NeuroToxicology 31 (2010) 277–290

Quantitative assessment of neurite outgrowth in human embryonic stemcell-derived hN2TM cells using automated high-content image analysis§

Joshua A. Harrill a, Theresa M. Freudenrich a, Dave W. Machacek b, Steven L. Stice b,c, William R. Mundy a,*a Systems Biology Branch, Integrated Systems Toxicology Division, National Health and Environmental Effects Research Laboratory, United States Environmental Protection Agency,

Research Triangle Park, NC 27711, United Statesb ArunA Biomedical, Athens, GA 30602, United Statesc Regenerative Bioscience Center, University of Georgia, Athens, GA 30602, United States

A R T I C L E I N F O

Article history:

Received 14 December 2009

Accepted 17 February 2010

Available online 25 February 2010

Keywords:

Neurite outgrowth

High-content analysis

Human embryonic stem cell-derived neural

culture

A B S T R A C T

Throughout development neurons undergo a number of morphological changes including neurite

outgrowth from the cell body. Exposure to neurotoxic chemicals that interfere with this process may

result in permanent deficits in nervous system function. Traditionally, rodent primary neural cultures

and immortalized human and non-human clonal cell lines have been used to investigate the molecular

mechanisms controlling neurite outgrowth and examine chemical effects on this process. The present

study characterizes the molecular phenotype of hN2TM human embryonic stem cell (hESC)-derived

neural cells and uses automated high-content image analysis to measure neurite outgrowth in vitro. At

24 h post-plating hN2TM cells express a number of protein markers indicative of a neuronal phenotype,

including: nestin, bIII-tubulin, microtubule-associated protein 2 (MAP2) and phosphorylated

neurofilaments. Neurite outgrowth in hN2TM cells proceeded rapidly, with a majority of cells extending

one to three neurites by 48 h in culture. In addition, concentration-dependent decreases in neurite

outgrowth and ATP-content were observed following treatment of hN2TM cells with either

bisindolylmaleimide I, U0126, lithium chloride, sodium orthovanadate and brefeldin A, all of which

have previously been shown to inhibit neurite outgrowth in primary rodent neural cultures. Overall, the

molecular phenotype, rate of neurite outgrowth and sensitivity of hN2TM cells to neurite outgrowth

inhibitors were comparable to other in vitro models previously characterized in the literature. hN2TM

cells provide a model in which to investigate chemical effects on neurite outgrowth in a non-transformed

human-derived cells and provide an alternative to the use of primary rodent neural cultures or

immortalized clonal cell lines.

Published by Elsevier Inc.

Contents lists available at ScienceDirect

NeuroToxicology

1. Introduction

During the differentiation of precursor cells to a committedneuronal lineage, newly formed neurons undergo a series ofextensive morphological changes as they mature includingemergence of neurites, neurite outgrowth, neurite branchingand establishment of cell–cell contacts (i.e. synaptogenesis).These morphological changes are necessary, although notsufficient, for the formation of the intricate network of neuralcircuits that facilitate nervous system function (Sanes et al.,

§ This manuscript has been reviewed by the National Health and Environmental

Effects Research Laboratory, U.S. Environmental Protection Agency, and approved

for publication. Approval does not signify that the contents reflect the views of the

Agency, nor does mention of trade names or commercial products constitute

endorsement or recommendation for use.

* Corresponding author at: USEPA, Integrated Systems Toxicology Division, B105-

06, Research Triangle Park, NC 27711, USA. Tel.: +1 919 541 7726.

E-mail address: mundy.william@epa.gov (W.R. Mundy).

0161-813X/$ – see front matter . Published by Elsevier Inc.

doi:10.1016/j.neuro.2010.02.003

2006). Early pre- and post-natal exposure to neurotoxiccompounds can interfere with these developmental events andcould potentially result in deficits in nervous system function inlater life stages (Rice and Barone, 2000; Costa et al., 2004;Grandjean and Landrigan, 2006). Neurite outgrowth, a criticalcomponent of this developmental chain of events, can berecapitulated in vitro using a variety of cell models, such asnervous system derived clonal cell lines and primary neuralcultures from the mammalian CNS. These models have becomevaluable tools for studying the molecular mechanisms thatcontrol neurite outgrowth (Zhang et al., 2009a,b; Yu and Malenka,2003; Redmond et al., 2002; Jin et al., 2003; Khaibullina et al.,2004) and for investigating the mechanism(s)-of-action forknown developmental neurotoxicants (Yamauchi et al., 2007;Lein et al., 2000; Howard et al., 2005; Audesirk et al., 1991). It hasalso been proposed that in vitro measures of neurite outgrowthcan be useful in high-throughput screening assays (Radio et al.,2008, 2010) as a means to identify potential developmentalneurotoxicants (Radio and Mundy, 2008).

J.A. Harrill et al. / NeuroToxicology 31 (2010) 277–290278

A number of recent publications advocate the use of in vitro cellculture models as tools for efficient identification and prioritiza-tion of chemicals that may be hazardous to humans (NRC, 2007;Coecke et al., 2007; Lein et al., 2005). Specifically, a report by theNational Academy of Sciences entitled ‘Toxicity testing in the 21stcentury: a vision and a strategy’ emphasizes the use of in vitro

models derived from human tissues (NRC, 2007). The impetus forthis point-of-view being that use of in vitro toxicity assays inhuman-derived cells, as opposed to cells from non-humanmammalian species, may decrease some of the uncertaintiesinvolved in evaluating the effects of chemicals and applying thatknowledge to address human risk (i.e. interspecies extrapolation)(NRC, 2007). In the case of assessing neurite outgrowth, there are anumber of immortalized and tumor-derived neural cell lines ofhuman origin currently available (Radio and Mundy, 2008; Harryand Tiffany-Castiglioni, 2005) as well as reliable methods for theculture of primary rodent neurons (Higgins and Banker, 1998).However, transformed clonal cell lines or rodent primary neuralcultures may not accurately represent human nervous systembiology (LePage et al., 2005; Allen et al., 2005). The response orsensitivity of neural cultures to toxic compounds may differ acrossspecies or across models as noted in previous reports examiningthe effects of ethanol, staurosporine and mercury on the processesof neural development in vitro (Breier et al., 2009; Moors et al.,2009; Cedrola et al., 2003). In the context of evaluating chemicalsas potential developmental neurotoxicants in humans, a humanstem cell-derived culture model may prove more informative thantransformed cell lines and also circumvent problems associatedwith primary human neural cell availability (McNeish, 2004).

Recent advances in stem cell biology have resulted in methodsin which embryonic stem cells of human origin can bedifferentiated along a neuronal lineage and grown in dissociatedcultures (Reubinoff et al., 2001; Zhang et al., 2001). Given theproper extracellular cues and growth substrate, the maturing cellscan display morphological characteristics and express a number ofprotein markers indicative of a neuronal lineage (Reubinoff et al.,2001; Zhang et al., 2001; Shin et al., 2006). These stem cell-derivedmodels may serve as valuable tools for examining chemical effectson neuronal maturation, including neurite outgrowth, usinghuman cells. In the context of high- to medium-throughputchemical screening, the use of stem cell-derived neuronal culturesalso has some potential caveats. Namely, the time- and labor-intensive process of differentiating a proliferative population ofstem cells (hESC) to a population of terminally differentiatedneurons, which can take weeks (Reubinoff et al., 2001; Zhang et al.,2001; Shin et al., 2005). The present study describes thephenotypic characteristics and measures neurite outgrowth inhN2TM cells, a novel, commercially available, hESC-derivedneuronal model which is provided in a pre-differentiated statefor rapid end user applications (ArunA Biomedical, Athens, GA).

The hN2TM cell line is derived from neuroepithelial cells ofWA09 hESC (Thomson et al., 1998) origin according to a previouslydescribed protocol (Shin et al., 2005, 2006). Importantly, asopposed to other methods of deriving neural progenitors throughthree-dimensional neurosphere and embryoid body formations(Reubinoff et al., 2001; Zhang et al., 2001), these adherentmonolayer cultures are uniformly exposed to growth factorsand/or morphogens throughout their propagation. Prior todifferentiation into hN2TM cells the population was confirmedkaryotypically normal, >95% nestin positive and <3% OCT-4positive (Shin et al., 2006). The cells were produced in bulk bypropagation for an additional 2 weeks beyond the neuroepithelialstage by removal of bFGF from the media and cryopreserved(ArunA Biomedical, Athens, GA) for end user applications. In thepresent study, the utility of dissociated hN2TM cultures as an in

vitro model for neurite outgrowth was assessed using automated

high-content image analysis (HCA). In addition, the molecularphenotype of these cells was examined using immunocytochem-ical staining.

2. Methods

2.1. Materials

hN2TM human neural cells, growth media and supplementswere obtained from ArunA Biomedical, Inc. (Athens, GA). Thegrowth substrates poly-L-lysine and laminin were purchased fromSigma–Aldrich (St. Louis, MO). Bisindolylmaleimide I (Bis1) andbrefeldin A were purchased from Calbiochem, Inc. (San Diego, CA).Dimethyl sulfoxide (DMSO, dosing vehicle), lithium chloride (LiCl)and sodium orthovanadate (Na3VO4) were purchased from Sigma–Aldrich (St. Louis, MO). U0126 was purchased from Promega Corp.(Madison, WI). Hoechst 33258 dye, immunocytochemical stainingbuffer (ISB), mouse monoclonal antibody against bIII-tubulin andDyLight1 488-conjugated rabbit anti-mouse IgG secondary anti-body were components of a Cellomics1 Neurite OutgrowthHitKitTM purchased from ThermoFisher Scientific, Inc. (Waltham,MA). Mouse monoclonal antibodies for microtubule-associatedprotein 2 (MAP2) and nestin were purchased from Millipore, Inc.(Billerica, MA). Mouse monoclonal antibody SMI-312 whichdetects phosphorylated forms of a variety of axonal neurofilamentswas purchased from Covance, Inc. (Princeton, NJ).

