Neuroectodermal differentiation from mouse multipotent adult progenitor cells

Abstract

We recently showed that a rare cell from murine bone marrow, which we termed multipotent adult progenitor cells (MAPCs), can be expanded for >120 population doublings. Mouse (m)MAPCs differentiate into mesenchymal lineage cells as well as endothelium and endoderm, and, when injected in the blastocyst, mMAPCs contribute to most if not all somatic cell lineages including the different cell types of the brain. Our results, reported herein, demonstrate that mMAPCs can also be induced to differentiate into cells having anatomical and electrophysiological characteristics similar to those of midbrain neurons. Differentiation to a neuronal phenotype was achieved by coculturing mMAPCs with astrocytes, suggesting that neuronal differentiation may require astrocyte-derived factors similar to what is required for the differentiation of embryonic stem cells and neural stem cells to neurons. Differentiation of mMAPCs to neuron-like cells follows similar developmental steps as described for embryonic stem cells and neural stem cells. MAPCs therefore may constitute a source of cells for treatment of central nervous system disorders.

Until recently it was thought that tissue-specific stem cells could only differentiate into cells of the tissue of origin. However, several recent studies suggest that tissue-specific stem cells may be able to differentiate into cells of different tissues. For instance, cells infused at the time of bone marrow (BM) transplantation contribute to skeletal muscle myoblasts (1, 2) and endothelium (3-7) acquire properties of hepatic (8-10), biliary duct (8-10), lung, gut, and skin epithelia (11) as well as neuroectoderm (12, 13). When injected in the heart, BM cells acquire a cardiac myoblast phenotype (6). Neural stem cells (NSCs) may repopulate the hematopoietic system (14, 15), and muscle cells may differentiate into hematopoietic cells (16, 17). What mechanism(s) is responsible for this apparent plasticity is not known.

We recently showed that a rare cell, which we termed the multipotent adult progenitor cell (MAPC), within mesenchymal stem cell cultures from rodent BM can be expanded for >120 population doublings (18-20). This cell differentiates not only into mesenchymal lineage cells but also endothelium and endoderm. We also showed that mouse (m)MAPCs injected in the blastocyst contribute, similar to embryonic stem (ES) cells, to most if not all somatic cell lineages including the brain. Within the brain, region-specific appropriate differentiation occurred (21). We show here that similar to mES cells, mMAPCs can also be induced to differentiate in vitro into cells with biochemical, anatomical, and electrophysiological characteristics of midbrain neuronal cells.

Methods

Generation of MAPCs. BM was collected from femurs of 3- to 4-week-old 129 × C57BL/6J ROSA26 mice according to guidelines from the University of Minnesota Institutional Animal Care and Use Committee. MAPCs were generated as described (20). To demonstrate that cells were MAPCs, they were induced to differentiate to endothelium and hepatocyte-like cells as described (18, 20). In addition, we showed that these cell populations contributed to most somatic cells of the mouse after blastocyst injection (20).

Neuroectodermal Differentiation of MAPCs. Base medium consisted of 60% DMEM-low glucose (GIBCO/BRL), 40% MCDB-201 (Sigma) with 1× insulin-transferrin-selenium, 1× linoleic acid BSA, 10^-9 M dexamethasone (Sigma), and 10^-4 M ascorbic acid 2-phosphate (Sigma), 100 units of penicillin, and 1,000 units of streptomycin (GIBCO) on fibronectin (Sigma). In some instances we also added N2 supplement (GIBCO). Cytokines added included 100 ng/ml basic fibroblast growth factor (bFGF), 100 ng/ml Sonic Hedgehog (SHH), 10 ng/ml FGF8, and 10 ng/ml brain-derived neurotrophic factor (BDNF) (all from R & D Systems). All experiments were repeated at least three times by using different subpopulations from a single MAPC cell line. MAPCs had been expanded for 50-70 population doublings before use in neuroectodermal differentiation.

Astrocyte Preparation. Mouse brain was dissected from embryonic day (E)16 fetuses in Hanks' balanced salt solution (HBSS, Sigma). The dissected brain was minced and incubated in 0.125% trypsin/0.05% DNase (Sigma) in HBSS at 37°C for 20 min. The tissue was triturated with a pipette and dissociated to a mixture of single cells and small cellular aggregates. After passing through a 70-μm nylon mesh, the cells were centrifuged at 300 × g for 5 min and resuspended in DMEM plus 10% FBS (HyClone). Cells were plated onto culture dishes, precoated with poly-d-lysine (100 μg/ml, Sigma) at 4°C overnight, at a density of 600,000 cells per cm² until confluent.

