Long-Term Proliferation of Human Embryonic Stem Cell–Derived

Neuroepithelial Cells Using Defined Adherent Culture Conditions

SOOJUNG SHIN,a MAISAM MITALIPOVA,a,d SCOTT NOGGLE,c DEANNE TIBBITTS,a,d ALISON VENABLE,b

RAJ RAO,a STEVEN L. STICEa

aRegenerative Bioscience Center and bDepartment of Biochemistry, University of Georgia, Athens, Georgia;cInstitute of Molecular Medicine and Genetics, Medical College of Georgia, Augusta, Georgia;

dBresaGen, Athens, Georgia, USA

Key Words. Embryonic stem cell • Differentiation • Neuroepithelial stem cell • Defined culture

ABSTRACT

Research on the cell fate determination of embryonic stemcells is of enormous interest given the therapeutic potentialin regenerative cell therapy. Human embryonic stem cells(hESCs) have the ability to renew themselves and differen-tiate into all three germ layers. The main focus of this studywas to examine factors affecting derivation and furtherproliferation of multipotent neuroepithelial (NEP) cells fromhESCs. hESCs cultured in serum-deprived defined mediumdeveloped distinct tube structures and could be isolatedeither by dissociation or adherently. Dissociated cells sur-vived to form colonies of cells characterized as NEP whenconditioned medium from human hepatocellular carcinomaHepG2 cell line (MEDII) was added. However, cells isolatedadherently developed an enriched population of NEP cellsindependent of MEDII medium. Further characterizationsuggested that they were NEP cells because they had asimilar phenotype profile to in vivo NEP cells and expres-

sion SOX1, SOX2, and SOX3 genes. They were positivefor Nestin, a neural intermediate filament protein, andMusashi-1, a neural RNA-binding protein, but few cellsexpressed further differentiation markers, such asPSNCAM, A2B5, MAPII, GFAP, or O4, or other lineagemarkers, such as muscle actin, � fetoprotein, or the plu-ripotent marker Oct4. Further differentiation of theseputative NEP cells gave rise to a mixed population ofprogenitors that included A2B5-positive and PSNCAM-positive cells and postmitotic neurons and astrocytes. Toproliferate and culture these derived NEP cells, idealconditions were obtained using neurobasal medium sup-plemented with B27 and basic fibroblast growth factor in5% oxygen. NEP cells were continuously propagated forlonger than 6 months without losing their multipotent cellcharacteristics and maintained a stable chromosomenumber. STEM CELLS 2006;24:125–138

INTRODUCTION

After human embryonic stem cells (hESCs) were established [1,

2], there was an immediate interest in differentiating these

pluripotent cell lines toward a neuronal cell fate as a promising

source for replacement cell therapy. The central nervous system

(CNS) contains endogenous stem cells that are capable of pro-

liferating; however, in many cases these cells are too few in

number or incapable of restoring function after neuronal damage

has occurred [3]. Neural tissues from fetuses, immortalized cell

lines, and embryonic stem cells (ESCs) are three main candidate

Correspondence: Steven Stice, Ph.D., Regenerative Bioscience Center, University of Georgia, Athens, Georgia 30605, USA.Telephone: 706-583-0071; Fax: 706-542-7925; e-mail: sstice@uga.edu Received July 2, 2004; accepted for publication July 5, 2005;first published online in STEM CELLS EXPRESS. ©AlphaMed Press 1066-5099/2006/$12.00/0 doi: 10.1634/stemcells.2004-0150

Embryonic Stem Cells

STEM CELLS 2006;24:125–138 www.StemCells.com

sources for replacement cells. Fetus-derived neural tissue has

been transplanted in humans, and encouraging results were

obtained [4]. The outcome varied, however, depending on the

age of the graft cells or the presence of subculture [5]. In

addition, the supply of fetal neural tissue is limited because of

ethical concerns.

Neuronal stem cells derived from cancer cell lines have been

considered a potential alternative cell source with unlimited

capability for cell proliferation, but there is significant concern

that cancer cells may be unstable and prone to tumorigenesis [6].

Furthermore, it has been shown that the range of cell types

derived from immortalized cells may be quite small [7]. In

contrast, ESCs have a unique advantage because they can pro-

liferate and maintain their pluripotency for years [1] and can

differentiate into virtually any cell type in the body. Addition-

ally, there is no decrease in plasticity, which is shown in neural

stem cells isolated from fetal tissue [7, 8]. Mouse ESCs that

have been expanded and differentiated into oligodendrocyte

precursors and then transplanted into an animal model of human

myelin disease have resulted in effective remyelination of host

axons and functional recovery [9].

Neuroepithelial (NEP) stem cells are self-renewing multi-

potent cells that can differentiate into neurons, oligodendro-

cytes, and astrocytes [10]. These undifferentiated nonlineage-

committed cells express Nestin but not the differentiated cell

markers A2B5 and PSNCAM [11]. In humans, NEP cells form

the neural tube during the third and fourth weeks of gestation

[12]. These cells divide symmetrically or asymmetrically to give

rise to all the cells that comprise the mammalian CNS, including

various types of neurons and glial cells [13].

Neural developmental pathways can be delineated through

ESC studies. Neuronal development in rodents is a well-docu-

mented stepwise process, much like hematopoietic stem cell

differentiation. Mouse neurectoderm forms the earliest pluripo-

tent neural stem cells, called NEP cells, which then differentiate

further into neuronal-restricted precursor cells or glial-restricted

precursor cells [14]. PSNCAM and A2B5 are used as critical

lineage markers of rodent neuronal and glial lineages, respec-

tively. Human NEP cells can be isolated from the fetus [11] and

also from ESCs [15]. These cells form neural rosettes and are

Nestin and Musashi 1 positive.

Nestin has been the primary antigen used as a marker of

NEP cells [15, 16]. However, Nestin expression is not exclusive

to NEP cells but is widely expressed in developing embryos. For

example, Nestin is expressed in endocrine progenitor cells,

vascular endothelial cells [17], testis [18], and skeletal muscle

[19]. In vivo expression studies in mouse and chicken indicated

that SOX1, SOX2, and SOX3 are predominantly expressed in

the undifferentiated cells of NEP cells in CNS [20, 21]. Human

and mouse SOX genes are highly conserved to have over 95%

homology, and SOX1 is moderately abundant in human embry-

onic brain [22]. Also, it has been shown that SOX2 and SOX3

expression was modulated during neural differentiation of hu-

man embryonic carcinoma cell line NTERA2 [23]. Therefore,

expression of SOX genes can serve as conservative criteria for

NEP characterization.

A variety of methods have been used to derive NEP cells

from ESCs [15, 16, 24, 25]. However, most of these methods

have used cell aggregation or embryoid bodies (EBs), which

allows stochastic differentiation into all three germ layers, in-

cluding NEP cells. When either mouse ESCs [26] or nonhuman

primate ESCs [24] were cultured with conditioned medium from

the human hepatocellular carcinoma HepG2 cell line (MEDII),

they developed preferentially into neurectoderm. In this study,

factors required for the neural differentiation of hESCs were

examined and conditions allowing further proliferation were

optimized. We show that adherent cultures of hESCs in serum-

deprived medium without feeder layers gave rise to a rosette-

enriched population. Characterization of this population showed

that the cells were multipotent NEP cells with proper phenotype

marker and SOX genes expression profiles and that they were

able to differentiate further to both A2B5-positive and

PSNCAM-positive precursor cells. Thus, this study demon-

strates that derived NEP cells can be cultured more than 6

months in optimized conditions without the cells losing their

capacity for neural and glial differentiation while maintaining a

stable chromosome number.

