STEM CELLS AND DEVELOPMENT 15:200–208 (2006)© Mary Ann Liebert, Inc.

Original Research Report

Uniform Adherent Neural Progenitor Populations from Rhesus Embryonic Stem Cells

DEANNE TIBBITTS,1,4 RAJ R. RAO,2,4 SOOJUNG SHIN,3,4 FRANKLIN D. WEST,4and STEVEN L. STICE4

ABSTRACT

Rhesus and human embryonic stem cells (ESCs) are similar, making rhesus ESCs an appropriatepreclinical allograft model for refining stem cell therapies. Use of rhesus ESC-derived neural pro-genitors (NPs) in preclinical applications will be enhanced if the neural derivation process is scal-able and free from contaminating ESCs or nonneural cells. In this study, we have quantified tem-poral gene expression changes of rhesus ESC differentiated to uniform NPs using simple feeder-freeadherent cultures. NPs exhibited a significant up-regulation of neural-specific genes and a down-regulation of pluripotency genes. Additionally, expression of Hu, MAP2, and Tuj1, shows that NPscan form post-mitotic neurons. This study represents a simple and scalable means of producing ad-herent primate NPs for preclinical testing of neural cell-based therapy.

INTRODUCTION

HUMAN AND RHESUS EMBRYONIC STEM CELLS (ESCs)share many characteristics, including morphology,

surface marker expression, and developmental potential(1). Rhesus ESCs can thus play an important intermedi-ate translational role as stem cell differentiation strate-gies are transitioned from rodent systems to human clin-ical applications. Clinical trials for human ESCs as atherapy will be greatly enhanced by data gathered first inan allograft primate transplant model. Rhesus ESCs pro-vide the opportunity to refine differentiation protocols ina species in which allogeneic transplants of these deriv-atives can be used to test in vivo efficacy. The methodof propagating cells of interest becomes important whenconsidering the scalability and application of these cells

in the last preclinical large animal models, such as non-human primates. For any clinical application, the abilityto generate large numbers of cells is critical, use of bulkpassage methods, and a feeder-free system will facilitateprogress. Manual dissection of rhesus embryoid bodies(EBs) has been employed to select neural rosettes or tubesof columnar cells (2,3); however, any technique that re-quires manual selection and propagation may hamperscalable expansion of the cells for large animal studies.

Previously, most ESC differentiation studies derivedthe cell type of interest through an EB intermediate. Dif-ferentiation via this three-dimensional structure allowsfor cell–cell communication that is not possible in an ad-herent cell culture setting. Rhesus ESC-derived EBs pro-duced multiple cell types. Kuo and colleagues (3) usedthe EBs to initiate differentiation of rhesus ESCs and

1Department of Molecular and Medical Genetics, Oregon Health and Science University, Portland, OR 97239.2Department of Chemical and Life Science Engineering, Virginia Commonwealth University, Richmond, VA 23284-3028.3Stem Cell Biology Section, Laboratory of Neurosciences, National Institute on Aging, Baltimore, MD 21224.4Regenerative Bioscience Center, The University of Georgia, Athens, GA 30602

200

RHESUS ESC-DERIVED NEURAL PROGENITORS

201

showed that this method produced cells expressing mark-ers from all three germ layers. However, the difficulty inprecisely controlling differentiation and the subsequentappearance of multiple cell types are inherent problemsof this method (4). Dang and colleagues compared EBdifferentiation cultures to adherent differentiation cultureand reported that cell number limitation was not a factorin adherent differentiation cultures. In addition, theyshowed that adherent differentiation seemed to excludecell differentiation toward hematopoietic development.Further studies also demonstrated that adherent differen-tiation with mouse ESCs produced efficient neural com-mitment (5). In addition, we have previously demon-strated through immunocytochemistry that greater than90% of the adherent neural progenitor cells derived fromhuman ESC were nestin and Musashi positive and noneof the cells were Oct-4 positive (6). Together, recent stud-ies have focused on directed differentiation of ESC to adesired cell type or progenitor populations; however si-multaneous quantitative gene expression analysis of bothneural and nonneural genes has not been determined inrhesus monkey ESCs.

