SEARCH

SEARCH BY CITATION

Keywords:

  • Trans-differentiation;
  • Stem cell;
  • Adult bone marrow;
  • Tissue culture;
  • Artifact;
  • Neural differentiation

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. MaterialsandMethods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

Trans-differentiation is a mechanism proposed to explain how tissue-specific stem cells could generate cells of other organs, thus supporting the emerging concept of enhanced adult stem cell plasticity. Although spontaneous cell fusion rather than trans-differentiation may explain some unexpected cell fate changes in vivo, such a mechanism does not explain potential trans-differentiation events in vitro, including the generation of neural cell types from cultured bone marrow-derived stem cells. Here we present evidence that shows that cultured bone marrow-derived stem cells express neural proteins and form structures resembling neurons under defined growth conditions. We demonstrate that these changes in cell structure and neural protein expression are not consistent with typical neural development. Furthermore, the ability of bone marrow-derived stem cells to adopt a neural phenotype in vitro may occur as a result of cellular stress in response to removing cells from their niche and their growth in alternative environmental conditions. These findings suggest a potential explanation for the growth behavior of cultured bone marrow-derived stem cells and highlight the need to carefully validate the plasticity of stem cell differentiation.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. MaterialsandMethods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

Stem cells have been isolated from numerous adult tissues, including the central nervous system [1, 2], bone marrow [3, 4], and skin [5]. These so called “adult stem cells” have a more restricted developmental potential compared with embryonic stem cells, generating only differentiated cell types of the same lineage as the organ in which they reside. Traditionally, cell commitment has been viewed as consisting of a series of irreversible steps, which involve increasing commitment to particular cell lineage. However, this model of irreversible and restricted differentiation has been challenged by recent experimental findings.

Transplantation experiments suggest that increased cell plasticity may occur under certain physiological conditions in which at least a subpopulation of adult stem cells are capable of generating cells of a different embryonic germ layer, a process referred to as “trans-differentiation.” This term denotes an alteration in the differentiation potential of a cell already programmed to a given lineage [4, 6]. For example, when donor bone marrow cells are transplanted into lethally irradiated recipients, genetic markers of the donor cells can be detected in various adult tissues outside the hematopoietic lineages, including skeletal muscle [7], liver [8], heart [9], and brain [10, [11]12]. Reports of cell plasticity have not been confined to bone marrow-derived stem cells; for example, neural stem cells were shown to differentiate into virtually all cell types when injected into blastocysts [13].

These findings have been met with considerable controversy, with reports of conflicting results and low reproducibility. Some investigators have found little or no evidence to support the trans-differentiation of adult stem cells using similar or identical experimental paradigms [14, [15]16]. Others argue that cell fusion and not trans-differentiation may offer an explanation for unexpected cell fate changes in vivo [17, [18]19].

A multitude of studies have been published that demonstrate an increase in adult stem cell plasticity following in vitro cultivation [20, [21]22]. In cases where coculture has been used to promote differentiation toward a particular cell lineage, some investigators have been able to show that differentiation has been independent of cell fusion [23, 24]. These results imply either that certain adult stem cells have an intrinsic capacity for differentiation beyond their organ of residence with such plasticity suppressed in situ, or that adult such cells can be reprogrammed in the their differentiation potential toward specific cell lineages by exogenous cues.

Mesenchymal stem cells (MSCs) derived from bone marrow (BM) are an example of one adult stem cell population proposed to demonstrate increased plasticity following cultivation in vitro [25, [26]27]. MSCs are particularly good candidates for cell therapy because of their accessibility and capacity for ex vivo expansion [28]. While retaining their capacity for mesoderm differentiation, MSCs have been shown to differentiate into endodermal and ectodermal derivatives under defined culture conditions independent of cell fusion [22, 26]. One of the most striking examples of this has been the demonstration that MSCs can form neuroectodermal derivatives in vitro [22, 23, 29, [30]31] or following transplantation into the nervous system [11, 12, 32, 33].

In this study, we investigated whether non-neural tissue-specific stem cells have an intrinsic capacity to generate neural derivatives by trans-differentiation. Specifically, we examined the neurogenic potential of cultured rat MSCs isolated from the adult BM and questioned the mechanism by which such cells adopt an apparently neural phenotype.

MaterialsandMethods

  1. Top of page
  2. Abstract
  3. Introduction
  4. MaterialsandMethods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

Materials

Tissue culture reagents and other materials were acquired from Sigma-Aldrich (Poole, U.K., http://www.sigmaaldrich.com) unless otherwise stated. All substances were of the appropriate chemical, molecular biological, or tissue culture grade. Cytochalasin-B and colcemid (COL) were dissolved in ethanol and used at final concentrations of 10 and 1 μg/ml, respectively, unless otherwise stated. The broad-spectrum protein kinase C inhibitors staurosporine and chelerythrine chloride were dissolved in dimethyl sulfoxide (DMSO) and used at final concentrations of 10 and 50 μM, respectively. The mitogen-activated protein (MAP) kinase inhibitors were purchased from Calbiochem (Merck Biosciences Ltd., Nottingham, U.K., http://www.merckbiosciences.co.uk), reconstituted in DMSO and used at the following concentrations: PD98059 (75 nM, selective mitogen-activated protein kinase kinase [MEK] inhibitor), SB 202190 (10 nM, a potent inhibitor of p38), SB 203580 (10 nM, a highly selective inhibitor of p38), and SP 600125 (10 nM, a potent inhibitor of c-Jun N-terminal kinase [JNK]) were reconstituted in DMSO. MAP kinase and protein kinase C (PKC) inhibitors were added to the culture 30 minutes prior to any further treatment.