Astrocyte-Conditioned Medium. After astrocytes had been cultured in DMEM plus 10% FBS for 8 days, culture medium was switched to serum-free medium supplemented with N2 supplement. Three days later, the medium was collected as astrocyte-conditioned medium.

Coculture of MAPC-Derived Neurons with Fetal Brain Astrocytes. Glass coverslips were coated with 500 μg/ml poly-d-lysine overnight at 4°C. Fetal brain astrocytes that have been cultured for 8 days were trypsinized, replated on glass coverslips, and allowed to grow to confluency. Once confluent, the coverslips were evaluated for the presence of neurons by staining with antibodies against neurofilament (NF)-200 (see methods below) or placed upside down in cultures of MAPC-derived neuron-like cells. Cultures were maintained in serum-free medium supplemented with N2 supplement and without additional cytokines for 5-12 days.

Lentivirus Transduction. mMAPCs were seeded at 10⁵ cells per well in six-well tissue-culture plates. One milliliter of MAPC culture medium, 1 ml of supernatant of the 293 cell line transfected with a third generation VSV-g-pseudotyped eGFP lentivirus (10⁷ infectious particles per ml) (a kind gift from Thierry Vandendriesche, Katholieke Universiteit Leuven, Leuven, Belgium) and polybrene (8 μg/ml final concentration), was added to the wells. After 6 h of incubation at 37°C and 5% CO₂, the medium was replaced with fresh MAPC medium. Transduction of MAPCs was repeated three times. Transduction efficiency of the final population was 28% as determined by counting 200 cells.

Quantitative RT-PCR for Neuronal Transcription Factors and Genes. RNA was extracted from MAPCs, MAPCs differentiated for 5, 7, 10, 14, and 21 days, and brain from E18 or adult mice by using the RNeasy kit (Qiagen, Valencia, CA). Contaminating DNA was eliminated by two sequential steps of DNase (Invitrogen) treatment. mRNA was reverse-transcribed, and cDNA underwent 40 rounds of amplification (ABI PRISM 7700, Perkin-Elmer/Applied Biosystems) with reaction conditions of 40 cycles of a two-step PCR (95°C for 15 min and 60°C for 60 min) after initial denaturation (95°C for 10 min) with 2 μl of DNA solution and 1× SYBR green PCR master mix reaction buffer (Applied Biosystems). Controls consisted of amplifications without reverse transcription and reactions without addition of cDNA template. The authenticity and size of PCR products were confirmed by melting curve analysis (using software provided by Perkin-Elmer) and gel analysis. Primers used and the size of expected products are shown in Table 1. mRNA levels were normalized by using GAPDH as housekeeping gene and compared with levels in E18 or adult mouse brain.

Table 1. Primers used for quantitative RT-PCR.

Gene	Forward	Reverse	Size, bp
Sox-1	AAGATGCACAACTCGGAGATCAG	TGTAATCCGGGTGTTCCTTCAT	51
Otx-2	CCATGACCTATACTCAGGCTTCAGG	GAAGCTCCATATCCCTGGGTGGAAAG	211
Otx-1	AGGCGCTGTTCGCAAAGA	CCTCCTCGCGCATGAAGAT	50
Pax-2	CCAGGCATCAGAGCACATCA	CGTCTGTGTGCCTGACACATT	141
Pax-5	AAACGCAAGAGGGATGAAGGT	AACAGGTCTCCCCGCATCT	100
Ptx-3	TGTGTGGCACCTGGAGTTCA	CACCCTCAGGAACAGAGTGACTT	107
cRet	GAGGAAATGTACCGTCTGATGCT	TCTTGACCATCATCTTCTCCAGATC	102
Nurr-1	TGAAGAGAGCGGAGAAGGAGATC	TCTGGAGTTAAGAAATCGGAGCTG	255
Nestin	GAGAAGACAGTGAGGCAGATGAGTTA	GCCTCTGTTCTCCAGCTTGCT	113
GFAP	GAGGAGTGGTATCGGTCTAAGTTTG	GCCGCTCTAGGGACTCGTT	165
MBP	GTGCAGCTTGTTCGACTCCG	ATGCTCTCTGGCTCCTTGGC	153
GABA	AGGTTGACCGTGAGAGCTGAAT	TGGGCAGGCATGGGC	68
DAT	GCAATCATCACCACCTCCATTA	ATGGGCACATTGTGCTTCTG	100
TH	AGTTCTCCCAGGACATTGGACTT	ACACAGCCCAAACTCCACAGT	100
TrH	GGATGGAGTCTGATGTCACCAA	TGACGTTTCTCAGGCATTAAGC	120
DBH	TTCCAATGTGCAGCTGAGTC	GGTGCACTTGCTTGTGCAGT	242

