Neurosphereforming cell (NSFC) lines have been established from cultures of human adult olfactory neuroepithelium. Few of these cells ever express mature neuronal or glial markers in minimal essential medium supplemented with 10% fetal bovine serum or defined medium. However, these neural progenitors have the potential to differentiate along glial or neuronal lineages. To evaluate the potential of NSFCs to form motoneurons, transcription factors Olig2, Ngn2, and HB9 were introduced into NSFCs to determine if their expression is sufficient for motoneuron specification and differentiation, as has been shown in the early development of the avian and murine central nervous systems in vivo. NSFCs transfected with Olig2, Ngn2, and HB9 alone exhibited no phenotypic lineage restriction. In contrast, simultaneous transfection of Ngn2 and HB9 cDNA increased the expression of Isl1/2, a motoneuron marker, when the cells were maintained in medium supplemented with retinoic acid, forskolin, and sonic hedgehog. Furthermore, a population of Olig2-expressing NSFCs also expressed Ngn2. Cotransfection of NSFCs with Olig2 and HB9, but not Olig2 and Ngn2, increased Isl1/2 expression. Coculture of NSFCs trans-fected with Ngn2-HB92 or Olig2 and HB9 with purified chicken skeletal muscle demonstrated frequent contacts that resembled neuromuscular junctions. These studies demonstrate that transcription factors governing the early development of chick and mouse motoneuron formation are able to drive human adult olfactory neuroepithelial progenitors to differentiate into motoneurons in vitro. Our long-term goal is to develop cell populations for future studies of the therapeutic utility of these olfactory-derived NSFCs for autologous cell replacement strategies for central nervous system trauma and neurodegenerative diseases.
Olfactory neuroepithelium contains neural stem cells, which can regenerate and give rise to olfactory receptor neurons and supporting cells throughout life [1– 4]. Stem cells from the olfactory neuroepithelium provide a unique source of adult neural progenitor cells, which can be obtained from a patient's olfactory neuroepithelium. Human olfactory neuroepithelium is located within the nasal cavity and can be obtained by nasal sinus surgery via endoscopic biopsy . Therefore, it is possible to obtain these cells for autologous transplantation without highly invasive surgery. Previously, our laboratory has shown that cultures of olfactory epithelium can be grown to yield a population of neurosphereforming cells (NSFCs) . Thus, they may have the potential to be used therapeutically to treat neurological disorders [7–9].
Basic helix-loop-helix (bHLH) transcription factor Olig2 has been shown to be essential in the generation of oligodendrocytes and motoneurons in vivo . Olig2 is expressed by motoneuron progenitors, has a key role in specifying panneuronal properties of developing motoneurons, and precedes the expression of MNR2. Olig2 also directs the expression of motoneuron transcription factors, including Islet1/2 (Isl1/2), LIM-homeodomain (HD) gene, and the MNR2/HB9 homeobox gene, in neural progenitor cells . Isl1 and HB9 are defined markers for motoneurons and their progenitors . The bHLH gene Ngn2 is required for the generation of mouse motoneurons  and the development of the mouse cranial sensory ganglia  and functions as a vertebrate neuronal lineage factor [14, 15]. The bHLH proteins Ngn2 and NeuroM coupled with Isl1 and Lhx3 specify motoneurons in the embryonic chicken spinal cord and mouse P19 embryonic carcinoma cells . Olig2 and Ngn2 were specifically coexpressed in motoneuron progenitor cells and resulted in cell-cycle exit and panneuronal properties of developing chicken motoneuron differentiation [10, 11]. Lineage-restriction transcription factors HB9 and MNR2 are specifically expressed in all somatic motoneurons. MNR2 is expressed in presumptive motoneuron progenitor cells, and HB9 only in postmitotic motoneurons in the chick [17–19]. In contrast, HB9 is expressed in both motoneuron progenitors and post-mitotic motoneurons in the mouse [18 –21]. Transcription factor HB9 has been evolutionarily conserved, expressed selectively by embryonic motoneurons, and found essential for motoneuron development and differentiation . Coexpression of Ngn2 and MNR2 increased the number of Isl1/2-positive neurons in the chicken spinal cord compared with MNR2 alone . Mature motoneurons are characterized by their cytoskeletal components, ability to release neurotransmitters, surface receptors, and their ability to form synapses [22, 23].