2.2. Cell culture

Costar1 96-well polystyrene cell culture dishes (Corning, Inc.,Corning, NY) were coated with a solution of 50 mg/ml poly-L-lysinein sterile H2O for 2 h (37 8C), rinsed once with sterile H2O and thencoated with a solution of 20 mg/ml laminin in sterile phosphate-buffered saline (PBS) for 2 h. Plates were then rinsed once withwarm PBS prior to plating of hN2TM cells. Cells were stored at�70 8C and thawed at time of use. After thawing at 37 8C, cells weresuspended in serum-free ArunA basal medium supplemented withArunA Neural Supplement (ANSTM), leukemia inhibitory factor (LIF,10 ng/ml), penicillin (50 U/ml), streptomycin (50 mg/ml) and2 mM L-glutamine. A small aliquot of cells were then stainedwith 0.4% trypan blue and counted on a hemocytometer. Live cellyields post-thawing ranged from 60 to 80%. Cells were plated atdensities ranging from 2500 to 10,000 cells/well (8.33 � 103 to3.33 � 104 cells/cm2, respectively) based on the number of livecells counted. The number of cells per cm2 (i.e. plating density) wascalculated by dividing the number of cells per well by the well area(0.3 cm2). Cells were maintained in a humidified incubator at 37 8Cwith a 95% air/5% CO2 atmosphere.

2.3. Chemical treatment

Concentration ranges for the five test compounds were asfollows: brefeldin A (0.01, 0.03, 0.1, 0.3, 1 mM), Bis1 (0.1, 0.3, 1, 3,10 mM), U0126 (0.3, 1, 3, 10, 30 mM), sodium orthovanadate (1, 3,10, 30, 100 mM) and lithium chloride (0.3, 1, 3, 10, 30 mM).Guidance for concentration range selection was based onpreviously published works cited in Table 1. Stock solutions(1000�) of the highest tested concentration of Bis1, brefeldin A andU0126 were prepared in pure DMSO and stock solutions for theremainder of the concentration ranges were prepared by serialdilution in DMSO. Dosing solutions for each chemical concentra-tion were prepared by diluting stock solutions 1:100 in ArunAbasal media. Stock and dosing solutions of LiCl and Na3VO4 wereprepared using the same method, save that stock solutions wereprepared in ArunA basal medium as opposed to DMSO. 10 ml ofdosing solutions were then added to the cell culture wells

Table 1Chemicals with evidence of neurite outgrowth inhibition in rodent primary neural cultures.

Compound Study Cell type Concentration range Effects

Bis1 Radio et al. CGC (PND7)a,b 1 nM–100 mM # in neurite outgrowth

U0126 Radio et al. CGC (PND7) 1 nM–100 mM # in neurite outgrowth

LiCl Takahashi et al. Hippocampal (E18)c 2–15 mM # ratio of axon length to cell body diameter

Munoz-Montano et al. CGC (PND7) 1–20 mM Biphasic " then # in % cells with long neurites

Hollander and Bennett DRG (E8-9)d 25 mM # in neurite length

Na3VO4 Mandel and Banker Hippocampal (E18) 25–100 mM # length of longest neurite

Bref A Jareb and Banker Hippocampal (E18) 0.14–3.57 mM # in % of neurons with axons

a PNDx = post-natal day x.b CGC = cerebellar granule cells.c Ex = embryonic day x; designates the age of mouse or rat pups at the time of culture.d DRG = dorsal root ganglia.

J.A. Harrill et al. / NeuroToxicology 31 (2010) 277–290 279

containing growing hN2TM cells in 90 ml of media to achieve thenominal media concentrations listed above. Final DMSO concen-trations were 0.1% for all treatment wells containing varyingconcentrations of brefeldin A, Bis1, U0126 and correspondingvehicle control wells. Dose solutions were applied to the cells 2 hafter plating. Cells were then returned to the incubator. ATP-content and neuronal morphology were then examined at either 2,6, 24 or 48 h as detailed below. The effects of DMSO on ATP-contentand neuronal morphology were examined by preparing dosingsolutions of 0.5–50% DMSO in ArunA basal media and applying totissue culture wells at a 1:10 dilution (final concentration rangeexamined: 0.05–5% DMSO).

2.4. Immunocytochemistry

Cell cultures containing 100 ml of media volume were removedfrom the incubator and fixed in situ with 100 ml of a warm (37 8C)solution of 8% paraformaldehyde (PFA)/8% sucrose and 0.1%Hoechst 33258 dye in PBS for 20 min. This fixation methodeffectively preserved the fine morphological features of thecultures. Fixative was then gently aspirated and cells washedthree times with immunocytochemical staining buffer (ISB).Primary antibodies diluted in ISB were then applied as follows:bIII-tubulin (1:800), MAP2 (1:800), nestin (1:400) and panaxonalneurofilament SMI-312 (1:200) for 1 h at room temperature. Theentire antibody panel was used to characterize the neuronalphenotype of the hN2TM cells, while bIII-tubulin was specificallyused to label cell bodies and neurites for high-content imageanalysis (HCA). Following incubation in primary antibodies, cellswere washed three times with ISB and incubated with a 1:500dilution of DyLight1 488-conjugated rabbit anti-mouse IgGsecondary antibody in ISB for 1 h at room temperature, protectedfrom light. Cells were then washed twice with ISB, twice withDulbecco’s phosphate-buffered saline (PBS), and stored at 4 8Cprior to image acquisition and analysis.

2.5. Measurements of hN2TM morphology

bIII-Tubulin stained cell cultures were allowed to warm to roomtemperature. Plates were then loaded into a Cellomics ArrayScanVTI HCS reader high-content imaging system (ThermoFisherScientific, Waltham, MA) for automated image acquisition andmorphometric analyses. This system consists of an epifluorescentmicroscope with an EXFO X-citeTM 120 metal-halide arc lamp,motorized imaging objectives, stage and excitation/emission filterwheel and a 12-bit high-resolution CCD camera connected to a DellIntel1 XenonTM computer terminal with 2 GHz processor. Imageacquisition and storage was performed using the vHCS Scansoftware package, version 6.6.1.4. Matched fluorescent images ofHoechst-stained nuclei (Fig. 1A) and bIII-tubulin/DyLight1 488immunolabeled cells (Fig. 1B) were acquired using 365/515

(channel 1) and 475/515 (channel 2) nm excitation/emission filtercouplings, respectively, with a 20� objective (Zeiss, Inc., Thorn-wood, NY). Fixed integration times for image acquisition in eachchannel were determined by manual sampling of control-treatedwells across multiple plates. A matching pseudocolored compositeimage of Hoechst-stained nuclei (blue) and bIII-tubulin/DyLight1

488 labeled cell bodies and neurites (green) is shown in Fig. 1C.Image analysis was performed in real-time with a manually

optimized version of the Cellomics Neural Profiling Bioapplicationv3.5. Optimization of nucleus and cell body selection criteria, aswell as cell body masking and neurite tracing parameters, weredetermined a priori by using representative images from untreatedcultures following 24 h of growth at a density of 7500 cells perwell. Manual comparison of representative images from untreatedcontrol wells to matched tracing overlays was performed duringoptimization to insure the algorithm settings provided an accuratetrace. A full listing of parameters for the algorithm used herein isavailable from the authors upon request.

The Neural Profiling Bioapplication performs automated imageanalysis in a sequential manner as follows. Briefly, nuclei wereidentified in channel 1 as bright objects on a dark background(Fig. 1D). Nuclei with size and intensity values outside of theranges determined a priori for viable cells were identified in thechannel 1 image and rejected from further analyses (Fig. 1D,objects circled in orange). Spatial coordinates from the channel 1image were then superimposed on the matching channel 2 image.Cell body masks in channel 2 were then cast based on positionaldata from channel 1 nuclei and a set of user-defined geometric andsignal intensity-based parameters (Fig. 1E, blue and red traces).Cell bodies corresponding to valid neurons were then selected(Fig. 1E, blue traces) and invalid cell bodies rejected (Fig. 1E, redtraces). Parameters for valid cell body selection include thepresence of exactly one nucleus within the cell body mask, arequirement that the nucleus met the gating criteria imposed inchannel 1, a requirement that at least 25% of the nucleus perimeteris bounded by DyLight1 488 labeled cytoplasm and a requirementthat the total cell body area not exceed 4000 mm2.

Neurites emerging from the selected cell bodies were thenindividually traced and measured (Fig. 1F). For this study, neuriteswere defined as processes >10 mm in length. Neurites wereseparated from cell bodies at points when the half-width of thelabeled cytoplasm was less 3.6 mm across. In the case of neuriteswith an ambiguous origin (i.e. appearing to emerge from or contactmultiple cell bodies) the Neural Profiling Bioapplication traced theneurite from all potential origin points and retained the longestneurite for measurements of length and number of neurites perneuron. This effectively prevented repeated sampling of the sameneurite segment within each image.

Morphometric data from high-content image analysis (HCA)included measurements of the average number of neurites perneuron and total neurite length per neuron. Data for both

Fig. 1. Automated measurement of neurite outgrowth in hN2TM cells. Cells were grown in 96-well titer plates at a density of 7500 cells/well (2.5 � 104 cells/cm2) for 24 h, fixed

and fluorescently labeled. Cells were then imaged and neurite outgrowth measured. Images in panels A–F are 20�magnification high-resolution images obtained using the

ArrayScan VTI. (A) Nuclei labeled with Hoechst 33258 and visualized in channel 1. (B) Cell bodies and neurites labeled with bIII-tubulin/DyLight1 488 in channel 2. (C)

Pseudocolored composite image. (D) Nuclei are identified as bright objects on a dark field and masked; blue trace = selected nuclei, orange trace = rejected nuclei. An

expanded view of nuclei bounded by the yellow box is given in the panel D inset to better illustrate nuclei traces. (E) Cell body masks based on fluorescent intensity of bIII-

tubulin/DyLight1 488 labeling and position of channel 1 nuclei; blue trace = accepted, red trace = rejected cell body. Yellow arrows denote cells with cell body size and shape

parameters inside the accepted range for valid cells but are rejected due to a rejected nucleus in channel 1. Rejected nuclei with no discernable cell body do not generate a

mask in channel 2 tracing. These cells are not included in the final measurements of average number of neurites per neuron or total neurite length per neuron. (F) Neurites

emerging from accepted cell bodies are traced (light blue, green and purple lines) and quantified. Scale bars = 50 mm in all panels. (For interpretation of the references to color

in this figure legend, the reader is referred to the web version of the article.)

