Marrow Stromal Cells, Mitosis, and Neuronal Differentiation: Stem Cell and Precursor Functions


  • Guillermo Muñoz-Elías,

    1. University of Medicine and Dentistry of New Jersey/Robert Wood Johnson Medical School, Piscataway, New Jersey, USA
    Search for more papers by this author
  • Dale Woodbury,

    1. University of Medicine and Dentistry of New Jersey/Robert Wood Johnson Medical School, Piscataway, New Jersey, USA
    Search for more papers by this author
  • Ira B. Black M.D.

    Corresponding author
    1. University of Medicine and Dentistry of New Jersey/Robert Wood Johnson Medical School, Piscataway, New Jersey, USA
    • Department of Neuroscience and Cell Biology, UMDNJ/Robert Wood Johnson Medical School, 675 Hoes Lane, CABM Building, Room 342, Piscataway, New Jersey 08854-5635, USA. Telephone: 732-235-5388; Fax: 732-235-4990
    Search for more papers by this author


To define relationships among marrow stromal cells (MSCs), multipotential progenitors, committed precursors, and derived neurons, we examined differentiation, mitosis, and apoptosis in vitro. Neural induction medium morphologically converted over 70% of MSCs to typical neurons, which expressed tau, neuronal nuclear antigen, neuron-specific enolase, and TUC-4 within 24 hours. A subset decreased fibronectin expression, consistent with mesenchymal to neuroectodermal conversion. More than 35% of differentiating neurons incorporated bromodeoxyuridine (BrdU) and divided, increasing cell number by 60%, while another subpopulation differentiated without incorporating BrdU or dividing. Inhibition of mitosis and DNA synthesis did not prevent neural differentiation, with 70% of blocked cells expressing tau and displaying neuronal morphologies. By deoxynucleotidyl transferase-mediated dUTP-biotin nick end labeling assay, less than 1% of cells underwent apoptosis at 36 and 72 hours, suggesting differentiation without cell-selective mechanisms. Apparently, MSCs may directly differentiate into neurons without passing through a mitotic stage, suggesting that distinctions among stem cells, progenitors, and precursors are more flexible than formerly recognized.


The standard model of development involves the progressive, sequential commitment of undifferentiated cells to a lineage, and ultimately to a terminally differentiated cell type. Specifically, totipotent stem cells are thought to convert to multipotent stem cells and become progressively restricted in potentiality as they pass through progenitor and committed precursors to fully differentiated stages. In contrast, recent studies have defined remarkable flexibility in stem and progenitor cell differentiation, leading to the emerging concept of developmental and stem cell plasticity, encompassing transgerminal differentiation and even trans- and de-differentiation [13]. These recent observations raise fundamental questions regarding differentiation, in general, and the relationship of stem to progenitor to precursor cells, in particular. We have developed an in vitro system to investigate these issues using adult bone marrow stromal cell (MSC) differentiation into presumptive neurons.

Stem cells, exhibiting the classical traits of self-renewal and multipotentiality, have been detected, unexpectedly, in multiple organs in the adult, challenging traditional concepts of terminal differentiation. In addition to long-recognized stem cells of the adult marrow lymphohematopoietic and stromal mesenchymal lineages [4, 5], stem cells have been identified in liver, muscle, central nervous system, and skin [611]. Moreover, differentiation is apparently not restricted to derivatives of the host tissue. Rather, adult stem cells can differentiate into derivatives of other embryonic germ layers. For example, neural stem cells have been differentiated into blood [12] and myogenic cells [13], and hematopoietic cells have been differentiated into epithelial cells of the liver, lung, gastrointestinal tract, and skin [4]. In another striking example of plasticity, differentiated pancreatic exocrine-like cells have been directly differentiated into hepatocytes in the absence of mitosis, without passing through a stem cell stage, through a process of transdifferentiation [2].

MSCs isolated from the bone marrow of adult animals proliferate in culture and can differentiate into multiple mesenchymal derivatives (e.g., chondrocytes, osteoblasts, and adipocytes) following in vitro manipulation [14]. We have recently differentiated adult MSCs into neural elements in vitro using fully defined medium in a simple protocol [15]. In our studies, 80% of the mesodermally derived adult rat and human MSCs differentiated into neuroectodermal neuron-like cells within 5 hours of exposure to neuronal induction medium (NIM), exhibiting typical neuronal morphology and expressing multiple neuron-specific genes [15]. The differentiated cells exhibited spherical, refractile cell bodies, long processes, and terminal expansions typical of neuronal growth cones with filamentous filopodia. They expressed a variety of neuronal genes, including neuron-specific enolase (NSE), tau, neurofilament M, neuronal nuclear antigen (NeuN), β-III tubulin, and synaptophysin [15, 16]. The cells transiently expressed nestin at day 1, but the intermediate filament, characteristic of neuroepithelial precursors, disappeared by day 6, as in normal development [15]. We now employ this system to examine mechanisms of differentiation and the relationship of stem cells, progenitors, and precursors.