The signaling molecule retinoic acid (RA) has been reported to play important roles in neurogenesis [24 –28], neurite elongation , and synaptic plasticity [30, 31]. Forskolin (FN), a molecule that increases intracellular cAMP, can stimulate axonal elongation [32, 33]. Sonic hedgehog (Shh) regulates neuronal specification (such as motoneurons and dopaminergic neurons) in the ventral region of the embryonic central nervous system (CNS) and can also direct human embryonic stem cells [34 –37]. RA and Shh have been shown to determine the expression of HD and bHLH transcription factors, which act on neuronal fate, especially motoneurons in the ventral spinal cord and mouse embryonic stem cells [26, 27, 38, 39].
The role that these genes and modulatory molecules play in neuronal growth and differentiation of human adult-derived neural stem cells has not been demonstrated. Thus, the purpose of this study was to investigate the roles of Olig2, Ngn2, and HB9 genes in human neuronal lineage specification and differentiation in vitro using adult human olfactory neuroepithelial progenitors. Here we report that the simultaneous transfection of NSFCs with Ngn2-HB9 or Olig2-HB9 combined with the treatment with RA, FN, and Shh can lead to the expression of motoneuron lineage-restricted markers.
Materials and Methods
The two different NSFC lines used in this study were obtained from adult olfactory neuroepithelium from a male (96-year-old) cadaver  and from a female patient (34 years old) . Frozen stock (passages 3 through 6) of each line was thawed rapidly and plated at a density of 5 × 105 cells per flask in minimal essential medium (MEM) with 10% heat-inactivated fetal bovine serum (FBS) (Gibco, Grand Island, NY, http://www.invitrogen.com) and 100 μg/ml gentamycin (MEM10) in flasks (25 cm2, Corning Incorporated, Corning, NY, http://www.corning.com) in humidified 5% CO2/95% air (37°C) for 24 hours. The NSFCs were adapted to the absence of serum via serial dilution of serum for 1 week until the cells were finally cultured in DFBNM (Dulbecco's modified Eagle's medium/F12 supplemented with 0.5% N2 and 1% B27) and 100 μg/ml gentamycin for 1 week [40, 41] and then used for in vitro differentiation analyses between passages 10 through 20. Parallel experiments were performed on NSFCs with both lines. Since equivalent results were obtained with these two different lines, data from only one line have been presented.
Construction of Expression Vectors
Full-length mouse Olig2 cDNA, Ngn2, and HB9 were cloned into the pIRES2-EGFP expression vector (Clontech, Palo Alto, CA, http://www.clontech.com) individually (O-E, N-E, and H-E). For the Olig2 and Ngn2 coexpression vector, Ngn2 cDNA was cloned into pIRES (Clontech) between NheI and EcoRI, and Olig2 cDNA was inserted between XbaI and SalI. For coexpression of Ngn2 and HB9, the Ngn2 cDNA was cloned into the pLHCX expression vector (Clontech), and H-E was used (N-H). For coexpression of Olig2 and HB9, the Olig2 cDNA in pCAAGS vectors and H-E were used (O-H). The pIRES2-EGFP and pIRES expression vectors served as controls (C-V). All expression vectors were verified by extensive DNA sequencing.
Transfection and Selection
All plasmid constructs were introduced into the NSFCs by liposomal transfection. The cells were plated on glass coverslips in six-well plates (3 × 104 cells per 35-mm well) in DFBNM without antibiotics 1 day before transfection. NSFCs were transfected with each plasmid (4 μg/well) for 48 hours according to the manufacturer's protocol (Life Technologies, Rockville, MD, http://www.lifetech.com). A further control was provided by lipofectamine alone. Two days after transfection, the cells were fixed or fed with DFB27M and selected with G418 (50 μg/ml; Gibco)  and/or hygromycin (50 μg/ml; Gibco) for 7 days in vitro (DIV).
The viability of the NSFCs after 2 days of transfection and 7 days of selection in flasks (25 cm2, Corning) was measured with a MTT kit (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com). Cells were seeded in the flasks in DFBNM without transfection and selection as controls. Cells were plated at a density of 5 × 104 cells/well in 24-well plates (Falcon; Becton, Dickinson and Company, Franklin Lakes, NJ, http://www.bd.com), and MTT solution was added to each well. Mitochondrial dehydrogenases in living cells metabolized MTT into formazan crystals, the concentration of which was determined spectrophotometrically at a wavelength of 570 nm as described previously .