J.A. Harrill et al. / NeuroToxicology 31 (2010) 277–290280

endpoints were collected on cell-by-cell basis. The number ofneurites and the cumulative length of all neurites associated witheach cell body (i.e. total neurite length) were calculated for eachcell meeting the selection criteria outlined above. Cell-levelmeasurements were then averaged to obtain a mean measurementfor the average number of neurites per neuron and total neuritelength per neuron for the cell populations sampled within each

well. These well-level averages are reported in the present studyand were treated as the statistical unit for analysis of neuronalmorphology. In addition, the average number of neurons per fieldwas measured as an indicator of cell health in treated cultures.Only cells that met the criteria for valid cell body selection, as listedabove, are included in this measure of neuron density. Given that auniform number of viable cells are plated in each well at the

Fig. 2. Growth of hN2TM cells over time. Images of live cells grown for (A) 2, (B) 6 or (C)

24 h at a density of 7500 cells/well (2.5 � 104 cells/cm2). Images were taken at 20�magnification on a Nikon Eclipse TE200 microscope equipped with Hoffman

modulation contrast optics. The arrow in panel (A) points to an emerging neurite.

Scale bars = 100 mm.

J.A. Harrill et al. / NeuroToxicology 31 (2010) 277–290 281

initiation of the cultures, a relative decrease in neuron density isinterpreted as a decrease in cell health in the present concentra-tion-response experiments.

At 20� magnification, Cellomics Arrayscan VTI can sample 81unique fields of view and quantify the number of neurons sampledper well in real-time. For time course, cell-density gradient andconcentration-response studies, a sufficient number of fields weresampled so that at least 350 neurons were measured within eachwell. This was sufficient to minimize the variation observed inaverage morphometric measurements across wells (data notshown).

2.6. ATP-content

ATP-content in chemical treated cultures was measured using aCellTiter-Glo1 Luminescent Cell Viability Assay Kit (Promega,Madison, WI). This assay uses a luciferase-catalyzed reaction tomeasure the level of adenosine 50-triphosphate (ATP) in each well,which is produced by metabolically active cells. Radio et al. (2008)demonstrate that the amount of ATP present is proportional to thenumber of living cells in a well. Briefly, 100 ml of luminescentreagent was added to each well 22 h after exposure (i.e. 24 h afterplating). Plates were gently mixed on an orbital shaker and stored,protected from light, for 30 min at room temperature. Luminescentsignal was then quantified using a FLUOstar Optima plate reader(BMG LABTECH, Durham, NC).

2.7. Statistics

In experiments measuring basal neurite outgrowth over time orevaluating the effects of plating density on neurite outgrowth, rawvalues from morphological measurements were analyzed. All timecourse and plating density experiments were performed twiceusing independent cultures with n = 4–6 wells per condition perculture. For concentration-response experiments, neurite out-growth data were normalized within experiment to correspondingcontrol wells prior to statistical analysis. In experiments with non-organic molecules (LiCl, Na3VO4) or DMSO alone, data werenormalized to untreated control wells. In experiments withorganic compounds prepared in DMSO (Bis1, U0126, brefeldinA) data were normalized to vehicle control wells. In ATP-contentexperiments, luminescent signals were normalized to appropriatecontrols within each plate and analyzed across experiments. Foreach concentration-response examined, experiments wererepeated two to three times using independent cultures asdescribed in figure captions. Neurite outgrowth data and ATP-content data were analyzed using a one-way ANOVA with asignificance threshold of p < 0.05. This was followed by a Dunnett’spost hoc mean contrast test (p < 0.01) to determine if treatmentgroup means were significantly different from correspondingcontrol means, as described. Neurite outgrowth, neuron densityand ATP-content data are presented as % change from control. Rawmean values � standard deviations for neurite outgrowth measure-ments are provided throughout the text. Statistical analysis wasperformed using Graphpad Prism1 v5 (La Jolla, CA).

3. Results

3.1. Characterization of hN2TM cells

Neurite outgrowth in hN2TM cells progresses rapidly followingplating on polylysine and laminin coated 96-well plates. At thetime of plating the cells appear spherical in shape with no apparentneurite growth. Within the first 2 h after plating, thin neuritesbegin to emerge from the cell body of a small proportion of cells(Fig. 2A). By 6 h many cells have neurites that are longer than the

J.A. Harrill et al. / NeuroToxicology 31 (2010) 277–290282

broadest diameter of their respective cell bodies (Fig. 2B) and at24 h a majority of viable cells have between one and three longneurites (Fig. 2C).

At 24 h post-plating, hN2TM cells are positive for a number ofprotein markers indicative of a neuronal phenotype. The neuronalmicrotubule protein bIII-tubulin was present in both the cell bodiesand all neurites of viable cells (Fig. 3A). Similarly, cell bodies andneurites were positive for nestin, an intermediate filament proteinexpressed in progenitor and newly differentiated cells of the neuralepithelial lineage (Wiese et al., 2004) (Fig. 3B). In addition, cellsgrown for 24 h were immunolabeled with the SMI-312 primaryantibody, which is targeted against phosphorylated neurofilamentproteins and specifically labels axons in human fetal tissue (Ulfig etal., 1998). The resulting immunofluorescent signal was prominentin the long, thin processes emerging from the cell body of the cells(Fig. 3C). Less intense immunolabeling was also apparent withinthe cell body, but only at the points where neurites are emerging.Some thin processes not immunolabeled with SMI-312 were alsoobserved upon comparison of fluorescent images to matchingdifferential interference contrast images (Fig. 3C and G, arrows). At24 h, the cells were also positive for the expression of MAP2(Fig. 3D), a neuronal microtubule-associated protein enriched indendrites (Caceres et al., 1986). Similar to the SMI-312 axonalmarker, several neurites were observed that were not immuno-labeled with the MAP2 antibody (Fig. 3D and H). For studies ofneurite outgrowth in hN2TM cells using automated high-contentmicroscopy, the bIII-tubulin primary antibody was used as itindiscriminately labeled all neurites, as well as cell bodies.

3.2. Measurement of neurite outgrowth in hN2TM cells using HCA

Similar to the qualitative observations made during live cellimaging, automated measurement of hN2TM cells demonstrated arapid increase in neurite outgrowth during the first 48 h afterplating (Fig. 4). During this early growth phase, the averagenumber of neurites per neuron increased significantly (Fig. 4A,0.15 � 0.08 at 2 h; 1.28 � 0.07 at 24 h). The population distributionfor this endpoint indicated that only 8.6% of cells had at least 1 neuriteat 2 h, whereas >60% of cells had at least 1 neurite at 24 h (Fig. 4B).Very few cells (<2.5%) developed more than three primary neuritesduring the time period sampled (Fig. 4B). There was no significantdifference in the average number of neurites per neuron between 24and 48 h.

Total neurite length per neuron significantly increased duringthe initial 48 h growth period (Fig. 4C). Total neurite length perneuron increased by more than 10-fold between 2 and 24 h(3.12 � 2.14 mm at 2 h; 56.5 � 2.3 mm at 24 h). Unlike the averagenumber of neurites per neuron, neurite total length per neuroncontinued to increase after 24 h (73.10 � 7.9 mm at 48 h). For neuriteoutgrowth and ATP-content concentration-response experiments,cells were fixed and sampled at 24 h. At this time point, measure-ments of total neurite length had not yet reached an asymptote. Thisprovided a dynamic range in which both chemically inducedincreases and decreases in neurite outgrowth could be detected.

Changing the plating density of hN2TM cells between 2500 and10,000 cells/well did not significantly affect neurite outgrowthmeasurements (Fig. 4D–F) at 24 h. For neurite outgrowth and ATP-content concentration-response experiments, an intermediate cellnumber of 7500 cells/well was used. This cell number providedenough separation of cells along the plating surface for resolutionof individual neurites with a 20� imaging objective.

At 24 h, the population of hN2TM cells within any given culturewell had heterogeneous morphological characteristics. The num-ber of neurites per neuron, as well as total neurite length perneuron, varied from cell to cell. The ArrayScan VTI HCS readerallows the user to define a minimum number of neurons to be

sampled in each well (sampling threshold), and therefore controlthe number of cells used to calculate well-level averages of neuritecount and total neurite length per neuron. Retrospective analysesof hN2TM cells at a density of 7500 cells/well demonstrated that theacross well coefficient of variation (C.V.) for these two endpoints(n = 6 wells) dropped from �20% to less than 10% when thesampling threshold was increased from 10 cells (sampling a singlefield) to 100 cells (sampling 6–7 fields). Increasing the samplingthreshold greater than 100 neurons per well did not appreciablylower the across well C.V.s further (data not shown). For all studiesshown here, a sampling threshold of 350 neurons per well wasused. In control experiments using this sampling threshold,measurements of the average number of neurites per neuronand total neurite length per neuron were very reproduciblebetween cultures, varying by less than 6% at 24 h.