Immunophenotypic Analysis. Cells were fixed with 4% paraformaldehyde (Sigma) for 4 min at room temperature followed by methanol (Sigma) for 2 min at -20°C. For nuclear ligands, cells were permeabilized with 0.1 M Triton X-100 (Sigma) for 10 min. Slides were incubated sequentially for 30 min each with primary antibody and fluorescein or Cy3- or Cy5-coupled anti-mouse-, goat-, or rabbit-IgG antibodies. Between each step, slides were washed with PBS plus 1% BSA (Sigma). Cells were examined by confocal fluorescence microscopy (Confocal 1024 microscope, Olympus AX70, Olympus Optical, Tokyo). To assess the frequency of different cell types in a given culture, we counted the number of cells staining positive with a given antibody in four visual fields (50-200 cells per field).

Antibodies against myelin basic protein (MBP, 1:20), NF-200 (1:400), microtubule-associated protein (1:400), tyrosine hydroxylase (TH, 1:1,000), dopa-decarboxylase (DDC, 1:100), tryptophan hydroxylase (TrH, 1:250), γ-aminobutyric acid (GABA, 1:500), control mouse, rabbit, and rat IgGs, and FITC- and Cy3-labeled secondary antibodies were from Sigma. Antibodies against Nestin (1:150) and Nurr1 (1:250) were from BD Transduction Laboratories (Lexington, KY). Antibodies against glial fibrillary acidic protein (GFAP, 1:400) were from DAKO or Santa Cruz Biotechnology. Anti-dopamine antibodies (1:2,000) were from Abcam (Cambridge, U.K.). Polyclonal antibodies against Tau (1:400) were from Santa Cruz Biotechnology, and Cy5-labeled secondary antibodies were from Chemicon.

Electrophysiology. Standard whole-cell patch-clamp recording methodologies were used to examine the physiological properties of cultured BM stem cells. Voltage-clamp and current-clamp recordings were obtained by using a Dagan 3900A patch-clamp amplifier (Dagan, Minneapolis), which was retrofitted with a Dagan 3911 expander unit. Patch pipettes, made from borosilicate glass, were pulled on a Narishige pipette puller (model PP-83). The pipettes were filled with an intracellular saline that consisted of 142.0 mM KF, 7.0 mM Na₂SO₄, 3.0 mM MgSO₄, 1.0 mM CaCl₂, 5.0 mM Hepes, 11.0 mM EGTA, 1.0 mM glutathione, 2.0 mM glucose, 1.0 mM ATP (magnesium salt), and 0.5 mM GTP (sodium salt) (Sigma). For most recordings, the fluorescent dye 5,6-carboxyfluorescein (0.5 mM, Eastman Kodak) was also added to the pipette solution to visually confirm by using fluorescence microscopy that the whole-cell patch-recording configuration had been achieved. Pipette resistances ranged from 11 to 24 MΩ. The standard extracellular recording saline was comprised of 155 mM NaCl, 5.0 mM KCl, 1.8 mM CaCl₂, 1.0 mM MgCl₂, 10 mM Hepes, and 5 mM glucose (Sigma). For some experiments, 1 μM tetrodotoxin (TTX) was added to the standard control solution. The pH of the intracellular and extracellular recording solutions was adjusted to 7.4 and 7.8, respectively. Unless otherwise noted, all chemical compounds were obtained through Sigma. PCLAMP 8.0 (Axon Instruments, Foster City, CA) was used to run experiments and collect and store data. The data presented herein were corrected for an 8.4-mV liquid junctional potential, which was calculated by using the JPCALC software package (22). Off-line analyses and graphical representations of the data were constructed by using CLAMPFIT 8.0 (Axon Instruments) and PRISM (GraphPad, San Diego).

Results

Undifferentiated mMAPCs Did Not Have Neuroectodermal Characteristics. No staining was seen with antibodies against nestin, GFAP, NF-200, MBP, or neurotransmitters (data not shown). By quantitative RT-PCR, mMAPCs did express low levels of c-Ret, Otx-2, and nestin mRNA but no mRNA for sox1, otx2, Pax2, Pax5, Nurr1, GFAP, MBP, dopamine, TH, GABA, TrH, or DBH (Table 2).

Table 2. mRNA levels on days 5, 7, 10, and 14 of differentiation of mMAPCs to neuroectoderm.