The NSFCs (3 × 104 cells/well) were plated on 22-mm round glass coverslips in six-well plates (Falcon) and incubated at 37°C in 5% CO2/95% air for 24 hours, transfected for 2 days, and either immediately fixed or selected for 7 days combined with 4 days of treatment with 1 μM RA and 5 μM FN and another 3 days of treatment with 1 μM RA and 15 nM Shh (RFS) before fixation for immunofluorescence. Cultures were incubated with 4′,6-diamidino-2-phenylindole dihydrochloride (1:1,000, 2 mg/ml, Molecular Probes, Eugene, OR, http://probes.invitrogen.com) for 30 minutes at 37°C for vital labeling of DNA when nuclear staining was desired. The coverslips were rinsed with cytoskeletal buffer (CB) twice and fixed in 3% paraformaldehyde in CB (10 minutes), treated with 0.2% Triton X-100 (10 minutes, Sigma-Aldrich) when permeabilization was desired, and incubated (1 hour) in 3% bovine serum album (BSA) in Tris-buffered saline (TBS). Primary antibodies (see Supplementary Table 1) were applied overnight (4°C). After washing (1 hour) in TBS three times, the cells were incubated with secondary antibodies: Texas red-conjugated goat anti-rabbit immunoglobulin G (IgG), Texas red-conjugated goat anti-mouse IgG, and Cy2-conjugated goat anti-mouse IgG (all diluted 1:100, Cy2, Jackson Immunoresearch Laboratories, West Grove, PA, http://www.jacksonimmuno.com; Texas red, Molecular Probes). Experiments were performed in triplicate; first and second antibody omission controls were performed with each experiment to ensure the specificity of staining .
NSFCs were plated on glass coverslips in six-well plates (3 × 104 cells/35-mm well) in DFBNM. After transfection and selection combined with the treatment of RFS for 7 DIV, the number and length of neurites were determined. Cells (500 to 1,000) were sampled systematically from standardized fields (total magnification, ×200) with the aid of an eyepiece reticule under constant magnification with phase-contrast optics. Only the primary neurites originating directly from the soma of the NSFCs that were greater than the diameter of the cell body were evaluated in a double-blind study [41, 42].
Western Blot Analysis
Western blot analysis was used to support the immunofluorescence studies. Proteins from NSFCs cultured in DFBNM without selection, NSFCs transfected with control vectors, as well as NSFCs transfected with the vectors plus each combination of transcriptions factors (Olig2, Ngn2, HB9, Olig2-Ngn2, Ngn2-HB9, and Olig2-HB9), selected and combined with RFS treatment in all groups, were collected in lysis buffer. After 10 minutes of incubation (4°C), samples were centrifuged at × 12,000g (20 minutes) and the protein concentration of each supernatant was determined. The protein samples (20 μg/well) were electrophoresed on 10% to 14% SDS-polyacrylamide gels along with standardized-molecular-size marker proteins in an adjacent lane and transferred from gel to nitrocellulose paper. Nonspecific binding was blocked (1 hour) with 5% nonfat dry milk in TBS-Tween (TBST) buffer. Blots were incubated (4°C overnight) after addition of primary antibodies (Supplementary Table 1). Blots were washed with TBST buffer four times for 20 minutes and then three times for 10 minutes. Washed blots were incubated (1 hour) with polyclonal horseradish peroxidase-labeled anti-rabbit IgG (1:1,000) as well as monoclonal horseradish peroxidase-labeled anti-mouse IgG (1:1,000). Chemiluminescence Western blotting detection (Bio-Rad, Hercules, CA, http://www.bio-rad.com) was used to identify bound antibodies. Densitometry of the protein bands was carried out on a Molecular Dynamics gel scanner (Molecular Dynamics, Sunnyvale, CA). Data were analyzed using the Image Quant software programs supplied by the manufacturer.