3.3. Chemical effects on hN2TM neurite outgrowth

DMSO at concentrations between 0.05 and 0.5% had no effect onATP-content, neuron density, the average number of neurites perneuron or total neurite length per neuron in hN2TM cells following22 h of exposure (Fig. 5A–C). At a concentration of 1%, DMSOsignificantly decreased total neurite length per neuron by 21.7%(control: 44.1 � 6.6 mm, treated: 27.8 � 7.5 mm, Fig. 5C) with nosignificant effects on ATP-content, neuron density or the averagenumber of neurites per neuron (Fig. 5A and B). At a concentration of5% DMSO, a significant decrease in ATP-content (51.7%) and neurondensity (63.2%) was observed coupled with significant decreases(>97%) in the average number of neurites and total neurite length perneuron. In the present study, DMSO was used to dissolve Bis1, U0126and brefeldin A prior to preparation of dosing solutions. The finalconcentration of DMSO in tissue culture wells treated with thesecompounds did not exceed 0.1%. At this concentration of DMSO, nosignificant effects on ATP-content, neuron density, or neuriteoutgrowth measurements were observed.

A set of five compounds with well defined molecularmechanisms of action was identified from the literature as havingeffects on neurite outgrowth in primary rodent neural cultures.Descriptions of these studies are listed in Table 1. For eachcompound a five point concentration-response curve was exam-ined. Concentration ranges were based on exposure levels used inprevious research as described in the literature (Table 1).

Following a 22 h exposure, the protein kinase C (PKC) inhibitorBis1 had no significant effects on ATP-content or neuron density atany of the concentrations examined (Fig. 6A). The thresholdconcentration of Bis1 for decreasing ATP-content or neuron densitycould not be determined from these data. In contrast, concentra-tion-dependent decreases in neurite outgrowth was observedfollowing exposure to Bis1 (Fig. 6B). At 10 mM a significant (22.3%)decrease in the average number of neurites per neuron (control:1.14 � 0.22, treated: 0.97 � 0.13) as well as a significant (26%)decrease in total neurite length per neuron (control: 38.8 � 11.8 mm,treated: 31.9 � 4.5 mm) was observed. Significant decreases inneurite outgrowth were not observed at concentrations of Bis1below 3 mM. These data demonstrate a specific inhibition of neuriteoutgrowth by 10 mM Bis1, a concentration that does not affectindicators of cell health.

In contrast to Bis1, significant decreases in neurite outgrowthwere accompanied by concurrent decreases in ATP-content andneuron density following exposure to U0126, an inhibitor ofmitogen-activated protein kinase/extracellular-regulated kinase(MEK) signaling (Fig. 6C and D). There were no concentrations ofU0126 that affected neurite outgrowth without a concurrent effecton either ATP-content or neuron density. Between 3 and 30 mM ofU0126, ATP-content was significantly decreased by approximately20% from control values. A significant decrease in neuron density

Fig. 3. Antigenic characterization of hN2TM cells. Cells grown in 96-well titer plates at 7500 cells/well (2.5 � 104 cells/cm2) for 24 h and immunocytochemically labeled for (A)

bIII-tubulin, (B) nestin, (C) phosphorylated neurofilaments (pNFs) or (D) MAP2. Panels E–H are modulation contrast images corresponding to fluorescently imaged fields to

the immediate left each panel. All images were taken at 20�magnification on a Leica DMI6000 microscope. All cell bodies and neurites were positive for both bIII-tubulin and

nestin at 24 h. Arrows in panels (C) and (G) correspond to neurites that are not positive for pNFs, a marker of neuronal axons. Likewise, arrows in panels (D) and (H) correspond

to neurites that are not positive for MAP2, a marker of neuronal dendrites. Scale bars = 100 mm.

J.A. Harrill et al. / NeuroToxicology 31 (2010) 277–290 283

Fig. 4. Effects of time and plating density on neurite outgrowth in hN2TM cells. (A–C) Cells were grown for either 2, 6, 24 or 48 h at a density of 2500 to 10,000 cells/well (8.33 � 103

and 3.33 � 104 cells/cm2) and measured. (A) Average number of neurites per neuron. (B) Normalized histogram of the average number of neurites per neuron. (C) Total

neurite length per neuron. Values for each measurement significantly increased over time between 2 and 48 h (p < 0.05, one-way ANOVA). (D–F) Cells were grown for 24 h at

densities of 2500, 5000, 7500 or 10,000 cells/cm2 and measured. (D) Average number of neurites per neuron. (E) Normalized histogram of the average number of neurites per

neuron. (F) Total neurite length per neuron. Values for each measurement were not significantly different across this range of plating densities. n = 10–12 wells from two

independent cultures for both time course and plating density experiments.

J.A. Harrill et al. / NeuroToxicology 31 (2010) 277–290284

(29.2%) was observed only at 30 mM, 10-fold higher than theminimum concentration which decreased ATP-content (Fig. 6C).Decreases in neurite outgrowth were first observed at 10 mMU0126 and continued up to 30 mM, the highest concentrationtested (Fig. 6D). At 30 mM, significant decreases of 45% (control:1.14 � 0.22, treated: 0.72 � 0.33) and 53.2% (control:38.8 � 11.8 mm, treated: 22.6 � 12 mm) were observed for theaverage number of neurites per neuron and total neurite lengthper neuron, respectively. Concentrations of U0126 between 0.3 and1 mM had no significant effects on ATP-content, neuron density orneurite outgrowth measurements.

Exposure to millimolar concentrations of lithium chloride (LiCl)also affected hN2TM ATP-content, neuron density and neuriteoutgrowth (Fig. 6E and F). At 10 mM and 30 mM LiCl, ATP-contentwas significantly decreased by 27.1 and 45.6%, respectively,compared to untreated controls. Neuron density was alsosignificantly decreased by 20.3 and 56.3% at these concentrations,respectively. At 10 mM, significant decreases of 49.1% (control:1.13 � 0.23, treated: 0.59 � 0.15) and 62% (control: 38.6 � 12.1 mm,treated: 14.8 � 4.3 mm) were observed for the average number ofneurites per neuron and total neurite length per neuron, respectively.At 30 mM, both the average number of neurites per neuron and totalneurite length per neuron were significantly decreased >90% fromvehicle treated control values. Similar to the patterns observed withU0126, there were no concentrations of LiCl that affected neuriteoutgrowth without concurrent effects ATP-content or neuron density.

Similarly, the fungal antibiotic brefeldin A significantlydecreased neurite outgrowth but only at concentrations whereconcurrent significant decreases in ATP-content and neurondensity were observed (Fig. 6G and H). This compound disruptsmembrane transport of proteins from intracellular organelles (i.e.endoplasmic reticulum and Golgi) to the cell membrane. Theconcentration-response for brefeldin A on both neurite outgrowthendpoints was steep, with no effects observed between 0.01 and0.03 mM and significant decreases (>85%) at concentrations

ranging from 0.1 to 1 mM. In contrast, significant decreases inATP-content and neuron density were observed at the sameconcentrations (0.1–1 mM) but only to �50% of control levels.

The effects of the broad-spectrum phosphatase inhibitorNa3VO4 were unique among the compounds examined in thatsignificant concentration-dependent decreases in all endpointswere observed with marked differences in the lowest effectiveconcentrations that inhibited neurite outgrowth, neuron densityand ATP-content, respectively (Fig. 6I and J). Significant decreasesin neurite outgrowth measurements were first observed at 3 mMNa3VO4, whereas significant decreases in hN2TM neuron densityand ATP-content were first observed at 10 and 30 mM, respec-tively. Following exposure to 10 mM Na3VO4, significant decreasesof 36.8% (control: 1.23 � 0.14, treated: 0.77 � 0.08) and 47.8%(control: 42.7 � 8.2 mm, treated: 21.9 � 3.2 mm) were observed forthe average number of neurites and total neurite length per neuron,respectively (Fig. 6J). Neuron density significantly decreased by10 mM Na3VO4 (22.1%) and ATP-content was not affected at thisconcentration (Fig. 6I). Neurite outgrowth measurements weresignificantly decreased by >94% by 30 and 100 mM Na3VO4 withconcurrent, significant decreases in ATP-content (47.7 and 73%,respectively) and neuron density (65.0 and 76.8%, respectively).Collectively, these data demonstrate that neurite outgrowth mea-surements were more sensitive than indicators of cell health fordetecting concentration-dependent effects of Na3VO4 on hN2TM cells.

4. Discussion

The present study characterized the molecular phenotype ofhESC-derived hN2TM cells grown in dissociated culture andquantified neurite outgrowth in these cells. At 24 h after initialplating, the cells expressed protein markers characteristic ofmaturing neurons. This is consistent with development along aneural lineage. Automated high-content image analysis (HCA)demonstrated that neurite outgrowth progressed rapidly once

Fig. 5. Effects of DMSO on hN2TM neurite outgrowth. Cells were grown at a density of

7500 cells/well and exposed to either 0.05, 0.1, 0.5, 1 or 5% DMSO at 2 h after

plating. Measurements were made at 24 h. (A) ATP-content (white bars) and neuron

density (i.e. the average number of neurons per field; gray bars) were measured as

indicators of cell health. (B) Average number of neurites per neuron. (C) Total

neurite length per neuron. All data are presented as % change from untreated

control wells �standard deviation (S.D.). ATP-content data is from three separate

experiments using independent cultures (control group, n = 27 wells total; treatment

groups, n = 11–17 wells total). Neuron density and neurite outgrowth data is from two

separate experiments using independent cultures (control group, n = 24 wells total;

treatment groups, n = 10–16 wells total). A significant effect of concentration was

observed for each endpoint (p < 0.05, one-way ANOVA). *Concentration is significantly

different from control values (p < 0.01, Dunnett’s post hoc test).

J.A. Harrill et al. / NeuroToxicology 31 (2010) 277–290 285

cells were plated on a poly-L-lysine/laminin substrate. Within 24 hof initiating the cultures, a large proportion of cells had developedone to three neurites. Finally, exposure to a panel of chemicalsknown to affect the growth of neurites in rodent primary neuralcultures produced concentration-dependent decreases in neuriteoutgrowth in these human neural cells. Collectively, these dataindicate that hN2TM cells have morphological and phenotypiccharacteristics of maturing neurons and are amenable tomeasurement of neurite outgrowth using medium-throughputHCA methods.