Gene	Day 0	Day 5	Day 7	Day 10	Day 14
Sox-1*	0	0.57; 1.79; 2.22	0.58; 1.15; 2.14	0.68; 0.59; 5.58	0.9; ND; 1.74
Otx-2*	0	0.11; 0.25; 3.4	0.21; 0.06; 4.2	0.12; 0.08; 2.4	ND; ND; 3.5
Otx-1*	0.01	0.33; 1.18; 7.06	0.36; 0.69; 32.1	0.26; 0.33; 24	1.67; ND; 5.31
Pax-2*	0	6.0; 8.2; 3.07	4.47; 7.06; 5.43	2.39; 1.64; 2.05	5.5; ND; 4.99
Pax-5*	0	0.11; 0.14; 0.9	0.13; 0.16; 3.75	0.05; 0.03; 1.84	0.48; ND; 1.33
En-1*	4.87	0.59; 1.45; 4.36	0.2; 0.8; 11.4	0.15; 0.32; 8	0.4; ND; 1.08
cRet†	0.14	2.84; 5.46; 15.7	2.19; 5.08; 72.8	2.53; 4.47; 54.2	24, 17; ND; 36.9
Nurr-1†	0	0.55; 2.4; 0.38	1.12; 1.47; 1.11	1.41; 3.59; 1.6	ND; ND; 1.68
Nestin*	0.52	27.3; 98.0; 50.4	10.2; 88.6; 70.8	5.9; 12.2; 243	6.48; ND; 14.7
GFAP†	0	0.56; 0.52; 1.05	0.36; 0.52; 3.26	1.32; 5.76; 21.0	11.75; ND; 9.19
MBP†	0	0.004; 0.006; 0.013	0.003; 0.008; 0.006	1.7; 4.08; 1.8	1.8; ND; 2.33
GABA†	0	1.36; 1.89; 6.23	0.99; 1.74; 17.5	7.8; 19.77; 94.0	61.8; ND; 69.6
DAT†	0	1.58; 4.39; 19.9	0.5; 2.83; 22.71	1.81; 5.66; 119	13.4; ND; 23.67
TH†	0	0.67; 1.31; 0.36	0.5; 1.2; 2.86	1.27; 3.94; 2.54	1.7; ND; 2.29
TrH†	0	0.49; 1.6; 0.46	0.51; 1.04; 4.99	1.56; 3.56; 7.14	3.92; ND; 3.81
DBH*	0	0	0	0	0

RNA was extracted from MAPCs, MAPCs differentiated for 5, 7, 10, and 14 days, and brain from E18 or adult mice. Contaminating DNA was eliminated by two sequential steps of DNAse treatment, mRNA was reverse-transcribed, and cDNA was amplified for 40 rounds of amplification with SYBR green PCR master mix. Controls consisted of amplifications without reverse transcription and reactions without addition of cDNA template. mRNA levels were normalized by using GAPDH as housekeeping gene and compared with levels in E18 fetal brain or adult mouse brain.

^*Relative abundance in cultured cells compared with fetal brain.

^†Relative abundance in cultured cells compared with adult brain.

Culture Conditions for Neuroectodermal Differentiation from MAPCs. Because neuroprogenitors can be expanded with platelet-derived growth factor (PDGF)-BB and induced to differentiate by the removal of PDGF and addition of bFGF (23), we initially replated undifferentiated MAPCs at 10,000 cells per cm² on fibronectin-coated plates or chamberslides, removed epidermal growth factor, PDGF, and leukemia inhibitory factor, and added 100 ng/ml bFGF. Quantitative RT-PCR of mMAPCs treated with bFGF for 5 and 7 days demonstrated acquisition of neuroectodermal transcripts. On days 5 and 7, mRNA for sox1, otx2, otx1, pax2, pax5, and Nurr1 could be detected at levels between 0.1- and 7-fold those seen in fetal brain (three independent experiments) (Table 2). By days 5 and 7, nestin mRNA levels increased to between 7- and 100-fold those in fetal brain (Table 2). Immunohistochemistry showed that by day 5, cells started to express nestin protein. By day 7, 65 ± 11% of cells stained positive for nestin; 23 ± 8% of nestin-positive cells also expressed Nurr1 (a representative example is shown in Fig. 1; see also Table 3). By day 10, 62 ± 7% of cells expressed NF-200, 15 ± 5% GFAP, and 11 ± 3% MBP (a representative example is shown in Fig. 1), consistent with the finding that mRNA for GFAP and MBP increased to 1- to 4-fold over that detected in fetal brain. Double immunohistochemistry showed that GFAP, MBP, and NF-200 were never detected in the same cells. When cultures were maintained in the presence of bFGF alone, cells started dying by days 10-14.