5-Bromo-2′ Deoxyuridine Incorporation
The NSFCs (3 × 104 cells/well) were plated on 22-mm round glass coverslips in six-well plates (Falcon) and incubated at 37°C in 5% CO2/95% air. To examine mitotic activity of NSFCs after transfection and selection combined with RA1FN5 Shh treatment, 5-bromo-2′ deoxyuridine (BrdU) (10 μM, Sigma-Al-drich) was added to the cells for 24 hours before fixation. The cells were rinsed with cytoskeletal buffer twice and fixed in 3% paraformaldehyde in CB (10 minutes), treated with 0.2% Triton X-100 (10 minutes, Sigma-Aldrich) when permeabilization was desired, and incubated in 0.6% H2O2 in TBS for 30 minutes. Cells were incubated in 2 N HCl for 30 minutes at 37°C. Acid was removed by washing with TBS twice and neutralized with 0.1 M sodium borate (Sigma-Aldrich) for 10 minutes. Cells were incubated (1 hour) in 3% BSA in TBS. Primary antibody anti-BrdU was applied overnight (4°C). After washing (1 hour) in TBS three times, the cells were incubated with the following secondary antibodies: Cy2-conjugated goat anti-mouse IgG or Texas red-conjugated goat anti-mouse IgG (1:100, Cy2, Jackson Immunoresearch Laboratories). Experiments were performed in triplicate; first and second antibody omission controls were performed with each experiment to ensure the specificity of staining.
Chicken skeletal pectoral muscles were removed from 12-day embryos and dissociated with 0.25% trypsin at 37°C for 15 minutes and plated in six-well plates (5 × 105 cells/well) with defined medium (DF) plus 10% FBS. From 2 DIV, the cells were treated with cytosine β-D-arabinofuranoside (5 μM; Sigma-Aldrich) to eliminate dividing cells and therefore substantially reduce the presence of fibroblasts. After 7 DIV, cells were trypsinized (0.05% trypsin in Ca2+/Mg2+-free Hanks' balanced salt solution) and plated on glass coverslips in six-well plates with DF plus 2% FBS for 3 days. NSFCs were transfected with Ngn2-HB9 or Olig2-HB9 and treated with 1 μM RA and 5 μM FN and selection for 4 days (RF). The established muscle straps were cocultured with these NSFCs in DFBNM supplemented with 1 μM RA and 15 nM Shh for an additional 3 days in DFBNM (RS).
Statistical analysis (GraphPad Prism) was carried out using analysis of variance (significance level, p < .05). Cells (500 to 1,000) were sampled systematically from standardized fields (total magnification, ×200) of cells stained for each marker. The mean and standard deviation of triplicate samples repeated a minimum of three times were determined for each of the two NSFC lines. Since there were no detectable differences between the two cell lines, data have been reported without reference to which line was evaluated.
The heterogeneous nature of the NSFC population just before transfection was demonstrated by immunolocalization of fixed triton-treated cells. More than 97% of the cells were β-tubulin III+ and peripherin+; 43.6% ± 3.5% were nestin+ in the absence of triton; 25.4% ± 1.9% were A2B5+; and 67.3% ± 5.2% were NCAM+ before transfection in DF-BNM. In contrast, no cells were detected that were reactive for galactosylceramide (GalC), myelin basic protein (MBP) (an oligodendrocyte marker), glial fibrillary acidic protein (GFAP) (an astrocyte marker), OX42 (a microglia marker), a mature neuron marker (NeuN), vesicular acetylcholine transporter (VAChT), choline acetyltransferase (ChAT), acetylcholine, tyrosine hydroxylase, the rate-limiting enzyme for the synthesis of dopamine (TH), Olig2, Ngn2, HB9, or Isl1/2 (data not shown). No differences in phenotypic expression were detected between cells selected from the two different lines used in these studies.
