The first two authors contributed equally to this work.
Cancer Cell Biology
Mesenchymal stem cell secretion of chemokines during differentiation into osteoblasts, and their potential role in mediating interactions with breast cancer cells
Article first published online: 11 NOV 2008
Copyright © 2008 Wiley-Liss, Inc.
International Journal of Cancer
Volume 124, Issue 2, pages 326–332, 15 January 2009
How to Cite
Molloy, A. P., Martin, F. T., Dwyer, R. M., Griffin, T. P., Murphy, M., Barry, F. P., O'Brien, T. and Kerin, M. J. (2009), Mesenchymal stem cell secretion of chemokines during differentiation into osteoblasts, and their potential role in mediating interactions with breast cancer cells. Int. J. Cancer, 124: 326–332. doi: 10.1002/ijc.23939
- Issue published online: 14 NOV 2008
- Article first published online: 11 NOV 2008
- Manuscript Accepted: 12 AUG 2008
- Manuscript Received: 27 MAR 2008
- National Breast Cancer Research Institute (NBCRI)
- Health Research Board
- Science Foundation Ireland
- bone metastases;
- breast cancer;
- mesenchymal stem cells;
- monocyte chemotactic protein-1
Over 70% of patients with advanced breast cancer will develop bone metastases for which there is no cure. Mesenchymal Stem Cells (MSCs) and their derivative osteoblasts are subpopulations of cells within the bone marrow environment, postulated as potential interacting targets for disseminating cancer cells because of their ability to secrete a range of chemokines. This study aimed to investigate chemokine secretion throughout MSC differentiation into osteoblasts and their effect on the breast cancer cells. Primary MSCs and osteoblast progenitors were cultured in appropriate conditions to induce differentiation into mature osteoblasts. Chemokines secreted throughout differentiation were detected using ChemiArray and ELISA. Migration of breast cancer cells in response to the bone-derived cells was quantified using Transwell inserts. Breast cancer cells were cocultured with MSCs, retrieved using magnetic beads, and changes in CCL2 expression were analyzed. MSCs secreted a range of factors including IL-6, TIMP-1 and CCL2, the range and level of which changed throughout differentiation. CCL2 secretion by MSCs increased significantly above control cells as they differentiated into mature osteoblasts (p < 0.05). The bone-derived cells stimulated migration of breast cancer cells, and this was inhibited (21–50%) in the presence of a CCL2 antibody. CCL2 gene expression in breast cancer cells was upregulated following direct coculture with MSCs. The varying levels of chemokines secreted throughout MSC differentiation may play an important role in supporting tumor cell homing and progression. These results further highlight the distinct effect MSCs have on breast cancer cells and their potential importance in supporting development of metastases. © 2008 Wiley-Liss, Inc.
Over 70% of patients with advanced breast cancer will develop bone metastases for which there is no cure.1 This aspect of disease progression results in a myriad of complications including pain, spinal cord compression and fractures with an average 5-year survival of 20% of the patients.2 The preferential metastasis of breast cancer cells to bone is poorly understood, with many other cancers unable to cause skeletal metastases and their subsequent complications.3, 4 The microenvironment associated with bone is ideal for tumor progression.5 Bone is a highly vascular mineral that produces adhesion molecules and is a source of angiogenic and bone resorbing molecules, all of which are conducive to the spread anddevelopment of tumors.6, 7 It also contains immobilized growth factors, which, when released, further enhance tumor cell proliferation.8, 9
Normal bone displays a dynamic environment involving tight regulation of reciprocating osteoblast and osteoclast activity, each complimenting each other in a continuous cycle of cell turnover.1, 10 Osteoblasts are associated with bone mineralization and calcium deposition. The incidence of osteoblastic or mixed metastases from breast cancer varies in the literature from 20–50%.11, 12 Clinically, osteoblastic metastases result in the increased production of structurally unstable poorly woven bone which, when stressed, results in pathological fractures.10, 13 It appears that osteoblasts also play a significant role in the development of osteolytic bone metastases with reports showing the regulation of osteoclast activity by chemokines secreted by osteoblasts.14–16
