner (Fierro et al., 2004; Klopp et al., 2010; Rhodes et al., 2010; Sasser et al., 2007a). These effects may be mediated by IL-6, VEGF and/or SDF-1/CXCL12 as secreted by MSCs (Fig. 3) (Fierro et al., 2004; Sasser et al., 2007b). MSCs or similarly IL-6 induced the phosphorylation of STAT3 (signal transducer and activator of transcription 3) on tyrosine-705 in MCF-7 cells (Sasser et al., 2007b). Incubation of MSCs with TGF, a ligand of the EGFR (epidermal growth factor receptor), further stimulated the secretion of IL-6 and other factors (De Luca et al., 2010). This suggests that TGF-primed MSCs would even be more effective in promoting proliferation of MCF-7 cells. MSCs also enhanced the tumorigenic activity of

RANTES/CCL5 (Karnoub et al., 2007).

**4.2 Modulatory effects of MSCs on breast cancer cell function** 

Fig. 3. Paracrine actions of MSCs on breast cancer cells (BCCs). The effects of MSCs on luminal A and basal/mesenchymal subtype BCCs, on murine BCCs and on cancer stem cells (CSCs) are separately displayed. IL-6/-17B(R) = interleukin-6/-17B (receptor), VEGF = vascular endothelial growth factor, SDF-1 = stromal-derived factor-1, DKK = dickkopf, INF-1 = interferon 1, TGF = transforming growth factor , TNF = tumor necrosis factor , EGFR = epidermal growth factor receptor, NF-B = nuclear factor kappa-light-chain-enhancer of activated B-cells, Stat3 = signal transducer and activator of transcription 3, ADAM10 = a disintegrin and metalloprotease 10, EMT = epithelial-to-mesenchymal transition.

MCF-7 cells. In mouse xenografts, MCF-7 tumor formation and growth were fostered by MSCs (Klopp et al., 2010). Though ER-positive MCF-7 cells are dependent on estrogen for growth, MSCs may even trigger estrogen-independent proliferation of MCF-7 cells (Rhodes et al., 2009). The estrogen-independent growth may nevertheless be dependent on ER, as the proliferation-promoting effect of MSCs on MCF-7 cells was found to be blocked by ERspecific inhibitor ICI 182780 (Rhodes et al., 2010). It is thought that, by a yet unknown mechanism, MSCs activate ER which, in turn, stimulates the expression of SDF-1/CXCL12, a chemokine shown to trigger the proliferation of MCF-7 cells. Growth-stimulating effects of MSCs were also found on other ER-positive breast cancer cell lines, including T47D, BT474 and ZR-75-1 (Sasser et al., 2007a) and may be dependent on similar mechanisms as those on MCF-7 cells. There are also two reports that show that MSCs are able to inhibit the proliferation of MCF-7 cells (Goldstein et al., 2010; Qiao et al., 2008a). Dickkopf-1 (DKK-1) secreted by MSCs and known to block differentiation and to promote proliferation of MSCs

Involvement of Mesenchymal Stem Cells in Breast Cancer Progression 257

secrete IFN- have also a suppressing effect on growth of MDA-MB-231 cells in mouse xenografts (Studeny et al., 2004). It may well be that the ratio of tumor-suppressing vs. tumor-promoting factors as secreted by MSCs determine whether MSCs promote or inhibit