MATERIALS AND METHODS

Human Embryonic Stem Cell CultureHuman ESC lines BG01 and BG02 used in this experiment were

cultured on mouse embryonic fibroblasts (MEF) layer, inacti-

vated by mitomycin C [27]. Because there were no differences

in experimental results due to ESC lines in this study, data from

both cell lines were pooled. The cells were cultured in ES

medium of Dulbecco’s modified Eagle’s medium (DMEM)/F12

medium (Gibco, Grand Island, NY, http://www.invitrogen.com)

supplemented with 15% serum and 5% knockout serum replace-

ment (KSR) (Gibco), 2 mM L-glutamine, 0.1 mM minimal

essential medium nonessential amino acids, 50 U/ml penicillin,

50 �g/ml streptomycin, 4 ng/ml basic fibroblast growth factor

(bFGF) (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.

com), and 10 ng/ml leukemia inhibitory factor (LIF) (Chemicon,

Temecula, CA, http://www.chemicon.com). For passage, ideal

colonies were mechanically dissected into small pieces and

replated on mitotically inactivated MEF and the medium

changed every other day as described [27]. These cell lines have

maintained their distinct stem cell morphology and karyotype

and remain Oct-4–positive and SSEA4-positive [27].

126 Adherent Differentiation of hESCs into NEP Cells

Conditioned Medium PreparationHuman hepatocellular carcinoma (HepG2) cells (ATCC HB-8065)

were seeded at a density of 9.4 � 104 cells/cm2 and proliferated for

3 days in DMEM/F12 medium supplemented with 10% fetal calf

serum, 2 mM L-glutamine, 50 U/ml penicillin, and 50 �g/ml

streptomycin. To produce conditioned medium, cells were washed

twice with phosphate-buffered saline (PBS), and DMEM/F12 me-

dium without serum supplement was added at a ratio of 0.285

ml/cm2. In 3 days, conditioned medium was collected and stored at

4°C for less than 5 weeks as MEDII.

Antibodies and ImmunocytochemistryCells plated on polyornithine/laminin-coated permanox slides were

washed in PBS and fixed with 4% paraformaldehyde/4% sucrose in

PBS for 15 minutes. Fixed cells were washed two times with PBS

before staining. Permeabilization and blocking was carried out in

blocking buffer consisting of 0.1% Triton, 3% goat serum in Tris

buffer for 40 minutes. For cell-surface antigen, permeabilization

was excluded. Primary antibodies were applied in blocking buffer

for 2 hours at room temperature and washed three times in blocking

buffer before secondary antibody application. Secondary antibodies

of goat anti-mouse Alexa-conjugated, goat anti-rabbit Alexa-con-

jugated (Molecular Probes Inc., Eugene, OR, http://probes.invitro-

gen.com) were diluted at 1:1,000 in blocking buffer and applied to

cells for 40 minutes at room temperature. After two washes in PBS,

4�,6�-diamidino-2-phenylindole was applied for nuclear staining

for 10 minutes, and cells were observed under the fluorescence

microscope. For flow cytometry application, cells were harvested

by trypsinization and suspended in PBS to be fixed and stained

using the same procedure coupled with serial centrifugation at

3,000 rpm and resuspension in PBS. For negative controls, first

antibodies were omitted and the same staining procedure was

followed. Primary antibodies and dilutions used included the fol-

lowing: mouse anti-Nestin (1:100; R&D Systems Inc., Minneapo-

lis, http://www.rndsystems.com), rabbit anti-Nestin (1:200; Chemi-

con), rabbit anti-Musashi 1 (1:500; Chemicon), mouse anti-beta III

tubulin (1:400; Sigma), rabbit anti-Tuj1 (1:500; Covance, Prince-

ton, NJ, http://www.covance.com), mouse anti-Hu (1:50; Molecu-

lar Probes), mouse anti-muscle actin (1:50; DAKOCytomation,

Glostrup, Denmark, http://www.dakocytomation.com), mouse an-

ti–� feto protein (1:50; DAKOCytomation), rabbit anti-GFAP (1:

50; Sigma), mouse anti-O4 (1:10; Chemicon), mouse anti-

PSNCAM (1:400; Abcys, Paris, http://www.abcysonline.com), and

mouse anti-A2B5 (1:100; a gift from Mayor Proschel).

Effect of ES, DN2, and MEDII Media onDifferentiation of hESCs in a Three-Stage ProcessThe differentiation procedure is outlined in Figure 1 and divided

into three stages to assist in characterizing the progression of in

vitro neural differentiation. After manual passage onto fresh

feeder cells, hESCs were allowed to proliferate in ES medium

for 7 days (stage 1). Cell differentiation was then induced with

either DN2, MEDII, or ES medium for another 7 days (stage 2).

DN2 medium is DMEM/F12-based medium supplemented with

N2 (Gibco), L-glutamine, penicillin/streptomycin (P/S), and 4

ng/ml bFGF. MEDII medium for this study is DN2 medium

supplemented at 50% (unless otherwise noted) with conditioned

medium (described above). To understand and follow the dif-

ferentiation steps applied here, phenotype marker expression

was examined at the time intervals described in Figure 1. At

stages 1, 2, and 3, populations were harvested and the markers

Musashi-1, Nestin, and Oct-4 were observed. Immunocyto-

chemical analysis was also performed on the adherent cell

population. The cells at both stages were double-stained with

Nestin and Oct-4 and observed under the fluorescence micro-

scope for immunocytochemical examination associated with

morphology. Groups that displayed phenotypic difference were

then subjected to quantitative analysis for these same markers

using flow cytometry. All experiments were replicated three

times unless otherwise noted.

Effect of ES, DN2, and MEDII Media onDifferentiation of Stage 2 Cells in Adherent CellCulture Without Feeder CellsTo improve NEP cell derivation, a method using adherent dif-

ferentiation was exploited. It was possible to isolate subpopu-

lations of stage 2 cells that had infiltrated under the feeder layer

Figure 1. Procedure for adherent derivation of human embryonicstem (ES) cells into neuroepithelial cells.

127Shin, Mitalipova, Noggle et al.

www.StemCells.com

to attach firmly on culture plates. To test the effect of ES, DN2,

and MEDII media on this derivation method, the mouse feeder

layer was physically removed from each group of stage 2 cells

in calcium/magnesium-free PBS. The remaining cells were cul-

tured another 3 days in respective media as described in Figure

1 (stage 3). At stage 3, populations were harvested from each

group, and morphology and phenotype marker expression of

Oct-4, Nestin, and Musashi 1 was observed as described before

using flow cytometry and immunocytochemistry for Oct-4, Nes-

tin, and Musashi-1.