Mesenchymal stem cells (MSCs) are a subset of nonhaematopoietic cells within the bone marrow stroma. They possess the capacity to differentiate into cells of connective tissue lineages and are capable of self-renewal. When exposed to the appropriate environmental conditions, MSCs differentiate into mature osteoblasts.17–19 This process of osteogenesis is believed to occur in 3 distinct phases, (i) differentiation of MSCs and the subsequent proliferation of osteoprogenitor cells (ii) extracellular matrix deposition and finally (iii) mineralization.20, 21 There has been increased interest in MSCs in the cardiac, orthopedic and rheumatological fields as a result of their potential role in regenerative medicine.22, 23 However, recent studies have implicated MSCs as a potential interacting target for disseminating epithelial breast cancer cells partly as a result of their ability to secrete a range of chemokines.24–29
Coculture of MCF-7 breast cancer cells with MSCs has been shown to induce changes in morphology, proliferative capacity and aggregation pattern of the breast cancer cells.24 Interaction of the same cell line with MSCs has also been shown to downregulate E-cadherin and epithelial specific antigen (ESA) expression.25 More recently, it has been reported that bone marrow stromal cells enhance breast cancer cell growth in cell-line dependent manner.28 The role of MSCs in metastases has been further investigated in an extensive study by Karnoub et al.26 This study highlighted the intricate role that MSCs play in metastatic progression of circulating tumor cells. MSCs were shown to increase the rate of tumor spread and also to confer more invasive properties upon the circulating tumor cells. This was shown to be mediated in part through secretion of chemokines, including RANTES (CCL5).
Monocyte chemotactic protein-1 (MCP-1/CCL2) is a chemokine involved in chemotaxis of various immune cells and monocyte activation.27, 30 Several cell types produce CCL2 including mononuclear cells, T cells, fibroblasts and tumor cells.31 CCL2 has been implicated as an active participant in the tumor microenvironment, influencing angiogenesis and metastasis, and may potentially play a role in development of a premetastatic niche for circulating cancer cells.3, 32–34
There remains an urgent need for an increased understanding and insight into the regulatory events of both the metastatic and osteogenic cascades.26 The aim of this study was to identify factors secreted by MSCs and osteoblast progenitors during differentiation into mature osteoblasts and investigate their effects on breast cancer cells. Further understanding of the dynamic bone microenvironment may provide insights into factors controlling preferential metastases of breast cancer cells to this site and provide novel targets for therapeutic intervention.
Material and methods
Breast cancer cell lines MDA-MB-231, BT-474 and T47D were cultured in Leibowitz-15 (L-15) and RPMI 1640 medium (BT-474 and T47D), respectively. Both media types were supplemented with 10% fetal bovine serum (FBS), 100 IU/ml Penicillin/ 100 μg/ml Streptomycin and 1% L-glutamine. Human osteoblast progenitor (NHOst) cells (Cambrex) were maintained in osteoblast growth medium (OGM) containing 10% FBS, gentamycin and ascorbic acid (Bullet Kit, Cambrex, U.K). All cells were cultured in a humidified atmosphere at 37°C and 5% CO2.
Human MSCs were obtained through the Regenerative Medicine Institute (REMEDI) at NUI Galway. Briefly, following ethical approval and written informed consent, bone marrow was aspirated from the iliac crest of a healthy donor according to an approved clinical protocol. MSCs were then isolated by direct plating followed by culture for 12–14 days to deplete the nonadherent haematopoietic cell fraction. The cells were routinely cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with preselected FBS (10%) and penicillin G (2 units/ml)/streptomycin sulfate (100 mg/ml). The ability of cells to differentiate into osteoblasts, adipocytes and chondrocytes was confirmed before use.23, 35, 36
Direct coculture of breast cancer cells on a monolayer of MSCs was also performed. MSCs (1.5 × 106) were seeded into a T75 cm2 flask and allowed to adhere overnight. Media was then removed and a suspension of MDA-MB-231 or T47D cells (1 × 106) was added directly to flasks containing MSCs. Each cell population was also cultured alone. All conditions were repeated in triplicate, and following a 72-hr incubation, collection of conditioned medium (CM) and cell separation for RNA extraction were performed as described below.
To stimulate differentiation into mature osteoblasts, NHOst cells were cultured in OGM supplemented with 10% FBS, 10 mM β-glycerophosphate and 100 nM hydrocortisone (Differentiation Kit, Cambrex, UK). MSCs were cultured in DMEM supplemented with 10% FBS, 100 nM dexamethasone, 50 μM ascorbic acid,10 mM β-glycerophosphate, 100 U/ml penicillin and 100 μg/ml streptomycin to induce osteogenesis.