There is growing evidence that, in accordance with the hierarchical model of cancer development (Visvader & Lindeman, 2008), breast cancer is driven by cancer stem cells (CSCs) (Liu & Wicha, 2010). Breast CSCs are characterized by high expression of surface marker CD44 and low expression of CD24 (Fillmore & Kuperwasser, 2007). Another useful breast CSC marker is ALDH-1 (Ginestier et al., 2007). A recent study showed that MSCs increase the pool of CSCs in breast cancer lines, including MCF-7, SUM149 and SUM159 cells (Liu et al., 2011). Interestingly, bone marrow-derived MSCs themselves show also a hierarchical organization with only a minority of cells expressing the stem cell marker ALDH-1. And only those ALDH-1 positive MSCs were able to interfere with the CSC pool. Wicha and his co-workers showed that the MSC/breast cancer interaction generated a cytokine network which is initiated by IL-6 as secreted by breast cancer cells (Liu et al., 2011). IL-6 induces the production of CXCL7 in MSCs which, in turn, triggers the expression of a number of other cytokines and chemokines, namely IL-6, IL-8, CXCL6 and CXCL5, in both MSCs and breast cancer cells. This mixture of secreted factors then stimulates the expansion of the CSC pool. In line with the observation that MSCs induce the CSC pool to expand is the finding that MSCs stimulated mammosphere formation of normal mammary epithelial cells (Klopp et al., 2010). Evidence has been accumulated suggesting that the generation of mammospheres depends on the presence of mammary stem cells (Dontu et al., 2003). Hence, the number of mammospheres formed is supposed to be a measure of the number of mammary stem cells present (Charafe-Jauffret et al., 2008). A recent study on breast cancer patients support the notion of a link between MSCs and breast cancer stem cells (De Giorgi et al., 2011). It showed that the relative number of disseminated CD44+/CD24low/-/ALDH-1+ breast cancer stem cells correlated with the relative number of MSCs in the bone marrow. Recently, it has been found that epithelial cells after having undergone full EMT display stem cell-like characteristics, including expression of CD44 and ALDH-1 (Mani et al., 2008; May et al., 2011). EMT is linked to E-cadherin loss and the expression of mesenchymal markers. As mentioned above, MSCs have been shown to reduce E-cadherin expression or to induce E-cadherin shedding in luminal A-type, epitheloid breast cancer cells, such as MCF-7 and T47D. In T47D, this downregulation of Ecadherin was accompanied with increased expression of mesenchymal markers suggesting that MSCs may induce at least partial EMT of breast cancer cells. Hence, MSCs may not only be able to trigger the CSC pool to expand, but also to force new CSCs to be generated from the pool of non-CSC breast cancer cells by EMT. These MSC-induced new CSCs may have other features than the CSCs of the existing pool and may further contribute to tumor heterogeneity and progression (Visvader & Lindeman, 2008). Another interesting observation is that the gene expression profile of mesenchymal (basal-type) breast cancer cells show similarities to the expression profile of MSCs (Marchini et al., 2010) suggesting that this type of breast cancer cell and MSCs have also common functions. In support of this notion, a recent study showed that mesenchymal breast cancer cells generated by

tumor growth. This ratio could be different among different MSC isolates.