Cell Culture

Rat mesenchymal stem cells (rMSCs) were isolated from the femurs and tibiae of 6–8-month-old Wistar rats. The BM was aspirated with 20 ml of collection medium (RPMI-1640 supplemented with 10% fetal calf serum [FCS] [Invitrogen, Paisley, U.K., http://www.invitrogen.com], 100 U/ml penicillin, 100 μg/ml streptomycin, and 12 μM l-glutamine) into a T75-cm2 flask to allow stromal cells to adhere to the culture surface. Adherent cells were then washed and maintained in complete culture medium (CCM; Dulbecco's modified Eagle's medium supplemented with 10% FCS, 100 U/ml penicillin, 100 μg/ml streptomycin, and 12 μM l-glutamine and 1× nonessential amino acids) at 37°C in 5% CO2. Isolation of rMSCs was verified by their capacity to differentiate into mesodermal derivatives (bone and fat) and their cell surface expression of MSC markers SH2 (CD105), CD166, and CD90 and by the absence of hematopoietic markers CD45 and CD11b (data not shown). Passage 8 (approximately 25 population doublings) cells were used in the experiments described herein. To induce a presumptive neural phenotype, rMSCs (CD90+, CD105+, CD166+, CD45, CD11b) were grown to >80% confluence in CCM. Unless stated otherwise, rMSCs were subsequently incubated in serum-free Dulbecco's modified Eagle's medium (DMEM) Ham's F-12 medium supplemented with N2 supplement on tissue culture plastic. For some experiments, the medium was also additionally supplemented with basic fibroblast growth factor (bFGF) at a final concentration of 20 ng/ml. Rat dermal fibroblasts were cultured in α-minimal essential medium containing 10% FCS and 2 mM l-glutamine.

Intracellular Staining by Flow Cytometry

Suspensions of rMSCs were fixed in 2% paraformaldehyde (PFA) in phosphate-buffered saline (PBS) and permeabilized in Triton X-100 (0.2% in PBS). Nonspecific binding was blocked by incubation with 5% goat serum. Cells were incubated with primary antibody (TuJ-1 [Covance Research Products Inc., Berkley, CA, http://www.covance.com], 1:500), Nestin (Chemicon [Chemicon International, Temecula CA, http://www.chemicon.com]; 1:100), Vimentin (Chemicon; 1:200), Synaptophsin (Sigma-Aldrich; 1:100), GAP-43 (Sigma-Aldrich; 1:200), NeuN (Chemicon, 1:50), NeuroD1 (Abcam [Abcam, Cambridge, U.K., http://www.abcam.com], 1:100), and fibronectin (Sigma-Aldrich; 1:200) for 60 minutes 4°C in antibody buffer (PBS, 1% goat serum, 0.1% bovine serum albumin), followed by incubation with fluorescein isothiocyanate (FITC)-conjugated goat antibodies against mouse IgG were used as secondary antibodies (1:100). Samples of cells were passed through a Coulter EPICS XL flow cytometer (Coulter Corp., Hialeah, FL, http://www.coulter.com), and the level of immunofluorescence was determined.

Cell Morphology Assays

Digital images of phase contrast microscopy were acquired using a Nikon inverted microscope and camera (Nikon CM200). Measurements of cell footprint area (surface area occupied by the cell defined by its outer periphery of the cell) were determined by examining 10 nonoverlapping microscopic fields (>20 cells per field) for three independent experiments using Image J software (NIH). The number of arborized cells was also counted for each visual field and treatment condition. For time-lapse microscopy, phase images of the same field of view were recorded at 15-minute intervals using a Nikon inverted microscope equipped with a temperature-controlled stage.

Immunofluorescence Microscopy

To preserve the structure of the microtubule cytoskeleton, cells were washed in PBS and extracted with 1% Triton X-100 in microtubule stabilizing buffer (PEM: 1 mM MgCl2, 5 mM EGTA, 80 mM K-1,4-piperazinediethanesulfonic acid [Pipes], pH 6.8). After permeabilization, cells were fixed with 0.5% gluteraldehyde in PBS. Free aldehyde groups were blocked by sodium borohydride (10 minutes) and lysine (2% solution) for 1 hour. For Nestin staining, cells were washed with PBS and fixed in 4% PFA solution for 30 minutes, followed by postfixation and permeabilization in 0.5% Triton X-100 in PBS for 15 minutes. All samples were subsequently rinsed three times in blocking/wash buffer (2% PFA in PBS), incubated with monoclonal mouse antibody directed against either α-Tubulin (DMIA; Sigma-Aldrich; 1:100) or Nestin (Chemicon; 1:200). FITC-conjugated goat antibodies against mouse IgG (Sigma-Aldrich; 1:100) were used as secondary antibodies. To visualize the F-Actin cytoskeleton, cells were stained with rhodamine phalloidin according to the manufacturer's instructions (Invitrogen). Image acquisition was performed by using a Nikon 330 fluorescence microscope, and all images were captured using the same collection parameters for quantitative comparisons. The fraction of positive cells was determined for each culture condition by counting 10 nonoverlapping microscopic fields (>20 cells per field) for each condition in at least three independent experiments.