The aim of this study was to quantitate the temporalchanges in gene expression of adherent cultures of rhe-sus monkey ESCs undergoing differentiation in condi-tions conducive to generating neural phenotypes. We al-lowed rhesus ESC to differentiate spontaneously onMatrigel for 7 days before transferring them to a neural-permissive substrate and medium that selectively propa-gated neural progenitors. Using real-time PCR, we dem-onstrate that these cells exhibit between 1.5- and 17-foldup-regulation of neural developmental gene expressionand a significant down-regulation of pluripotent gene ex-pression with little contamination from other germ lay-ers. As expected, gene expression gradually changed overtime, with the most dramatic change observed at day 17.The cells were then propagated enzymatically for threepassages (nestin positive) and upon further differentia-tion also exhibited expression of key mature neuronalmarkers (HU, MAP2, and Tuj). This population of rhe-sus neural progenitors is uniquely suited for further stud-ies due to the simplicity of its derivation, the uniformityof the cultures (no feeder cells), and the ease of its prop-agation.

MATERIALS AND METHODS

ESC culture

Rhesus ESC were routinely cultured in Dulbecco’smodified Eagle medium (DMEM)/F12 base medium(Gibco) supplemented with 20% knockout serum re-placement (Gibco), 2 mM L-glutamine (Gibco), 0.1 mMnonessential amino acids (Gibco), 50 units/ml peni-

cillin/50 �g/ml streptomycin (Gibco), 4 ng/ml basic fi-broblast growth factor (bFGF) (Sigma-Aldrich), and 0.1mM �-mercaptoethanol (Sigma-Aldrich). Cells were pas-saged every 3–4 days using 0.05% trypsin-EDTA(Gibco). Cells were cultured on inactivated mouse em-bryonic fibroblasts (MEFs) plated 3 or more days priorto use. The cells were checked for normal karyotype priorto experimental use.

Neural differentiation

Rhesus ESCs were grown for 7 days on Matrigel-coateddishes in ES growth medium. Cells were trypsinized andreplated on dishes coated with poly-L-ornithine andlaminin and grown for 10 days in neural-permissivemedium. Neural permissive medium was comprised ofDMEM/F12 supplemented with 2 mM L-glutamine, 50units/ml penicillin/50 �g/ml streptomycin, 20 ng/ml bFGF,and 1 � N2 supplement (Gibco). 1 � N2 supplement iscomposed of human transferrin (100 �g/ml), bovine in-sulin (5 �g/ml), progesterone (6.3 ng/ml), putrescine(16.11 �g/ml), and selenite (5.2 ng/ml). These cells werepassaged every 5–7 days by brief exposure to 0.05%trypsin-EDTA solution. At each time point, cells were har-vested for flow cytometry and gene expression analysis.Neural progenitor cells were further differentiated by ex-posure to neural differentiation medium for 7 days andstained for a panel of neural markers.

Real-time PCR

In preparation for real-time PCR analysis, total RNAwas isolated from the crude homogenate using the Qi-ashredder kit (Qiagen) and RNeasy kit (Qiagen) accord-ing to the manufacturer’s instructions. The integrity ofthe RNA produced from all samples used was verifiedand quantified using a RNA 600 Nano Assay (AgilentTechnologies) and the Agilent 2100 Bioanalyzer. TotalRNA (5 �g) was reverse-transcribed using the cDNAArchive Kit (Applied Biosystems Inc.) according to man-ufacturer’s protocols using the MultiScribe™ ReverseTranscriptase. Reactions were incubated initially at 25°Cfor 10 min and subsequently at 37°C for 120 min. Quan-titative PCR (Taqman™) assays were chosen for the tran-scripts to be evaluated from Assays-On-Demand™ (ABI)a prevalidated library of human-specific QPCR assaysand incorporated into 384-well Micro-Fluidic Cards™.Two microliters of the cDNA samples (diluted to 50 �l)along with 50 �l of 2 � PCR master mix were loadedinto respective channels on the microfluidic card fol-lowed by a brief centrifugation. The card was then sealed,and real-time PCR and relative quantitation was carriedout on the ABI PRISM 7900 Sequence Detection Sys-tem (Applied Biosystems Inc.). After excluding failed re-actions (reactions called as ‘undetermined’ by the SDS

software) and genes for which fewer than two replicatesexisted from further analysis, �Ct values were calculated.For calculation of relative fold change values, initial nor-malization was achieved against endogenous 18S ribo-somal RNA using the ��CT method of quantification(ABI). Average fold changes from three independent runswere calculated as 2���CT. Data analysis for differentialexpression between the different populations was con-ducted in triplicate. Significance was determined by run-ning t-tests for each gene at a 95% confidence interval(p value � 0.05) between day 0 and day 7, day 0 and day17, and day 7 and day 17. �Ct and fold change valueswere calculated within each replicate, and then averagedto produce relevant standard deviation (SD) values.