Fig. 1.

mMAPCs were cultured sequentially for 7 days with 100 ng/ml bFGF, 10 ng/ml FGF8 and 100 ng/ml SHH, and 10 ng/ml BDNF on fibronectin-coated chamberslides. After 7, 10, and 21 days, cells were fixed and stained with antibodies against nestin and Nurr1 followed by secondary Cy5- and Cy3-coupled antibodies, respectively (d7); NF-200 and GFAP followed by secondary Cy3- and Cy5-coupled antibodies, respectively (1) and NF-200 and MBP followed by secondary Cy3- and Cy5-coupled antibodies, respectively (2) (d10); and GABA and DDC followed by secondary Cy5- and Cy3-coupled antibodies, respectively (1), TrH and TH followed by secondary Cy5- and Cy3-coupled antibodies, respectively (2), and microtubule-associated protein and Tau followed by secondary Cy3- and Cy5-coupled antibodies, respectively (3) (d21).

Table 3. Percentage of cells expressing neuronal markers during differentiation.

	Day 7	Day 10	Day 21	Day 30
Nestin	65 ± 11	NA	NA	NA
Nurr1	23 ± 8	NA	NA	NA
NF200	0	62 ± 7	90 ± 5	92 ± 6
GFAP	0	15 ± 5	3 ± 2	2 ± 2
MBP	0	11 ± 3	2 ± 2	2 ± 3
DDC/TH/Dopa	0	0	25 ± 7	22 ± 8
TrH	0	0	18 ± 3	23 ± 2
GABA	0	0	52 ± 5	52 ± 3

Cells were fixed with 4% paraformaldehyde (Sigma) for 4 min at room temperature followed by methanol (Sigma) for 2 min at − 20°C. For nuclear ligands, cells were permeabilized with 0.1 M Triton X-100 (Sigma) for 10 min. Slides were incubated sequentially for 30 min each with primary antibody and FITC or Cy3- or Cy5-coupled anti-mouse-, goat-, or rabbit-IgG antibodies. Between each step, slides were washed with PBS plus 1% BSA (Sigma). Cells were examined by confocal fluorescence microscopy (confocal 1024 microscope, Olympus AX70). To assess the frequency of different cell types in a given culture, we counted the number of cells staining positive with a given antibody in four visual fields (50-200 cells per field). Results shown are mean ± SEM from three independent differentiations from MAPCs to neuroectodermal cells evaluated after 7-30 days.

When mMAPCs were cultured sequentially with 100 ng/ml bFGF for 7 days followed by a combination of 10 ng/ml FGF8 and 100 ng/ml SHH for 7 days and finally 10 ng/ml BDNF for 7 days, the latter in medium also supplemented with N2 medium, a more mature phenotype was seen. Quantitative RT-PCR demonstrated that by days 10 and 14, levels of GABA, dopamine, and TH, and TrH mRNA increased between 1.7- and 120-fold. Immunophenotypic analysis on day 21 showed that 25 ± 7% of cells expressed markers of dopaminergic neurons (shown are DDC and TH; also dopamine), 18 ± 3% expressed markers of serotonergic (TrH) neurons, and 52 ± 5% expressed markers of GABA-ergic (GABA) neurons (a representative example is shown in Fig. 1; see also Table 3). Double immunohistochemistry showed that GABA, TrH, and TH (or DCC or dopamine) were never detected in the same cells. Neuron-like cells became polarized, because Tau and microtubule-associated protein were expressed in axonal and somatodendritic compartments, respectively. Fewer than 10% of cells stained positive for astrocytes or oligodendrocyte markers (data not shown). Consistent with this immunohistology, levels of MBP and GFAP mRNA decreased by day 21 (data not shown). A significant proportion of cells died when maintained in BDNF for >7 days.

Based on studies by Wagner et al. (24) and Song et al. (25), we next tested whether cultured neuron-like cells could be maintained in vitro for more extended periods of time to allow further maturation when cultured in the presence of fetal brain astrocytes. Astrocytes were cultured from E16 fetal brain in 10% FCS. After several passages, no neural cells could be detected by immunofluorescence microscopy (data not shown). In initial studies, astrocyte-conditioned medium was added to the developing neuroectodermal cells generated from MAPCs. However, no significant further morphologic maturation was observed (data not shown). In subsequent experiments, astrocytes were plated onto coverslips and allowed to grow to confluence. Coverslips were placed upside down in chamberslides in which enhanced GFP (eGFP)-transduced MAPCs had been cultured for 7 days with bFGF, 7 days with FGF8 plus SHH, and 7 days with BDNF in N2 medium. After an additional 5-12 days in culture, eGFP-expressing MAPC-derived neuron-like cells again were evaluated by immunofluorescence, and we demonstrated that eGFP-positive cells continued to express markers of dopaminergic neurons (≈25% TH and dopamine), serotonergic neurons (≈25% TrH), and GABA-ergic neurons (≈50% GABA) and acquired a much more mature neural morphology with a more elaborate array of axons (a representative example is shown in Fig. 2).