Is neuronal differentiation of MSCs restricted to stem cell functions in the population, requiring mitosis and subsequent withdrawal from the cell cycle? Alternatively, are subpopulations capable of neuronal differentiation in the absence of cell division, consistent with the direct conversion of MSCs, or differentiation of progenitors, and direct instructive mechanisms? Is there, consequently, evidence of plastic interconversion of stem and progenitor functions in MSCs? In the absence of mitosis, do derived neurons express the full panoply of neuron-associated genes and neuronal morphological characters or is the phenotype restricted? Is neuronal differentiation of MSCs characterized by the downregulation of mesenchymal gene products as neuronal genes increase expression? Do any of these observations alter the standard model of differentiation as a progressive, sequential, unidirectional process through stem, progenitor, and precursor stages?

Materials and Methods

Cell Culture

Bone marrow stromal cells were isolated from adult, female, Sprague-Dawley rats as previously described [17]. Cells were cultured in 20% fetal bovine serum (FBS; Atlanta Biological; Norcross, GA; in Dulbecco's modified Eagle's medium (DMEM; Invitrogen Life Sciences; Carlsbad, CA;, pH 8, at 37°C and passaged at least 10 times before being used for experimentation. MSCs were obtained with a protocol and procedure approved by the Institutional Animal Care and Use Committee.

Neuronal Induction

In preparation for induction, cultured MSCs were enzymatically and mechanically detached and suspended in DMEM (pH 8.0)/20% FBS at 8,000 cells/ml. Two milliliters of cell suspension were plated on concanavalin A- (Sigma; St. Louis, MO; or polylysine- (Sigma) coated 35-mm dishes and incubated at 37°C until subconfluent. Twenty-four hours prior to induction, the medium was replaced with DMEM (pH 8)/20% FBS supplemented with 5 ng/ml basic fibroblast growth factor. Cells were induced by replacing pretreatment medium with the NIM consisting of 100 μM butylated hydroxyanisole, 10 μM forskolin, N2-supplement (Invitrogen) or 5 μg/ml insulin (Sigma), 2% dimethylsulfoxide, 2.0 mM valproic acid, 5 nM K252A, and 10 mM KCl, in DMEM (pH 7). Cells were incubated at 31°C.


Cells were fixed with 4% paraformaldehyde at room temperature (RT) for 1 hour. After fixing, cells were washed with phosphate-buffered saline (PBS), blocked with serum in 0.3% Triton X-100 in PBS, and incubated for 1 hour at RT with primary antibody antiserum: polyclonal anti-tau (rabbit, 1:500, Sigma), polyclonal anti-NSE (rabbit, 1:500, Sigma), monoclonal anti-NeuN (mouse, 1:200, Chemicon; Temecula, CA;, polyclonal anti-TUC-4 (rabbit, 1:1,000, Chemicon), polyclonal anti-fibronectin (rabbit, 1:500, Chemicon). After rinsing, cells were incubated with 1:100 Texas Red-conjugated secondary antibody (Vector Laboratories; Burlingame, CA; overnight at 4°C. Similar results were obtained when cells were incubated overnight in primary antibody at 4°C followed by 1-hour incubation with Texas Red-conjugated secondary antibody at RT. For imaging, cells were shielded with Vectashield (Vector Laboratories) and a glass coverslip and visualized using a fluorescence microscope. The DAB Peroxidase Substrate system (Sigma) was used for NeuN immunocytochemistry.

For bromodeoxyuridine (BrdU) immunocytochemistry, cells were pulsed with 10 μM BrdU in PBS. After fixing, cells were incubated in 2N HCl for 30 minutes at RT. Washing and blocking steps were followed by incubation with monoclonal mouse anti-BrdU antibody (1:35, BD Biosciences Clontech; Palo Alto, CA; for 1 hour at RT. After washing, cells were incubated overnight in the dark at 4°C with fluorescein isothiocyanate (FITC)-conjugated secondary antibody (1:100, Vector Laboratories). For imaging, cells were covered with Vectashield and a cover slip. The signal was visualized using a fluorescent microscope. All cells were counterstained by incubating with DAPI (1.0 μg/ml in dH2O) for 3 minutes at RT followed by washing steps.