Immunocytochemistry

Cells were seeded onto Permanox chamber slides (Nunc)coated with Matrigel or poly-L-ornithine/laminin and fixedwith 4% paraformaldehyde/4% sucrose for 20 min. Anti-bodies were directed against SSEA-4 (Chemicon, 1:1000),Oct-4 (Santa Cruz, 1:500), nestin (Neuromics, 1:100),smooth muscle actin (Dako, 1:50), MAP2abc (Sigma-Aldrich, 1:200), HuC/D (Molecular Probes, 1:50), and Tuj1(Covance Research Products, 1:500). Antibodies were de-tected using secondary antibodies conjugated to AlexaFluor 488 or 594 (Molecular Probes, 1:1,000).

Flow cytometry

Cells were fixed in 2% paraformaldehyde/2% sucrosefor 15 min, washed twice with Dulbecco’s phosphatebuffered saline (DPBS), and held at 4°C in 1% bovineserum albumin (BSA) in DPBS until all samples werecollected. For detecting surface markers, cells wereblocked in 1% BSA in DPBS and stained with eitherSSEA-4 (1:2000; Chemicon) or mouse immunoglobulinG (IgG) isotype control (1:2,000; Sigma). Antibody wasdetected using a fluorescently conjugated secondary an-tibody (goat anti-mouse IgG Alexa Fluor 488, 1:1,000;Molecular Probes). Cells were analyzed using a Beck-man Coulter FC500. A total of 10,000 events werecounted for analysis.

RESULTS

Differentiation of rhesus ES cells to neural progenitors

To generate an adherent population of neural progeni-tors, we took a two-step approach. Rhesus ES cells wereplated on Matrigel in ES growth medium and allowed todifferentiate spontaneously for 7 days. The cells werebriefly exposed to trypsin and replated on a polyor-

nithine/laminin substrate in a neural-permissive medium.To determine the percentage of cells that were differenti-ating in response to the treatment, we collected cells at threetime points and subjected them to flow cytometry. Table 1shows the decrease in expression of SSEA-4, a cell-surfaceglycolipid and marker of pluripotency, as the ES cellsmoved toward a neural phenotype. After 10 days in cultureon polyornithine/laminin these cells had developed dis-tinctive neural progenitor morphology (Fig. 1A,B). Im-munocytochemistry revealed these cells to be nestin posi-tive (Fig. 1C) and SSEA-4 and Oct-4 negative (Fig. 1D,E).

Temporal characterization of gene expressionusing real-time PCR

To quantitate the differentiation process further in thepopulation of cells obtained, cells were harvested for real-time PCR analysis at day 0, day 7, and day 17. Thesethree samples were assayed for the expression of a panelof genes that are indicative of pluripotency, ectoderm,mesoderm, or endoderm. As detailed in the Methods sec-tion, genes whose expression met specific stringent cri-teria (see Methods) were considered for further analysis.Table 2 summarizes genes with a significant fold changeat either day 7 or day 17.

Culture of neural progenitors

After their derivation, the cells were passaged enzy-matically every 5–7 days with �95% viability. After twopassages, the cells were cryopreserved, and subsequentthawing resulted in recovery of cultures. The cells re-tained the same morphology after three passages as theyhad after derivation.

Differentiation of neural progenitors to mature neurons

To assess the ability of these cells to differentiate tomature neurons, the neural progenitors were grown for 7days in a serum-free differentiation medium with reducedbFGF. The cells were stained for Tuj1, Hu, and MAP2(Fig. 2), all markers of post-mitotic neurons. Both MAP2and Tuj1 staining localized cytoplasmically in cell bod-

TIBBITTS ET AL.

202

TABLE 1. QUANTITATIVE SSEA4 EXPRESSION BASED

ON FLOW CYTOMETRY ANALYSIS, AT SPECIFIC

TIME POINTS (DAYS 0, 7, AND 17) OF

NEURAL PROGENITOR DIFFERENTIATION

% SSEA-4Timepoint expression

Day 0 (ES cells) 91.41Day 7 (cells on Matrigel) 22.45Day 17 (cells on polyornithine/laminin) 0.55

RHESUS ESC-DERIVED NEURAL PROGENITORS

203

ies and neurites, whereas Hu staining was confined to thenucleus, indicating the ability of these neural progenitorsto differentiate into more mature neurons.

