H9 hESCs (cat. no. WA09; WiCell Research Institute, Madison, WI, USA) were maintained on Matrigel (BD Biosciences, San Diego, CA, USA) in mTeSR1 (StemCell Technologies, Vancouver, BC, Canada) as previously described (14,15).

The hNSCs were differentiated through the formation of human neuroectodermal spheres (hNESs) as previously reported with minor modifications (2,16). The H9 hESCs (cat. no. WA09; WiCell Research Institute) were maintained on Matrigel (BD Biosciences) in mTeSR1 (StemCell Technologies) as previously described (14). Human embryoid bodies (hEBs) were generated by culturing hESCs in hEB medium consisting of knockout DMEM supplemented with 10% knockout serum replacement, 1% non-essential amino acids, 1 mM L-glutamine (all from Invitrogen; Thermo Fisher Scientific, Inc., Waltham, MA, USA) and 0.1 mM β-mercaptoethanol (Sigma-Aldrich; Merck KGaA, Darmstadt, Germany) on non-coated Petri dishes. The resulting hEBs were then cultured in NES/NSC medium consisting of DMEM/F12, 1X N2/B27 (both from Invitrogen; Thermo Fisher Scientific, Inc.), 20 ng/ml basic fibroblast growth factor (bFGF; R&D Systems, Inc., Minneapolis, MN, USA), 20 ng/ml epidermal growth factor (EGF) and 10 ng/ml leukemia inhibitory factor (both from PeproTech, Inc., Rocky Hill, NJ, USA). The hNESs were sub-cultured every week using a Mcllwain tissue chopper (Mickle Engineering, Surrey, UK), and the medium was replaced every 2 days. The hNESs were passaged at least five times without disturbing the formation of neural rosettes. For terminal differentiation, each hNES was allowed to attach to a Matrigel-coated coverslip and was maintained without growth factors for 2 weeks, as previously described (17,18). To count the total number of cells within each hNES, the hNESs were dissociated into single-cell suspensions with 0.1% trypsin-EDTA (Invitrogen; Thermo Fisher Scientific, Inc.) for 3 min. Live cell numbers were counted using trypan blue (Invitrogen; Thermo Fisher Scientific, Inc.) exclusion under an Olympus fluorescence microscope (IX51; Olympus Corp., Tokyo, Japan).

ReNcell CX cells derived from the cortical region of human fetal brain tissue (cat. no. SCC007; EMD Millipore, Temecula, CA, USA) were cultured according to the manufacturer's protocol. The ReNcell CX cells were maintained in ReNcell NSC maintenance medium supplemented with 20 ng/ml EGF and 20 ng/ml bFGF (all from EMD Millipore) on laminin-coated tissue culture dishes (BD Biosciences). The culture medium was replaced every 2 days. For terminal differentiation, the ReNcell CX cells were cultured for 5 days without growth factors, as previously described (19).

Immunocytochemistry was performed as previously described (20). In brief, the cells were fixed in 4% formaldehyde and then permeabilized with PBS containing 0.1% Triton X-100. Following blocking with 3% bovine serum albumin (Sigma-Aldrich; Merck KGaA), the cells were incubated at 4°C overnight with anti-neuron-specific class III β-tubulin (TUJ1; 1:500; cat. no. PRB-435P; Covance, Inc., Princeton, NJ, USA), anti-NESTIN (1:100; cat. no. MAB5326), anti-microtubule-associated protein 2 (MAP2; 1:500; cat. no. MAB3418), anti-glial fibrillary acidic protein (GFAP; 1:200; cat. no. MAB3402) and anti-Ki67 (1:500; cat. no. AB9260; Chemicon) (all from EMD Millipore), followed by incubation with Alexa Fluor 488-conjugated anti-mouse IgG (1:1,000; cat. no. A21202), Alexa Fluor 594-conjugated anti-mouse IgG (1:1,000; cat. no. A21203), Alexa Fluor 488-conjugated anti-rabbit IgG (1:1,000; cat. no. A21441) or Alexa Fluor 594-conjugated anti-rabbit IgG (1:1,000; cat. no. A21442) (all from Molecular Probes, Eugene, OR, USA) as secondary antibodies for 1 h at room temperature. DAPI (1 mg/ml; Invitrogen; Thermo Fisher Scientific, Inc.) was added to visualize the nuclei. The slides were examined using an Axiovert 200M microscope (Carl Zeiss AG, Gottingen, Germany).

Total RNA was extracted from cells with an RNeasy kit (Qiagen, Inc. Hilden, Germany) and reverse transcribed using a Superscript IV First-Strand Synthesis System kit (Invitrogen; Thermo Fisher Scientific, Inc.) as previously described (21). The resulting cDNA was diluted 1:10 with deionized water, and 1 µl of the diluted cDNA was added to Accupower™ PCR PreMix (Bioneer Corp., Daejeon, Korea), 10 pmol/l of specific primers and deionized water to a final volume of 20 µl. The RT-PCR analysis was performed under the following conditions: 5 min at 95°C; 30–40 cycles of 30 sec at 95°C, 30 sec at 60°C, 30 sec at 72°C, and 5 min extension at 72°C. GAPDH was used as an internal control. The relative expression of target genes was determined using the 2^−ΔΔCq method (22). The primers used in this study are listed in Table I.