**4.3 MSCs, EMT and breast cancer stem cells** 

by an autocrine mechanism (Pinzone et al., 2009) may be responsible for this effect (Qiao et al., 2008a). As an inhibitor of the Wnt/-catenin pathway, DKK-1 was shown to downregulate -catenin activity and, concomitantly, to reduce the expression of proliferation-promoting proteins c-Myc and NF-B (nuclear factor kappa-light-chainenhancer of activated B-cells) in MCF-7 cells (Qiao et al., 2008a; Qiao et al., 2008b). Why some studies showed stimulatory while others demonstrated inhibitory effects of MSCs on MCF-7 cell proliferation is not clear yet. Qiao et al. used human fetal dermal tissue as a source to isolate MSCs for their study (Qiao et al., 2008a). In this case, the different MSC sources may have accounted for the contradictory results. Since MSCs are a heterogeneous population (Uccelli et al., 2008), a different environment may drive the selection of a certain subtype of MSCs with features distinct to the bone-marrow MSC population. In particular, types and amounts of chemokines/cytokines these MSC populations secrete might be different. The importance of environmental conditions for the ability of MSCs to interfere with breast cancer functions is nicely demonstrated in a study that compared serumexposed MSCs with serum-deprived MSCs (Sanchez et al., 2011). Serum-deprived MSCs were found to be more effective than serum-exposed MSCs in protecting MCF-7 cells from apoptotic death by secreting pro-survival factors. MSCs also modulate the functions of highly aggressive ER-negative MDA-MB-231 cells (Fig. 3). Two studies demonstrated that MSCs increase the invasive and metastatic behavior of these breast cancer cells (Goldstein et al., 2010; Karnoub et al., 2007). In one study, this effect was found to be mediated by IL-17B (Goldstein et al., 2010). In the other study, the chemokine RANTES/CCL5 was shown to be responsible (Karnoub et al., 2007). Paracrine feedback loops between breast cancer cells and MSCs seem to be important for these effects. It could be shown that MDA-MB-231 cells stimulate the expression of RANTES/CCL5 in MSCs by secreting osteopontin which binds to MSC surface integrins which then leads to the activation of AP-1, a transcription factor able to induce the transcription of the RANTES/CCL5 gene (Mi et al., 2011). MCP-1/CCL2 is another chemokine whose secretion can be stimulated when MSCs are co-cultured with MDA-MB-231 cells (Molloy et al., 2009). MCP-1/CCL2 belongs to those chemokines that enhance the motility of MDA-MB-231 cells. Other migration-promoting chemokines are CXCR3 ligands CXCL9, CXCL10 and CXCL11 (Shin et al., 2010). These CXCL chemokines also increase the activity of Rho GTPases and the expression of MMP-9 (matrix metalloprotease-9). These chemokines may be of particular importance when MSCs are exposed to TNF which was found to induce CXCL gene transcription through a mechanism involving NF-B (Fig. 3). One group also demonstrated inhibitory effects of MSCs on MDA-MB-231 cells (Sun et al., 2009; Sun et al., 2010). According to their data, MSCs suppress the proliferative, migratory, tumor-initiating and metastatic activities of MDA-MB-231 cells and induce apoptosis of these cells by interfering with the AKT/mTOR (mammalian target of rapamycin) pathway. Different to the other investigations, these studies were performed with MSCs isolated from human umbilical cord blood or adipose tissue. Hence, as discussed above, source-dependent features of MSC isolates may be responsible for these contradictory results. Also murine metastatic 4T1 breast cancer cells were shown to be affected by MSCs (Ling et al., 2010). Using a syngeneic, immunocompetent murine model, Ling and colleagues demonstrated that murine MSCs enter 4T1 tumors to deliver IFN- to the tumor. This factor then inhibited cancer growth by inducing the inactivation of STAT3, Src and AKT and by triggering the downregulation of c-Myc and MMP-2 (matrix metalloprotease-2). Interestingly, human MSCs engineered to secrete IFN- have also a suppressing effect on growth of MDA-MB-231 cells in mouse xenografts (Studeny et al., 2004). It may well be that the ratio of tumor-suppressing vs. tumor-promoting factors as secreted by MSCs determine whether MSCs promote or inhibit tumor growth. This ratio could be different among different MSC isolates.