Osteoid mineralization was detected using Von Kossa staining to confirm the presence of mature osteoblasts following 21 days culture in differentiating conditions. Cells were fixed in 4% paraformaldehyde for 15 min, washed 3 times with Phosphate Buffered Saline (PBS) and then stained in 3% silver nitrate for 15 min under UV-light. Cells were then washed in PBS and incubated in 5% sodium thiosulphate for a further 5 min. The cells were then washed and counter-stained with haematoxylin for 3 min, rinsed in distilled water and examined for the presence of calcium deposits which appear brown/black in color.
Collection of conditioned medium
Cell conditioned medium (CM) was collected from both control and differentiating NHOst cells and MSCs, at 72-hr intervals up to a 21-day time point. The medium was harvested, sterile filtered and stored at −20°C until required for chemokine analysis.
ChemiArray™ angiogenesis antibody arrays (Chemicon International) were used to detect angiogenic agents secreted by MSCs at 3 timepoints during differentiation into mature Osteoblasts. This array permitted simultaneous detection of 20 secreted factors in each sample.37 Cell CM was collected as described and applied to the membranes according to manufacturers' instructions. Representative medium from each cell line, not exposed to the cells themselves, was also analyzed as a negative control for each experiment. Chemiluminescent images were acquired using FlourChem™ imaging system (Alpha Innotech) and analyzed with Alpha Ease software. Regions of interest (ROI) of a fixed area on the ChemiArray membranes were selected and signal intensity measured by densitometry. Following the subtraction of a background reading, relative density units (RDUs) were used to measure the relative level of each chemokine. Comparison of the relative levels of chemokines, detected on separate membranes, was done following normalization against the positive controls on each membrane to take differences in exposure times into account. These are biotin-conjugated IgG areas on the dot blot, which create an area of high signal intensity, used for both dot blot orientation and to normalize the results on different membranes for comparison. ChemiArray analysis was performed on conditioned medium harvested from differentiating MSCs at the 3 different timepoints (Day 3, 14 and 21) to identify potential targets of interest. The differentiation protocol and medium harvest at a greater range of timepoints (Day 3, 7, 10, 14, 18 and 21) was performed 3 times, and the level of specific targets of interest (CCL2 and VEGF) was determined on triplicate samples (not pooled) using Quantikine® enzyme linked immunosorbent assays (ELISA) (R&D systems), with absorbance's detected on a platereader (Multiscan RC). Each individual sample was analyzed in duplicate.
Transwell™ inserts were used to assess breast cancer cell migration in response to factors secreted by MSCs or NHOst cells. Conditioned medium for these experiments was harvested from MSCs and NHOsts seeded at a fixed density (2.5 × 105 cells/well in 6-well plate) in a fixed volume of media (2 ml). Following a 72-hr incubation, media was harvested, sterile filtered and transferred to a well for use as a chemoattractant for migration experiments. Briefly, an insert containing a porous membrane (8 μm pore size) was suspended above each well and a suspension of breast cancer cells were placed in the insert (75,000 cells per insert). Cells could then adhere to or migrate through the porous membrane in response to the chemoattractants in the well below. As a control to confirm directional migration, cells were also resuspended in the same chemoattractant used in the lower compartment. Cell migration was allowed to proceed for 18 hr after which the membranes were harvested and migrated cells were counted as previously described.38
Because CCR2 is not the sole receptor for CCL2, the potential role of CCL2 in breast cancer cell migration was investigated using a monoclonal antibody to the chemokine itself. The antibody was added at a concentration of 150 pg/ml to the cells 1 hr prior to commencement of migration experiments.
Following direct coculture of MSCs with breast cancer cell lines, cells were trypsinized and dispersed into a single cell suspension. The human EpCAM selection kit (Stem Cell Technologies) was then used to retrieve epithelial cancer cells from the mixed cell population according to manufacturer's instructions. This involved use of a teterameric antibody complex which specifically binds to the epithelial cell adhesion molecule (EpCAM) found on viable epithelial cells and also has an anti-dextran portion. Magnetic beads coated with dextran were then added and bound to the labeled epithelial cells. Application of a magnetic force permitted retrieval of the magnetic bead-bound labeled cells. Retrieved cells were then centrifuged at 1,000 rpm and the pellet was stored at −80°C until required for RNA extraction.