Western Blot Analysis

Protein extracts (30 μg per lane) were separated by electrophoresis and transferred onto polyvinylidene difluoride membrane (Amersham Biosciences, Piscataway, NJ, http://www.amersham.com). For immunoblotting, membranes first incubated in blocking solution (10 mM Tris-HCl [pH 8.0], 150 mM NaCl containing 5% milk powder, and 0.2% Tween 20) for 1 hour followed by primary (TuJ1 [Covance], 1:5,000; Nestin [Chemicon], 1:1,000; and β-Actin [Sigma-Aldrich], 1:5,000), and secondary mouse or rabbit IgG-HRP (Amersham Pharmacia Biotech UK Ltd., Little Chalfont, U.K., http://www.amershambiosciences.com; 1:1,000) antibody. Protein-antibody binding was detected on film (Hyperfilm ECL; Amersham) using chemiluminescence (Amersham).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. MaterialsandMethods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

Effect of Serum on the Induction of Nestin Expression by rMSCs

To identify conditions that may promote the differentiation of MSCs toward a neural lineage, we cultured rMSCs for 24 hours in bFGF (10 ng/ml)-supplemented media either in the presence of 10% FCS in DMEM F-12 or under serum-free (SF) conditions using DMEM F-12 + N2 supplement or DMEM F-12 alone. We examined these cultures for the expression of Nestin, a class III intermediate filament protein commonly used as a marker of neural progenitors [34]. In the present study, we found that addition of 10 ng/ml bFGF for 24 hours was sufficient to induce Nestin expression in rMSC cultures. However, growth factor treatment in the presence of serum (10% FCS) resulted in only minimal induction of Nestin expression (9%–10%) compared with induction following growth factor treatment in serum-free cultures (DMEM F-12 + N2 and DMEM F-12, 28% and 34%, respectively) (Fig. 1A). This finding suggests that serum removal is required for increased induction of Nestin expression and implied that serum may itself have a regulatory role in the acquisition of a neural phenotype.

thumbnail image

Figure Figure 1.. Expression of neural proteins by rat mesenchymal stem cells is regulated by the presence of FCS. (A): Effect of FCS on FGF-induced Nestin expression. Cells were grown with the inclusion of FGF (10 ng/ml) for 24 hours in the presence (white bars) or absence (gray bars, SF + N2; black bars, SF) of 10% FCS, and the proportion of Nestin-positive cells was recorded (*, p < .05 compared with FCS-treated cells; n = 3; mean ± SEM). (B): Effect of FCS withdrawal alone on Nestin expression. Cells were grown in the presence (white bars) or absence (gray bars, SF + N2; black bars, SF) of 10% FCS, and the proportion of Nestin-positive cells was recorded (*, p < .05 compared with FCS-treated cells; n = 3; mean ± SEM). (C): Immunostaining and corresponding phase images showing cells maintained in FCS-supplemented media possessed a flat fibroblastic morphology, which was apparently unchanged following the addition of FGF. However, withdrawal of serum induced the appearance of Nestin bright cells with a neural-like morphology (scale bar = 10 μm). (D): Western analysis showing significant increased levels of the neural antigens TuJ1 and Nestin in response to FGF, but particularly in response to serum withdrawal. Abbreviations: FCS, fetal calf serum; FGF, fibroblast growth factor; SF, serum-free.

Download figure to PowerPoint

Consistent with this observation, Nestin expression was virtually undetectable in cells continually maintained in serum-supplemented media (9%–10% Nestin-positive cells, low expression), and removal of serum from the culture was sufficient to induce Nestin expression in a fraction of rMSCs (24% Nestin bright) (Fig. 1B). No significant difference was found between the level of Nestin expression in serum-free cultures and cultures in which serum had been substituted by N2 supplement (DMEM F-12 + N2, 31%), indicating that serum removal was a critical determinant of the level of Nestin expression within the culture.

The intensity of Nestin expression was heterogeneous during growth factor treatment but correlated with the acquisition of a neural-like morphology (Fig. 1C). Cells maintained in media supplemented with 10% FCS had a fibroblastic morphology consistent with a stromal phenotype. These cells stained only very weakly for Nestin protein and within the confines of this study were counted as Nestin-negative. However, 31%–35% of cells expressed high levels of Nestin following 24 hours of serum-free culture, and all these cells developed a neural-like morphology. The same pattern of Nestin expression was observed for growth factor treatment in the presence or absence of serum (Fig. 1C). Western analysis confirmed that the induction of Nestin was significantly higher in serum-free cultures and that such regulation in expression was also evident for the neuronal marker TuJ-1 (Fig. 1D).