Deoxynucleotidyl transferase-mediated dUTP-biotin nick end labeling (TUNEL) assay was performed with the ApoAlert DNA Fragmentation Assay Kit according to the manufacturer's guidelines (BD Biosciences Clontech). Briefly, after fixing and washing, cell cultures were incubated in chilled 0.2% Triton/PBS for 5 minutes on ice and then washed twice for 5 minutes with PBS at RT. One hundred microliters of equilibration buffer were added and equilibrated for 10 minutes at RT using a glass coverslip to ensure even distribution of solution. After removal of equilibration buffer, 50 μl of terminal transferase (TdT) incubation buffer were added to the center of the dish with a glass coverslip to ensure even distribution of the TdT incubation buffer. All the following steps were done in the dark. The cell cultures were placed in a humidified 37°C incubator for 1 hour. After removal of the coverslip, cultures were washed with 2X standard saline citrate for 15 minutes at RT. After washing, the cells were protected with Vectashield and a coverslip. A fluorescent microscope equipped with a FITC filter was used to visualize the signal. For negative controls, a TdT-minus control incubation buffer was prepared by replacing the TdT enzyme with deionized water.


Photographs for the time-lapse analysis were captured using a Zeiss microscope equipped with a FUJIX Digital Camera (HC-300Z). All other pictures were obtained using a Leitz Aristoplan fluorescent microscope equipped with a Clientpro digital imaging system.


Neuronal Morphologic Differentiation and Mitosis

We previously found that rat MSCs exposed to the NIM exhibit typical neuronal morphological changes after 3 hours [15]. The rapidity of this transformation suggested that morphological differentiation occurred in the absence of cell division. To assess relationships between presumptive neuronal morphologic differentiation and mitotic activity, we used time-lapse, phase-contrast imaging during the first 24 hours of NIM treatment.

Exposure of MSCs to NIM resulted in the expected changes in morphology as responsive cells assumed forms typical of neurons after 3 hours, consistent with previous findings [15]. Seven and a half hours after NIM treatment, cells exhibited typical neuronal or stromal morphologies (Fig. 1A). A representative responsive MSC (white arrow) at 7.5 hours exhibited a multipolar, refractile cell body bearing many process-like extensions. Over the next 2 hours, the cell body condensed, became increasingly spherical and refractile, and acquired a typical neuronal perikaryal appearance. At 10.5 hours, however, the same cell underwent division, yielding a doublet through cytokinesis. By 15.5 hours, the daughter cells separated while continuing to exhibit refractile, process-bearing cell bodies.

Figure Figure 1..

Cell division of MSC-derived presumptive neurons.A) Time-lapse, phase-contrast analysis of induced MSCs after 7.5 hours of NIM treatment. Cells exhibiting neuronal (white arrow) or MSC (black arrow) morphology underwent cell division. A differentiating MSC exhibiting neuronal morphology 7 1/2 hours after NIM treatment (white arrow) underwent cytokinesis by 10.5 hours, yielding two daughter cells 5 hours later. However, other cells exhibiting neuronal morphology did not divide during this period (white arrowhead). B-C) Quantitative analysis of mitotic activity. B) There was a significant increase in total cell number from 0 to 24 hours after NIM treatment (*differs significantly from 0 hours at p < 0.01). The increase in total cell number shown here is representative of three independent experiments. The results are expressed as mean ± standard error (SE) of total cells for each timepoint (C). BrdU labeling index (LI; percent of total cells that are BrdU+). Induced MSCs were pulsed with BrdU 2 hours prior to indicated postinduction time. LI at 0 and 2 hours differed significantly from the LI at 19 and 24 hours after NIM treatment (*p < 0.01). The decrease in the LI shown here is representative of two independent experiments. The results are expressed as percent mean ± SE of LI for each timepoint. (B-C) For each experiment, we counted the number of cells in each of 10 random fields for each of two separate dishes. The total number of cells per dish was calculated by multiplying the average number of cells/field by the ratio total area of dish/area of field.

In the same field, cells acquired neuronal morphologic traits and then underwent mitosis. For example, an undifferentiated cell exhibiting distinctive MSC characters at 7.5 hours (black arrow), began contracting its cytoplasm toward the nucleus and over time acquired a condensed, refractile soma, extending multiple processes. At 10.5 hours, the cell entered late anaphase or early cytokinesis with a refractile cell body bearing multiple processes. In contrast, other cells that developed neuronal forms at 7.5 hours did not divide over the 24-hour period (arrowhead). In sum, morphologic differentiation developed before, after, and in the absence of cell division.