Analysis of gene expression
Total RNA was isolated from individual cell populations using the RNeasy® mini kit according to manufacturer's instructions. RNA concentration and integrity were determined using the NanoDrop® ND-1000 Spectrophotometer (NanoDrop Technologies, Wilmington, DE), and 2100 BioAnalyser (Agilent Technologies, Waldbronn, Germany), respectively. RNA (1 μg) was reverse transcribed using SuperScript III reverse transcriptase enzyme. cDNA samples were amplified and analyzed by real-time quantitative PCR (RQ-PCR) using the ABI Prism 7000. Taqman® universal master mix was used as well as Taqman® gene expression assays designed for the target gene CCL2, its principle receptor CCR2 (targeting predominant isoform B, which localizes to cytoplasmic membrane) and endogenous RNA reference genes PPIA and MRPL19. The comparative CT method for relative quantification was used, permitting determination of the quantity of the CCL2 and CCR2 gene in each sample population normalized to the endogenous controls (PPIA and MRPL19), and levels in epithelial cells cultured alone expressed relative to cells retrieved following coculture with MSCs. Results were expressed in a linear form using the formula 2−ΔΔCT.39
Confirmation of cell differentiation
Von Kossa staining was carried out to confirm differentiation of MSCs into mature osteoblasts through detection of calcium deposition (Fig. 1). Following 21 days culture of MSCs in differentiating conditions, calcium deposits, which appear in a typical black/brown color, were detected throughout the culture (Fig. 1a). No calcium deposits were detected in MSCs cultured in parallel in normal growth medium (Fig. 1b).
ChemiArray™ analysis of cell conditioned medium was carried out at Days 3, 14 and 21 of MSC differentiation into mature osteoblasts (Fig. 2). This revealed secretion of a number of angiogenic factors including Angiogenin, GRO, RANTES, IL-6, IL-8, TIMP-1, TIMP-2 and CCL2, the range and level of which changed throughout the differentiation process (Fig. 2a). Neither Angiogenin nor GRO were detected at Day 3 but were present at Days 14 and 21, while the other factors detected were present at variable intensities at all 3 timepoints.
ROI of a fixed area for each detected target were selected and the relative levels detected at each time point were compared by densitometry, following subtraction of background levels and normalization of variance between membranes using the positive control readings. Results were expressed in RDU (Fig. 2b). Although the levels of most factors detected peaked at Day 14 of differentiation, relative quantities of CCL2, TIMP-1 and TIMP-2 increased throughout the differentiation process. The relative level of CCL2 detected at Day 21 was higher than any other chemokine included in the analysis, highlighting this as a potential target for further investigation.
Quantification of CCL2 levels at 72-hr intervals over the 21 Day time course of MSC differentiation into mature osteoblasts was carried out using ELISA (Fig. 3a). There was a steady increase in the levels of CCL2 secreted during the differentiation process, from 303 ± 11 pg/ml (Day 3, Mean ± SEM) to its peak of 11,055 ± 711 pg/ml (Day 21), reflecting the pattern of increased relative intensity seen in ChemiArray™ analysis. The most significant increase was seen following Day 14, when the levels of CCL2 secreted by the differentiating cells trebled (2,275 ± 134 pg/ml to 7,354 ± 297 pg/ml, p < 0.05 compared to control cells). Control cells were cultured in parallel on the 6-well plates in standard MSC medium to retain their undifferentiated state. The level of VEGF secreted by differentiating MSCs was also analyzed at the same time points during differentiation (Fig. 3b). Although there was a gradual increase in VEGF secretion during differentiation, from 139 ± 31 pg/ml (Day 3) to a peak of 945 ± 55 pg/ml (Day 21), levels detected were not significantly different from those of undifferentiated MSCs (Day 21, 1,217 ± 89 pg/ml).
Quantification of CCL2 and VEGF secretion during NHOst osteoblast progenitor cell differentiation into mature osteoblasts was also carried out using ELISA (results not shown). A similar trend was observed to that seen in the differentiation of MSCs. CCL2 increased from 1,479 ± 92 pg/ml (Day 3) to 6,607 ± 232 pg/ml (Day 21), whereas the level secreted by control cells peaked at 3,682 ± 696 pg/ml. Levels of VEGF secreted did not show any significant variation throughout the NHOst differentiation, ranging from 635 ± 126 pg/ml (Day 3) to 1,119 ± 79 pg/ml (Day 21), with similar levels secreted from control cells (1,147 ± 186 pg/ml at Day 21).