Changes in Protein Expression Profile by rMSCs Cultured in Serum-Free Media

Neural development is a controlled process in which the temporal and spatial expression of neural genes is tightly regulated [1]. Using immunofluorescent detection of selected proteins and flow cytometry, we examined the expression of neural proteins to determine whether serum withdrawal induced a true transition toward the neural lineage with the concomitant downregulation of mesodermal markers commonly expressed by MSCs (data not shown). Cells cultured in serum-containing media were negative for Nestin and NeuroD1, but a fraction of rMSCs did express TuJ-1, NeuN, and Synaptophysin at low levels. GAP-43 was expressed on 95% of cells but also at relatively low levels. Culturing rMSCs in serum-free media for 24 hours resulted in an upregulation in the expression of the neural proteins tested, namely Synaptophysin, NeuN, GAP-43, TuJ-1, Nestin, and NeuroD1, together with a reduction in the expression of the mesoderm-associated proteins fibronectin and Vimentin (data not shown).

Changes in the Morphology of rMSCs in Response to Serum-Free Culture

The morphology of rMSCs grown in serum-free conditions was recorded at regular time intervals using phase imaging (Fig. 2). Cells were measured for footprint area and whether they resembled neurons. Upon removal of serum, 23%–25% of rMSCs adopted a neural-like morphology within 5 hours (Fig. 2A). Such a response was significantly enhanced when rMSCs were exposed to 2% DMSO in serum-free culture, where 91% of cells displayed a neural-like morphology (Fig. 2A). The structural features that underlie the change in cell morphology from a flat fibroblastic morphology to a neural-like morphology were demonstrated by time-lapse imaging (Fig. 2B). The cytoplasm of responsive cells shrank toward the nucleus, and this retraction left behind cytoplasmic extensions radiating from the nucleus to where the cell periphery had originally been located. Not all cells responded in the same fashion, and there were varying degrees of cell shrinkage in response to serum withdrawal. Responsive rMSCs became increasingly spherical and highly refractile, resembling typical neuronal perikarya.

thumbnail image

Figure Figure 2.. Morphological response of rat mesenchymal stem cells (rMSCs) following withdrawal of serum. (A): Phase images showing the morphology of rMSCs under various growth conditions. Cells maintained in 10% FCS-supplemented media have a characteristic stromal morphology (control; scale bar = 80 μm). Removal of FCS from the culture for 5 hours resulted in approximately 25% of cells adopting a neural-like morphology (scale bars = 50 μm [×20] and 25 μm [×40]). Other cells in culture displayed extensive membrane ruffling or cell rounding. Addition of 1% DMSO to SF cultures resulted in the majority of cells forming neural-like morphologies (scale bar = 40 μm). (B): Time-lapse imaging of rMSCs during the first 5 hours of serum-free culture. Images from two representative experiments (Example 1 and Example 2) are shown at three time points (0, 180, and 300 minutes; scale bar = 25 μm). (C): The percentage of cells displaying a neural-like morphology progressively increased following the removal of serum until reaching a maximal response between 3 and 5 hours. This response is unaffected by the presence of N2 supplement and was potentiated by the addition of 1% DMSO (**, p < .01 compared with cells maintained in SF media; n = 3; mean ± SEM). (D): The footprint area of individual cells was significantly reduced 5 hours after serum withdrawal (white bars) and to an equivalent extent following addition of 1% DMSO (5 hours) in SF media (light gray bar) and SF media plus N2 (dark gray bar) compared with control (10% FCS, black bar). *, p < .05 compared with control cells (n = 3; mean ± SEM). Abbreviations: DMSO, dimethyl sulfoxide; FCS, fetal calf serum; min, minutes; SF, serum-free.

Download figure to PowerPoint

Reorganization of the rMSC Cytoskeleton During the Formation of Neural-Like Cells

We examined the structure of the cytoskeleton in rMSCs during their transformation into neural-like cells. Rat MSCs were cultured in the presence of 2% DMSO in the absence of serum to induce a high percentage of cells to undergo cellular collapse. Staining for F-Actin and α-Tubulin over a 5-hour period showed that these cytoskeletal elements appeared to collapse toward the nucleus (data not shown). These results, together with those from the time lapse imaging experiments (Fig. 2B), suggest that rMSCs acquire neuronal morphologies through the collapse of the cytoskeleton and not through the active process of cytoplasmic extension and neurite outgrowth, as traditionally observed in a growing neuron.