To determine whether mitosis led to true proliferation (with survival), cell number was assessed. In fact, cell number increased by 60% during the first 24 hours, indicating that mitosis was leading to proliferation (Fig. 1B). Measurement of the BrdU labeling index (LI; the ratio of BrdU-labeled to total cells) after acute BrdU administration revealed that most of the incorporation occurred during the first 12 hours after NIM treatment of MSCs (Fig. 1C). The LI decreased sharply from a maximum plateau of approximately 40% to a minimum of 5% at 24 hours post induction (Fig. 1C).

Short-Term Characterization of Gene Expression, Morphology, and Mitosis

In addition to neuronal morphological differentiation, our previous work indicated that 5 hours after NIM treatment MSCs expressed a host of neuron-specific gene products [15]. In light of the mitotic activity within 24 hours of induction, we examined the hypothesis that mitosis and expression of neuron-specific gene products are not mutually exclusive. Specifically, we characterized the NIM-treated MSCs using neuron-specific antibodies, while monitoring incorporation of the S-phase marker BrdU, a thymidine analogue that is incorporated into newly synthesized DNA [18]. To identify cells at different stages of maturation, we performed immunocytochemistry examining multiple gene products typical of differentiating neurons (tau and NSE), as well as postmitotic and mature neurons (TUC-4 and NeuN). To characterize the foregoing relationships, experiments were performed at different times within the 24-hour postinduction period.

To define overall entry into S-phase and neuronal differentiation, we examined BrdU incorporation and tau expression during the entire initial 24-hour period. Tau, a neuron-specific, microtubule-associated gene product, is expressed in differentiating neurons [19, 20]. MSCs were induced and exposed to BrdU from 0–24 hours (Fig. 2). Cells exhibiting neuronal and MSC morphologies (Fig. 2A) were identified by phase contrast. Approximately 60% of the cells differentiated into presumptive neurons after induction, exhibiting neuronal morphologies and expressing tau (Fig. 2A–B). Representative cells exhibiting polarized neuronal morphologies, with a principle process (arrowhead) and several secondary dendrite-like extensions (arrows, Fig. 2A) expressed tau (arrow, Fig. 2B) while in metaphase, as indicated by BrdU incorporation (arrowhead, Fig. 2C). Thirty-five percent of the tau+ neurons incorporated BrdU during the 24-hour period (Fig. 3). In contrast, 25% of the presumptive neurons expressed tau without BrdU incorporation (arrows; Fig. 2D, Fig. 3). Homogeneous distribution of tau intracellularly is characteristic of the immaturity of these developing neurons. Positive controls, using E18.5 (embryonic day) hippocampal neurons cultured 15 days in vitro, revealed typical mature tau distribution (Fig. 2E–F), while negative controls, omitting the primary antibody, yielded no immunopositivity (Fig. 2G–H).

Figure Figure 2..

Mitosis and tau expression in MSC-derived presumptive neurons.Relationship among changes in morphology, tau expression, BrdU incorporation, and mitosis. Differentiating MSCs express tau and exhibit neuronal morphology while undergoing cell division. A) MSC-derived presumptive neurons exhibit polarized morphology with several dendritic-like processes (arrows) and a main principal axon-like extension (arrowhead), expressing tau (B, red) with a BrdU+metaphase plate (C, green). D) Tau is expressed in morphological neurons in cells that did (arrow-head) and did not (arrows) incorporate BrdU. Primary E18.5 (embryonic day) hippocampal neurons cultured 15 days in vitro (DIV) exhibited neuronal morphologies by phase contrast (E) and revealed typical mature tau distribution (positive control) (F). NIM-treated MSCs acquiring neuronal morphologies (G) yielded no tau immunopositivity when the primary antibody (Ab) was omitted (negative control) (H).

Figure Figure 3..

Quantification of tau expression and BrdU incorporation.There was no significant difference in the percent of cells incorporating BrdU during 24 hours of exposure to the S-phase marker between NIM-treated and control cells. However, there was a significant difference in the percent of tau expressing cells between conditions (p < 0.001). The data from two independent experiments, each of which consisted of three separate dishes for each condition, were pooled into a single data set for each condition. Cells in 10 random fields/dish were counted and the total number of cells per dish was calculated by multiplying the average number of cells/field by the ratio total area of dish/area of field. The results are expressed as percent mean ± SE of cells that incorporated BrdU (BrdU+), expressed tau (tau+), or incorporated BrdU and expressed tau simultaneously (BrdU+/tau+). The mean percentages for each condition were compared to determine significant differences.