Migration of breast cancer cell lines in response to factors secreted by MSCs and NHOst cells was quantified using Transwell inserts as described. The number of cells migrating in response to basal medium was not significantly different from the number that migrated when cells were suspended in chemoattractant, which served as a baseline/control to differentiate between chemotaxis and chemokinesis. There was a significant increase in migration of all 3 breast cancer cell lines in response to factors secreted by the bone derived cells (CM harvested at 72 hr), with examples of MDA-MB-231 migration to MSCs, and BT-474 migration to NHOsts shown here (Fig. 4). Expression of the principle CCL2 receptor, CCR2, was confirmed in T47D and MDA-MB-231 cells as shown previously40 and was also detected, at the highest level, in the BT-474 cells. There was a 5-fold increase in MDA-MB-231 migration toward MSC conditioned medium (105 ± 7 cells) when compared to controls (16 ± 5 cells) (p < 0.05). To determine the role of CCL2 in cell migration, a monoclonal antibody to the chemokine was used. Inclusion of the CCL2 antibody in the cell-conditioned medium caused a significant inhibition in MDA-MB-231 cell migration (50% reduction). There was also a significant increase in migration of BT-474 cells in response to NHOst cell-conditioned medium (99 ± 5 cells) (p < 0.05) when compared to controls, with a 21% inhibition in migration in the presence of a monoclonal antibody to CCL2 (78 ± 9 cells).
Coculture experiments: CCL2 secretion and expression
Secretion of CCL2 by breast cancer cell lines and MSCs was determined when the cells were cultured individually and when they were incubated in coculture conditions. The quantity of CCL2 (pg/ml) secreted by breast cancer cell lines cultured alone was relatively low, ranging from 31 ± 4 pg/ml (MDA-MB-231) to 95 ± 25 pg/ml (T47D) (Table I). CCL2 levels secreted by MSCs were quantified at 2,286 ± 347 pg/ml. Following 72-hr coculture of breast cancer cell lines on a monolayer of MSCs, an additive effect was observed with CCL2 levels significantly higher than that seen in the individual populations (T47D + MSC: 6,568 ± 745 pg/ml, MDA-MB-231 + MSC: 11,828 ± 1,999 pg/ml, p < 0.05 Students t-test).
|Cell Population||CCL2 (pg/mL)|
|T47D||95 ± 24|
|MSC||2288 ± 347|
|MSC + T47D||6568 ± 745|
|MDA-MB-231||31 ± 4|
|MSC + MDA-MB-231||11828 ± 1999|
Following coculture of MSCs with breast cancer cell lines, the epithelial tumor cells were retrieved using EpCAM beads, and total RNA was extracted from individual epithelial cell populations. The level of expression of CCL2 was determined in the cells grown individually and those retrieved with EpCAM beads following 72 hr in coculture (Fig. 5). Levels of CCL2 expression were determined relative to endogenous control genes PPIA and MRPL19, and log values are expressed relative to levels detected in the breast cancer cells cultured alone in parallel (log 2−ΔΔCT, Fig. 5). Gene expression levels reflected the protein secretion trend with CCL2 significantly upregulated in both breast cancer populations following coculture with MSCs, compared to those cultured alone (T47D 1.7 log fold increase p < 0.05, MDA-MB-231 4.1 log fold increase p < 0.005, Students paired t-test). CCR2 expression levels remained unchanged in MDA-MB-231 cells following coculture with an elevation seen in the T47D cells (1 log fold increase, results not shown).