NSFCs Transfected with Ngn2-HB9 Had Increased Expression of Mature Neuronal and Motoneuronal Markers and Elevated Neuritogenic Activity
To examine the phenotypic expression of NSFCs after transfection and selection, representative cultures and controls were compared. Nontransfected NSFCs or those transfected with lipofectamine alone died within 1 week after selection with 50 μg/ml G418 and/or 50 μg/ml hygromycin. Furthermore, there was no difference in cell viability between the various experimental groups, as determined by the MTT assay (data not shown). Transfection with control vectors, single genes, or Olig2-Ngn2 combined with RFS treatment resulted in no morphologic changes compared with RFS treatment alone or as indicated for Olig2-EGFP, Ngn2-EGFP, HB9-EGFP alone, or Olig2-Ngn2 (Figs. 1A–1F, 2). In contrast, the morphology of NSFCs transfected with both Ngn2 and HB9 and selection combined with RFS treatment of 7 days underwent dramatic changes that resulted in a phenotype characteristic of neurons (Figs. 2A, 2B). Immunofluorescent analysis demonstrated that NSFCs transfected with Ngn2-HB9 combined with the RFS treatment had expression rates of greater than 97% for each of the following: Ngn2 (Fig. 3A), HB9 (Fig. 3B), β-tubulin III (Fig. 3C), and peripherin (data not shown). Increased expression was also noted for the neuronal markers NF68 (Fig. 3D), NF160 (Fig. 3E), NF200 (Fig. 3F), NeuN (Figs. 2A, 4A), and ChAT (Fig. 4B) and the motoneuronal transcription factor Isl1/2 (Figs. 2A, 4C). In contrast, the expression of nestin and BrdU labeling was decreased (Figs. 2B, 4D, 4E); no markers for other cell lineages were detected, including GalC (data not shown), MBP, GFAP, or OX42 in the transfected cells (Figs. 4F– 4H).
Furthermore, the neurite lengths and numbers of NSFCs were increased after transfection, selection, and treatment with RFS compared with cells exposed to RFS without transfection (p < .01).
Western Blot Analysis of the Effects of Transcription Factors Olig2, Ngn2, and HB9 on the Neuronal Lineage Restriction of NSFCs
Although no Olig2+ cells were detected by immunohistochemistry, Western blot results indicated the presence of a low level of endogenous Olig2 expression in the groups without transfection with Olig2 (Figs. 5A, 5B). However, the expression of Olig2 was much higher after transfection with Olig2 (Figs. 5A, 5B). Similarly, the expression of Ngn2 and HB9 was detected after transfection with Ngn2 and HB9, respectively (Figs. 5A, 5B). The neuronal phenotype induced by the transcription factor Ngn2-HB9 was further confirmed by Western blot analysis. In NSFCs cultured in defined medium (DFBNM, controls) or transfected by control vectors, Olig2-EGFP, Ngn2-EGFP, HB9-EGFP alone, or Olig2-Ngn2 with the RFS treatment, VAChT, ChAT, and TH were not increased compared with cells exposed solely to RFS (Figs. 6A, 6B). However, in the cells cotransfected with both Ngn2 and HB9 with the RFS treatment, the expression was increased (p < .01, Figs. 6A, 6B) and VAChT+ (37.6% ± 3.8% to 41.2% ± 3.9%) and ChAT+ (34.8% ± 3.2% to 36.9% ± 3.3%) cells were increased compared with RFS treatment alone (VAChT+, 16.8% ± 1.4%; ChAT+, 14.1% ± 1.2%). Furthermore, no TH+ cells coexpressed VAChT or ChAT (Supplemental Table 2).
Olig2 Can Mimic the Effects of Ngn2 in Inducing Neuronal Phenotype in Collaboration with HB9
Despite the low level of expression of Olig2, there was a strong induction of Ngn2 expression after transfection with Olig2 (Figs. 5A, 5B). Because a high level of expression of Ngn2 in collaboration with HB9 could induce increased expression of neuronal antigens and neurite formation of NSFCs when combined with the treatment of RFS, Olig2 and HB9 cDNAs were introduced into NSFCs to determine whether they could mimic the effects of Ngn2-HB9. Immunohistochemistry and Western blot results confirmed the expression of Olig2 and HB9 after transfection with Olig2-HB9 (Figs. 5A, 5B, 7A, 7B). Cotransfection with Olig2-HB9 in combination with the RFS treatment induced the expression of neuronal markers NeuN, VAChT, ChAT, and TH and the motoneuronal marker Isl1/2 (Fig. 2A, 2B, 6A, 6B, 7C), as well as increased neurite lengths (p < .05) and numbers (p < .01). In contrast, the expression of nestin and BrdU labeling were decreased (Figs. 2A, 2B).