### **4.3 MSCs, EMT and breast cancer stem cells**

256 Breast Cancer – Focusing Tumor Microenvironment, Stem Cells and Metastasis

by an autocrine mechanism (Pinzone et al., 2009) may be responsible for this effect (Qiao et al., 2008a). As an inhibitor of the Wnt/-catenin pathway, DKK-1 was shown to downregulate -catenin activity and, concomitantly, to reduce the expression of proliferation-promoting proteins c-Myc and NF-B (nuclear factor kappa-light-chainenhancer of activated B-cells) in MCF-7 cells (Qiao et al., 2008a; Qiao et al., 2008b). Why some studies showed stimulatory while others demonstrated inhibitory effects of MSCs on MCF-7 cell proliferation is not clear yet. Qiao et al. used human fetal dermal tissue as a source to isolate MSCs for their study (Qiao et al., 2008a). In this case, the different MSC sources may have accounted for the contradictory results. Since MSCs are a heterogeneous population (Uccelli et al., 2008), a different environment may drive the selection of a certain subtype of MSCs with features distinct to the bone-marrow MSC population. In particular, types and amounts of chemokines/cytokines these MSC populations secrete might be different. The importance of environmental conditions for the ability of MSCs to interfere with breast cancer functions is nicely demonstrated in a study that compared serumexposed MSCs with serum-deprived MSCs (Sanchez et al., 2011). Serum-deprived MSCs were found to be more effective than serum-exposed MSCs in protecting MCF-7 cells from apoptotic death by secreting pro-survival factors. MSCs also modulate the functions of highly aggressive ER-negative MDA-MB-231 cells (Fig. 3). Two studies demonstrated that MSCs increase the invasive and metastatic behavior of these breast cancer cells (Goldstein et al., 2010; Karnoub et al., 2007). In one study, this effect was found to be mediated by IL-17B (Goldstein et al., 2010). In the other study, the chemokine RANTES/CCL5 was shown to be responsible (Karnoub et al., 2007). Paracrine feedback loops between breast cancer cells and MSCs seem to be important for these effects. It could be shown that MDA-MB-231 cells stimulate the expression of RANTES/CCL5 in MSCs by secreting osteopontin which binds to MSC surface integrins which then leads to the activation of AP-1, a transcription factor able to induce the transcription of the RANTES/CCL5 gene (Mi et al., 2011). MCP-1/CCL2 is another chemokine whose secretion can be stimulated when MSCs are co-cultured with MDA-MB-231 cells (Molloy et al., 2009). MCP-1/CCL2 belongs to those chemokines that enhance the motility of MDA-MB-231 cells. Other migration-promoting chemokines are CXCR3 ligands CXCL9, CXCL10 and CXCL11 (Shin et al., 2010). These CXCL chemokines also increase the activity of Rho GTPases and the expression of MMP-9 (matrix metalloprotease-9). These chemokines may be of particular importance when MSCs are exposed to TNF which was found to induce CXCL gene transcription through a mechanism involving NF-B (Fig. 3). One group also demonstrated inhibitory effects of MSCs on MDA-MB-231 cells (Sun et al., 2009; Sun et al., 2010). According to their data, MSCs suppress the proliferative, migratory, tumor-initiating and metastatic activities of MDA-MB-231 cells and induce apoptosis of these cells by interfering with the AKT/mTOR (mammalian target of rapamycin) pathway. Different to the other investigations, these studies were performed with MSCs isolated from human umbilical cord blood or adipose tissue. Hence, as discussed above, source-dependent features of MSC isolates may be responsible for these contradictory results. Also murine metastatic 4T1 breast cancer cells were shown to be affected by MSCs (Ling et al., 2010). Using a syngeneic, immunocompetent murine model, Ling and colleagues demonstrated that murine MSCs enter 4T1 tumors to deliver IFN- to the tumor. This factor then inhibited cancer growth by inducing the inactivation of STAT3, Src and AKT and by triggering the downregulation of c-Myc and MMP-2 (matrix metalloprotease-2). Interestingly, human MSCs engineered to