Fig. 2.

eGFP-transduced mMAPCs (28% transduction efficiency) were cultured on fibronectin-coated chamberslides sequentially for 7 days with 100 ng/ml bFGF, 10 ng/ml FGF8 and 100 ng/ml SHH, 10 ng/ml BDNF, and finally with E16 fetal mouse brain astrocytes plated on coverslips that were placed upside down in the chamberslides. After a total of 28 days, cells were fixed and stained. Slides were analyzed for the presence of GFP-positive cells and cells costaining with Cy3- or Cy5-labeled antibodies. (A) Cells labeled with antibodies against GABA and DDC. (A1-A3) Single fluorescence color analysis of cells stained with antibodies against GABA followed by secondary Cy3-coupled antibody, eGFP-labeled cells, and cells stained with antibodies against DDC followed by secondary Cy5-coupled antibody, respectively. (A4-A6) Overlay pictures of GFP/anti-GABA-Cy3, anti-GABA Cy3/anti-DDC-Cy5, and GFP/anti-DDC-Cy5, respectively. Shown is that GFP-positive cells acquired morphological and phenotypic features of GABA-ergic and dopaminergic neurons, whereas a fraction of cells with morphological and phenotypic features of GABA-ergic and dopaminergic neurons was GFP-negative. (B) Cells labeled with antibodies against TrH and dopamine. (B1-B3) Single fluorescence color analysis of cells stained with antibodies against TrH followed by secondary Cy3-coupled antibody, eGFP-labeled cells, and cells stained with antibodies against dopamine followed by secondary Cy5-coupled antibody, respectively. (B4-B6) Overlay pictures of GFP/anti-TrH-Cy3, anti-TrH-Cy3/anti-dopamine-Cy5, and GFP/anti-dopamine-Cy5, respectively. Shown is that GFP-positive cells acquired morphological and phenotypic features of serotonergic and dopaminergic neurons, whereas a fraction of cells with morphological and phenotypic features of serotonergic and dopaminergic neurons was GFP-negative.

MAPC-Derived Neuron-Like Cells Acquire Functional Voltage-Gated Sodium Channels. Patch-clamp recordings were obtained from 50 MAPCs from five independent cultures. Recordings were made from cells that were cultured for 7 days each with bFGF, FGF8b and SHH, and BDNF followed by either coculture with astrocytes for 5 days (9 neuron-like cells were tested from a single experiment), 7 days (9 neuron-like cells were tested from a single experiment), 8 days (14 neuron-like cells were tested from a single experiment), 9 days (9 neuron-like cells were tested from a single experiment), and 12 days (2 neuron-like cells were tested from a single experiment) or cells that were incubated for 7 days with conditioned medium from cultures of fetal brain astrocytes (7 neuron-like cells were tested from a single experiment). At all time points, the resting membrane potentials (RMP) of cells cocultured with astrocytes were variable, ranging between -8.4 and -55.4 mV. However, RMPs tended to become more negative as a function of time in culture with astrocytes. The median RMPs were -27.4, -33.7, -41.9, and -44.4 mV after 5, 7, 8, and 9 days, respectively, in coculture. Input resistance also varied considerably across cells (range = 0.133-9.8 GΩ); however, no trend was apparent in the value of input resistance as a function of time the cells spent in culture with astrocytes (median input resistance = 2.4, 1.6, 2.4, and 1.2 GΩ after 5, 7, 8, and 9 days in culture with astrocytes, respectively). Current-clamp recordings demonstrated that spiking was observed in cells that were cocultured with astrocytes at all time points examined. Fig. 3A illustrates an example of spiking behavior evoked from a cell that had been cocultured with astrocytes for 8 days. Interestingly, the proportion of cells studied that were capable of generating action potentials increased dramatically after day 5 in culture with astrocytes. Twenty-two percent of cells that were cocultured with astrocytes for only 5 days spiked. In contrast, after day 5, spiking cells represented between 71% and 100% of cells studied at each time period. Voltage-clamp experiments showed that spiking MAPCs expressed a rapidly inactivating inward current. This transient inward current generally could not be elicited from cells, which failed to produce a spike in current-clamp experiments (a small transient inward current was observed in only one of the nonspiking MAPCs). Fig. 3B shows the inward sodium current that was elicited from the same cells as in Fig. 3A. Approximately 67% of spiking cells could be made to spike repetitively in response to depolarizing current-injection steps; the other 33% of cells generated only a single action potential with varying amounts of depolarizing current stimulation. Where examined, the spiking behavior and transient inward currents were blocked by TTX (see Fig. 3 A and B, TTX). All cells examined that were cocultured with astrocytes had outward currents; however, the identities of those currents remain to be determined. Voltage and current traces from our patch-clamp recordings also suggested the occurrence of synaptic events (Fig. 3C, see arrows).