To more precisely delineate the temporal relationship of incorporation and neuronal gene expression, we exposed cells to BrdU for periods ranging from 2–4 hours at different times during the first 12 hours after NIM treatment. We examined relationships among expression of NSE, a protein expressed in immature and differentiating neurons [21], changes in cell morphology, and mitotic activity. Twelve hours after initiation of NIM treatment, MSCs were exposed to a 2-hour BrdU pulse and were examined immunocytochemically. Cells with contracted, multipolar, process-bearing cell bodies (Fig. 4A) were immunopositive for NSE (arrow; Fig. 4B, 4b′). Subsets of these presumptive neurons incorporated BrdU consistent with simultaneous mitotic activity and expression of the neuronal marker (arrowhead, Fig. 4C). However, another subpopulation of presumptive neurons did not incorporate the S-phase marker, suggesting that expression of NSE was not preceded by cell division in these cells (arrow, Fig. 4C).

Figure Figure 4..

Differentiating MSCs exhibiting neuronal forms and expressing NSE incorporate BrdU during acute exposure.A) Phase-contrast image of MSC-derived neurons. B) MSC-derived neurons express NSE. C) NSE/BrdU double labeling demonstrating that some NSE+cells incorporate BrdU (arrowhead) while others do not (arrow). (b), (c) are higher magnifications of boxed areas in (B) and (C), respectively, demonstrating NSE and BrdU colocalization.

Identification of tau/BrdU, and NSE/BrdU double-labeled cells exhibiting neuronal morphologies prompted us to examine mitosis and expression of genes characteristic of more mature developing neurons. We extended characterization of the mature neuronal phenotype by analyzing expression of the NeuN, a marker of mature, postmitotic neurons [22]. MSCs were treated with NIM at 0 hours and were acutely exposed to BrdU 4 hours later. Eight hours post induction, the cells were analyzed immunocytochemically. The presumptive neurons, exhibiting spherical, refractile, process-bearing somata, stained positively for NeuN (Fig. 5A). Higher magnification (Fig. 5a′) demonstrated that NeuN immunoreactivity was exclusively nuclear and perinuclear (white arrow), and did not extend into the process-like extensions (white arrows) or putative varicosities (blue arrows). Omission of primary antibody eliminated all NeuN staining, indicating specificity (Fig. 5C). A subset of presumptive neurons, exhibiting neuronal morphologies and expressing NeuN, incorporated BrdU (arrow; Fig. 5B, 5b′). Furthermore, identification of BrdU/NeuN-labeled doublets suggests that these cells were indeed undergoing mitosis (picture not shown). However, not all of the NeuN+ cells incorporated BrdU (arrowhead), suggesting that NeuN was also expressed in MSCs that did not enter S phase during the BrdU-exposure period (Fig. 5A–B). The acute exposure to BrdU immediately prior to fixation excluded the possibility that these cells began expressing NeuN postmitotically.

Figure Figure 5..

Expression of mature neuronal proteins in mitotically active presumptive neurons.NeuN: Differentiating MSCs expressing NeuN (A, white arrow) and exhibiting neuronal morphologies incorporated BrdU (B, red arrow) during acute exposure to the S-phase marker. Some NeuN+cells exhibiting neuronal morphologies (A, blue arrowhead) did not incorporate BrdU (B, blue arrowhead). (a) High magnification of boxed area in (A) highlights the nuclear and perinuclear pattern of NeuN expression. Note that the prominent process is devoid of staining (white arrowhead), and has two varicosity-like swellings (blue arrows). (b) High magnification of boxed area in (B) shows nuclear BrdU staining (red arrow). (C) Induced MSC cultures not exposed to primary anti-NeuN antibody did not stain positively for NeuN (black arrowhead), despite the presence of cells exhibiting neuronal morphology. Tuc-4: (A) Phase-contrast image illustrates cells with neuronal (arrow and double arrow) or stromal morphology (arrowhead). (B) Double-pass filter image showing colocalization of TUC-4 (red) and BrdU (green) in the same cell (arrow) and TUC-4 expression without BrdU incorporation in a second cell (double arrow). (b) Higher magnifications of TUC-4+/BrdUcell (white double arrow in B). (b) Higher magnification of TUC-4+/BrdUcell (from boxed area in B).