There is growing evidence that, in accordance with the hierarchical model of cancer development (Visvader & Lindeman, 2008), breast cancer is driven by cancer stem cells (CSCs) (Liu & Wicha, 2010). Breast CSCs are characterized by high expression of surface marker CD44 and low expression of CD24 (Fillmore & Kuperwasser, 2007). Another useful breast CSC marker is ALDH-1 (Ginestier et al., 2007). A recent study showed that MSCs increase the pool of CSCs in breast cancer lines, including MCF-7, SUM149 and SUM159 cells (Liu et al., 2011). Interestingly, bone marrow-derived MSCs themselves show also a hierarchical organization with only a minority of cells expressing the stem cell marker ALDH-1. And only those ALDH-1 positive MSCs were able to interfere with the CSC pool. Wicha and his co-workers showed that the MSC/breast cancer interaction generated a cytokine network which is initiated by IL-6 as secreted by breast cancer cells (Liu et al., 2011). IL-6 induces the production of CXCL7 in MSCs which, in turn, triggers the expression of a number of other cytokines and chemokines, namely IL-6, IL-8, CXCL6 and CXCL5, in both MSCs and breast cancer cells. This mixture of secreted factors then stimulates the expansion of the CSC pool. In line with the observation that MSCs induce the CSC pool to expand is the finding that MSCs stimulated mammosphere formation of normal mammary epithelial cells (Klopp et al., 2010). Evidence has been accumulated suggesting that the generation of mammospheres depends on the presence of mammary stem cells (Dontu et al., 2003). Hence, the number of mammospheres formed is supposed to be a measure of the number of mammary stem cells present (Charafe-Jauffret et al., 2008). A recent study on breast cancer patients support the notion of a link between MSCs and breast cancer stem cells (De Giorgi et al., 2011). It showed that the relative number of disseminated CD44+/CD24low/-/ALDH-1+ breast cancer stem cells correlated with the relative number of MSCs in the bone marrow. Recently, it has been found that epithelial cells after having undergone full EMT display stem cell-like characteristics, including expression of CD44 and ALDH-1 (Mani et al., 2008; May et al., 2011). EMT is linked to E-cadherin loss and the expression of mesenchymal markers. As mentioned above, MSCs have been shown to reduce E-cadherin expression or to induce E-cadherin shedding in luminal A-type, epitheloid breast cancer cells, such as MCF-7 and T47D. In T47D, this downregulation of Ecadherin was accompanied with increased expression of mesenchymal markers suggesting that MSCs may induce at least partial EMT of breast cancer cells. Hence, MSCs may not only be able to trigger the CSC pool to expand, but also to force new CSCs to be generated from the pool of non-CSC breast cancer cells by EMT. These MSC-induced new CSCs may have other features than the CSCs of the existing pool and may further contribute to tumor heterogeneity and progression (Visvader & Lindeman, 2008). Another interesting observation is that the gene expression profile of mesenchymal (basal-type) breast cancer cells show similarities to the expression profile of MSCs (Marchini et al., 2010) suggesting that this type of breast cancer cell and MSCs have also common functions. In support of this notion, a recent study showed that mesenchymal breast cancer cells generated by

Involvement of Mesenchymal Stem Cells in Breast Cancer Progression 259

et al., 1988). Defined as myofibroblasts, CAFs share features with both smooth muscle cells and fibroblasts (Mueller & Fusenig, 2004). CAFs are found in many cancers, including breast cancer (Chauhan et al., 2003), and are linked to tumor invasion and proliferation (De Wever & Mareel, 2003; Tlsty & Coussens, 2006). They are responsible for a phenomenon called desmoplasia and promote angiogenesis and inflammation (Orimo et al., 2005; Tlsty & Coussens, 2006). Interestingly, CAFs secrete factors that are also produced by MSCs. In particular, CAFs and MSCs both secrete IL-6 and SDF-1/CXCL12, cytokines able to induce the proliferation of luminal A-type MCF-7 breast cancer cells (Bhowmick et al., 2004; Fierro et al., 2004; Mishra et al., 2008; Orimo et al., 2005; Sasser et al., 2007b). In addition, both cell types were found to interfere similarly with the response of MCF-7 and MDA-MB-231 breast cancer cells to inhibitors of mTOR and B-RAF (Dittmer et al., 2011). Differentiation of MSCs to CAFs requires the exposure of MSCs to conditioned medium from tumor cells over several weeks (Mishra et al., 2008; Spaeth et al., 2009). Conditioned medium from MDA-MB-231 breast cancer cells and from Skov-3 ovarian cancer cells were similar effective in inducing a MSC/CAF conversion which was accompanied by increased expression of CAF markers, such as tenascin-C, -smooth muscle actin and IL-6. What are the consequences of this finding? As soon as MSCs enter a tumor, they will be bombarded with a cocktail of cytokines as produced by the tumor cells and may receive additional signals by direct cellcell contacts. This may then force MSCs to lose their stemness and to undergo differentiation towards CAFs. By converting to CAFs, MSCs may not further be able to act also suppressive on tumor cells and may only keep their potency to promote tumor progression. Hence, the differentiation of MSCs to CAFs may be as much of a benefit for a progressing tumor as is the differentiation of MSCs to particular cells for an injured tissue to be repaired (Dittmer,