Although thought to involve a complex cascade of cell–cell interactions, the mechanisms underlying breast cancer metastasis to bone remain poorly understood.41, 42 The “seed and soil” hypothesis, as postulated by Paget,5 is still at the centre of current theories for the development of breast cancer metastases with both the disseminating tumor cells and site for metastases believed to contribute to disease spread. Formation of micrometastasis in bone is thought to be dependent upon the rich blood flow in the red marrow, tumor cells producing adhesive molecules and the production of angiogenic and bone-resorbing factors.1
Healthy bone consists of a dynamic microenvironment undergoing constant remodelling. Increased bone turnover is believed to attract cancer cells and promote their colonization in bone43 with suppression of bone turnover by bisphosphonates shown to inhibit development of metastases.9 Cancer patients with Pagets disease of bone exhibit increased frequency of bone metastases compared to uninvolved sites, and tooth extraction in breast cancer patients increases incidence of metastases in sockets in which extensive bone remodelling is taking place.43
The role of differentiation in creating or influencing the premetastatic niche has yet to be fully investigated. Although a previous report showed variations in breast cancer cell migration in response to bone marrow stromal cells based on phase of differentiation, the factors involved were not investigated.44 The results suggested secretion of potent factors able to direct tumor cell migration toward bone undergoing remodeling. With chemokines already strongly implicated in current concepts of metastatic development, differentiation may be a potential influencing factor in the ability of bone derived cells to attract circulating tumor cells. Various cytokines and growth factors are secreted from bone derived cells and are believed to play an integral role in the metastatic cascade. VEGF is known to have a role in tumor angiogenesis, whereas IL-6 has been described as having a more important role in tumor cell proliferation.45, 46 Osteoprotegerin produced by bone marrow stromal cells has also been found to be sufficient to protect cancer cells from undergoing TRAIL induced apoptosis,47 and CCL5 (RANTES) has recently been implicated as playing a pivotal role in the interaction between breast cancer cells and MSCs.26
In this study, we report that both the range and quantity of angiogenic factors secreted by MSCs varied throughout differentiation into mature osteoblasts. Interleukins detected were shown to peak at the midpoint of differentiation, whereas TIMP-1, TIMP-2 and CCL2 secretion were shown to steadily increase to peak at the culmination of the process. While there was no difference in secretion of VEGF in control and differentiating MSCs, CCL2 levels were significantly higher in differentiating cells. Further investigation confirmed breast cancer cell migration in response to factors secreted by both MSCs and lineage committed osteoblast progenitors (NHOsts). Migration was shown to be mediated at least in part by CCL2, with an antibody to the chemokine reducing MDA-MB-231 breast cancer cell migration by up to 50%. These results suggest a potentially important role for CCL2 in recruitment of circulating breast cancer cells to the bone microenvironment.
The effect of MSCs on breast cancer cell expression of CCL2 in direct coculture was also determined. Following 72-hr coculture, an additive effect was observed with a significant increase in CCL2 secretion by the mixed cell populations, compared to those grown individually. This was reflected by a trend toward increased CCL2 gene expression in breast cancer cells retrieved following coculture with MSCs.
Bisphosphonates target bone turnover, and while they are effective at reducing initiation of bone metastases, have been reported to have little effect on established metastases. This suggests that following interactions with bone-derived cells, breast cancer cells may become independent.43 The results presented here show that following coculture with MSCs, breast cancer cells had increased expression of CCL2. Considering the potential role of this chemokine in tumorigenesis,3, 32–34 upregulation may play an important role in both the development and progression of breast cancer metastases.
It is clear that crosstalk between metastatic cancer cells and bone-derived cells is critical to the development and progression of bone metastases.43 Recent studies have demonstrated that interaction with MSCs results in changes in breast cancer cell proliferation, morphology, aggregation and adherence, conferring upon the cells more invasive and metastatic properties.24–26, 28 The results presented here further highlight the distinct effect MSCs have on breast cancer cells and the potential importance of this bone-derived population in supporting development of metastases, mediated at least in part through secretion of CCL2. Further understanding of the mechanisms involved in these cellular interactions will support development of effective and specific therapeutics to interfere with this microenvironmental support.
R.M. Dwyer is supported by a research project grant from the Health Research Board. J.M. Murphy, F.P. Barry and T. O'Brien are supported by a CSET award from Science Foundation Ireland (SFI).
- 9Monitoring metastatic behavior of human tumor cells in mice with species-specific polymerase chain reaction: elevated expression of angiogenesis and bone resorption stimulators by breast cancer in bone metastases. J Bone Miner Res 2001; 16: 1077–91., , , , , .
- 36Reevaluation of in vitro differentiation protocols for bone marrow stromal cells: disruption of actin cytoskeleton induces rapid morphological changes and mimics neuronal phenotype. J Neurosci Res 2004; 77: 192–204., , , , , .