The discovery that MSC demonstrate a unique tropism for the tumor microenvironment has led to a great deal of interest in understanding the function of MSC within tumors 
. MSC have been shown to increase the growth of certain cancers when injected together with MSC and can increase the incidence of breast xenograft metastasis 
. However, a great deal remains unknown about the interaction of MSC and breast cancers. In this study, we report that MSC increase the capacity of normal mammary epithelial cells and established breast cancer cell lines (MCF-7 and SUM149) but not short-term primary breast cells (MDA-IBC-3) to form primary mammospheres. Both direct exposure to MSC, as well as MSC-conditioned media, promoted mammosphere formation from HMEC, MCF7, and SUM149 demonstrating that MSC secrete a sphere-promoting factor to which that these lines are sensitive. We hypothesize that this may occur through an increase in EMT in breast cancer cells, since MSC provoke a cadherin switch, characteristic of EMT, in normal and malignant (SUM149) breast mammospheres. Decreased full-length E-cadherin expression was observed in vivo
or in vitro
in all lines examined. In cells that express N-cadherin at baseline, this downregulation of E-cadherin was accompanied by an increase in mesenchymal proteins including N-cadherin and fibronectin.
The mesenchymal phenotype has previously been reported to increase primary mammosphere formation. Mani et. al.
demonstrated that expression of EMT regulating transcription factors, snail
, increased mammosphere formation 
. Furthermore, mammospheres from snail
over-expressing cells contained a higher percentage of cells with stem cells markers suggesting that EMT generates cells with stem cells properties 
. Our findings are consistent with previous studies in which MSC have been shown to interact with breast cancer cells in monolayer cell culture to promote epithelial-mesenchymal transition (EMT) 
We examined whether exposure to MSC-CM could be enriching for the growth of cells with stem cell properties. However, while primary sphere formation was reproducibly increased in a dose dependent manner in established cell lines, secondary sphere formation was not increased and neither sphere formation nor CD44+
cells were increased in MDA-IBC-3 tumors cultured with MSC. This suggests that the expansion of mammospheres may represent an amplification of non-self-renewing progenitors involved in the observed tumor growth promotion and increased invasion 
rather than self-renewing tumor stem cells. Rigorous in vivo
serial passaging experiments will be required to determine if MSC influence the cancer stem cell population.
Dittmer et. al.
recently reported that MSC can integrate into formed breast cancer spheroids, altering E-cadherin expression and increasing breast cancer cell invasion. In contrast to our work, these authors observed that MSC containing aggregates were more disorganized, while we have observed that spheroids with MSC appear more spherical and well organized than non-MSC containing spheres 
. This difference may be explained by the fact that Dittmer et. al.
were examining the effects of MSC added to aggregates of spontaneously floating cells in monolayer serum-containing culture as compared to our study in which the effect of MSC was evaluated in spheres grown from single cells plated at clonal density in stem cell promoting serum-free media. The process of MSC integration into established spheres may result in disrupted morphology. This may imply a differential effect on TIC secondary to less reliance on E-cadherin for adhesion in TIC.
Strikingly, while the down-regulation of E-cadherin was observed in vitro
or in vivo
in all cells examined, there are distinct differences across cell lines in expression of other pro-invasion and mesenchymal proteins. E-cadherin was decreased in both the estrogen receptor (ER) positive luminal E-cadherin positive cells (MCF-7) and in the ER negative E-cadherin positive cells (SUM149). E-cadherin is a cell adhesion protein which functions as a tumor supressor gene with forced expression of E-cadherin resulting in reduced tumor cell invasion 
. High levels of E-cadherin are associated with better clinical outcomes in patients with non-IBC 
. However, E-cadherin has been proposed to have a unique role in IBC where E-cadherin expression is present in both the primary and in the tumor emboli found in the angiolymphatics in the breast 
, prompting the hypothesis that E-cadherin expression is dynamic, and potentially only transiently down-regulated during metastasis 
. In the Mary-X IBC mouse model, it has been shown that the aggregates of tumor in emboli are facilitated by a functional E-cadherin/β-catenin axis 
and that knock-down inhibits aggregates, and that further these aggregates metastasize as E-cadherin positive clusters through a passive process rather than hemotogenous spread 
. To that end, dominant negative E-cadherin in SUM149 cells inhibits invasion as expected in non-IBC tumors. Here, however, we demonstrate for the first time that MSC can promote the growth of an IBC cell line, MDA-IBC-3, and that E-cadherin is down-regulated in the larger MSC co-injected MDA-IBC-3 xenograft. Given that IBC clearly can develop metastatic disease via hematogenous spread as well as potentially passive spread via the angiolymphatic channels we propose that IBC cells are capable of both E-cadherin positive non-hemotogenous spread as well as more well-described E-cadherin negative promoting invasive behavior. We can not determine from this model if E-cadherin is re-expressed after metastases are established.
Our findings demonstrate that MSC provoke breast cancer cells to form mammospheres and assume a more mesenchymal phenotype and that MSC integrate into breast cancer mammospheres and decrease E-cadherin expression in both ER positive luminal E-cadherin positive cells and ER negative E-cadherin positive IBC cells. This suggests that MSC may represent a novel therapeutic target either independently or by inhibiting the effects of MSC on cadherin expression in breast cancer cells.