Although the cytoplasmic processes in responsive cells contained both actin and tubulin, we were unable to determine which cytoskeletal system was principally the cause of this morphological change. To determine the cytoskeletal compartment primarily involved in such cell shrinkage, we selectively disrupted F-Actin and the microtubule systems independently using cytochalasin-B and COL, respectively (Fig. 3). Rat MSCs treated with cytochalasin-B in serum-supplemented media resulted in a characteristic arborized morphology in >90% of cells. Disruption of the actin cytoskeleton in this way resulted in cells with morphology closely resembling the structure of cells treated with DMSO or those cells that underwent collapse following serum withdrawal. In contrast, the selective disruption of the microtubule system using COL (a microtubule depolymerizing agent) did not result in an arborized morphology. Although no polymerized α-Tubulin was detected, staining for F-Actin appeared to remain unaffected. These data suggest that F-Actin is principally disrupted in response to serum withdrawal and DMSO treatment and that the structure of the microtubule system changes as a result of F-Actin retraction.

thumbnail image

Figure Figure 3.. Effect of selective cytoskeletal disrupting agents on the morphology and cytoskeleton of rat mesenchymal stem cells maintained in media containing 10% FCS. (A): Phase images showing the morphology of cells following exposure to 10 μM CB or 10 μg/ml COL for 5 hours. CB treatment resulted in a characteristic arborized phenotype in 75%–81% of cells, whereas exposure to COL resulted in only partial cell shrinkage and extensive membrane ruffling. (B): The cellular footprint area (μm2) and number of fully arborized cells (% total cells per field) were determined (n = 10 fields of view; +20 cells per field; mean ± SEM). Treatment with either CB or COL resulted in a significant reduction in the cellular footprint area compared with control cells (10% FCS alone), but only CB treatment resulted in an arborized phenotype (*, p < .05 compared with control cells [FCS]; n = 3; mean ± SEM). (C): Effect of CB and COL treatment on the organization of the F-Actin and microtubule cytoskeleton. Disruption of F-Actin gave rise to a highly arborized phenotype, whereas depolymerization of the microtubule system with COL for 20 hours did not appear to affect the F-Actin cytoskeleton, and these cells more or less maintained their fibroblastic cell shape (scale bar = 50 μm). Abbreviations: CB, cytochalasin-B; COL, colcemid; FCS, fetal calf serum.

Download figure to PowerPoint

To verify this conclusion, we treated rMSCs with cytochalasin-B for 5 hours to induce cells with neural-like morphologies and then subsequently removed serum from the culture for 5 hours (data not shown). No additional morphological response was apparent as a result of serum removal. Similarly, the addition of COL for 5 hours post-cytochalasin-B treatment did not result in any further morphological changes. This supported the notion that F-Actin disruption was the principal cause of morphological changes following serum withdrawal. Moreover, F-Actin alone was sufficient to maintain the neurite-like processes, whereas the growth and extension of true neurites is dependent on a dynamic microtubule cytoskeleton [35]. Interestingly, removal of serum or addition of cytochalasin-B following COL treatment also did not result in any cell shape changes, despite an intact F-Actin system being present (data not shown). This suggests that microtubules may provide the retractile force for F-Actin collapse.

Reorganization of the Cytoskeleton and the Expression of Neural Genes

We next investigated whether the increased expression of neural proteins that occurs subsequent to induction of a neural-like phenotype is related to the collapse of the cytoskeleton. We chose to examine the expression of Nestin, since this protein was not found in untreated rMSCs and its increased expression appeared to correlate with the formation of neural-like cells in response to serum withdrawal (Fig. 1). Varying degrees of F-Actin disruption were induced by treatment with cytochalasin-B over a range of concentrations (Fig. 4A). Merged images show that it was the arborized cells that expressed Nestin. A direct correlation existed between the extent of disruption and the number of Nestin expressing cells (Fig. 4B). In addition, as amount of Actin disruption increased (as a consequence of increased cytochalasin-B concentration), the cellular footprint area reduced together with an increase in the number of arborized cells. Although microfilament disruption was slightly evident at 0.1 μM cytochalasin-B treatment, no significant upregulation in Nestin expression was detected by immunocytochemistry. Accordingly, disruption of the Actin cytoskeleton by treatment with cytochalasin-B resulted in neural-like cells that express high levels of Nestin in a concentration-dependent manner. Western analysis performed on samples of cells treated with different concentrations of cytochalasin-B, ruled out the possibility that the appearance of Nestin bright cells was an artifact of cellular collapse and not simply the result of an increased amount of antigen per surface area (data not shown). We also recorded that the effect of cytochalasin-B on Actin disruption and the induction of expression for certain other neural proteins appeared to be related to cytoskeletal disruption and reached maximal levels after 5 hours of treatment (Fig. 5). Treatment of cultures with 10 μM cytochalasin-B resulted in cellular arborization in which 92% of cells possessed a fully arborized phenotype by 5 hours (Fig. 5A), which correlated with a reduced cell footprint area over the same period (Fig. 5B). Using flow cytometry, we found that percentage of cells expressing NeuroD1 and Nestin increased in a similar fashion to the number of cells changing shape over time, whereas levels of Vimentin decreased over this period (Fig. 5C).

thumbnail image

Figure Figure 4.. Regulation of Nestin expression in response to treatment with CB (0.1–10 μM) for 5 hours. (A): Cells incubated in 10% FCS had a normal fibroblastic morphology and Actin filaments organized in classic stress fiber patterns. The F-Actin microfilaments became progressively more disrupted as CB concentration increased and as cells progressively acquired an arborized phenotype. High levels of Nestin expression were confined only to those cells with a fully arborized phenotype (see merge images) (scale bar = 80 μm). (B): Measurement of cell footprint area (μm2) and percentage of Nestin expressing cells (number of Nestin bright cells per total cells per visual field) further demonstrated that the expression of Nestin only occurred at higher concentrations of CB (1–10 μM), correlating with the proportion of fully arborized cells and the disruption of the F-Actin cytoskeleton. Data represented as mean ± SEM (n = 10; +20 cells per field). ∗, p < .05. Abbreviations: CB, cytochalasin-B; FCS, fetal calf serum.