Effect of MEDII Medium and Low Cell Density onCell Survival of Stage 2 Differentiating CellsThe effect of MEDII medium was examined using single-cell

passage of stage 2 cells in the medium supplemented with four

different concentrations of MEDII. As shown in Figure 1, stage

2 MEDII-cultured cells were obtained. The resulting adherent

cells were dissociated in 0.02 M EDTA containing PBS, and 104

cells/cm2 were then plated on polyornithine- and laminin-coated

dishes in different concentrations of MEDII medium (0%, 25%,

50%, 100%). After 10 days of culture in respective media, cells

were harvested and derivation efficiency (resulting cell number/

starting cell number � 100) was determined over four repli-

cates. In addition, TDT-mediated dUTP nick-end labeling

(TUNEL) assay was performed at 6 and 24 hours after plating in

0% or 50% MEDII to determine levels of apoptosis in these

cultures. The TUNEL assay was performed according to the

manufacturer’s instructions (Molecular Probes), and cells were

subsequently analyzed by flow cytometry.

Characterization and Examination of DifferentiationCapacity of Derived NEP CellsRosette-forming NEP cell populations from stage 3 cells derived

in DN2 and MEDII media either by adherent feeder removal or

by dissociated culture were characterized by immunocytochem-

istry. The markers included early neural stem cell markers

(Nestin, Musashi1) for positive expression and mesodermal

marker muscle actin, endodermal marker �-fetoprotein, pluri-

potent marker Oct-4, and late-stage neuronal and glial markers

(A2B5, PSNCAM, � III tubulin, Hu, GFAP, O4) for negative

expression. For terminal differentiation, NEP cells were cul-

tured in neurobasal medium (Gibco) and supplemented with

B27 (Gibco), L-glutamine, and penicillin/streptomycin without

bFGF for 14 days. For oligodendrocyte differentiation, NEP

cells were exposed to 5 �g/ml platelet-derived growth factor

(Upstate, Lake Placid, NY, http://www.upstate.com) and 50 �M

3T3 (Sigma) for 6 days before terminal differentiation. Differ-

entiated cells were characterized using the restricted progenitor

markers PSNCAM, A2B5, and the postmitotic neural marker

Hu, the neuron-specific tubulin, and �-III tubulin, the oligoden-

drocyte marker O4, and the astrocyte-specific marker GFAP.

Effect of Medium, Supplement, Growth Factor, andOxygen Conditions on Proliferation and Viability ofSubcultures of Derived NEP Cells

Effect of Culture MediumTo obtain a more uniform subculture system, two different kinds

of base media—DMEM/F12 (D) and neurobasal medium (N)—

were tested with supplements of N2-, B27-, or MEDII-condi-

tioned media. Stage 3 NEP cells were allocated into four dif-

ferent media: DN2, NN2 (neurobasal medium supplemented

with N2), NB27 (neurobasal medium supplemented with B27),

and 50% MEDII in DN2 medium, as described above with the

same supplement of L-glutamine, P/S, and 4 ng/ml bFGF. After

12 days of culture, cells were harvested and examined for

morphology and viability using the Guava ViaCount (Guava

Technologies, Hayward, CA, http://www.guavatechnologies.

com) flow cytometry assay. Briefly, the Guava ViaCount re-

agent combines two different DNA dyes. One dye binds to the

nucleus of every cell to give a total cell number, and the other

dye binds differentially to only nonviable cells. The data col-

lected include total cell number and viability of the sample.

Subculture of NEP CellsNEP cells derived from either DN2 or MEDII were further

propagated in NB27 with L-glutamine, P/S, 10 ng/ml LIF, and

20 ng/ml bFGF on poly-ornithine–coated and laminin-coated

dishes. Cells were continuously passaged either by mechanical

trituration or by trypsin (1 � 105/cm2) to be replated. After more

than 6 months in culture, NEP cells were characterized as

described above, metaphase spreads were prepared using stan-

dard protocols, and chromosomes were counted. Briefly, cells

were treated with 0.02 �g/�l colcemide for 1.5 hours and

harvested to be hydrated and fixed. Chromosomes were stained

with Giemsa and then counted (15 cells).

Effect of LIF and bFGF on Subcultured NEP CellsTwo groups of cultured NEP cells, one less than 1 month and the

other approximately 6 months in NB27 (described in previous

section), were dissociated by 0.05% trypsin to obtain a single-

cell suspension, and 50,000 cells/cm2 were plated in one of the

subculture media on polyornithine- and laminin-coated dishes.

Two concentrations of two growth factors (LIF, 0 or 10 ng/ml;

bFGF, 0 or 20 ng/ml) in NB27 were applied to cells. Cells were

harvested from each group, and nuclei were counted by flow

cytometry on days 1 and 14. Plating efficiency rate was calcu-

lated as the ratio of cells harvested to cells plated on day 1.

Proliferation was measured on day 14. For each replicate,

128 Adherent Differentiation of hESCs into NEP Cells

counted nuclei from the four treatment groups were added to

obtain an overall total. The total cell number within each group

was then divided by the overall total cell number and expressed

as a percent. This data conversion was carried out to reduce

biological variation due to replicate preparation.

Effect of Oxygen Concentration on SubculturedNEP CellsTo examine the effect of oxygen concentration on cell prolifer-

ation and viability, the subcultured NEP cells (described above)

were dissociated by 0.05% trypsin, and 2 � 105 cells/cm2 were

plated and propagated using the NEP subculture process, except

one group was cultured at oxygen concentration of 20% and the

other group was cultured at 5% O2. After 7 days of culture, cells

were harvested to calculate total cell number and viable cell

number, as described previously.

Expression of SOX Genes in Freshly Derived andLong-Term Subcultured NEP CellsAlong with their differentiation potential to neuron and glial cells

and NEP marker expression, expression of SOX genes was exam-

ined both in freshly derived (early) and long-term subcultured (late)

NEP cells. To examine expression of SOX genes, RNA was

isolated from early and late NEP cells using Trizol. For reverse

transcription–polymerase chain reaction (PCR), 2 �g of total RNA

from each sample was treated with DNase (Promega, Madison, WI,

http://www.promega.com). RNA 1 �g was converted to cDNA by

using the Superscript III kit (Invitrogen, Carlsbad, CA, http://

www.invitrogen.com) using oligo dT as a primer, and 1 �g was

prepared without reverse transcription to serve as control for ex-

clusion of genomic amplification. ReadyMix REDTAQ (Sigma)

was used, and 50 ng of cDNA was added for the PCR reaction for

35 cycles with denaturing at 95°C for 30 seconds, annealing at

60°C for 30 seconds, and elongation at 72°C for 30 seconds. For

SOX1, commercial primer and probe for real-time PCR were used,

and 25 ng of cDNA was subjected to real-time PCR (Applied

Biosystems, Foster City, CA, http://www.appliedbiosystems.com)

according to the manufacturer’s instructions. After amplification,

products were separated on 2% agarose gel and visualized using

ethidium bromide (EtBr) staining under UV light. Primer se-

quences (forward and reverse), size of the product, and PCR

condition were as follows: SOX2 (5�-AGT CTC CAA GCG ACG

AAA AA-3� and 5�-GCA AGA AGC CTC TCC TTG AA-3�, 141

bp); SOX3 (5�-GAG GGC TGA AAG TTT TGC TG-3� and

5�-CCC AGC CTA CAA AGG TGA AA-3�, 131 bp); � actin

(4326315E, Applied Biosystems); SOX1 (Hs00534426 s1, Ap-

plied Biosystems).

Statistical AnalysisFor each parameter, significance of main effects was determined

using the GLM procedure of SAS 8.01. Significance of differ-

ences among individual treatment means was determined by the

least-square means method. Differences were considered signif-

icant at p � .05.