Immunocytochemical staining demonstrated that at 24 h afterinitial plating, hN2TM cells expressed markers indicative ofneuroepithelial lineage. Prior to initiation of cultures by the end

user, the cells have been subjected to culture conditions designedto promote neuronal differentiation. Accordingly, during HCAanalysis, the average number of cell nuclei per field did notincrease over time (data not shown) indicating that these cellswere in a non-proliferative state, a feature of differentiated ESCs. Inaddition, the cells expressed nestin (Fig. 2), an intermediatefilament protein expressed in human neuroepithelium in vivo andhESC-derived neuroepithelial cells in culture (Tohyama et al.,1992; Gilyarov, 2008). In the developing embryonic nervoussystem in vivo, nestin expression decreases as cells become post-mitotic (Dahlstrand et al., 1995). Similarly, nestin expression alsodecreases in vitro as ESCs mature into a non-proliferative state(Rolletschek et al., 2001). However, in vitro, some cells may remainpositive for nestin protein for up to several weeks following stimulithat promote neuronal differentiation (Shin et al., 2006; Roll-etschek et al., 2001; Nat et al., 2007). Therefore, positive nestinexpression suggested that hN2TM cells are of a neuroepitheliallineage, but does not indicate which stage of neural differentiationthese cells were in shortly after initiation of cultures from frozenstocks.

In addition to nestin, hN2TM cells were also positive for bIII-tubulin, MAP2 and phosphorylated neurofilaments (Fig. 2), all ofwhich are expressed in mature neurons in vitro (Fletcher andBanker, 1989; Caceres et al., 1986; DeFuria and Shea, 2007).Expression of bIII-tubulin and MAP2 increases in ESCs upondifferentiation along a neural lineage and have been found to be co-expressed with nestin following a differentiating stimulus (Natet al., 2007; Baharvand et al., 2007). In addition, the patterns ofMAP2 and phosphorylated neurofilament immunolabeling(Fig. 2E–H) indicated that hN2TM cells grown for 24 h displayedearly features of neuronal polarity, a fundamental functionalproperty of mature neurons. Neuronal polarization facilitates theunidirectional flow of electrical activity from synaptic contacts onthe dendritic arbor and cell body to neurotransmitter release sitesin the axon (Arimura and Kaibuchi, 2007). Expression of MAP2protein is restricted to the dendritic compartment while thepresence of a number of phosphorylated neurofilaments (detectedby the SMI-312 antibody) is specific to axons (Ulfig et al., 1998;Kosik and Finch, 1987). Fig. 3 demonstrates that some hN2TM cellsgrown for 24 h had some neurites which were MAP2-positive andothers which were MAP2-negative. Likewise, some neurites werepositive for pNFs while others were negative. These stainingpatterns demonstrated cytoplasmic compartmentalization andindicate that these cells may have been developing toward apolarized neuronal phenotype. Dual-immunolabeling experimentsusing these cells are needed to clarify if neurites selectively labeledwith MAP2 and pNFs can develop from the same cell. Overall, it isclear from the present data that hN2TM cells plated from frozenstocks had a molecular phenotype consistent with developingneurons. Staining patterns indicate that these cells are in adevelopmental period between the onset of mitotic quiescence andterminal differentiation into a mature, polarized neuron at 24 hafter plating.

Dissociated cultures of hN2TM cells immunostained for bIII-tubulin were amenable to measurement of neurite outgrowthusing automated HCA. In order to accurately perform neuriteoutgrowth measurements using HCA, cells must be grown at a lowenough density so that the morphological features of individualcells can be easily resolved (Dragunow, 2008). In the presentexperiment, increasing the plating density from 2500 to10,000 cells/well had no significant effect on measurements ofneurite outgrowth (Fig. 4). Within this range the morphologicalfeatures of individual cells could be easily resolved and thevariability in population mean measurements was low, both fromwell-to-well and experiment-to-experiment. Above 10,000 cells/well, the complexity of the image prevented accurate assignment

Fig. 6. Chemical effects on neurite outgrowth in hN2TM cells. Cells were grown at a density of 7500 cells/cm2 and exposed to varying concentrations of Bis1 (A and B), U0126 (C

and D), LiCl (E and F), brefeldin A (G and H) or Na3VO4 (I and J) 2 h after plating. Measurements were made at 24 h after plating. ATP-content (open circles) and neuron density

(i.e. average number of neurons per field, open squares) are presented in the left-hand column (A, C, E, G, and I). Average number of neurites per neuron (open circles) and total

neurite length per neuron (open squares) are presented in the right-hand column (B, D, F, H, and J). Note, the x-axis for LiCl (panels G and H) is in mM as opposed to mM. All

data are presented as % change from untreated control (LiCl, Na3VO4) or controls containing 0.1% DMSO (Bis1, U0126, brefeldin A) � standard deviation (S.D.). For Bis1, U0126,

LiCl and Na3VO4 ATP-content, neuron density and neurite outgrowth data is from 2 separate experiments using independent cultures (control group, n = 9–15 wells total; treatment

groups, n = 6–9 wells total). For brefeldin A, data is from 3 separate experiments using independent cultures (control group, n = 24–27 wells; treatment groups, n = 5–11 wells). A

significant main effect of concentration was found for each endpoint for all compounds tested, except ATP-content and neuron density with Bis1 (p < 0.05, one-way ANOVA).

Symbols above, below or immediately beside data points denotes that the response at a given concentration is significantly different from control values for ATP-content (*), neuron

density (**), average number of neurites per neuron (#), total neurite length per neuron (§) using Dunnett’s post hoc test (p < 0.01).

J.A. Harrill et al. / NeuroToxicology 31 (2010) 277–290286

J.A. Harrill et al. / NeuroToxicology 31 (2010) 277–290 287

of neurites to associated cell bodies (data not shown) andprecluded automated measurements. These data demonstratedthat during the early growth phase, the density of the cultures hadno influence on the rate of neurite outgrowth. Similarly, Radio et al.(2010) demonstrated that neurite outgrowth in an NGF-stimulatedPC12 cell clone (Neuroscreen-1, NS-1) was not affected by changesin plating density whereas neurite outgrowth in a primary mixedneural culture was dramatically affected. In the latter type ofculture model, neurite outgrowth is influenced by cell–cellcontacts as well as soluble factors and extracellular matrixproteins secreted by supportive glial cell populations (Hattenet al., 1988; Ibata et al., 1991). The cultures in the present studyappeared to be of a homogeneous neuronal phenotype.

The morphological changes observed in hN2TM cells over timewere also consistent with these cells developing toward a neuronalphenotype. Neurite outgrowth progressed rapidly during the first48 h of growth (Fig. 4). Greater than 60% of the cells had developedbetween one and three neurites by 24 h after plating. The rate andtotal amount of neurite outgrowth was comparable to that ofclonal neural cell lines and primary neural cultures cited in theliterature (Radio et al., 2010; Bearer et al., 1999; Hong et al., 2003;Das et al., 2004; Vutskits et al., 2006; George et al., 2009). Fromthese data, it is clear that morphological development of hN2TM

cells progressed rapidly and was similar to that observed inpreviously characterized neural culture models. Therefore, thesetypes of human stem cell-derived neural cells provide analternative to rodent primary cultures or immortalized clonal celllines for investigating the mechanisms of neurite outgrowth.

Neurite outgrowth in hN2TM cells was inhibited by a variety ofchemicals previously shown to affect neurite outgrowth in primaryneural culture models (Table 1). In developing neurons, neuriteoutgrowth is dependent upon a number of inter-connected cellularprocesses including membrane and protein trafficking, reorgani-zation of the cytoskeleton and a variety of intracellular signalingcascades (Larsson, 2006; Yoshimura et al., 2006; Tang, 2001; Dentand Gertler, 2003; Ditlevsen et al., 2008). In the present study, adiverse set of test chemicals that disrupt these processes wereexamined in order to compare and contrast neurite outgrowth inhESC-derived neural cells with that of previously establishedmodels. Overall, the present data demonstrated that these hESC-derived cells were sensitive to agents that disrupt neuriteoutgrowth in primary rodent neural cultures. In addition, thesedata demonstrated that the molecular mechanisms that mediateneurite outgrowth in neural clonal cell lines and rodent primarycultures may also play a role in the developmental biology of hESC-derived neurons (Table 2).

In the present study two measures were used as generalindicators of cell health: ATP-content as a measure of metabolicstatus and neuron density (i.e. the average number of neurons perfield) as a measure of cell viability. These data were coupled withmeasurements of neurite outgrowth to provide a context in whichto evaluate of the overall health of the cultures when treatment-

Table 2Summary of concentration-response data.

Chemical Concentration range Lowest effective concentrati

ATP content Neu

Brefeldin A 0.01–1 mM 0.1 0.1

Bis1 0.1–10 mM n.d. n.d

U0126 0.3–30 mM 30 3

Na3VO4 1–100 mM 30 10

LiCl 300–30,000 mM 10,000 10,

a Values are the lowest effective concentration (in mM) for which a significant chan

(p<0.01). n.d. = no change detected.b Average number of neurons per field.

related decreases in neurite outgrowth were observed. In thecontext of quantitative screening for neurotoxicants, measure-ments of cell health or cytotoxicity alone may not be sufficient to:(1) identify neurotoxic compounds or (2) provide accurateinformation on the potency for eliciting a neurotoxic response.For example, in the former case, ATP-content measurements wereunable to positively identify trans-retinoic acid, a known devel-opmental neurotoxicant (DNT), as having any detrimental effectson PC12 cells in the absence of neurite outgrowth data (Radio et al.,2008, 2010). In the latter case, the threshold for effects on ATP-content for the known developmental neurotoxicant methyl-mercury (MeHg) was 10,000-fold greater than the minimumeffective concentration needed to affect neurite outgrowth in PC12cells (Radio et al., 2008, 2010). These data suggest that measure-ments of neurite outgrowth may be more sensitive than cell healthassays, and thus have a greater capability for accurate identifica-tion of potential developmental neurotoxicants.