The microarray experiments were performed using the Low RNA input linear amplification kit, cRNA cleanup module and one-color (Cy3) Whole Human Genome Microarray 4X44K, according to the manufacturer's protocol (Agilent Technologies, Inc., Santa Clara, CA, USA) as previously described (23). The raw data was normalized using global scale normalization and processed using GeneSpring software version 11.0 (Agilent Technologies, Inc.,). Heatmap and hierarchical clustering of genes was generated using MeV v. 4.9.0 software (http://www.tm4.org). Gene functions were annotated using the GeneCard database (http://www.genecards.org/). The principal component analysis (PCA) was performed with GeneSpring software. Biological processes and protein classes were described using Protein Analysis Through Evolutionary Relationships (PANTHER; http://www.pantherdb.org/). Kyoto Encyclopaedia of Genes and Genomes (KEGG) analysis and functional annotation clustering were performed using David Bioinformatics Resources 6.8 with the Database for Annotation, Visualization and Integrated Discovery (DAVID; http://david.abcc.ncifcrf.gov).

In the present study, H9 hESCs were differentiated into hNSCs based on the previously described hNES formation method (17). The aggregates of hESCs were cultured in hEB medium for 5 days, followed by transfer into NES/NSC medium to drive neuronal fate commitment and promote neuronal differentiation (Fig. 1A). During differentiation, following the first subculture, hNESs containing neural rosette structures, a key structure representing NSCs, appeared and retained the potential to form neural rosette structures (Fig. 1A; red dotted circle). The hNESs were dissociated into single NSCs by trypsin digestion and were cultured as adherent monolayers. As reported in our previous study (1,17,24), the hNESs generated using this method are characterized as NSCs as they have the potential to differentiate into neuronal and glial cell types, and can be serially passaged to form new hNESs.

Immortalized hNSC lines are in increasing demand due to the inherent limitations of primary hNSCs, including limited availability, poor expandability and associated ethical issues (5). The ReNcell CX cell line, which is a commercially available immortalized fetal cortical NSC line, was used as a reference hNSC type in the present study. Under normal growth conditions, ReNcell CX cells exhibited immature neural morphology, similar to that of monolayer-cultured hNSCs (Fig. 1). The ReNcell CX cells grew rapidly as a monolayer on laminin, with a doubling time of ~24 h due to the c-MYC-based immortalization. Therefore, there are safety concerns, including the risk that oncogenic c-MYC may render this hNSC line tumorigenic following transplantation (19,25).

To compare hNESs derived from hESCs and ReNcell CX cells for use as an hNSC model, and examine the mechanisms underlying lineage commitment in NSCs, microarray analyses were performed in undifferentiated hESCs, hEBs (intermediate cells in hNSC differentiation), hESC-derived hNESs and ReNcell CX cells. A heatmap showing the hierarchical clustering results from the whole-genome expression profiles indicated that differentiated hNESs preferentially clustered with ReNcell CX cells (Fig. 2A). The principal component analysis (PCA) also confirmed that hNESs and ReNcell CX cells were distinctly separated from undifferentiated hESCs and hEBs (Fig. 2B). Accordingly, only ~12.4% of all the genes were differentially expressed, with a fold-change threshold of 2.0 between the hNESs and ReNcell CX cells. These data indicated that the global transcription of differentiated hNESs is similar to that of ReNcell CX cells.

To obtain the overall profile regarding common aspects of hNSC identity, a total of 1,711 commonly upregulated and 856 commonly downregulated genes between hNESs and ReNcell CX cells with a fold-change of >5 were identified as the hNSC-specific transcriptome and analyzed using the PANTHER classification system. The top biological process term was cellular process (26.9%) (Fig. 2C). Other major processes corresponding to these hNSC-related genes included metabolic process (20.4%), response to stimulus (8.7%), developmental process (8.3%), biological regulation (7.6%), localization (7.3%), multicellular organismal process (6.3%) and cellular component organization or biogenesis (5.6%), as shown in Fig. 2C. In addition, the predominant protein class was nucleic acid binding (12.9%), followed by hydrolase (8.9%), enzyme modulator (8.6%), signaling molecule (8.5%), transcription factor (8.3%), receptor (7.0%), transporter (6.6%), transferase (6.5%) and cytoskeletal protein (5.6%), as shown in Fig. 2D. A significant number of genes were identified as transcription factors, which are known to have prominent roles in lineage specification and developmental processes (26). Therefore, the contribution of these transcription factors to the hNSC identity was dissected. The important enriched transcription factor categories were zinc finger transcription factor (38.4%), helix-turn-helix transcription factor (25.6%), transcription cofactor (18.4%), basic helix-loop-helix transcription factor (8.0%), HMG box transcription factor (4.8%), and nuclear hormone receptor binding (4.0%), as shown in Fig. 2D. The expression of several transcription factors from the microarray data were analyzed further, and the transcription factor expression levels were similar between the hNESs and ReNcell CX cells (Fig. 2E). Representative genes are shown in Fig. 3.