We next investigated the expression of TUC-4, a protein expressed in postmitotic neurons [23]. MSCs were treated with NIM at 0 hours and received a 2-hour BrdU pulse 3.5 hours later. A subpopulation of cells exhibiting neuronal forms were TUC-4+ (arrow and double arrow; Fig. 5A–b′). In contrast, cells with MSC morphologies did not express TUC-4 (arrowhead), as expected. A subpopulation of TUC-4+ morphological neurons incorporated BrdU (arrow), suggesting that these presumptive neurons were mitotic. In contrast, another subpopulation exhibiting similar neuronal morphologies and patterns of TUC-4 expression (double arrow) did not incorporate BrdU, demonstrating that TUC-4 was expressed in both mitotic and nonmitotic presumptive neurons.

Cell Cycle Arrest and Expression of the Neuronal Phenotype

To further explore the relationship of mitosis to neuronal differentiation, we examined the effect of the DNA synthesis inhibitor cytosine arabinoside (Ara-C). At zero time, medium was replaced with NIM containing BrdU (Fig. 6A, panels a–c) or BrdU and Ara-C (Fig. 6A, panels d–f). Twenty-four hours later presumptive neurons were examined immunocytochemically. Ara-C treatment did not inhibit the assumption of neuronal morphologies or tau expression (panels d, e, respectively) but did abolish BrdU incorporation (panel f). Ara-C treatment inhibited the increase in total cell number, as expected (Fig. 6B). Importantly, Ara-C treatment had no significant effect on the percent of putative neurons exhibiting neuronal morphologies and expressing tau (Fig. 6C), indicating that cell division was not required for neuronal differentiation.

Figure Figure 6..

Neuronal differentiation with blockade of mitosis.Cell cycle arrest with Ara-C did not prevent neuronal phenotypic expression. MSCs were treated with NIM+BrdU (A, panels a–c) or NIM+BrdU+Ara-C (A, panels d–f) for 24 hours. Immunocytochemical analysis revealed that 70% of the MSC-derived morphological neurons (panel a, arrow) expressed tau (panel b) and more than 50% of total cells incorporated BrdU (panel c). Induction in the presence of Ara-C yielded morphological neurons (panel d, arrow) that expressed tau (panel e) in the complete absence of BrdU incorporation (panel f). B) Induction in the presence of Ara-C (Ind+Ara-C) resulted in a significant decrease in the total cell number compared with induction in the absence of the inhibitor (Ind) (p < 0.003). Whereas there was a significant difference in the percent of BrdU+cells between MSCs induced in the absence (1=NIM+BrdU) or presence of Ara-C (2=NIM+BrdU+Ara-C) (p < 0.0001), the inhibitor had no significant effect on the percent of tau+morphological neurons in each group (p > 0.9). (B–C) The results presented here are representative of two independent experiments, each of which consisted of three separate dishes for each condition. Cells in 10 random fields/dish were counted and the total number of cells per dish was calculated by multiplying the average number of cells/field by the ratio total area of dish/area of field. Significant differences between conditions were determined by comparing the mean of the total number of cells for each condition using analysis of variance. The results are expressed as percent mean ± SE of cells for each condition.

The Role of Cell Death in Selection of the Neuronal Phenotype

To examine the potential role of programmed cell death in selection of the neuronal phenotype after cessation of mitotic activity, we used the TUNEL assay to identify apoptotic cells 36 and 72 hours after initiation of differentiation. Analysis 36 hours after NIM treatment revealed cells exhibiting neuronal (arrowheads and arrows) or stromal morphologies (double arrow, Fig. 7A). Less than 1% of the total cells were TUNEL positive, suggesting that 99% of the cells were not undergoing apoptosis. Note, for example, the two sample cells with neuronal morphologies (arrows, Fig. 7A) that exhibited neither condensed nor cleaved chromatin (Fig. 7B) and were TUNEL negative (Fig. 7C). A small percentage of cells appeared to be apoptotic (arrowhead), demonstrating condensed chromatin via DAPI staining (Fig. 7B, b′) and intense TUNEL labeling (Fig. 7C, c′). Similar results were obtained 72 hours post induction (data not shown). The minimal incidence of apoptosis suggests that selection of the neuronal phenotype through cell death is not a central regulatory mechanism in this system.

Figure Figure 7..