RESULTS

Effect of ES, DN2, and MEDII Media onDifferentiation of ES Cells Cultured withFeeder CellsAfter 7 days of culture, hESCs in ES medium (stage 1) prolif-

erated to form multicell layers. These cells expressed both

Nestin (Fig. 2A) and Musashi-1 and the pluripotent marker

Oct-4. When expression was quantitated for each phenotype

marker using flow cytometry, 74.9%, 77.5%, and 88% of total

cells were positive for Oct-4, Nestin, and Musashi-1, respec-

tively (Table 1). These results showed that in ES medium, ESC

transition to NEP cells occurred gradually, with intermediate

stages expressing both Oct-4 and the Musashi-1, Nestin. This

overlap in expression was observed using both flow cytometry

and immunocytochemistry, including double-staining for both

Nestin and Oct-4 (Fig. 2A).

When the stage 1 cells were cultured for an additional week

in either DN2, MEDII, or ES media (stage 2), resulting colony

morphologies were compared and differences were observed

between ES medium and DN2 or MEDII media. DN2- and

MEDII-cultured stage 2 cells developed neural tube–like struc-

tures (Fig. 3A), whereas ES medium–cultured stage 2 cells

failed to form these structures (Fig. 3B). When cells were

examined under the microscope, nuclear staining indicated the

distinct cell arrangement (neural tube–like structures) developed

in MEDII- and DN2-derived populations that was not seen in

ES-derived populations. There was no morphological difference

between DN2- and MEDII-derived stage 2 cells; therefore,

quantitative data were obtained only for ES- and MEDII-derived

stage 2 cells (Table 1). The pluripotent cell expression marker

Oct-4 decreased in both groups from 74.9% (stage 1) to 32.6%

and 18.8% for ES and MEDII stage 2 groups, respectively (p �

.05). Both cell lines (BG01 and BG02) exhibited similar mor-

phological changes.

Effect of ES, DN2, and MEDII Media onDifferentiation of Stage 2 Cells in AdherentCell Culture Without Feeder CellsSimilar to results from stage 2 cells, we found differences for

stage 3 cells cultured in ES medium compared with cells cul-

tured in MEDII or DN2 media after feeder cell removal. After

feeder cell removal, cell culture gave rise to enriched rosette

formation in MEDII or DN2 media, characteristic of NEP cell

formation (Fig. 3C), but ES medium–derived cell culture re-

sulted in cells with large nucleus-to-cytoplasmic ratios, charac-

129Shin, Mitalipova, Noggle et al.

www.StemCells.com

teristic of ESCs (Fig. 3D). Both MEDII and DN2 groups devel-

oped a similar differentiation pattern with distinct structure of

neural tube–like formation [15] and further rosette-enriched

populations.

In addition to microscopic examination, quantitative data

obtained from whole populations indicated differences between

cell populations. When cells were differentiated in MEDII me-

dium, the percent of cells expressing Oct-4 was decreased

dramatically (74.9% at stage 1 versus 17.4% at stage 3). Fur-

thermore, stage 3 MEDII-cultured cell populations with rosette

structures showed expression of Nestin and Musashi-1, markers

found in early neural stem cells. However, most stage 3 ES

medium–cultured cell populations retained their Oct-4 expres-

sion even after spontaneous differentiation (74% at stage 1

versus 62.8% at stage 3). In accordance with the flow cytometry

results, immunocytochemistry demonstrated that for cells cul-

tured in ES medium, stage 3 cell populations were positive for

both Nestin and Oct-4 (Fig. 2C), whereas rosette-forming stage

3 cells cultured in MEDII medium had only increased Nestin

staining without Oct-4 expression (Nestin�/Oct-4�; Fig. 2B).

These results indicate that in adherent cell cultures without

feeder cells, DN2 and MEDII medium promote differentiation

to NEP cells whereas ES medium does not.

Effect of MEDII Medium and Low Cell Density onCell Survival of Stage 2 Differentiating Cells (Tube-Like Structure–Forming Cells)To obtain enriched populations of the desired cells (Nestin�/

Oct-4�), we attempted single-cell passaging to propagate the

differentiating cells in various concentrations of MEDII. A 50%

MEDII medium was used based on previous mESC MEDII

neural differentiation studies [28]. However, no previous reports

have tested different concentrations of MEDII on single-cell or

clonal propagation of NEP cells.

In an attempt to propagate stage 2 cleaner populations, these

cells were single-passaged in one of four concentrations of

MEDII serum–deprived medium in feederless cultures (Table

2). Regardless of treatment, significant cell death was observed;

without MEDII, few cells survived and/or propagated (1.9% �

1.2% cell survival). However, when these cultures were supple-

mented with as little as 25% MEDII-conditioned medium, there

was a tenfold increase in surviving colony-forming cells (22.3%

cell survival). Cell survival and cell propagation were further

improved and optimized at the 50% MEDII level, with 40,200

(40.2%) of the original cells surviving or propagating over the 5

days in culture. Although MEDII treatment significantly in-

creased the number of cells at 10 days of culture, it was obvious

that most cells passaged in this manner were lost during the first

24 hours of culture. Therefore, a TUNEL assay was used to

determine if these cells were undergoing apoptosis. At 6 hours

Figure 2. (A-C): Phenotype marker expression of cells counter-stained with Oct-4 (green), Nestin (red), and 4�,6�-diamidino-2-phenylindole (blue). (A): Stage 1 cells double stained both by Oct4and Nestin. (B): Stage 3 cells developed in DN2 medium–enrichedrosette formation (MEDII-cultured cells were similar, so the dataare not shown). (C): Stage 3 cells developed in embryonic stem(ES) medium. Bar � 100 �m. Stage 1 cells are ES cells that haveproliferated for 7 days in ES medium. Stage 2 cells are stage 1 cellsthat have been further subjected to either ES or MEDII medium for7 days. Stage 3 cells are stage 2 cells that have been further culturedfor 3 days in respective medium with the feeder layer removed.

130 Adherent Differentiation of hESCs into NEP Cells

of culture, 25% of both the 0% MEDII and 50% MEDII single-

passaged cell cultures had undergone apoptosis. The apoptotic

population increased to 36% and 38% for 0% and 50% MEDII

groups, respectively, by 24 hours of culture.

Characterization and Examination of DifferentiationCapacity of Derived NEP CellsRosette-forming NEP cells were enriched in DN2 and MEDII

stage 3 groups and from clonally passaged cells. To characterize

Figure 3. (A, B): Phase-contrast image of stage 2 cells (A). Cells cultured in MEDII medium (DN2-cultured cells were similar, so the dataare not shown). (B): Cells cultured in embryonic stem (ES) cell medium. (C–D): Phase-contrast image of stage 3 cells. (C): Neuroepithelialcells in adherent cell culture without feeder cells in MEDII medium (DN2-cultured cells were similar, so the data are not shown). (D): Cellscultured in ES medium. Bar � 100 �m. Stage 2 cells are stage 1 cells that have been further subjected to either ES or MEDII medium for7 days. Stage 3 cells are stage 2 cells that been further cultured for 3 days in respective medium with the feeder layer removed.