DISCUSSION

To our knowledge, this is the first report of quantita-tive real-time PCR temporal analysis of initial neural dif-

ferentiation events from a starting rhesus ESC popula-tion. The derivation of the neural progenitors wasachieved using a simple adherent ES cell differentiationmethod. Previously most neural progenitors were derivedusing the EB to foster differentiation driven by three-di-mensional cell–cell interactions (3,6–8). When we com-bined the Matrigel adherent substrate with a definedmedium, rhesus ESC spontaneously differentiated for 7days. ESCs at this 7-day stage were harvested to moni-

TABLE 2. GENES ANALYZED BY REAL-TIME PCR WITH SIGNIFICANT FOLD CHANGE (p VALUE � 0.05)

Gene Category Expression FC � SDa

Day 0 versus day 7CD9 Pluripotent ↓ 4.51 � 0.27CER1 Pluripotent ↓ 89.61 � 0.95FGF13 Pluripotent ↓ 2.68 � 0.07FOXH1 Pluripotent ↓ 7.26 � 0.005EBAF Pluripotent ↓ 5.70 � 0.72FN1 Neural-related ↑ 3.61 � 0.25GATA3 Neural-related ↑ 21.96 � 0.64PTCH Neural-related ↑ 3.89 � 0.18BMP4 Mesoderm-related ↑ 3.04 � 0.09GSC Mesoderm-related ↓ 17.75 � 0.09HOXB4 Mesoderm-related ↓ 4.78 � 0.66KDR Mesoderm-related ↓ 4.65 � 0.88TBX6 Mesoderm-related ↓ 3.18 � 0.39GATA6 Endoderm-related ↓ 13.24 � 0.75

Day 0 versus day 17CD9 Pluripotent ↓ 3.80 � 0.16CER1 Pluripotent ↓ 18.83 � 0.82DNMT3B Pluripotent ↓ 36.99 � 0.23FOXH1 Pluripotent ↓ 146.38 � 0.29GABRB3 Pluripotent ↓ 4.62 � 0.55FN1 Neural-related ↑ 6.47 � 0.11GATA3 Neural-related ↑ 16.50 � 0.48HEY1 Neural-related ↑ 9.68 � 0.19LMO2 Neural-related ↑ 3.77 � 0.20MADH2 Neural-related ↑ 1.98 � 0.12MCAM Neural-related ↑ 2.72 � 0.23MSI1 Neural-related ↑ 5.31 � 0.06NCAM Neural-related ↑ 9.01 � 0.44NEDD5 Neural-related ↑ 1.45 � 0.11NEFL Neural-related ↑ 10.00 � 0.11PTCH Neural-related ↑ 8.20 � 0.80THBS1 Neural-related ↑ 2.66 � 0.61VCAM Neural-related ↑ 17.16 � 0.70OTX2 Ectoderm-related ↓ 2.51 � 0.20FLT1 Mesoderm-related ↓ 40.23 � 0.32GSC Mesoderm-related ↓ 34.48 � 0.23TBX6 Mesoderm-related ↓ 3.41 � 0.74

Comparative analysis was conducted between samples at day 7 (rhesus ES cells on Matrigel) and day 17 (rhesus neural pro-genitors on polyornithine/laminin) against day 0 (rhesus ES cells).

aFC, Fold change; SD, standard deviation.

tor changes in gene expression. Without the addition ofMEF conditioned medium, Matrigel alone has beenshown to be insufficient to maintain the pluripotency ofhuman ESCs (9), thus making it an appropriate substratefor this protocol. After 7 days, the cells were transferredto a serum-free medium containing N2 and 4 ng/ml bFGFon polyornithine/laminin-coated dishes. bFGF has longbeen a recognized mitogen of neural precursor cells (10)and polyornithine and laminin have previously been usedtogether to support neural progenitors (11). The cells de-rived from this protocol developed an appropriate neuralprogenitor morphology and were nestin positive (Fig. 2)and further characterized by immunocytochemistry andreal-time PCR. In this study, we found that 7 days of dif-ferentiation without serum was sufficient to induce neuraldifferentiation. An additional 10 days of neural inductionresulted in further increased neural-associated gene ex-pression and decreases in pluripotent gene expression.

The panel of real-time PCR probes and primers arrayedon the ABI microfluidics card were optimized for the hu-man sequence of each gene; it was unknown how wellany of these would recognize rhesus monkey transcripts.Overall, 28.6% of the total reactions yielded a resultcalled ‘undetermined’ by the SDS 2.1 software. This wasdue to inefficient amplification of the target sequence,which is caused by either insufficient availability ofcDNA or weak homology to the probe/primer set. Be-cause an equal amount of cDNA was loaded in all wellsof the card, the latter is the more likely cause of any failedreactions. We believe on the basis of the stringency ofour criteria, the genes retained for final analysis reflectthe gene expression profile of the population of cells un-dergoing neural differentiation.