Download figure to PowerPoint

thumbnail image

Figure Figure 5.. Rat mesenchymal stem cells form neural-like cells and express neural proteins in response to CB over time. Cells were treated with CB (10 μM) to disrupt F-Actin and subsequently fixed at various time points ranging from 0 to 300 minutes. The percentage of arborized cells (A) (data represented as mean ± SEM [n = 3]) and mean cellular footprint area (B) (for 10 fields of view, +20 cells per field; *, p < .05 compared with vehicle-treated cells) were measured at each time point. These data show that the morphological response to CB treatment occurred progressively during the 5-hour treatment period. (C): Expression of neural and mesodermal proteins was determined by flow cytometric analysis of fixed and permeabilized cells at each time point. The number of cells positive for neural proteins NeuroD1 and Nestin correlated with the increased number of arborized cells over time. In contrast, expression of the mesodermal-associated marker Vimentin progressively decreased over time. The percentage of cells positive for TuJ-1 did not change in response to CB. Data shown represent mean ± SEM; n = 3. Abbreviations: CB, cytochalasin-B; FCS, fetal calf serum; Veh, vehicle.

Download figure to PowerPoint

Cell Signaling Molecules That Mediate the Cytoskeletal Regulation of Neural Protein Expression in rMSCs

Perturbation of cell shape by compounds that disrupt the cytoskeleton has been shown to alter the activity of several signaling molecules [36, [37]38]. For example, elements of the Ras, Raf, and mitogen-activated protein kinase (MAPK) cascade are understood to associate with a microfilament signaling particle that is thought to mediate MAPK activation by the cytoskeleton [39, 40]. To investigate the molecular mechanism(s) underlying the induction of neural-like phenotype and expression of neural proteins by rMSCs, we used selective inhibitors to antagonize particular signaling molecules during this process. The PKC signaling pathway has previously been implicated in the regulation of both cell morphology and neural gene expression. We found that the upregulated expression of NeuroD1 and Nestin by rMSCs cultured in SF media or treated with cytochalasin-B in serum-containing media for 24 hours was significantly inhibited by broad-spectrum PKC inhibitors (Fig. 6A; Table 1). However, this attenuation may not be a direct effect on the inhibition of signaling molecules involved in neural protein induction, since the presence of the inhibitors also reduced the morphological response to SF and cytochalasin-B treatments, an effect that may account for the reduced expression of neural protein in these cultures. In contrast, MAPK inhibitors did not affect the morphological response of rMSCs to Actin disruption (Fig. 6B). The induction of neural proteins in response to cytoskeleton disruption did not appear to involve MEK-extracellular signal-regulated kinase (MEK-ERK) signaling but was partly dependent on both JNK and p38 signaling (Table 1). Inhibition of either the JNK pathway or the p38 pathway significantly reduced the level of neural protein; however, this reduction was partial in both cases. This may be because other signaling molecules operate that substitute for JNK and p38 signaling.

thumbnail image

Figure Figure 6.. Effect of broad-spectrum protein kinase C (PKC) and mitogen-activated protein kinase (MAPK) inhibitors on changes in cell shape induced by CB treatment. Cells were treated with 10 μM CB for 5 hours in serum-containing media before being fixed and permeabilized for immunocytochemistry. The mean cellular footprint area was determined for cells cultured under each condition (mean ± SEM; 10 fields of view; +20 cells per field). (A): The footprint area of CB-treated cells was significantly reduced compared with vehicle-treated cells (FCS + vehicle [control]). The PKC inhibitors Str and ChCl partially inhibited the CB-induced reduction in the footprint area in response to CB treatment; however, the reduction was still significant compared with vehicle-treated controls. (B): Addition of MAPK inhibitors had no significant effect on the CB-induced reduction in the cellular footprint area. (*, p < .05 compared with vehicle-treated cells; +, p < .01 compared with FCS+CB-treated cells). Abbreviations: CB, cytochalasin-B; ChCl, chelerythrine chloride; FCS, fetal calf serum; Str, staurosporine.