Table 1. Phenotype marker expression changes over time

Stage 1a Stage 2b Stage 2b Stage 3c Stage 3c

Marker/group ES medium ES medium MEDII ES medium MEDII

Oct-4 74.9 � 3.0 32.6 � 3.5 18.8 � 8.4 62.8 � 3.5a 17.4 � 8.3d

Musashi 1 88.0 � 2.9 53.3 � 2.2d 76.6 � 6.1d 76.9 � 4.0 66.0 � 4.9Nestin 77.5 � 7.4 30.9 � 11.9 50.14 � 3.3 79.7 � 5.9 70.9 � 5.5

Cells positive to each phenotype marker were calculated to obtain a percent (means � S.E.) of total cell number.aStage 1 cells are ES cells that have been proliferated for 7 days in ES medium.bStage 2 cells are stage 1 cells that have been further subjected to either ES or MEDII medium for 7 days.cStage 3 cells are stage 2 cells that been further cultured for 3 days in respective medium with the feeder layer removed.dDifferent superscripts within each parameter and stage are significantly different; p � .05.Abbreviation: ES, embryonic stem.

131Shin, Mitalipova, Noggle et al.

www.StemCells.com

NEP cells, rosette structures were examined by using a combi-

nation of positive and negative markers. Nearly 100% of rosette-

forming cells were positive for the early NEP markers Nestin

and Musashi 1 (Figs. 4A, 4B) and negative for later stages of

differentiation markers A2B5, PSNCAM, � III tubulin, Hu,

GFAP, and O4. In addition, they did not express the mesodermal

marker muscle actin, the endodermal marker � fetoprotein, or

the pluripotent marker Oct4. Removal of FGF and LIF from the

culture medium resulted in further differentiation of NEP cells

to form intermediate precursors staining positive for A2B5 or

PSNCAM (Figs. 4C, 4D). After 14 days of culture in neurobasal

medium supplemented with B27 without bFGF, terminally dif-

ferentiated cell cultures contained neurons positive for Hu and

Tuj1 (Fig. 4E), astrocytes stained with GFAP (Fig. 4F), and

oligodendrocytes stained with O4 (Fig. 4G).

Effect of Medium, Supplement, Growth Factor, andOxygen Conditions on Proliferation and Viability ofSubcultures of Derived NEP Cells

Effect of Culture MediumThe effects of base media and supplements on survival of stage

3 NEP cells were determined to establish the most effective

subculture conditions. A higher percentage of cells cultured in

NN2 survived compared with cells cultured in DN2 (33.8%

DN2 versus 75.4% NN2, p � .05), indicating that derived NEP

cells survived better in neurobasal medium than DMEM me-

dium with N2 supplement. Furthermore, all three groups of

NN2-, DN2-, and MEDII-supplemented cultures developed ro-

sette structures. Also, the addition of MEDII to DN2 medium

increased cell survival rate from 33.8% to 77.6% (p � .05). In

contrast, there was no difference in survival rate or the mor-

phology of cells between N2 and B27 supplement when added

to the neurobasal medium (75.4% NN2 versus 74.6% NB27;

p � .05).

Subculture of NEP CellsThese derived NEP cells have been cultured for more than 6

months without losing this characteristic and maintained a

normal chromosomal number. Cells retained expression of

Nestin and Musashi-1 (Fig. 4H), and when terminally differ-

entiated in medium lacking bFGF and LIF, the cell popula-

tion included both neurons and glial cells (data not shown).

To further characterize freshly derived and subcultured NEP

cells, we examined SOX1, SOX2, and SOX3 gene expres-

sions in Oct-4 –negative early and late NEP cells. Both

groups expressed SOX2 and SOX3 (Fig. 5A). By using

real-time PCR, the SOX1 gene was amplified and the ampli-

con was visualized by EtBr staining. The SOX1 gene was

also expressed in both cell groups (Fig. 5B). When subcul-

tured NEP cell metaphase spreads were visualized by Giemsa

staining, all 15 samples examined were stable with 46 XY

chromosome numbers.

Effect of LIF and bFGFNEP cells propagated in NB27 for approximately 1 or 6 months

were subjected to different concentrations of LIF and bFGF, and

cell survival as well as cell proliferation was determined at 14

days (Table 3). For early NEP cells (1 month), the addition of

LIF, bFGF, or LIF � bFGF had no effect on plating efficiency,

which was only approximately 50%, indicating a relatively high

rate of cell death. In contrast, the presence of bFGF increased

cell proliferation more than fourfold (8.9% versus 38.5%; p �

.05), whereas LIF had no effect on proliferation of NEP cells

either in the presence or absence of bFGF. After 6 months in

LIF-supplemented culture, LIF, bFGF, and the combined groups

exhibited a higher plating efficiency than the control. bFGF had

a greater effect on cell proliferation than LIF (p � .05) for both

the short-term (�1 month) and long-term (6 months) NEP

cultures. However, only long-term cultured NEP cells demon-

strated increased proliferation rate for both LIF and bFGF

individually and in combination.

Effect of Oxygen ConcentrationAfter 7 days of culture in NB27 medium, total NEP cell number

was approximately 25% greater in 5% oxygen compared with

20% oxygen (p � .05) (Table 4). Considering that the plating

efficiency was 50% when NEP cells were dissociated, we esti-

mated that there was an approximately 2.5-fold increase in cell

proliferation for 5% oxygen and a 1.96-fold increase for 20%

oxygen.

DISCUSSION

The overall objective of these experiments was to obtain effi-

cient neural differentiation of hESCs and to develop a defined

medium that would be supportive of NEP cells and allow

Table 2. Effect of MEDII supplement on percent cell survival of dissociated stage 2a cells in serum-deprived and feeder cell–deprivedculture conditions (means � S.E.)

0% MEDII 25% MEDII 50% MEDII 100% MEDII

1.9% � 1.2%b 22.3% � 8.1%b 40.2% � 10.9%b 32.6% � 12.1%b

aStage 2 cells are stage 1 cells that have been further subjected to either embryonic stem cell or MEDII medium for 7 days.bDifferent superscripts within each parameter (row) are significantly different; p � .05.

132 Adherent Differentiation of hESCs into NEP Cells

enzymatic passage, thereby facilitating more controlled and

refined future studies. In contrast to previous reports, we used

both immunocytochemistry and flow cytometry analysis to ob-

tain both quantitative and morphological information on NEP

Figure continues on following page.

Figure 4. (A, B): Rosette-forming neuroepithelial cells stained with Nestin (A) or Musashi (B). (C, D): Intermediate precursor cells afterremoval of basic fibroblast growth factor (bFGF) and leukocyte inhibitory factor from the culture medium. (C): Cells stained for A2B5 (red)and 4�,6�-diamidino-2-phenylindole (DAPI) (blue). (D): Cells stained for PSNCAM (red) and DAPI (blue). (E–G): Terminally differentiatedneurons and astrocytes after 14 days of culture in neurobasal medium supplemented with B27 and L-glutamine, without bFGF. (E): Neuronsdouble stained for Hu C/D (green), Tuj1 (red), and DAPI (blue). (F): Astrocyte stained with GFAP (red) and DAPI (blue). (G):Oligodendrocyte stained with O4 (green) and DAPI (blue). (H): Long-term (10 months) cultured neuroepithelial cells stained with Nestin(green), Musashi (red), and DAPI (blue). Bar � 100 �m.