As shown in Table 2, many of the genes up-regulatedat day 17 have been previously described as genes re-lated to neural development. These genes encode extra-cellular matrix proteins (FN1, THBS1), cell adhesionmolecules (VCAM, NCAM, MCAM), transcription fac-tors (GATA3, HEY1, LMO2, MADH2), an RNA-bind-ing protein (Musashi-1), a transmembrane receptor(Patched), and cytoskeletal proteins (NEFL, NEDD5).Numerous studies have demonstrated the relevance ofthese genes to neural development. Fibronectin-1 (FN-1)expression increased 3.6- and 6.5-fold on days 7 and 17,respectively. Fibronectin is a fundamental component ofthe neural crest migratory pathway (12), and fibronectin-deficient mice exhibit increased neuronal apoptosis (13).In developing mice (E9), thrombospondin-1 (2.7-foldchange on day 17) was found to be densely deposited be-tween neuroepithelial cells (14).

The vascular cell adhesion molecule VCAM was moreup-regulated than any other gene at day 17; its expres-sion increased 17-fold compared to day 0 (ESC). In mice,VCAM has been shown to be co-expressed with nestinin neural progenitors of the central nervous system (15).

Other cell adhesion molecules were significantly up-reg-ulated (MCAM, 2.7-fold; NCAM, 9-fold), though to alesser extent. MCAM is strongly expressed in the chickin both the neural epithelium of the developing forebrainand neural crest cells (16), whereas NCAM is expressedin neural progenitors as they migrate from the ventricu-lar zone of the developing brain (17).

Neural-related transcription factors related to neuraldevelopment were also up-regulated. Hey-1 has been im-plicated in the maintenance of neural precursors in thebrain and is highly expressed in the ventricular zone ofthe developing brain and increased 9.7-fold on day 17.LMO2 is highly expressed in the developing mouse cen-tral nervous system (18). GATA3 is expressed in the ven-tral neural tube and developing mouse hindbrain (19),whereas MADH2 transcripts are present in the neurecto-derm of the forebrain and hindbrain of mouse embryos(E10.5) (20). Overall, we observed that these neural-as-sociated transcription factors were more highly expressedafter the neural phenotype was observed at day 17.

Other neural-related genes were also up-regulated.Musashi, an RNA-binding protein, was up-regulated 5-fold at day 17 and has been shown to be initially expressed in the neural tube and later restricted to theventricular zone (21). Patched, the receptor for SonicHedgehog, was up-regulated 8-fold at day 17 in our study.Localization of Patched varies temporally, but in thechick, expression is first restricted to the neural tube, ex-tending from the caudal end through the diencephalon(22). Using a LacZ reporter under the control of the neu-rofilament light-chain (NEFL) promoter, Yaworsky andcolleagues (23) were able to localize expression of thepromoter to neuroepithelial cells in transgenic mouse em-bryos (E8.5), indicating the relevance to neural develop-ment. NEDD5 is one of a group of genes discovered inmouse neural precursor cells that are down-regulated dur-ing development of the brain (24), pointing to a role inearly neural development. We found that both NEFL andNEDD5 were up significantly up-regulated at day 17. Intotal, 13 neural developmentally related genes were upregulated by day 17 and SSEA4 was absent based on flowcytometry analysis.

The only gene significantly up-regulated at day 7 wasBMP4, which is an important factor in nascent mesoderminduction and patterning (25). Transient expression ofBMP4 at day 7 but not at day 17 is likely indicative ofthe mixed population of cells present at the day 7 timepoint. Its regression to near baseline levels at day 17 ispotentially a result of directed neural differentiation inthe overall population of cells.

Most genes significantly down-regulated at day 7 andday 17 are markers of pluripotent human ESCs. Many ofthe analyzed genes down-regulated in our study havebeen found to be highly expressed by human ES cells inother large-scale gene expression studies (26–28). In an

TIBBITTS ET AL.

204

205

FIG. 2. Immunocytochemical staining for (A,B) MAP2 (green) and DAPI (blue); (C,D) Tuj1 (red), HuC/D (green), and DAPI(blue) of rhesus neural progenitors differentiated for 7 days. Scale bar, 100 �m.