Download figure to PowerPoint

Table Table 1.. Effect of specific kinase inhibitors on the expression of neural proteins by rat mesenchymal stem cells in response to cytochalasin-B treatment and serum withdrawal
Thumbnail image of

Reversal of Morphological and Protein Expression Responses to Serum Removal or Actin Disruption

The differentiation of multipotent stem cells involves a progressive restriction in cell fate with increasing commitment to a specific cell lineage. We hypothesized that if rMSCs cultured under defined conditions could be instructed to adopt a neural fate, then such reprogramming would involve the progressive loss of mesodermal-specific genes accompanied by a gene expression profile consistent with a neural lineage. To test whether the expression of neural proteins represented a firm commitment to the neural lineage or simply a transient reversible expression, we examined the levels of neural and mesodermal proteins following the reintroduction of serum or removal of cytochalasin-B (data not shown). Cells cultured in serum-free media or in the presence cytochalasin-B upregulated their expression of Nestin, NeuroD1, and TuJ-1, with the concomitant downregulation of the mesoderm marker Vimentin. This change in the protein expression profile in both cytochalasin-B-treated cells and cells cultured in serum-free media was completely reversible following the removal of cytochalasin-B or reintroduction of serum to serum-free cultures. The protein expression profile of these previously treated cells was comparable with rMSCs maintained in serum-containing media. This reversal in the expression of neural proteins was accompanied by a reversal of the morphology of the cells from an arborized phenotype to a flat fibroblastic morphology (data not shown).

This reversible behavior suggested that rather than a differentiation response, the expression of neural proteins may be a cellular stress response to cytoskeleton disruption. Aberrant expression of proteins in response to cellular stress is often transient [41]. We have also shown that the expression level of the neural marker TuJ-1 was initially increased in response to serum-free culture conditions but subsequently returned to lower levels over the next 5–6 days when maintained in the same type of growth media (data not shown). This reduction in TuJ-1 expression was also associated with a progressive reduction in the number of cells displaying an arborized phenotype. Therefore, the morphological and protein response of rMSCs to serum-free culture is a transient effect, with cells progressively recovering over time.

Expression of Neural Proteins by Dermal Fibroblasts in Response to Actin Disruption

We examined the behavior of rat dermal fibroblasts under growth conditions identical to those used with rMSCs, to determine whether the induction of neural protein expression in response to Actin disruption was specific to rMSCs. Rat dermal fibroblasts adopted an arborized morphology and underwent cytoskeletal collapse in response to cytochalasin-B treatment in serum-containing media and serum-free media (data not shown). This morphology was highly comparable to that observed following such treatment in rMSCs. Fibroblasts also showed upregulated expression of Nestin and TuJ-1 in response to cytochalasin-B, expression that increased from 5 hours to 24 hours. These effects on rat fibroblasts were completely reversible (data not shown).

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. MaterialsandMethods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

Stem cells reside in specialized niches where they are subject to spatial and temporal regulation in respect to their developmental potential. Removal of stem cells from their normal microenvironment and their subsequent culture may be a potential explanation for a potential increase in their plasticity. Multilineage adult progenitor cells have been isolated from mammalian bone marrow and shown to differentiate into tissues representative of all three germ layers [42]. To achieve these differentiation capabilities, cells are removed from their niche and maintained ex vivo under highly defined and selective culture conditions. Recent studies have reported that MSCs isolated from adult BM have the capacity to differentiate into neuroectodermal derivatives independent of cell fusion following cultivation ex vivo [22, [23]24]. This implied that although such adult stem cells are predisposed to differentiate into cells of a particular lineage, they may be much more “plastic” than previously appreciated, and that the developmental potential of these cells may be dictated by niche-specific signals.

Despite the importance of these observations, little has been reported on the mechanisms that underlie the unexpected potency of MSCs in vitro. MSCs may have an intrinsic capacity to differentiate into neural cell fates, a property that could be suppressed in the BM niche but not in culture or in other tissues in vivo. In the current study, we attempted to elucidate the mechanism by which cultured MSCs acquire a presumptive neuronal phenotype. Culturing MSCs in the absence of serum was a critical determinant in the formation of Nestin-positive cells with neural-like morphology. Previous studies have explored what may be involved in the regulation of neural protein expression by cultured MSCs [43, 44]. However, these investigations are compounded by the diversity of procedures and inductive agents used to induce phenotypic changes in these cells. It is therefore of significance that we found that serum played a crucial regulatory role regardless of other inductive factors (i.e., growth factors) being present. This is consistent with longer-term experiments on MSCs that have shown that the removal of serum plays an important role in the development of neural morphology and the expression of neural proteins [23, 24].

Neurons have a unique architecture characterized by dynamic neurite outgrowth. We have shown, however, that the formation of neural-like cells by MSCs was not the result of typical neuronal development but the result of cellular shrinkage. Targeted disruption of individual cytoskeletal compartments revealed that this morphological response was principally the result of F-Actin disruption, although an intact microtubule system is required to provide the retractile force for Actin collapse. Alteration to the architecture of a cell has previously been linked to changes in gene transcription. For example, genes involved in tissue remodeling are closely associated with dynamic changes in cell morphology induced by stress and shape-changing physiologic processes [45, [46], [47]48]. These findings implicate a direct involvement of the cytoskeleton in the cell signaling apparatus. Consistent with this hypothesis, targeted reorganization of cell morphology with microfilament disrupting agents can activate the transcription of shape-responsive genes [46, 49]. Furthermore, drug-induced alterations in both the microfilament and microtubule networks mobilize intracellular signaling elements activating the ERK, JNK, and p38 MAPKs, which have been shown to result in changes in gene transcription [36, 37, 50, [51], [52]53]. In this study, we provide evidence that PKC signaling is potentially involved in mediating the changes in morphology and protein expression by MSCs during the formation of neural-like cells in response to serum withdrawal and disruption of the Actin cytoskeleton.