In the present study, a majority of the chemicals tested (3 out of5 as outlined above) produced decreases in neurite outgrowthmeasurements only at concentrations that also affected ATP-content and/or neuron density. The relative sensitivity of these twoendpoints varied across compounds. However, the patterns ofconcentration-dependent effects were consistent across endpointsfor each compound tested. In instances where effects on neuriteoutgrowth were detected the relative magnitude of the effect wasgreater than that observed with ATP-content or neuron densitymeasurements. This supports the hypothesis that measurementsof neurite outgrowth are more sensitive than measures of cellhealth in the context of neurotoxicity screening. In addition, theeffects observed following Na3VO4 and Bis1 treatment confirmsthat a neuronal cell-type specific toxicity response can be observedin the absence of acute changes in indicators of cell health in hESC-derived cells.

Concentration-response data suggested that the PKC and MEK/ERK signaling pathways may mediate neurite outgrowth in hN2TM

cells. The PKC inhibitor Bis1 decreased the average number ofneurites and total neurite length per neuron with no observableeffects on ATP-content or neuron density in the concentrationrange examined (Fig. 6A and B). U0126, an inhibitor of MEK/ERKsignaling, also decreased both measures of neurite outgrowth, butonly where concurrent decreases in ATP-content or neurondensity were observed (Fig. 6C and D). Both the PKC and theMEK/ERK signaling pathways have been shown to mediate neuriteoutgrowth in both clonal and primary neural culture modelsfollowing application of neurotrophic receptor ligands (i.e. nervegrowth factor, NGF) and cell adhesion molecules (Kim et al., 1997;Gerecke et al., 2004; Schmid et al., 1999; Burry, 1998; Kolkova etal., 2000). Specifically, Bis1 and U0126 dramatically inhibitedneurite outgrowth in NGF-stimulated PC-12 cells as well asprimary cerebellar granule cells (CGCs) and carbachol-stimulatedprimary hippocampal neurons (Radio et al., 2010; Das et al., 2004;VanDeMark et al., 2009). Importantly, Radio et al. (2010)

ona

ron densityb Average # of neurites

per neuron

Total neurite length

per neuron

0.1 0.1

. 10 10

10 10

3 3

000 10,000 10,000

ge from control was detected using a Dunnett’s many-to-one mean contrast test

J.A. Harrill et al. / NeuroToxicology 31 (2010) 277–290288

demonstrated that the characteristics of neurite outgrowthinhibition by these agents may vary from cell-type to cell-type.In PC12 (NS-1) cells Bis1 inhibited neurite outgrowth with noapparent decreases in ATP-content (i.e. cell health) indicating aspecific effect on this developmental process. In contrast, neuriteoutgrowth inhibition by Bis1 in primary CGCs occurred only atconcentrations where ATP-content was also decreased, indicatingthe inhibition was likely a secondary response to decreased cellhealth (Radio et al., 2010). These data demonstrate that parallelanalysis of neurite outgrowth with some measure of cell healthcan be critical for interpretation of neurite outgrowth data from in

vitro neurotoxicity studies. The present data support thatinhibition of hN2TM neurite outgrowth by 10 mM Bis1 is not asecondary response to a decrease in cell health and also indicatesthat PKC signaling is likely involved in neurite outgrowth in hESC-derived neurons. In contrast, it is unclear from the present datawhether U0126 decreased neurite outgrowth in hN2TM cells as aspecific result of MEK/ERK inhibition or whether the decreases inneurite outgrowth were the result of a generalized decrease in cellhealth in response to this compound. Collectively, the presentdata indicate that the PKC pathway likely plays a role in mediatingneurite outgrowth in hESC-derived neurons and that the MEK/ERKpathway is active in the maintenance of cell health. Additionalstudies are required to define and refine the specific role of the PKCand MEK/ERK signaling pathways in neurite outgrowth of hESC-derived neurons.

In addition to kinase inhibitors, the protein tyrosine phospha-tase inhibitor Na3VO4 also decreased neurite outgrowth in hN2TM

cells. Na3VO4 acts as a broad-spectrum tyrosine phosphataseinhibitor and decreases axonal outgrowth in hippocampal neuronsand neurite outgrowth in NGF-stimulated PC12 cells (Wu andBradshaw, 1993; Mandell and Banker, 1996). The present data areconsistent with these previous findings. In addition, decreases inneurite outgrowth were observed at concentrations of Na3VO4 thatdid not produce decreases in neuron density or ATP-content (Fig. 6Iand J). These data demonstrate that at 3 mM the effects of Na3VO4

were specific to the process of neurite outgrowth. A similaroutcome was observed with Bis1 at a concentration of 10 mM. Asthe concentration of Na3VO4 increased the effects on neuriteoutgrowth were accompanied by decreases in neuron density at10 mM and ATP-content at 30 mM. These data reflect a progressionin the severity Na3VO4 toxicity from a specific effect on adevelopmental process to a more generalized effect on cell health.These data demonstrated that coupling measurements of neuriteoutgrowth with indicators of cell health permits detection ofchemical effects on neuronal morphology that are not the result ofgeneralized decreases in cell health. These data also imply that forsome compounds neurite outgrowth may be a more sensitiveendpoint for detecting developmental neurotoxicity potential ascompared to cell counts, measures of ATP-content or otherindicators of cell health. From these data, Na3VO4 or Bis1 mayserve as an endpoint specific control for neurite outgrowth inhESC-derived neural cells for toxicity screening applications.

Brefeldin A, a fungal antibiotic that disrupts membranetrafficking, produced a decrease in neurite outgrowth in hN2TM

cells (Fig. 6G and H). Decreases in neurite outgrowth wereaccompanied by concurrent 30–40% decreases in ATP-content andneuron density. Brefeldin A acts by inhibiting a guanine nucleotideexchange factor that regulates the activity of ADP-ribosylationfactor (ARF). ARF controls membrane trafficking from the Golgi tothe plasma membrane (Donaldson et al., 1992). Treatment ofneurons with brefeldin A terminates membrane trafficking andeventually results in dissolution of the Golgi membrane network.This results in a reduction in the rate of neurite extension and, insome cases, neurite retraction (Jareb and Banker, 1997; Prager-Khoutorsky and Spira, 2009). The present data indicate that

membrane trafficking from the Golgi is necessary for promotingneurite outgrowth in hN2TM cells and maintaining cell health,which is consistent with observations from primary rodent neuralcell cultures.

Neurite outgrowth in hN2TM cells was also inhibited by 10–30 mM concentrations of lithium chloride (LiCl). Neurite out-growth was inhibited >49% from control values and wasaccompanied by concurrent decreases in ATP-content and neurondensity (Fig. 6E and F). This is consistent with data from Takahashiet al. (1999) which also demonstrates concurrent decreases inneurite outgrowth and cell viability in primary rat hippocampalcultures following LiCl exposure. In contrast, there are a number ofconflicting reports in the literature regarding the effect LiCl onneurite outgrowth. Some studies in primary neural cultures andclonal cell lines demonstrate an inhibition of neurite outgrowthwith mM concentrations of LiCl (Takahashi et al., 1999; Koike et al.,2006; Hollander and Bennett, 1991; Tamura and Ohkuma, 1991)while others demonstrate enhancement (Orme et al., 2003; Sayaset al., 2002) or even a biphasic effect (Munoz-Montano et al., 1999).The source of the disparate findings between studies is unknown,but could be due to differences in cell-types or culture conditionsacross studies. In the context of neurite outgrowth, the putativemolecular mechanism-of-action of LiCl is inhibition of kinases thatregulate cytoskeletal protein dynamics such as glycogen-synthasekinase 3b (GSK3b) (Yoshimura et al., 2006) and nemo-like kinase(NLK) (Ishitani et al., 2009). The expression patterns, activity levelsand sub-cellular distribution of GSK3b and NLK have not beencharacterized in hESC-derived neural cells. However, the inhibitionof neurite outgrowth by LiCl suggests a putative role for thesekinases in mediating this process. Alternatively, the decrease inneurite outgrowth observed here could be due a decrease ingeneral cell health and off-target effects caused by mM concen-trations of LiCl. Characterizing the expression of GSK3b and NLKand examining neurite outgrowth in response to protein knock-down or more specific pharmacological inhibitors would aid indefining the role of these proteins in development of hESC-derivedneural cells.

Interestingly, decreases in hN2TM neurite outgrowth wereobserved with both selective (Bis1, U0126) and non-selective(LiCl) kinase inhibitors as wells as a non-selective phosphataseinhibitor (Na3VO4). In general, protein phosphorylation acts amolecular switch controlling the activity of a large varietyintracellular signaling pathways and processes. In the case ofneurite outgrowth, phosphorylation of effector proteins can act toeither promote or inhibit neurite outgrowth depending upon thefunction of the protein. For example, phosphorylation of micro-tubule-associated protein 1B (MAP1B) downstream of GSK3b andNLK and phosphorylation of paxillin downstream of NLKpromotes neurite outgrowth in NGF-stimulated PC12 cells(Ishitani et al., 2009). In contrast, phosphorylation of collapsinresponse mediator protein 2 (CRMP2) downstream of GSK3binhibits neurite development in hippocampal neurons andcorrelates with a decrease in neurite outgrowth in NGF-stimulated PC12 cells (Yoshimura et al., 2005; Patrakitkomjornet al., 2008). Neurite outgrowth in hESC-derived neural cells maybe mediated by a similar phenomenon; i.e. the balance ofphosphorylation-dependent activity and inactivity of stimulatoryand inhibitory effector proteins. The present data support thatchemicals that interfere with phosphorylation and dephosphor-ylation of proteins may result in similar inhibitory effects onneurite outgrowth in hESC-derived neural cells, possibly bydisrupting this type of balanced intracellular signaling network.More detailed biochemical studies are required to identify thephosphorylation-dependent effectors of neurite outgrowth anddefine the molecular mechanisms whereby chemicals can inhibitthis developmental process in hESC-derived neural cells.