FIG. 1. Phase-contrast micrograph (A,B) and immunocytochemical staining (C) for nestin (green), DAPI (blue); (D) SSEA4(green), DAPI (blue); (E) Oct-4 (green), DAPI (blue) of rhesus neural progenitor cells. Insets are rhesus ES cells stained posi-tive for SSEA4 (D) and Oct-4 (E). Scale bar, 100 �m.

array comparing hES cells and human embryonic carci-noma (EC) cells to other tumor cell lines and normal tis-sues, Sperger and colleagues (29) identified GABRB3 asthe second most up-regulated receptor in hESCs, withDNMT3� and OTX2 as the third and twelfth most posi-tively significant genes in hESC overall, respectively. Bran-denberger and colleagues (30) identified Cerebrus (CER-1), FGF13, LeftyA, FoxH1, and Goosecoid as commonlyexpressed genes in their analysis of pooled mRNA fromthe hES cell lines. It has also been proposed that Goose-coid is directly regulated by FoxH1–dependent transcrip-tional activation (31). In our study, we found that all ofthese genes were significantly-down regulated by day 17of differentiation. Simultaneous down-regulation of bothpluripotent and candidate genes associated with mesodermdifferentiation in our study would seem to support that thisadherent culture system is specific for neural induction.

Other genes down-regulated in our study have also beenassociated with or involved in mouse ES cell maintenanceor mesoderm early development. In mouse ES cells, CD9was found to be regulated by the LIF/STAT3 pathway (32),and while it has recently been shown that LIF does notsupport maintenance of pluripotency in human ES cells(33), this has not nullified the use of CD9 as a marker ofpluripotency (34). Upon generation of GATA-6 knockoutmice, Koutsourakis and colleagues (35) found expressionof GATA-6 in a subset of cells in the inner cell mass of3.5-day preimplantation embryos, with null mutants ex-hibiting lethality at 5.5 days post coitus (dpc).

HoxB4 plays a role in very early hematopoiesis (36) butmay be active earlier in development; Kuliev and col-leagues (37) detected HoxB4 transcripts in cleavage-stagehuman embryos. Mesoderm-associated genes were signif-icantly down-regulated genes at both day 7 and day 17.Tbx6 is expressed in paraxial mesoderm and is requiredfor cells to choose between a neural or mesodermal fate(38). KDR and Flt-1 are receptors for both vascular endo-thelial growth factor (VEGF) and placental growth factor(39), two proteins important for angiogenesis, althoughthey play a role in placental development as well (40).

In this study, real-time PCR using the microfluidicsapproach offers the opportunity to quantitatively exam-ine a limited set of candidate genes believed to be im-portant in early development. Large-scale gene expres-sion profiling (cDNA and oligo microarrays) has becomean important and popular tool to compare human stemcell populations (26–29,41,42) as well as to monitor dif-ferentiation (43–45), and results are confirmed via quan-titative PCR or other comparable methods. However,large-scale rhesus monkey microarrays are not available,so we elected to use a more focused approach using real-time PCR. Discrepancies between optimal rhesus and hu-man probe/primer sets may have impacted our ability togenerate replicates with ‘tight’ standard deviations andeliminated some important genes from our final list. Un-

fortunately Pou5f1 was one of these genes. AlthoughPou5f1 was down-regulated, it produced variabilityamong replicates, thus not meeting our defined stringencycriteria, and had to be excluded. However, on the basisof the greatly reduced SSEA4 expression, there is littledoubt that there are very few pluripotent cells at day 17.

Taken together, the group of genes up-regulated at day17 is a strong indicator of a population of neural pro-genitors. The lack of significant up-regulation of non-neural genes at day 17 and the down-regulation ofpluripotent markers points to the selectivity of the dif-ferentiation process. For a clinical application, the con-sistency of the transplanted population of cells is critical,and we feel that this method of neural progenitor de-rivation is both simple and robust. Further experimentswill be required to examine the ability of these cells torespond to terminal differentiation cues to produce ther-apeutically relevant products. However, on the basis ofgene expression, morphology, and immunocytochemicalanalysis, the rhesus neural cells derived in this study aresimilar to neural stem-like cells derived from human EScells (6). This simple and scalable method of producingneural progenitors represents a step forward in theprogress toward a primate model for cell-based therapyof neurodegenerative diseases.

ACKNOWLEDGMENTS

We thank Don Wolf for the generous gift of rhesus EScells and Roger Nielsen for assistance with real-timePCR. This work was supported in part by funding fromthe National Institutes of Health (NIH-NS44208-2) andpartly by the Georgia Tech/Emory Center (GTEC) for theEngineering of Living Tissues, and ERC program of theNational Science Foundation under Award Number EEC-9731643.

REFERENCES

1. Wolf DP, HC Kuo, KY Pau and L Lester. (2004). Progresswith nonhuman primate embryonic stem cells. Biol Reprod71:1766–1771.