Cocultures of NSFCs Transfected with Ngn2-HB9 or Olig2-HB9 with Chicken Skeletal Muscle
A few axonal-muscular contacts were observed when RF-treated nontransfected NSFCs or NSFCs transfected with control vectors in DFBNM were plated on an established layer of chicken skeletal muscle (controls) for 3 days with RS treatment. In contrast, NSFCs transfected with Ngn2-HB9 or Olig2-HB9 after 4-day selection and treatment with 1 μM RA and 5 μM FN, cocultured with muscle for 3 days in the presence of 1 μM RA and 15 nM Shh (RS), formed multiple processes that often were observed in direct contact with the muscle cells (Figs. 7D–7I). As demonstrated with confocal microscopy, neurites of NSFCs and muscle cells formed apparent neuromuscular junctions, in which synapsin I (Figs. 7G–7I), ChAT (Fig. 7D), or acetylcholine (Figs. 7E, 7F) was colocalized.
Neural stem cells can differentiate into neurons and glia and therefore have the potential to provide cell populations for the treatment of various neurological diseases or injuries and for gene function and pharmacological evaluation [7–9]. NSFCs isolated from adult human olfactory neuroepithelium remained relatively undifferentiated despite exposure to a variety of media and trophic factors [6, 40]. This progenitor population seems to have a default characteristic of immature neurons, with > 97% of cells expressing β-tubulin III and peripherin, indicating that they may be different from neural stem cells isolated from embryos and/or other species [41, 43]. However, these NSFCs have the characteristics of neural progenitors [5–6, 44]; they can be driven to differentiate into oligodendrocytes with the transcription factors . In this study, NSFCs transfected with Ngn2-HB9 or Olig2-HB9 and treated with RFS gradually lost their progenitor characteristics, such as nestin expression, as previously reported for other cells , and gained the properties of mature neurons.
Recent reports have shown that several key lineage-specific inductive signals play important roles in directing progenitors to differentiate into specific cell types. RA and Shh play key roles in neuronal fate specification in embryonic avian and rodent ventral spinal cord [27, 46], forebrain [47–49], and hindbrain  and the development of dopaminergic neurons in the midbrain or in the human embryonic stem cells . Furthermore, RA interacts with Shh to direct the expression of transcription factors for neuronal differentiation [26, 27, 39]. Consistent with the roles of RA and Shh in vivo, there was no change after transfection with the transcription factors in the absence of RFS. Furthermore, RA cooperated with Shh in transfected NSFCs to induce the motoneuron differentiation, consistent with the result that simultaneous exposure of neural cells in embryonic chicken explants to RA and Shh in vitro promotes the formation of Olig2+ motoneuron progenitors . Molecular and genetic studies have shown that many transcription factors, when expressed in combination, promote lineage-specific differentiation during embryonic development . For instance, coexpression of Olig2 and Nkx2.2 promoted oligodendrocyte differentiation and maturation in vivo and in vitro, whereas expression of either Olig2 or Nkx2.2 alone was not sufficient to enable differentiation [41, 52]. In this study, Ngn2 and HB9 have been used to drive adult human progenitors to differentiate along the motoneuron lineage. Consistent with the previous observations in embryonic neural stem cells, neither Ngn2 nor HB9 alone could induce NSFCs to differentiate into neurons. However, NSFCs transfected simultaneously with Ngn2 and HB9 exhibited neuronal morphology and antigen expression, resembling in vivo patterns. Western blot analysis provided parallel independent assessment to complement the immunological studies. The two different approaches demonstrated that Ngn2 and HB9 or Olig2 and HB9, but not Olig2, Ngn2, or HB9 alone, possessed the capability of driving neuronal lineage-specific differentiation. Consistent with an earlier study that Olig2 is expressed in avian olfactory epithelium from E11.5 onward , a low level of endogenous Olig2 expression was detected in the NSFCs. This suggests that the low level of endogenous Olig2 expression was not sufficient to cooperate with HB9 to induce NSFC differentiation into neurons or motoneurons. No endogenous HB9 or Ngn2 expression was detected.