Fig. 3.

Spiking behavior and voltage-gated currents from MAPCs in coculture with fetal mouse brain astrocytes. (A) Current-clamp recordings from a MAPC that had been cocultured with astrocytes for 8 days. Illustrated in the bottom three panels are the voltage responses elicited by the current-injection protocol shown (a 17-pA current-injection step, Top). The repetitive spiking recorded in this cell was blocked reversibly by TTX. The current-injection protocol reports the current injected relative to a negative DC current that was injected into the cell to “hold” it near -100 to -130 mV. (B) Voltage-clamp recordings of leak-subtracted currents from the same cell shown in A. (Top) The voltage-clamp protocol used to elicit the families of currents shown in the bottom three panels. A large transient inward current was evident that could be blocked reversibly by TTX. (C) Current-clamp records obtained from a MAPC that had been in culture with astrocytes for 8 days. In this example, the cell produced only one spike in response to depolarizing current injections (Δ pA = 7). The arrows point to possible synaptic potentials.

Spikes could not be elicited from cells that had been treated with conditioned medium for 7 days. We also failed to observe sodium currents from cells in this treatment condition. Median resting membrane potential and input resistance were -20.9 mV and 2.25 GΩ, respectively. However, similar to cells that were cocultured with astrocytes, most cells (6/7 cells) had outward currents.

Discussion

We demonstrate here that stem cells obtained from BM can be induced to differentiate into cells with morphological and phenotypical characteristics of central nervous system neurons. Differentiation of marrow pluripotent stem cells to neuron-like cells follows similar developmental steps as described for ES cells and NSCs. Similar to neurons derived from NSCs or ES cells in vitro, in vitro-differentiated MAPCs also exhibit electrophysiological characteristics of neurons.

A large body of work exists in which differentiation of adult or fetal NSCs and mES cells to neurons, and specifically dopaminergic neurons, has been evaluated ex vivo and during embryonal life. The sox1 homeobox gene is expressed early and exclusively during neuroectoderm commitment. The Otx homeobox genes (Otx1 and Otx2) are widely expressed at early stages of neuroectoderm differentiation (26). Otx2 is expressed throughout the epiblast and subsequently restricted to the anterior neuroectoderm, where it is required for development of the forebrain and midbrain. Otx1 is first expressed in the dorsal telencephalon, and interactions between Otx1 and Otx2 are thought to specify the development of the midbrain. NSCs can be defined by the presence of the intermediate filament protein nestin (27, 28), which is also found in young neurons, reactive glial cells, and ependymal cells (29). Neuroprogenitors then acquire specific neurotransmitter fates through the action of cytokines. For instance, differentiation of dopaminergic and serotoninergic neurons in the midbrain and hindbrain is controlled by SHH and FGF8b, which leads to the selective activation of the transcription factor Nurr1 (30-32). In addition, differentiation requires activation of several other transcription factors including Pax2 and Pax5 (24, 33-36). Lee et al. (37) showed that generation of neuroprogenitors, and subsequently dopaminergic neurons from mES cells, occurs through sequential activation of these same transcription factors in response to the same external stimuli as have been identified for embryonal development and in vitro differentiation of NSCs.

Fetal- or adult brain-derived NSCs and neuroprogenitors are expanded with the mitogens epidermal growth factor and bFGF, and differentiation to neurons and glial cells occurs when bFGF or epidermal growth factor is withdrawn (38-40). Other studies have used PDGF to expand neuroprogenitors and induced differentiation by the removal of PDGF and addition of bFGF (23). Rodent or human neuroprogenitors can also be expanded with epidermal growth factor alone, while undergoing initial differentiation to astrocytes and oligodendrocytes but less neuronal cells, or bFGF alone, while undergoing initial differentiation to neurons and to a lesser extent glia (41). We demonstrate here that differentiation of mMAPCs to cells with neuroectodermal characteristics occurs by initial culture in the presence of bFGF as the sole cytokine, and that this is associated with the activation of transcription factors known to be important in neural commitment in vivo and differentiation from NSCs and mES cells in vitro. However, differentiation to mature neuron-like cells was not seen when bFGF was used as the sole differentiation agent.