MSCs display an astounding plasticity and have shown to differentiate to cells as different as neurons and epithelial cells. The main function of MSCs is likely to promote tissue regeneration after injuries and, since tumors are probably wounds that never heal, also to support repair of tumoral lesions. However, tumors may misguide MSCs and "misuse" them for their "own benefit". Primary tumors may particularly profit from MSCs when they differentiate to tumor-promoting CAFs. MSCs may further facilitate breast cancer to metastasize by helping breast cancer cells to enter the bone marrow as well as by increasing the pool of metastasizing breast cancer stem cells. Most of the interactions between MSCs and tumor cells are mediated by cytokines as secreted by both cell types. Paracrine feedback mechanisms may further increase cytokine concentrations at places where these cells communicate with each other and may attract other cell types, such as macrophages, that are known to support tumor progression. To interfere with the interaction between MSCs and breast cancer cells treatments may be considered involving the inhibition of the activities of key cytokines, such as IL-6 (Liu & Wicha, 2010), which are important for both attraction of MSCs to breast cancer and expansion of the breast cancer stem cell pool by MSCs. On the other hand, there is also evidence that MSCs may have suppressive effects on breast cancer. Different sources from which MSCs were isolated may partially account for these contradictory results. Further studies are necessary to clarify this controversy, before conclusions can be drawn in terms of treatment of breast cancer patients. Certainly, when

2010).

**5. Conclusions** 

transformation of human mammary epithelial cells by SV40 T-antigen and forced expression of EMT-inducing proteins had the potential to undergo adipogenic and chondrogenic differentiation. Also, these mesenchymal breast cancer cells were attracted to wounds and tumors, a feature typical for MSCs. The latter observation may shed a new light on a phenomenon called tumor-self seeding (Leung & Brugge, 2009). Tumor-self seeding describes the ability of metastasized cells to circulate back to the primary tumor. Chemoattraction to the primary tumor was shown to be driven by IL-6 and IL-8. As mentioned above, IL-6 is highly active on MSCs and triggers the production of a number of chemokines (Leung & Brugge, 2009). Based on these data, it is tempting to assume that MSCs are also generated from breast cancer cells (Fig. 1) and that these MSCs containing the mutations (and epigenetic changes) of the breast cancer cells they derived from play a role in breast cancer metastasis and tumor-self seeding. Nestin+ - MSCs have been reported to share with haematopoietic stem cells the same niche in the bone marrow (Mendez-Ferrer et al., 2010). This niche might also be available for breast cancer-derived MSCs and allow these cells to survive in this tissue. An exciting hypothesis would be to assume that, at least in some cases of breast cancer, dormancy (Pantel et al., 2009; Willis et al., 2010) is caused by breast cancer-derived MSCs that are caught in these niches. The niches could fulfill two functions in order to maintain dormancy, preventing the cells from proliferating while, at the same time, protecting them from death-inducing signals. The bone marrow seems to be an attractive tissue for circulating tumor cells to home and form micrometastasis which is an early event in breast cancer development (Pantel et al., 2009). The number of such disseminated breast cancer cells in bone has been linked to prognosis of breast cancer patients. MSCs may also play a role in this early entry of breast cancer cells into bone marrow (Corcoran et al., 2008). MSCs were shown to facilitate the migration of MCF-7 and T47D breast cancer cells across bone marrow endothelial cells *in vitro* and to be in close contact with bone-metastasized breast cancer cells *in vivo*. Evidence was presented that these MSC/breast cancer cell interactions may require the chemokine receptor CXCR4 as well as its ligand SDF-1/CXCL12. Hence, it might be possible that breast cancer-derived MSCs not only would be able to home to the bone marrow and induce tumor dormancy, but also to help other breast cancer cells to enter this tissue and form micrometastasis. The breast cancer cell may not be the only non-stem cell which may be able to convert to an MSC. Vascular endothelial cells have been shown to become MSC-like cells as well as displaying typical MSC features, such as the ability to differentiate to osteoblasts, chondrocytes and adipocytes, upon treatment with ALK2 (activin-like kinase-2), TGF2 or BMP4 (bone morphognetic protein-4) (Medici et al., 2010). This suggests that certain non-stem cells under certain conditions can be an additonal source for generating MSCs. These cells may have different features compared to those MSCs that derived from the bone marrow.