In the present study, the mechanism of induction of motoneurons by Olig2 and HB9 was investigated. Previous studies indicated that Olig2 determines the motoneuron identity; combined with RA signaling, Olig2 induces the downstream Ngn2 expression as well as other downstream hierarchical transcriptional cascades for the motoneuron differentiation in ovo [10, 11, 26, 27]. These findings suggest that Ngn2 may function downstream of Olig2 and mediate the function of Olig2 in collaboration with HB9 to control motoneuronal differentiation and maturation. In agreement with this observation, NSFCs transfected with Olig2 alone induced Ngn2 expression. The overexpression of Olig2 and HB9 in NSFCs achieved equivalent effects compared with that of Ngn2 and HB9. Expression of Olig2 in conjunction with HB9 when combined with the treatment of RFS resulted in increased expression of neuronal and motoneuronal antigens and neurite formation. NSFCs expressing Olig2 or HB9 alone did not exhibit these responses, possibly because the low level of Olig2 or Ngn2 was not sufficient to activate downstream transcription factor expression, and a high level of expression may be required to exert its function in concert with other transcription factors, such as HB9 and other inductive signals, including RA and Shh [10, 26, 27]. It has also been reported that coexpression of Olig2 and Ngn2 determined motoneuron progenitor cells and resulted in panneuronal properties of developing chicken motoneuron differentiation [10, 11] and that HB9 is critical for proper motoneuron differentiation in embryonic chicken and mouse spinal cord [17–19, 53]. In contrast, coexpression of Olig2 and Ngn2 or HB9 alone in NSFCs did not induce the neuronal differentiation, perhaps reflecting differences between species (human and rodent) [41, 43] and/or adult and embryonic tissues, although key qualitative properties were equivalent .
Previously it was reported that the murine embryonic stem cells and mouse P19 embryonic carcinoma cell lines differentiated into neurons and formed synapses after induction by RA. Although there were fewer individual synaptic contacts between pairs of P19 or embryonic stem cells, most embryonic stem cells had spontaneous synaptic activity . Synapses have also been reported to form between adult rat hippocampus stem cell–derived neurons and neonatal rat primary hippocampal neurons, although there were some differences in the synapses . It also been reported that synapses form between rodent neuroblastoma X glioma hybrid cells and mouse embryonic skeletal muscle cells  and rat pheochromocytoma PC12 cells and L6 rat skeletal muscle cells . Others reported that many closely related cell lines fail to form synapses with primary muscle cells in vitro [57, 58] but did form synapses when transplanted to the cerebellum of newborn mice . In the present study, NSFCs transfected with Ngn2-HB9 or Olig2-HB9 and treated with RFS expressed more mature motoneuronal markers and seemed competent to form synapses when cocultured with chicken skeletal muscle. Future studies will examine the ability of these transfected cells after transplantation to integrate and form synapses in various regions of rat spinal cord.
In summary, these studies reveal that the transcription factors (Ngn2 and HB9) that control neuronal, and especially motoneuronal, development in embryonic chick and rodent CNS are able to direct adult human olfactory-derived progenitors toward neuronal lineage. Second, in this model, Ngn2 and HB9 functioned cooperatively in the environment of a RFS supplement to produce neuronal differentiation. Neither Ngn2 nor HB9 alone or Ngn2-HB9 in the absence of RFS exhibited phenotypic changes. Third, NSFCs transfected with Olig2 induced the expression of the transcription factor Ngn2. Fourth, NSFCs transfected with Ngn2-HB9 or Olig2-HB9 had an elevated expression of neuronal-specific antigens, including VAChT, ChAT, NeuN, and the motoneuron transcription factor Isl1/2. Fifth, NSFCs transfected with Ngn2-HB9 or Olig2-HB9 and cocultured with chicken skeletal muscle formed multiple contacts that were consistent with the development of neuromuscular junctions.
The readily accessible location of adult human olfactory epithelium makes it possible to obtain endogenous neural progenitors without highly invasive surgery by endoscopic biopsy . Our studies suggest that transcription factors can be used to direct these progenitors to differentiate into specific neuronal or glial cell types and thereby expand their therapeutic potential as well as their utility for diagnostic evaluation and genetic studies.
We thank George Harding for his assistance with the confocal microscopy. This work was supported by NIH (1920RR15576 to F.J.R.), Kentucky Spinal Cord Head Injury Research Trust (to M.Q.), and National Multiple Sclerosis Society (FA1400-A-1 to J.C.).
The authors indicate no potential conflicts of interest.