After sequential stimulation with bFGF, FGF8b and SHH, and BDNF (42), a neurotrophin known to support survival of dopaminergic neurons in vitro and in vivo (43, 44), development of cells with a neuroectodermal morphology and staining pattern could be seen. Such neuron-like cells became polarized. As has been described in most studies in which ES cells or NSCs were differentiated in vitro to a midbrain neuroectodermal fate, ≈25% of cells stained positive for dopaminergic markers, 25% for serotonergic markers, and 50% for GABA-ergic markers (24, 25, 37, 45-47). Despite the addition of BDNF, neuron-like cells could not be maintained beyond 21 days in culture. Neuron-like cells expressed glutamate as well as glutamate receptors (data not shown). Because few if any astrocytes (which bind and remove glutamate) could be detected in these cultures, we speculate that cell death may be caused by glutamate toxicity.

A number of studies have found that terminal neuronal differentiation requires yet-to-be-characterized factors secreted by region-specific glial cells. For instance, Wagner et al. (24) found that coculture of Nurr1 neurons with type II astrocytes from primary E16 rat ventral mesencephalon, the age and region where endogenous neurons of the substantia nigra have just been born, yielded a significant numbers of functioning dopaminergic neurons. Panchision et al. (48) created a type II astrocyte line that supports terminal differentiation of dopaminergic neurons. Song et al. (25) demonstrated that neural differentiation in vitro occurred when NSCs were cocultured with brain-derived astrocytes, and that the type of neuron generated was dictated by the brain-specific region from which astrocytes were derived. We tested whether the addition of astrocyte-conditioned medium or astrocytes themselves would induce further maturation and prolonged survival of the mMAPC-derived neuron-like cells. Astrocytes were obtained from whole fetal brain at E16. The addition of astrocyte-conditioned medium did not affect the survival or maturation of mMAPC-derived neuron-like cells significantly. However, coculture with fetal brain astrocytes resulted in prolonged survival and further morphological maturation of the neuron-like cells. These findings are consistent with the findings from Wagner et al. (24) and may suggest that the effects exerted by astrocytes may be as simple as the removal of glutamate from the culture, as discussed earlier, leading to prolonged survival of neuronal cells and allowing further maturation. Alternatively, a direct cell-cell interaction event may be needed, or factors responsible for the neuronal maturation are unstable and therefore only present in low concentrations when astrocyte-conditioned medium is used rather than astrocytes themselves.

Aside from further morphological maturation observed after coculture of neuron-like cells with fetal brain astrocytes, we also observed acquisition of electrophysiological characteristics consistent with neurons. Occurrence of spiking behavior that can be attributed to voltage-gated sodium channels was found in 22% of cell cocultured for 5 days with fetal brain astrocytes and between 80% and 100% of cells cocultured with astrocytes for 7-12 days but not in cells exposed to astrocyte-conditioned medium. Current traces from our patch-clamp recordings also suggested the occurrence of synaptic events.

The mechanism underlying the finding that cells from BM can differentiate into neuron- and glia-like cells remains unknown. Several possible explanations have been suggested, including the possibility that multiple stem cells exist in postnatal tissues (17, 49). However, we have shown previously that single MAPCs differentiate in vitro and in vivo in multiple differentiated cells from mesoderm, endoderm, and ectoderm (20). A second possibility is that cells of one type (BM cells, for example) fuse with a second cell type, and the fused cell, which is mostly hyperdiploid, acquires the characteristics of the second cell type (50, 51). However, we show here that neuronal characteristics are acquired without coculture with brain-derived cells even though full maturation to neurons with electrophysiological characteristics requires coculture with fetal brain astrocytes. Whether dedifferentiation or reprogramming of a mesenchymal stem cell to a more pluripotent cell capable of differentiating to cells outside the mesoderm or whether a more pluripotent stem cell persists even after birth are questions that are unanswered at this time.

In conclusion, we demonstrate here that BM-derived MAPCs can be induced to differentiate to cells with biochemical, morphological, and electrophysiological characteristics of midbrain dopaminergic, serotonergic, and GABA-ergic neurons. If future studies demonstrate that such cells can engraft in vivo, they may be an excellent source of cells for treatment of neurodegenerative disorders.