133Shin, Mitalipova, Noggle et al.

www.StemCells.com

formation at various stages of in vitro differentiation and culture

conditions. Most studies investigating mouse and human ESC

differentiation to neural progenitors have used methods involv-

ing cell aggregation or EB formation. EB formation in serum-

containing medium included cells differentiated into NEP cells

[15, 29] but also led to stochastic differentiation yielding mul-

tiple cell lineages, thus limiting the overall yield of the desired

NEP cells [30]. Dang et al. [30] compared EB differentiation

cultures to adherent differentiation culture and reported that cell

number limitation was not a factor in adherent differentiation

cultures. In addition, they showed that adherent differentia-

tion seemed to exclude cell differentiation toward hemato-

poietic development. Ying et al. [31] used adherent differen-

tiation with mESCs and obtained efficient neural

commitment. In our study, hESCs were allowed to differen-

tiate adherently in serum-free medium, and our findings

indicate efficient production of NEP cells. In our system,

feeder cells were present during the first 14 days, allowing

hESCs to proliferate and differentiate. Subpopulations of

stage 2 cells infiltrated underneath the feeder cell layer to

attach firmly on culture plates. Serum deprivation apparently

is crucial for ectodermal derivation [32], and removal of the

feeder cell layer produced homogenous rosette formation

from homogenous spread of cells in adherent culture condi-

tions.

In an attempt to follow the spatial and temporal differenti-

ation of ESCs to neural lineages, we divided the process into

three stages. We found that Oct-4 expression gradually de-

creased with the onset of expression of Nestin and Musashi-1,

markers associated with NEP cells. At an initial stage (stage 1),

when cells were allowed to proliferate in ES medium, most cells

were positive for both pluripotent and NEP cell markers. This

Oct4 and Nestin double staining has not been reported in other

species. However, in mESCs, an intermediate cell status was

reported as primitive ectoderm-like cells [26]. It is not certain

whether hESCs go through this intermediate stage, although

further studies on Nestin and Oct4 double-staining populations

could help to answer this question. Further differentiation re-

sulted in morphological changes, including neural tube–like

structures, when cells were cultured in either DN2- or MEDII-

supplemented media but not in ES medium. Visual inspection

indicated that in both DN2 and MEDII groups, cell populations

developed rosette structures in more than 70% of the total

culture area, and there was little difference in rosette numbers or

appearance between these two groups. The neural tube–like

structures and rosettes have been previously identified as char-

acteristic morphology of NEP cells [15].

At stage 3, we found that removal of LIF, nonessential

amino acids, KSR, and undefined factors in serum forced ESCs

to choose a neurectodermal fate. Rosette formation was not

promoted when cells were cultured in ES medium with these

factors included. Instead, cells left in ES medium retained their

Oct-4 expression and delayed progression to a more differenti-

ated state. This finding is similar to that seen with spontaneous

differentiation. For example, Reubinoff et al. [16] showed that

more than 4 weeks of culture was required for ESCs to differ-

entiate into NEP cells, and their system also resulted in endoder-

mal and mesodermal differentiation [16]. Our results indicated

that the total cell number expressing Oct4 was higher for stage

3 than for stage 2 for the ES medium group. This surprising

result may be due the techniques used rather than a change in

Oct4 expression in this group. During feeder cell removal, cells

were separated into two populations, one removed with the

feeder layer and the other remaining to proliferate further in ES

medium. It is likely that spontaneously differentiating Oct4-

negative cells were removed, leaving Oct4-retaining cells be-

hind in the ES medium group.

MEDII added to DN2 medium did not improve tube-like

structure formation (stage 2) or subsequent progression to

stage 3 adherent colonies. The effect of MEDII was distinct,

however, on low-cell-density NEP cell derivation. When tube

Figure 4. (Continued)

134 Adherent Differentiation of hESCs into NEP Cells

structure–forming cells were dissociated and passaged in

DN2, more than 98% of cells died. This finding is similar to

results obtained with mouse cells. Tropepe et al. [32] re-

ported that just 0.2% of the starting cell population was able

to form neurospheres and that supplementing with LIF can

improve this process. When we supplemented the dissociated

cell cultures with MEDII medium, a higher proportion of

cells attached to the substrate and then subsequently prolif-

erated to form NEP colonies. This finding was expected,

because two of the known components of MEDII are fi-

bronectin and LIF [28]. Furthermore, TUNEL assay results

suggest that the MEDII does not decrease apoptosis but

prevents cell death or increases proliferation by some other

mechanism. When cells retained their cell contact and main-

tained attachment to the substrate, supplement of MEDII had

no beneficial effects over DN2 medium. Using just morpho-

logical analysis, when cells were not disaggregated and their

cell-to-cell contact remained, a more uniform and enriched

rosette formation was obtained after another 3 to 4 days of

culture in either DN2 or MEDII than cells passaged as single

cells. NEP is designated as an unrestricted neural cell popu-

lation based on Nestin expression, and these cells are non-

immunoreactive to any restriction markers, such as A2B5 and

PSNCAM [11]. Our results showed that the rosette-enriched

stage 3 NEP cells had the same phenotype profile as rodent

Figure 5. Both early and late neuroepithelial cells express SOX1, SOX2, and SOX3. Reverse transcription–polymerase chain reactionanalysis of the expression of SOX1 (B), SOX2, and SOX3 (A). Panels show 2% agarose gels stained with ethidium bromide. Genomiccontamination was monitored by sample prepared without reverse transcription (-). For size marker, 1-kb DNA ladder was used. The sizeof SOX2 is 141, and SOX3 is 131.

Table 3. Effect of basic fibroblast growth factor (bFGF) and leukemia inhibitory factor (LIF) supplementation on plating efficiency andproliferation of neuroepithelial cells (means � S.E.)

<1 month of culture

�/� bFGF/� �/LIF bFGF/LIF

Plating efficiency 51,267 � 13,487 53,733 � 11,293 50,767 � 11,305 51,400 � 8,713(% plated cell #)a (51.3 � 13.5) (53.7 � 11.3) (50.8 � 11.3) (51.4 � 8.7)Proliferation 61,516 � 10,155 308,274 � 68,538 40,365 � 4,303 347,927 � 79,011(% total cell #)a (8.9 � 1.9) (38.5 � 4.2) (6.9 � 2.0) (45.8 � 4.5)

6 months of culture

�/� bFGF/� �/LIF bFGF/LIF

Plating efficiency 70,480 � 2,500 93,013 � 8,623 92,072 � 876 100,326 � 8,573(% plated cell #)a (35.2 � 1.3) (46.5 � 4.3) (46.0 � 0.4) (50.2 � 4.3)Proliferation 123,154 � 3,398 501,150 � 37,743 278,611 � 4,585 75,3847 � 41,196(% total cell #)a (7.4 � 0.2) (30.3 � 2.4) (16.8 � 0.2) (45.5 � 2.4)

aDifferent superscripts within each parameter (row) are significantly different; p � .05.

Table 4. Effect of oxygen (O2) concentration on viability andproliferation of neuroepithelial cells (means � S.E.)

ViabilityCell number(% of total)

High O2 80% � 4%a 196,268 � 18,736(44.19% � 1.00%a)

Low O2 83% � 3%a 250,657 � 7,605(55.81% � 1.00%a)

aDifferent superscripts within each parameter (column) are signif-icantly different; p � .05.