2. Calhoun JD, NA Lambert, MM Mitalipova, SA Noggle, ILyons, BG Condie and SL Stice. (2003). Differentiation ofrhesus embryonic stem cells to neural progenitors and neu-rons. Biochem Biophys Res Commun 306:191–197.

3. Kuo HC, KY Pau, RR Yeoman, SM Mitalipov, H Okanoand DP Wolf. (2003). Differentiation of monkey embry-onic stem cells into neural lineages. Biol Reprod 68:1727–1735.

4. Dang SM, M Kyba, R Perlingeiro, GQ Daley and PW Zand-stra. (2002). Efficiency of embryoid body formation andhematopoietic development from embryonic stem cells indifferent culture systems. Biotechnol Bioeng 78:442–453.

TIBBITTS ET AL.

206

RHESUS ESC-DERIVED NEURAL PROGENITORS

207

5. Ying QL, M Stavridis, D Griffiths, M Li and A Smith.(2003). Conversion of embryonic stem cells into neuroec-todermal precursors in adherent monoculture. NatureBiotechnol 21:183–186.

6. Shin S, M Mitalipova, S Noggle, D Tibbitts, A Venable, RRao and SL Stice. (2006). Long-term proliferation of hu-man embryonic stem cell-derived neuroepithelial cells us-ing defined adherent culture conditions. Stem Cells 24(1):125–138.

7. Okabe S, K Forsberg-Nilsson, AC Spiro, M Segal and RDMcKay. (1996). Development of neuronal precursor cellsand functional postmitotic neurons from embryonic stemcells in vitro. Mech Dev 59:89–102.

8. Zhang SC, M Wernig, ID Duncan, O Brustle and JA Thom-son. (2001). In vitro differentiation of transplantable neuralprecursors from human embryonic stem cells. NatureBiotechnol 19:1129–1133.

9. Xu C, MS Inokuma, J Denham, K Golds, P Kundu, JDGold and MK Carpenter. (2001). Feeder-free growth of un-differentiated human embryonic stem cells. NatureBiotechnol 19:971–974.

10. Kitchens DL, EY Snyder and DI Gottlieb. (1994). FGF andEGF are mitogens for immortalized neural progenitors. JNeurobiol 25:797–807.

11. Pau KY and DP Wolf. (2004). Derivation and characteri-zation of monkey embryonic stem cells. Reprod Biol En-docrinol 2:41.

12. Perris R and D Perissinotto. (2000). Role of the extracel-lular matrix during neural crest cell migration. Mech Dev95:3–21.

13. Sakai T, KJ Johnson, M Murozono, K Sakai, MA Magnu-son, T Wieloch, T Cronberg, A Isshiki, HP Erickson andR Fassler. (2001). Plasma fibronectin supports neuronalsurvival and reduces brain injury following transient focalcerebral ischemia but is not essential for skin-wound heal-ing and hemostasis. Nature Med 7:324–330.

14. O’Shea KS and VM Dixit. (1988). Unique distribution ofthe extracellular matrix component thrombospondin in thedeveloping mouse embryo. J Cell Biol 107:2737–2748.

15. Sheppard AM, JJ McQuillan, MF Iademarco and DC Dean.(1995). Control of vascular cell adhesion molecule-1 genepromoter activity during neural differentiation. J Biol Chem270:3710–3719.

16. Pujades C, B Guez-Guez and D Dunon. (2002). Melanomacell adhesion molecule (MCAM) expression in the myo-genic lineage during early chick embryonic development.Int J Dev Biol 46:263–266.

17. Cai J, Y Wu, T Mirua, JL Pierce, MT Lucero, KH Alber-tine, GJ Spangrude and MS Rao. (2002). Properties of afetal multipotent neural stem cell (NEP cell). Dev Biol251:221–240.

18. Hinks GL, B Shah, SJ French, LS Campos, K Staley, JHughes and MV Sofroniew. (1997). Expression of LIMprotein genes Lmo1, Lmo2, and Lmo3 in adult mouse hip-pocampus and other forebrain regions: differential regula-tion by seizure activity. J Neurosci 17:5549–5559.

19. Pata I, M Studer, JH van Doorninck, J Briscoe, S Kuuse, JDEngel, F Grosveld and A Karis. (1999). The transcription fac-tor GATA3 is a downstream effector of Hoxb1 specificationin rhombomere 4. Development 126:5523–5531.

20. Dick A, W Risau and H Drexler. (1998). Expression ofSmad1 and Smad2 during embryogenesis suggests a rolein organ development. Dev Dyn 211:293–305.