J.A. Harrill et al. / NeuroToxicology 31 (2010) 277–290 289

5. Conclusions

In summary, this work demonstrates that the molecularphenotype and processes controlling neurite outgrowth in hN2TM

cells are similar to previously characterized neuronal cell models in

vitro. In contrast to rodent primary neural cultures, hN2TM cellsprovide a model in which to investigate the processes of neuriteoutgrowth in the context of human biology. In addition, the cellsare non-transformed and provide an alternative to the use oftumor-derived or transfected clonal cell lines. These cells haveseveral qualities that make them amenable for use in high- tomedium-throughput developmental neurotoxicity screening,including: cellular homogeneity, a rapid rate of neurite outgrowth,low inter-experiment variability in automated morphologicalmeasurements and the ability to be cultured at low densities.These data also demonstrate Na3VO4 or Bis1 could serve as anendpoint specific positive controls for inhibition of neuriteoutgrowth. Additional studies are needed to characterize themolecular pathways controlling neurite outgrowth in hESC-derived neural cells and whether these cells continue to matureand form functional networks in culture. In the context ofneurotoxicity screening, future efforts will include evaluation ofa larger training set of inert and known DNT compounds in order toassess the sensitivity of the hN2TM culture model as a DNTscreening tool.

Conflict of interest statement

The data included in this study was generated at the U.S.Environmental Protection Agency, National Health and Environ-mental Effects Research Laboratories. hN2TM cells and growthmedia were provided through Material Transfer Agreement #466-08 between the U.S. EPA and ArunA Biomedical, Inc. There is apotential conflict of interest in that one author (Steven L. Stice)currently serves as the Chief Scientific Officer of ArunA Biomedical,Inc. and another author (Dave Machacek) was employed by ArunABiomedical, Inc. during the study.

Acknowledgements

The authors wish to thank Mr. Brian Robinette for his technicalassistance and thank Drs. Will Boyes and Linda Kaltenbach for theircomments and suggestions on an earlier version of this manu-script.

References

Allen DD, Caviedes R, Cardenas AM, Shimahara T, Segura-Aguilar J, Caviedes PA. Celllines as in vitro models for drug screening and toxicity studies. Drug Dev Ind Pharm2005;31(8):757–68.

Arimura N, Kaibuchi K. Neuronal polarity: from extracellular signals to intracellularmechanisms. Nat Rev Neurosci 2007;8(3):194–205.

Audesirk T, Audesirk G, Ferguson C, Shugarts D. Effects of inorganic lead on thedifferentiation and growth of cultured hippocampal and neuroblastoma cells.Neurotoxicology 1991;12(3):529–38.

Baharvand H, Mehrjardi NZ, Hatami M, Kiani S, Rao M, Haghighi MM. Neural differ-entiation from human embryonic stem cells in a defined adherent culture condi-tion. Int J Dev Biol 2007;51(5):371–8.

Bearer CF, Swick AR, O’Riordan MA, Cheng G. Ethanol inhibits L1-mediated neuriteoutgrowth in postnatal rat cerebellar granule cells. J Biol Chem 1999;274(19):13264–70.

Breier JM, Gassmann K, Kayser R, Stegeman H, De Groot D, Fritsche E, et al. Neuralprogenitor cells as models for high-throughput screens of developmental neuro-toxicity: state of the science. Neurotoxicol Teratol 2009.

Burry RW. PKC activators (phorbol ester or bryostatin) stimulate outgrowth of NGF-dependent neurites in a subline of PC12 cells. J Neurosci Res 1998;53(2):214–22.

Caceres A, Banker GA, Binder L. Immunocytochemical localization of tubulin andmicrotubule-associated protein 2 during the development of hippocampal neu-rons in culture. J Neurosci 1986;6(3):714–22.

Cedrola S, Guzzi G, Ferrari D, Gritti A, Vescovi AL, Pendergrass JC, et al. Inorganicmercury changes the fate of murine CNS stem cells. FASEB J 2003;17(8):869–71.

Coecke S, Goldberg AM, Allen S, Buzanska L, Calamandrei G, Crofton K, et al. Workgroupreport: incorporating in vitro alternative methods for developmental neurotoxicityinto international hazard and risk assessment strategies. Environ Health Perspect2007;115(6):924–31.

Costa LG, Aschner M, Vitalone A, Syversen T, Soldin OP. Developmental neuropathologyof environmental agents. Annu Rev Pharmacol Toxicol 2004;44:87–110.

Dahlstrand J, Lardelli M, Lendahl U. Nestin mRNA expression correlates with the centralnervous system progenitor cell state in many, but not all, regions of developingcentral nervous system. Brain Res Dev Brain Res 1995;84(1):109–29.

Das KP, Freudenrich TM, Mundy WR. Assessment of PC12 cell differentiation andneurite growth: a comparison of morphological and neurochemical measures.Neurotoxicol Teratol 2004;26(3):397–406.

DeFuria J, Shea TB. Arsenic inhibits neurofilament transport and induces perikaryalaccumulation of phosphorylated neurofilaments: roles of JNK and GSK-3beta.Brain Res 2007;1181:74–82.

Dent EW, Gertler FB. Cytoskeletal dynamics and transport in growth cone motility andaxon guidance. Neuron 2003;40(2):209–27.

Ditlevsen DK, Povlsen GK, Berezin V, Bock E. NCAM-induced intracellular signalingrevisited. J Neurosci Res 2008;86(4):727–43.

Donaldson JG, Finazzi D, Klausner RD. Brefeldin A inhibits Golgi membrane-catalysedexchange of guanine nucleotide onto ARF protein. Nature 1992;360(6402):350–2.

Dragunow M. High-content analysis in neuroscience. Nat Rev Neurosci 2008;9(10):779–88.

Fletcher TL, Banker GA. The establishment of polarity by hippocampal neurons: therelationship between the stage of a cell’s development in situ and its subsequentdevelopment in culture. Dev Biol 1989;136(2):446–54.

George J, Dravid SM, Prakash A, Xie J, Peterson J, Jabba SV, et al. Sodium channelactivation augments NMDA receptor function and promotes neurite outgrowth inimmature cerebrocortical neurons. J Neurosci 2009;29(10):3288–301.

Gerecke KM, Wyss JM, Carroll SL. Neuregulin-1beta induces neurite extension andarborization in cultured hippocampal neurons. Mol Cell Neurosci 2004;27(4):379–93.

Gilyarov AV. Nestin in central nervous system cells. Neurosci Behav Physiol2008;38(2):165–9.

Grandjean P, Landrigan PJ. Developmental neurotoxicity of industrial chemicals. Lancet2006;368(9553):2167–78.

Harry GJ, Tiffany-Castiglioni E. Evaluation of neurotoxic potential by use of in vitrosystems. Expert Opin Drug Metab Toxicol 2005;1(4):701–13.

Hatten ME, Lynch M, Rydel RE, Sanchez J, Joseph-Silverstein J, Moscatelli D, et al. Invitro neurite extension by granule neurons is dependent upon astroglial-derivedfibroblast growth factor. Dev Biol 1988;125(2):280–9.

Higgins D, Banker G. Primary dissociated cell cultures. In: Stevens CF, Banker G, GoslinK, editors. Culturing nerve cells 2nd ed. Cambridge, MA: MIT Press; 1998.

Hollander BA, Bennett GS. Lithium chloride alters cytoskeletal organization in growing,but not mature, cultured chick sensory neurons. J Neurosci Res 1991;28(3):332–42.

Hong MS, Hong SJ, Barhoumi R, Burghardt RC, Donnelly KC, Wild JR, et al. Neurotoxicityinduced in differentiated SK-N-SH-SY5Y human neuroblastoma cells by organo-phosphorus compounds. Toxicol Appl Pharmacol 2003;186(2):110–8.

Howard AS, Bucelli R, Jett DA, Bruun D, Yang D, Lein PJ. Chlorpyrifos exerts opposingeffects on axonal and dendritic growth in primary neuronal cultures. Toxicol ApplPharmacol 2005;207(2):112–24.

Ibata Y, Toya S, Kohsaka S. Effects of astroglia on the development of cultured neuronsfrom embryonic rat cerebral cortex. Comp Biochem Physiol C 1991;98(1):139–46.

Ishitani T, Ishitani S, Matsumoto K, Itoh M. Nemo-like kinase is involved in NGF-induced neurite outgrowth via phosphorylating MAP1B and paxillin. J Neurochem2009;111(5):1104–18.

Jareb M, Banker G. Inhibition of axonal growth by brefeldin A in hippocampal neuronsin culture. J Neurosci 1997;17(23):8955–63.

Jin X, Hu H, Mathers PH, Agmon A. Brain-derived neurotrophic factor mediates activity-dependent dendritic growth in nonpyramidal neocortical interneurons in devel-oping organotypic cultures. J Neurosci 2003;23(13):5662–73.

Khaibullina AA, Rosenstein JM, Krum JM. Vascular endothelial growth factor promotesneurite maturation in primary CNS neuronal cultures. Brain Res Dev Brain Res2004;148(1):59–68.

Kim B, Leventhal PS, Saltiel AR, Feldman AR. Insulin-like growth factor-I-mediatedneurite outgrowth in vitro requires mitogen-activated protein kinase activation. JBiol Chem 1997;272(34):21268–73.

Koike T, Uno S, Ishizawa M, Takahashi H, Ikeda K, Yokota S, et al. The heat shock proteininhibitor KNK437 induces neurite outgrowth in PC12 cells. Neurosci Lett2006;410(3):212–7.