#### **4.4 MSCs and carcinoma-associated fibroblasts**

Besides differentiating to osteoblast, chondrocytes and adipocytes, MSCs are able to convert to neural cells or to undergo transdifferentiation to different kinds of epithelial cells (Uccelli et al., 2008; Wislet-Gendebien et al., 2005). In tumors, MSCs can also differentiate to the carcinoma-associated fibroblasts (CAFs) (Mishra et al., 2008; Spaeth et al., 2009). These cells are different to normal fibroblasts/myofibroblasts in that they are able to stimulate tumor progression (Olumi et al., 1999) and show higher proliferative and migratory activity (Schor et al., 1988). Defined as myofibroblasts, CAFs share features with both smooth muscle cells and fibroblasts (Mueller & Fusenig, 2004). CAFs are found in many cancers, including breast cancer (Chauhan et al., 2003), and are linked to tumor invasion and proliferation (De Wever & Mareel, 2003; Tlsty & Coussens, 2006). They are responsible for a phenomenon called desmoplasia and promote angiogenesis and inflammation (Orimo et al., 2005; Tlsty & Coussens, 2006). Interestingly, CAFs secrete factors that are also produced by MSCs. In particular, CAFs and MSCs both secrete IL-6 and SDF-1/CXCL12, cytokines able to induce the proliferation of luminal A-type MCF-7 breast cancer cells (Bhowmick et al., 2004; Fierro et al., 2004; Mishra et al., 2008; Orimo et al., 2005; Sasser et al., 2007b). In addition, both cell types were found to interfere similarly with the response of MCF-7 and MDA-MB-231 breast cancer cells to inhibitors of mTOR and B-RAF (Dittmer et al., 2011). Differentiation of MSCs to CAFs requires the exposure of MSCs to conditioned medium from tumor cells over several weeks (Mishra et al., 2008; Spaeth et al., 2009). Conditioned medium from MDA-MB-231 breast cancer cells and from Skov-3 ovarian cancer cells were similar effective in inducing a MSC/CAF conversion which was accompanied by increased expression of CAF markers, such as tenascin-C, -smooth muscle actin and IL-6. What are the consequences of this finding? As soon as MSCs enter a tumor, they will be bombarded with a cocktail of cytokines as produced by the tumor cells and may receive additional signals by direct cellcell contacts. This may then force MSCs to lose their stemness and to undergo differentiation towards CAFs. By converting to CAFs, MSCs may not further be able to act also suppressive on tumor cells and may only keep their potency to promote tumor progression. Hence, the differentiation of MSCs to CAFs may be as much of a benefit for a progressing tumor as is the differentiation of MSCs to particular cells for an injured tissue to be repaired (Dittmer, 2010).