Differentiation with minimal apoptosis.A) TUNEL staining revealed minimal apoptosis 36 hours after NIM treatment. Cells exhibiting neuronal (black arrows and double arrows) or stromal (black double arrow) morphology were present after 36 hours of NIM treatment. The majority of cells exhibiting neuronal morphologies did not exhibit cleaved DNA by DAPI staining (B, red arrows) and were TUNEL negative (C, red arrows). Less than 1% of total cells showed cleaved DNA by DAPI staining (b, arrowhead) and stained positively for TUNEL (c, arrowhead). Higher magnification images demonstrate apparent cleaved DNA via DAPI staining (b) and TUNEL staining (c) of the cell boxed in (b). Similar results were obtained 72 hours after induction (data not shown).

Expression of Mesenchymal Traits in Differentiating Neurons

We anticipated that commitment to the neuronal lineage would result in progressive downregulation of mesenchymal gene products, and we investigated the expression of fibronectin to test this contention. Forty hours post induction, MSCs demonstrated neuronal morphologies (arrow, arrowhead, Fig. 8A). While some MSC-derived neurons continued to express fibronectin (arrowhead), this mesenchymal marker was downregulated in a subset of cells exhibiting neuronal morphologies (arrow) 40 and 80 hours post induction (data not shown), consistent with the mesodermal to neuroectodermal conversion. In contrast, MSC-derived neurons continued to express NSE and tau for at least 4 days, the longest time examined (data not shown). These findings suggest that regulation of MSC-neuronal differentiation occurs through continued expression of neuronal genes and concomitant downregulation of mesenchymal gene products.

Figure Figure 8..

Commitment to the neuronal lineage: downregulation of mesenchymal fibronectin over time.A) Phase-contrast image of MSC-derived neurons 40 hours after induction. B) None of the cells in this field (blue nuclei are DAPI stained) have incorporated BrdU. C) A population of MSC-derived neurons continues to express fibronectin (arrowhead) while a second subpopulation has downregulated expression of this mesenchymal marker (arrow). Downregulation of fibronectin expression in the absence of cell division (BrdU incorporation) suggests that neuronal differentiation involves the continued expression of neuronal markers and downregulation of mesenchymal gene products in postmitotic cells.


Our overall goal in these studies was to begin elucidating the process of neuronal differentiation of adult MSCs and to define the relationship of stem cell, progenitor, and committed precursor functions in this population. The reigning model of development and differentiation involves the division of multipotent stem cells, yielding progressively more restricted progenitors and precursors committed to specific lineages [24]. Mitosis and subsequent withdrawal from the cell cycle are thought to precede commitment and differentiation in an obligatory fashion. Our observations, in contrast, indicate that neuronal differentiation of MSCs occurs before, during, after, and even in the absence of mitosis. MSCs assume typical neuronal forms, express a panoply of neural genes and downregulate mesenchymal gene expression independent of cell division and BrdU incorporation. Moreover, apoptosis is seen in less than 1% of this population, excluding a major role for programmed cell death and selection of the neuronal phenotype. We tentatively conclude that orthodox distinctions among stem cell, multipotent progenitor, and committed precursor functions may be less rigid than has been assumed.

Mitosis and Neural Differentiation

Our observations indicate that the adult MSCs and presumptive neurons are mitotically active. Cells exhibiting neuronal morphologic characteristics, including small, refractile somata and multiple long processes with terminal expansions, undergo cytokinesis, with the formation of doublets and subsequent preservation of the neuronal traits (Fig. 1A). Mitosis is functional, leading to true proliferation, with a 60% increase in cell number in 24 hours (Fig. 1B). Consequently, expression of neuronal morphological characters does not preclude cell division, consistent with observations of developing sympathetic neurons [25, 26]. Additionally, PC12 cells expressing dysfunctional p53 display mature neurite formation while in S phase, demonstrating that mitotic arrest and functional maturation can be uncoupled in this cell line [27].

Consistent with morphological evidence of differentiation, the putative neurons expressed diverse, specific neuronal gene products while dividing. Presumptive neurons expressed tau and NSE, typical of differentiating neurons, as well as TUC-4 and NeuN thought to be postmitotic neuronal markers, while incorporating the S-phase marker BrdU [Figs. 2–5Figure 5.]. Apparently, expression of multiple neuronal gene products and assumption of neuronal forms does not preclude mitosis. Neuronal traits were expressed before, during, and after mitosis. We tentatively conclude that neuronal differentiation from adult MSCs is not strictly a postmitotic phenomenon. Consequently, this differentiation process does not conform to a simple stem cell model of asymmetric division yielding a committed neural lineage precursor and a self-renewing stem cell [28].