To elucidate the signaling pathways and molecular mechanisms associated with the hNSC identity, the present study analyzed the hNSC-specific transcriptome. Pathway analysis based on the KEGG database showed that the commonly upregulated genes were significantly associated with the following pathways: Cytokine-cytokine receptor interaction, PI3K-Akt signaling, complement and coagulation cascades, axon guidance, focal adhesion, ECM-receptor interaction, Ras signaling, proteoglycans in cancer, regulation of actin cytoskeleton, and ABC transporters (Fig. 4A). Cell adhesion molecules, metabolic pathways, antigen processing and presentation, glycerophospholipid metabolism, T cell receptor signaling, cell cycle, oxytocin signaling, biosynthesis of antibiotics, ErbB signaling, and progesterone-mediated oocyte maturation were enriched for the commonly downregulated genes in hNESs and ReNcell CX cells (Fig. 4B).

To obtain a more comprehensive understanding of the functions of the hNSC-specific transcriptome, GO term enrichment analysis was performed through DAVID functional annotation clustering, which shows functionally linked groups by reducing the redundancy in the annotation (11,27). The following seven significant annotation clusters were identified in the hNSC-specific transcriptome, which were related to glycoprotein, cell junction (synapse), immunity, EGF-like domain, ECM-receptor interaction (fibrillar collagen), regulation of PI3K signaling and metal-binding based on statistical criteria (P<0.05 and an enrichment score of at least 1.7) (Fig. 4C).

Although the majority of the genes analyzed showed similar expression patterns, differentially expressed genes found only in a small portion of genes (12.4% of all genes given the 2.0-fold cutoff criterion) between hNESs and ReNcell CX cells were enriched in the following pathways: TGF-β signaling, acute myeloid leukemia, malaria, transcriptional misregulation in cancer, pathways in cancer, axon guidance, signaling pathways regulating pluripotency of stem cells, and melanoma (Fig. 5). The majority of the overrepresented pathways were identified as cancer-related pathways; this may be due to the oncogenic c-MYC having been functionally linked to cancer-related pathways (28,29).

To further characterize hNESs derived from hESCs molecularly, semi-quantitative RT-PCR analysis was performed for the expression of NSC markers. Transcripts of ESC markers, including OCT4 and NANOG, were decreased during differentiation (Fig. 6A). The levels of molecular markers for NSCs, including superoxide dismutase (SOX)1, SOX2, SOX3, musashi-1 (MSI1), TUJ1 and NESTIN, were increased in the hESC-derived hNESs (Fig. 6A). Dissociated hNES cells and ReNcell CX cells were also immunostained for NSC markers, including NESTIN and TUJ1 (Fig. 6B). To functionally characterize neuronal differentiation capacity, hNESs and ReNcell CX cells were differentiated following the withdrawal of growth factors. Following 15 days of differentiation, MAP2-positive neuronal cells and GFAP-positive glial cells were observed in the differentiated hNESs and ReNcell CX cells (Fig. 6B). Of note, the hESC-derived hNESs continued to expand over five passages without losing any of their features. It was possible to scale-up hNES production by ~662-fold in terms of cell number (mean values) following five passages, with the majority of cells undergoing active proliferation, as indicated by Ki-67 labeling at passage 5 (Fig. 6C).

The aim of the present study was to characterize hNSCs and to select a more suitable hNSC model for developing human-based platforms for applications in various neurological fields. The hNSC-specific transcriptome data from hESC-derived hNESs and ReNcell CX hNSCs were described. Global gene expression profiling enables a systems-based analysis of the biological processes through GO and pathway enrichment analyses, and of genes driving differentiation into hNSCs. The comparative analysis of the global gene expression showed that the hESC-derived hNESs were similar to the ReNcell CX hNSCs, as shown by PCA and hierarchical sample clustering. As described above, the hNESs derived from hESCs in the simple protocol exhibited differentiation potential, based on the specific terminal differentiation markers of neuronal and glial cells that were detected. The hNESs showed stable proliferation and were expanded for at least five passages without loss of NSC characteristics. These results indicated that hESC-derived hNESs may be used as a relevant hNSC model, similar to ReNcell CX cells, in several neurological research fields.

hNSC differentiation protocols remain inefficient, with poor yields of terminal differentiation and maturation into specific neuronal cell types. However, increased understanding of the mechanisms underlying hESC-based NSC differentiation through fine-tuning protocols for the efficient derivation, long-term maintenance and neuronal cell type-specific differentiation of hNSCs may provide novel insights into human neurodevelopment and the process of NSC fate specification.