135Shin, Mitalipova, Noggle et al.

www.StemCells.com

NEP or human NEP cells purified from fetal tissue. They

were not immunoreactive to restriction markers or to specific

differentiation cell markers of neurons or glial cells, but they

were immunoreactive to Nestin and Musashi-1. In addition,

the rosette-enriched population was not immunoreactive to

Oct-4 or mesodermal or endodermal markers. Mayer-Pros-

chel showed that neural cells derived from fetal tissue were

heterogeneous, with 50% of the population expressing A2B5

[11]. Another step of immunopanning was required to obtain

an enriched NEP population. In our study, enriched NEP cell

populations were obtained through an efficient differentiation

protocol. The flow cytometry results indicated that approxi-

mately 70% of the whole population expressed Nestin and

Musashi1, and immunocytochemistry showed almost 100%

of rosette structure expressed Nestin and Musashi1 without

Oct4 expression. Therefore, the combined immunocytochem-

istry and flow cytometry results suggested that 17% of Oct4

expression originated from nonrosette structure cells. As

differentiation progressed, cells expressing precursor mark-

ers of A2B5 or PSNCAM appeared (Figs. 4C, 4D), and

terminal differentiation resulted in neurons that expressed Hu

and Tuj1, oligodendrocytes that expressed O4, and astrocytes

that expressed GFAP (Figs. 4E– 4G). However, before being

classified as multipotent stem cell, clonal derivation must be

demonstrated and will require additional cell culture ad-

vances because our attempts to clonally propagate yielded

poor survival rates. Further studies were conducted to further

define medium requirements that would support NEP cells

and allow enzymatic passage and long-term culture of these

cells. We tested two base media, DMEM/F12, which has been

used for various cell cultures, including somatic cell lines and

ESC culture, and neurobasal medium, which was formulated

for long-term culture of rat hippocampal neurons [33]. We

also tested three supplements: MEDII, N2, and B27. N2 is a

chemically defined concentrate developed to support growth

of neural cell lines and includes insulin, transferrin, proges-

terone, putrescine, and selenite. B27 is an optimized serum

substitute for low-density plating and growth of CNS neu-

rons. We found that the serum-free base medium DN2 did not

support these NEP cells. In this medium, cells lifted off the

plate at approximately day 7 of subculture and were trypan

blue positive. Although cells cultured in DN2 supplemented

with MEDII showed increased viability, a complex condi-

tioned medium like MEDII can confound and limit the ex-

amination of candidate growth factor effects. In this study,

comparison of DMEM/F12 and neurobasal medium showed

that neurobasal medium supported NEP stem cell culture

when supplemented either with N2 or B27. It also supported

the survival of dissociated cells and allowed them to prolif-

erate. However, it was observed that clonal propagation was

less efficient with a low cell-survival rate and that cell

survival was improved when cell-to-cell contact was main-

tained either by high-density dissociation culture or by trit-

urated clump culture. Therefore, neurobasal medium supple-

mented with B27 was chosen as proliferation medium and

further experiments were conducted using NEP cells cultured

in this medium. This medium has been shown in previous

studies to support survival and expansion of both adult neural

stem cells and fetal and postnatal brainstem neurons in vitro

[34, 35].

We also tested the effects of the growth factors LIF and

bFGF on subculture of NEP cells. Mouse neural stem cells

have been shown to be dependent on bFGF [25], and it was

critical for neurosphere formation [32]. The presence of LIF

also supports and increases neurosphere formation; however,

whether it acts by inducing differentiation of ESCs or by

enhancing proliferation is not clear [32]. In fetus-derived

human neural stem cells, supplementing with both hLIF and

bFGF enhanced proliferation rate [36]. In our study done

with short-term cultured NEP cells (�1 month), bFGF

seemed to promote cell proliferation but supplement with LIF

had little effect, nor was there a synergistic effect when LIF

was combined with bFGF. Zhang et al. [15] reported that LIF

had no effect on proliferation of derived NEP after 14 days of

culture. However, we found that after 6 months, culture in

LIF-containing medium increased cell responsiveness and

cell proliferation was improved.

Physiological oxygen concentration does not exceed 5%;

however, in conventional cell culture, oxygen concentration

is maintained at 20%. In rat CNS stem cell culture, it has been

reported that reduced oxygen concentration helped to im-

prove cell proliferation and to reduce apoptosis [37]. We

tested whether reduced oxygen concentration produces the

same advantage on the growth of NEP cells derived from

hESCs. In agreement with this previous study, low oxygen

concentration improved cell proliferation rates approximately

25% after 1 week of culture. Because there was no difference

in viability as measured by flow cytometry, the increased cell

numbers do not seem to be due to increased initial cell

survival.

In this study, SOX genes were used to further characterize

derived NEP cells and long-term cultured NEP cells. Among

characterization markers, Nestin and Musashi1 have been pri-

mary phenotype markers for these cells [15, 16]. However, these

markers were not restricted to neural lineage. Along with their

negative expression for muscle actin and � fetoprotein, expression

of SOX genes was examined. In the mouse, SOX genes were

mainly expressed in developing nervous system, and SOX1 has

136 Adherent Differentiation of hESCs into NEP Cells

been used as target gene for neural stem cell isolation in mESC

differentiation [31]. Additionally, human SOX1, SOX2, and SOX3

are highly conserved [22, 38]. Among these SOX genes, SOX2 has

also been shown to be expressed in hESCs [39, 40], and we

observed that proliferating hESCs expressed SOX2 and SOX3

(unpublished data). In this study we showed that NEP cell cultures

expressed all three SOX genes. Both early and late NEP cells

expressed SOX1, SOX2, and SOX3, and there was no difference in

expression between the two populations. These results indicate that

expression of SOX genes in the absence of Oct4 can be used as

further verification for NEP cells.

CONCLUSION

In this study, we showed that NEP cells can be derived from

hESCs efficiently by adherent differentiation in defined me-

dium. Derived NEP cells were broadly characterized with phe-

notype markers and expression of SOX genes; in addition,

differentiation capacity was similar to that of in vivo purified

human NEP cells [11]. Further NEP cell subculture conditions

were optimized, and cells were propagated successfully for

more than 6 months without loss of differentiation potential or

stable chromosome number. Our efficient derivation and prolif-

eration of NEP cells demonstrates that this system can serve as

an in vitro model for the examination of human neural devel-

opment. A defined culture system would be ideal for further

studies of effects of extrinsic factors on neuronal cell fate

decision. In addition, long-term cultured NEP cells may be good

candidates for replacement cell therapy, with little possibility of

pluripotent cell contamination.

ACKNOWLEDGMENTS

We wish to thank Deb Weiler for preparing feeder layers; Karen

Jones and Olivia Wei, Kate Hodges and Allison Adam for flow

cytometry support; and Mary Anne Della-Fera for manuscript

preparation. This work was supported in part by BresaGen and

hESC Supplement to R21NS44208 (NIH).

DISCLOSURES

S.L.S., M.M., and D.T. have acted as consultants for Bresagen

within the last 2 years.

REFERENCES

1 Thomson JA, Itskovitz-Eldor J, Shapiro SS et al. Embryonic stem celllines derived from human blastocysts. Science 1998;282:1145–1147.

2 Reubinoff BE, Pera MF, Fong CY et al. Embryonic stem cell lines fromhuman blastocysts: somatic differentiation in vitro. Nat Biotechnol 2000;18:399–404.