Neuroblastoma cell lines undergo neuronal differentiation in response to serum withdrawal [54]. Expression of neural genes in these cells requires the nuclear accumulation of ERK, a process that is PKC-dependent [55]. Continued differentiation and expression of neural genes is, however, associated with a downregulation of PKC [56]. In contrast, inhibition of PKC signaling in MSCs attenuated the induction of Nestin and NeuroD1 proteins following serum-free culture or F-Actin disruption. However, this antagonism was associated with a significant inhibition in the morphological response of MSCs, implicating a central role for PKC in orchestrating F-Actin collapse in response to selective culture conditions. Signaling through JNK and p38 was required for induction of Nestin and NeuroD1 expression but not MEK-ERK signaling. The regulation of the signaling pathways studied in relation to neural protein expression by MSCs is not consistent with neural differentiation by neuronal cell lines but is consistent with a pattern of cell stress induced activation of gene expression.

The expression of neural proteins in naive (untreated) MSCs has gained recent interest with confirmation in several reports of neural gene expression in untreated cells [57, [58], [59], [60]61]. However, variability in the reproducibility of these observations between and within studies has been encountered, which may be explained by the sensitivity of different detection methods used and the low level of expression by MSCs. Our finding that primary rat dermal fibroblasts express TuJ-1 and Nestin at very low levels and upregulate these proteins in response to cytochalasin-B-induced actin disruption suggested that neural cell marker proteins may not be exclusively expressed by cells of the neural lineage. Nestin and NSE are two neural marker proteins, which have now been detected in a number of non-neural tissues [34, 62, 63]. Since fibroblasts have no known stem cell-like properties, the upregulation of neural proteins in response to actin disruption may be considered further evidence that such changes are not an authentic differentiation response and may be a consequence of aberrant growth in culture.

The expression of Vimentin is also associated with neuroprogenitor cells. However, Vimentin expression decreased, whereas Nestin expression increased. If MSCs were differentiating toward an early neuroprogenitor phenotype then we would expect higher levels of coexpression for these two proteins. In addition, Vimentin was downregulated as cells adopted a neural-like morphology with a concomitant decrease of the mesodermal marker fibronectin. Given that the same observation was made when cells were treated by cytochalasin-B in the presence of serum, it is likely that it was the disruption of actin filaments that leads to decreased Vimentin expression and not induction of neural differentiation. In support of this suggestion, it is known that the disruption of the actin cytoskeleton results in secondary disruption of the microtubular network, which causes perinuclear collapse of Vimentin, and the protein is lost [64]. This may explain the observed loss of Vimentin expression during growth under serum-free conditions. Furthermore, the downregulation in fibronectin expression is consistent with previous reports in fibroblasts, which show that cytochalasin-B treatment results in the loss of the fibronectin matrix [65].

Evidence for trans-differentiation by cultured MSCs into neurons has relied primarily on changes in cell shape and protein expression. Definitive proof of neural development requires assessment of neurological properties such as synapse formation, neuronal polarity, and electrophysiological characterization. Presumptive neurons derived from bone marrow-derived stem cells (BMDSCs) have shown the lack of Na+ and K+ channels and functional neurotransmitter receptors [66, 67] and have shown that BMDSCs possess atypical electrophysiological properties compared with primary neurons [31, 68]. Some investigators claim that the lack of maturity of MSC-derived neurons is the result of a restrictive growth environment [69, 70]. Nonetheless, functional neurons have been generated from embryonic [71] and neural stem cells [72] under similar growth culture conditions. In general, it is currently difficult to determine the exact nature of the neurophysiological phenotype of MSC-derived neurons due to the range of independent studies and variety of alternative growth conditions used.

Consistent with previous reports, we have shown that changes in cell morphology do not provide a reliable indicator of neural differentiation since such changes can result from Actin collapse and not neural development [73, 74]. Here we identify a potential mechanism by which Actin collapse in MSCs results in the expression of neural proteins. The regulation of this expression pattern does not compare with neural differentiation but may be a cell stress response induced by the culture conditions. Importantly, we have found that serum withdrawal alone is sufficient to induce alterations to cell shape and protein expression. We propose that these changes have been misinterpreted as trans-differentiation on several recent occasions in which serum-free media have been used for induction of MSC differentiation [22, 24, 75, 76]. In conclusion, morphological and molecular expression data are not sufficient indicators of trans-differentiation, and this highlights the need for greater caution during the interpretation of such observations.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. MaterialsandMethods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

This work was supported in part by awards to S.A.P. by the Wellcome Trust (Registered Charity 210183) and the Biotechnology and Biological Sciences Research Council.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. MaterialsandMethods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References