In addition to expression before and during mitosis, neuronal characteristics were expressed in cells that did not incorporate BrdU. Tau, NSE, TUC-4, and NeuN were all detected in cells that developed neuronal morphologies without incorporating the S-phase marker. To independently confirm that cell division is not required for neuronal differentiation, we employed the mitotic inhibitor, Ara-C. Exposure to Ara-C, with blockade of both BrdU incorporation and increases in cell number, did not prevent expression of the neuronal gene products. In sum, elaboration of neuronal forms and neuronal gene expression occurred before and during mitosis, and also in the absence of mitosis.

Neural differentiation was accompanied by the progressive decreased expression of fibronectin, a classical mesenchymal trait (Fig. 8). Decreased fibronectin occurred in the absence of BrdU incorporation. It may be concluded that neural differentiation, including the development of neuronal forms, the expression of multiple neural genes, and the decreased expression of mesenchymal genes, neither requires nor is precluded by mitosis. Rather, mitosis and neural differentiation appear to proceed independently in some populations, potentially suggesting that stem cells, multipotential progenitors, and committed precursors may all differentiate into neurons (see below).

Apoptosis and Neural Differentiation

We assessed the incidence of apoptosis in the differentiating population to determine whether cell selection potentially played a role in the development of the neuronal phenotype. Less than 1% of the total cells were TUNEL positive 36 and 72 hours after initiation of differentiation (Fig. 7). Similarly, we detected no evidence of condensed or cleaved chromatin (Fig. 7). We conclude that cell selection does not play a major role in the emergence of the neuronal phenotype in this MSC population under the conditions employed. Consequently, the neural induction medium apparently exerts instructive actions in neuronal differentiation of these MSCs. This contention carries a number of novel implications for the relationship of stem cell, progenitor, and committed precursor functions in the MSC to neuronal conversion.

Stem Cell, Multipotential Progenitor, and Committed Precursor Function

Our studies indicate that adult MSCs differentiate into cells that assume neuronal forms, express multiple neuronal gene products, and downregulate expression of mesenchymal genes both in the presence and absence of mitosis. These observations may help elucidate the relationship of stem cell, multipotent progenitor, and committed precursor functions in this population [29]. Neuronal differentiation in the absence of DNA replication and mitosis indicates that at least one subpopulation of the differentiating neurons arises directly from functional precursors, not stem cells that require mitosis for differentiation. These observations argue against an obligatory sequence in which stem cells divide to yield multipotential progenitors that subsequently divide and undergo neuronal differentiation. Rather, the neuronal induction medium apparently acts directly on at least some “primed” MSCs to elicit neural differentiation. This contention is consistent with our recent finding that undifferentiated mesenchymal MSCs express genes specific for all embryonic germ layers, including germline, endoderm, mesoderm, and ectoderm [16]. In this model, differentiation predominantly involves the quantitative modulation of already expressed genes in “multidifferentiated” MSCs. Stromal cells in the metanephric mesenchyme are known to express light chain neurofilament protein, a neuroectodermal marker [30]. This finding suggests that the multidifferentiation of pluripotent mesenchymal cells may not be a phenomenon limited to the adult marrow.

In sum, our observations indicate that MSCs can differentiate into presumptive neurons without passing through a stem cell stage requiring mitosis. This contention is consistent with the recently described transdifferentiation of pancreatic cells into liver cells by environmental dexamethasone alone, without intervening cell division or somatic mutation [2].

Nevertheless, other cells incorporated BrdU before, during, and after neuronal differentiation, consistent with orthodox stem cell differentiation. We tentatively conclude that MSC differentiation involves both mitotic and nonmitotic processes and, consequently, both stem cell and precursor functions. While these functions may represent different subpopulations within the MSC pool, they may also indicate diverse potentials of a single population. Moreover, increasing evidence suggests that marrow lymphohematopoietic stem cells reversibly and continually change phenotype, reflecting cell cycle-related chromatin changes that allow differential transcription factor access [3133]. This formulation is consistent with our observations and suggests that multipotential MSCs exhibit both stem cell and precursor functions allowing neuronal differentiation through both mitotic, stem cell, and nonmitotic precursor pathways. Consequently, stem cell, progenitor, and precursor functions may change in a reversible fashion and distinctions may be less rigid than has been assumed.


This work was supported by NICHD P01HD23315, CRPF IBC9901, Albert Zofchak Grant 181-01R, and the New Jersey Commission on Spinal Cord Research Grant 01-3009-SCR-S-0.

We thank Emanuel DiCicco-Bloom, Richard S. Nowakowski, and Laura B. Ngwenya for helpful discussions, and the members of the Cheryl Dreyfus, Emmanuel DiCicco-Bloom, and Ira Black labs for technical assistance.