Kolkova K, Novitskaya V, Pedersen N, Berezin V, Bock E. Neural cell adhesionmolecule-stimulated neurite outgrowth depends on activation of protein kinaseC and the Ras-mitogen-activated protein kinase pathway. J Neurosci 2000;20(6):2238–46.

Kosik KS, Finch EA. MAP2 and tau segregate into dendritic and axonal domains after theelaboration of morphologically distinct neurites: an immunocytochemical study ofcultured rat cerebrum. J Neurosci 1987;7(10):3142–53.

Larsson C. Protein kinase C and the regulation of the actin cytoskeleton. Cell Signal2006;18(3):276–84.

Lein P, Gallagher PJ, Amodeo J, Howie H, Roth JA. Manganese induces neurite outgrowthin PC12 cells via upregulation of alpha(v) integrins. Brain Res 2000;885(2):220–30.

Lein P, Silbergeld E, Locke P, Goldberg AM. In vitro and other alternative approaches todevelopmental neurotoxicity testing (DNT). Environ Toxicol Pharmacol 2005;19:735–44.

J.A. Harrill et al. / NeuroToxicology 31 (2010) 277–290290

LePage KT, Dickey RW, Gerwick WH, Jester EL, Murray TF. On the use of neuro-2aneuroblastoma cells versus intact neurons in primary culture for neurotoxicitystudies. Crit Rev Neurobiol 2005;17(1):27–50.

Mandell JW, Banker GA. Microtubule-associated proteins, phosphorylation gradients,and the establishment of neuronal polarity. Perspect Dev Neurobiol 1996;4(2–3):125–35.

McNeish J. Embryonic stem cells in drug discovery. Nat Rev Drug Discov 2004;3(1):70–80.

Moors M, Rockel TD, Abel J, Cline JE, Gassmann K, Schreiber T, et al. Human neuro-spheres as three-dimensional cellular systems for developmental neurotoxicitytesting. Environ Health Perspect 2009;117(7):1131–8.

Munoz-Montano JR, Lim F, Moreno FJ, Avila J, Diaz-Nido J. Glycogen synthase kinase-3modulates neurite outgrowth in cultured neurons: possible implications forneurite pathology in Alzheimer’s disease. J Alzheimers Dis 1999;1(6):361–78.

Nat R, Nilbratt M, Narkilahti S, Winblad B, Hovatta O, Nordberg A. Neurogenicneuroepithelial and radial glial cells generated from six human embryonic stemcell lines in serum-free suspension and adherent cultures. Glia 2007;55(4):385–99.

National Research Council, Board on Environmental Studies and Toxicology, Institutefor Laboratory Animal Research, Institute for Laboratory Animal Research. Toxicitytesting in the 21st century. A vision and a strategy. Washington, D.C.: NationalAcademies Press; 2007.

Orme MH, Giannini AL, Vivanco MD, Kypta RM. Glycogen synthase kinase-3 and Axinfunction in a beta-catenin-independent pathway that regulates neurite outgrowthin neuroblastoma cells. Mol Cell Neurosci 2003;24(3):673–86.

Patrakitkomjorn S, Kobayashi D, Morikawa T, Wilson MM, Tsubota N, Irie A, et al.Neurofibromatosis type 1 (NF1) tumor suppressor, neurofibromin, regulates theneuronal differentiation of PC12 cells via its associating protein, CRMP-2. J BiolChem 2008;283(14):9399–413.

Prager-Khoutorsky M, Spira ME. Neurite retraction and regrowth regulated by mem-brane retrieval, membrane supply, and actin dynamics. Brain Res 2009;1251:65–79.

Radio NM, Mundy WR. Developmental neurotoxicity testing in vitro: models for asses-sing chemical effects on neurite outgrowth. Neurotoxicology 2008;29(3):361–76.

Radio NM, Breier JM, Shafer TJ, Mundy WR. Assessment of chemical effects on neuriteoutgrowth in PC12 cells using high content screening. Toxicol Sci 2008;105(1):106–18.

Radio NM, Freudenrich TM, Robinette BL, Crofton KM, Mundy WR. Comparison of PC12and cerebellar granule cell cultures for evaluating neurite outgrowth using highcontent analysis. Neurotoxicol Teratol 2010;32(1):25–35.

Redmond L, Kashani AH, Ghosh A. Calcium regulation of dendritic growth via CaMkinase IV and CREB-mediated transcription. Neuron 2002;34(6):999–1010.

Reubinoff BE, Itsykson P, Turetsky T, Pera MF, Reinhartz E, Itzik A, et al. Neuralprogenitors from human embryonic stem cells. Nat Biotechnol 2001;19(12):1134–40.

Rice D, Barone S Jr. Critical periods of vulnerability for the developing nervous system:evidence from humans and animal models. Environ Health Perspect2000;108(Suppl. 3):511–33.

Rolletschek A, Chang H, Guan K, Czyz J, Meyer M, Wobus AM. Differentiation ofembryonic stem cell-derived dopaminergic neurons is enhanced by survival-promoting factors. Mech Dev 2001;105(1–2):93–104.

Sanes DH, Reh TA, Harriss WA. Axon growth and guidance. In: Development of thenervous system. London, UK: Elsevier Academic Press; 2006 p. 111–138.

Sayas CL, Avila J, Wandosell F. Glycogen synthase kinase-3 is activated in neuronal cellsby Galpha12 and Galpha13 by Rho-independent and Rho-dependent mechanisms.J Neurosci 2002;22(16):6863–75.

Schmid RS, Graff RD, Schaller MD, Chen S, Schachner M, Hemperly JJ, et al. NCAMstimulates the Ras-MAPK pathway and CREB phosphorylation in neuronal cells. JNeurobiol 1999;38(4):542–58.

Shin S, Dalton S, Stice SL. Human motor neuron differentiation from human embryonicstem cells. Stem Cells Dev 2005;14(3):266–9.

Shin S, Mitalipova M, Noggle S, Tibbitts D, Venable A, Rao R, et al. Long-termproliferation of human embryonic stem cell-derived neuroepithelial cells usingdefined adherent culture conditions. Stem Cells 2006;24(1):125–38.

Takahashi M, Yasutake K, Tomizawa K. Lithium inhibits neurite growth and tau proteinkinase I/glycogen synthase kinase-3beta-dependent phosphorylation of juveniletau in cultured hippocampal neurons. J Neurochem 1999;73(5):2073–83.

Tamura H, Ohkuma S. Induction of neurite outgrowth of PC12 cells by an inhibitor ofvacuolar H(+)-ATPase, bafilomycin A1. FEBS Lett 1991;294(1–2):51–5.

Tang BL. Protein trafficking mechanisms associated with neurite outgrowth andpolarized sorting in neurons. J Neurochem 2001;79(5):923–30.

Thomson JA, Itskovitz-Eldor J, Shapiro SS, Waknitz MA, Swiergiel JJ, Marshall VS, et al.Embryonic stem cell lines derived from human blastocysts. Science1998;282(5391):1145–7.

Tohyama T, Lee VM, Rorke LB, Marvin M, McKay RD, Trojanowski JQ. Nestin expressionin embryonic human neuroepithelium and in human neuroepithelial tumor cells.Lab Invest 1992;66(3):303–13.

Ulfig N, Nickel J, Bohl J. Monoclonal antibodies SMI 311 and SMI 312 as tools toinvestigate the maturation of nerve cells and axonal patterns in human fetal brain.Cell Tissue Res 1998;291(3):433–43.

VanDeMark KL, Guizzetti M, Giordano G, Costa LG. The activation of M1 muscarinicreceptor signaling induces neuronal differentiation in pyramidal hippocampalneurons. J Pharmacol Exp Ther 2009;329(2):532–42.

Vutskits L, Gascon E, Tassonyi E, Kiss JZ. Effect of ketamine on dendritic arbordevelopment and survival of immature GABAergic neurons in vitro. Toxicol Sci2006;91(2):540–9.

Wiese C, Rolletschek A, Kania G, Blyszczuk P, Tarasov KV, Tarasova Y, et al. Nestinexpression—a property of multi-lineage progenitor cells? Cell Mol Life Sci2004;61(19–20):2510–22.

Wu YY, Bradshaw RA. Effect of nerve growth factor and fibroblast growth factor onPC12 cells: inhibition by orthovanadate. J Cell Biol 1993;121(2):409–22.

Yamauchi J, Miyamoto Y, Murabe M, Fujiwara Y, Sanbe A, Fujita Y, et al. Gadd45a, thegene induced by the mood stabilizer valproic acid, regulates neurite outgrowththrough JNK and the substrate paxillin in N1E-115 neuroblastoma cells. Exp CellRes 2007;313(9):1886–96.

Yoshimura T, Kawano Y, Arimura N, Kawabata S, Kikuchi A, Kaibuchi K. GSK-3betaregulates phosphorylation of CRMP-2 and neuronal polarity. Cell 2005;120(1):137–49.

Yoshimura T, Arimura N, Kaibuchi K. Signaling networks in neuronal polarization. JNeurosci 2006;26(42):10626–30.

Yu X, Malenka RC. Beta-catenin is critical for dendritic morphogenesis. Nat Neurosci2003;6(11):1169–77.

Zhang SC, Wernig M, Duncan ID, Brustle O, Thomson JA. In vitro differentiation oftransplantable neural precursors from human embryonic stem cells. Nat Biotech-nol 2001;19(12):1129–33.

Zhang W, Smith A, Liu JP, Cheung NS, Zhou S, Liu K, et al. GSK3beta modulates PACAP-induced neuritogenesis in PC12 cells by acting downstream of Rap1 in a caveolae-dependent manner. Cell Signal 2009a;21(2):237–45.

Zhang Y, Wang YG, Zhang Q, Liu XJ, Liu X, Jiao L, et al. Interaction of Mint2 with TrkA isinvolved in regulation of nerve growth factor-induced neurite outgrowth. J BiolChem 2009b;284(18):12469–79.