3 McKay R. Stem cells in the central nervous system. Science 1997;276:66–71.

4 Bjorklund A, Lindvall O. Cell replacement therapies for central nervoussystem disorders. Nat Neurosci 2000;3:537–544.

5 Jain M, Armstrong RJ, Tyers P et al. GABAergic immunoreactivity ispredominant in neurons derived from expanded human neural precursorcells in vitro. Exp Neurol 2003;182:113–123.

6 Borlongan CV, Tajima Y, Trojanowski JQ et al. Transplantation ofcryopreserved human embryonal carcinoma-derived neurons (NT2Ncells) promotes functional recovery in ischemic rats. Exp Neurol 1998;149:310–321.

7 Morrison SJ. The last shall not be first: the ordered generation of progenyfrom stem cells. Neuron 2000;28:1–3.

8 Amit M, Carpenter MK, Inokuma MS et al. Clonally derived humanembryonic stem cell lines maintain pluripotency and proliferative poten-tial for prolonged periods of culture. Dev Biol 2000;227:271–278.

9 Liu S, Qu Y, Stewart TJ et al. Embryonic stem cells differentiate intooligodendrocytes and myelinate in culture and after spinal cord trans-plantation. Proc Natl Acad Sci U S A 2000;97:6126–6131.

10 Kalyani A, Hobson K, Rao MS. Neuroepithelial stem cells from theembryonic spinal cord: isolation, characterization, and clonal analysis.Dev Biol 1997;186:202–223.

11 Mayer-Proschel M. Human neural precursor cells–an in vitro character-ization. Clin Neurosci Res 2002;2:58–69.

12 Kennea NL, Mehmet H. Neural stem cells. J Pathol 2002;197:536–550.

13 Alvarez-Buylla A, Garcia-Verdugo JM, Tramontin AD. A unified hy-pothesis on the lineage of neural stem cells. Nat Rev Neurosci 2001;2:287–293.

14 Mujtaba T, Piper DR, Kalyani A et al. Lineage-restricted neural precur-sors can be isolated from both the mouse neural tube and cultured EScells. Dev Biol 1999;214:113–127.

15 Zhang SC, Wernig M, Duncan ID et al. In vitro differentiation oftransplantable neural precursors from human embryonic stem cells. NatBiotechnol 2001;19:1129–1133.

16 Reubinoff BE, Itsykson P, Turetsky T et al. Neural progenitors fromhuman embryonic stem cells. Nat Biotechnol 2001;19:1134–1140.

17 Lardon J, Rooman I, Bouwens L. Nestin expression in pancreatic stellatecells and angiogenic endothelial cells. Histochem Cell Biol 2002;117:535–540.

18 Frojdman K, Pelliniemi LJ, Lendahl U et al. The intermediate filamentprotein nestin occurs transiently in differentiating testis of rat and mouse.Differentiation 1997;61:243–249.

19 Sejersen T, Lendahl U. Transient expression of the intermediate filamentnestin during skeletal muscle development. J Cell Sci 1993;106:1291–1300.

20 Collignon J, Sockanathan S, Hacker A et al. A comparison of theproperties of Sox-3 with Sry and two related genes, Sox-1 and Sox-2.Development 1996;122:509–520.

21 Uwanogho D, Rex M, Cartwright EJ et al. Embryonic expression of thechicken Sox2, Sox3 and Sox11 genes suggests an interactive role inneuronal development. Mech Dev 1995;49:23–36.

22 Malas S, Duthie SM, Mohri F et al. Cloning and mapping of the humanSOX1: a highly conserved gene expressed in the developing brain.Mamm Genome 1997;8:866–868.

23 Stevanovic M. Modulation of SOX2 and SOX3 gene expression duringdifferentiation of human neuronal precursor cell line NTERA2. Mol BiolRep 2003;30:127–132.

137Shin, Mitalipova, Noggle et al.

www.StemCells.com

24 Calhoun JD, Lambert NA, Mitalipova MM et al. Differentiation ofrhesus embryonic stem cells to neural progenitors and neurons. BiochemBiophys Res Commun 2003;306:191–197.

25 Okabe S, Forsberg-Nilsson K, Spiro AC et al. Development of neuronalprecursor cells and functional postmitotic neurons from embryonic stemcells in vitro. Mech Dev 1996;59:89–102.

26 Rathjen J, Lake JA, Bettess MD et al. Formation of a primitive ectodermlike cell population, EPL cells, from ES cells in response to biologicallyderived factors. J Cell Sci 1999;112:601–612.

27 Mitalipova M, Calhoun J, Shin S et al. Human embryonic stem cell linesderived from discarded embryos. STEM CELLS 2003;21:521–526.

28 Rathjen J, Haines BP, Hudson KM et al. Directed differentiation of pluri-potent cells to neural lineages: homogeneous formation and differentiation ofa neurectoderm population. Development 2002;129:2649–2661.

29 Schuldiner M, Yanuka O, Itskovitz-Eldor J et al. From the cover: effectsof eight growth factors on the differentiation of cells derived from humanembryonic stem cells. Proc Natl Acad Sci U S A 2000;97:11307–11312.

30 Dang SM, Kyba M, Perlingeiro R et al. Efficiency of embryoid bodyformation and hematopoietic development from embryonic stem cells indifferent culture systems. Biotechnol Bioeng 2002;78:442–453.

31 Ying QL, Stavridis M, Griffiths D et al. Conversion of embryonic stemcells into neuroectodermal precursors in adherent monoculture. NatBiotechnol 2003;21:183–186.

32 Tropepe V, Hitoshi S, Sirard C et al. Direct neural fate specification fromembryonic stem cells: a primitive mammalian neural stem cell stageacquired through a default mechanism. Neuron 2001;30:65–78.

33 Brewer GJ, Torricelli JR, Evege EK et al. Optimized survival of hip-pocampal neurons in B27-supplemented Neurobasal, a new serum-freemedium combination. J Neurosci Res 1993;35:567–576.

34 Wachs FP, Couillard-Despres S, Engelhardt M et al. High efficacy ofclonal growth and expansion of adult neural stem cells. Lab Invest2003;83:949–962.

35 Kivell BM, McDonald FJ, Miller JH. Serum-free culture of rat post-nataland fetal brainstem neurons. Brain Res Dev Brain Res 2000;120:199–210.

36 Carpenter MK, Cui X, Hu ZY et al. In vitro expansion of a multipotentpopulation of human neural progenitor cells. Exp Neurol 1999;158:265–278.

37 Studer L, Csete M, Lee SH et al. Enhanced proliferation, survival, anddopaminergic differentiation of CNS precursors in lowered oxygen.J Neurosci 2000;20:7377–7383.

38 Stevanovic M, Lovell-Badge R, Collignon J et al. SOX3 is an X-linkedgene related to SRY. Hum Mol Genet 1993;2:2013–2018.

39 Ginis I, Luo Y, Miura T et al. Differences between human and mouseembryonic stem cells. Dev Biol 2004;269:360–380.

40 Carpenter MK, Rosler ES, Fisk GJ et al. Properties of four humanembryonic stem cell lines maintained in a feeder-free culture system.Dev Dyn 2004;229:243–258.

138 Adherent Differentiation of hESCs into NEP Cells