21. Sakakibara S, T Imai, K Hamaguchi, M Okabe, J Aruga,K Nakajima, D Yasutomi, T Nagata, Y Kurihara, S Ue-sugi, T Miyata, M Ogawa, K Mikoshiba and H Okano.(1996). Mouse-Musashi-1, a neural RNA-binding proteinhighly enriched in the mammalian CNS stem cell. Dev Biol176:230–242.

22. Marigo V and CJ Tabin. (1996). Regulation of patched bysonic hedgehog in the developing neural tube. Proc NatlAcad Sci USA 93:9346–9351.

23. Yaworsky PJ, DP Gardner and C Kappen. (1997). Trans-genic analyses reveal developmentally regulated neuron-and muscle-specific elements in the murine neurofilamentlight chain gene promoter. J Biol Chem 272:25112–25120.

24. Kumar S, Y Tomooka and M Noda. (1992). Identificationof a set of genes with developmentally down-regulated ex-pression in the mouse brain. Biochem Biophys Res Com-mun 185:1155–1161.

25. Dale L and CM Jones. (1999). BMP signalling in earlyXenopus development. Bioessays 21:751–760.

26. Rao RR, JD Calhoun, X Qin, R Rekaya, JK Clark and SLStice. (2004). Comparative transcriptional profiling of twohuman embryonic stem cell lines. Biotechnol Bioeng88:273–286.

27. Rao RR and SL Stice. (2004). Gene expression profilingof embryonic stem cells leads to greater understanding ofpluripotency and early developmental events. Biol Reprod71(6):1772–1778.

28. Sato N, IM Sanjuan, M Heke, M Uchida, F Naef and AHBrivanlou. (2003). Molecular signature of human embry-onic stem cells and its comparison with the mouse. DevBiol 260:404–413.

29. Sperger JM, X Chen, JS Draper, JE Antosiewicz, CH Chon,SB Jones, JD Brooks, PW Andrews, PO Brown and JAThomson. (2003). Gene expression patterns in human em-bryonic stem cells and human pluripotent germ cell tumors.Proc Natl Acad Sci USA 100:13350–13355.

30. Brandenberger R, I Khrebtukova, RS Thies, T Miura, CJingli, R Puri, T Vasicek, J Lebkowski and M Rao. (2004).MPSS profiling of human embryonic stem cells. BMC DevBiol 4:10.

31. Hoodless PA, M Pye, C Chazaud, E Labbe, L Attisano, JRossant and JL Wrana. (2001). FoxH1 (Fast) functions tospecify the anterior primitive streak in the mouse. GenesDev 15:1257–1271.

32. Oka M, K Tagoku, TL Russell, Y Nakano, T Hamazaki, EMMeyer, T Yokota and N Terada. (2002). CD9 is associatedwith leukemia inhibitory factor-mediated maintenance of em-bryonic stem cells. Mol Biol Cell 13:1274–1281.

33. Daheron L, SL Opitz, H Zaehres, WM Lensch, PW An-drews, J Itskovitz-Eldor and GQ Daley. (2004). LIF/STAT3 signaling fails to maintain self-renewal of humanembryonic stem cells. Stem Cells 22:770–778.

34. Rosler ES, GJ Fisk, X Ares, J Irving, T Miura, MS Raoand MK Carpenter. (2004). Long-term culture of humanembryonic stem cells in feeder-free conditions. Dev Dyn229:259–274.

35. Koutsourakis M, A Langeveld, R Patient, R Beddington

and F Grosveld. (1999). The transcription factor GATA6is essential for early extraembryonic development. Devel-opment 126:723–732.

36. Sauvageau G, U Thorsteinsdottir, CJ Eaves, HJ Lawrence,C Largman, PM Lansdorp and RK Humphries. (1995).Overexpression of HOXB4 in hematopoietic cells causesthe selective expansion of more primitive populations invitro and in vivo. Genes Dev 9:1753–1765.

37. Kuliev A, V Kukharenko, G Morozov, M Freidine, S Re-chitsky, O Verlinsky, V Ivakhnenko, V Gindilis, C Stromand Y Verlinsky. (1996). Expression of homebox-contain-ing genes in human preimplantation development and inembryos with chromosomal aneuploidies. J Assist ReprodGenet 13:177–181.

38. Chapman DL and VE Papaioannou. (1998). Three neuraltubes in mouse embryos with mutations in the T-box geneTbx6. Nature 391:695–697.

39. Autiero M, J Waltenberger, D Communi, A Kranz, LMoons, D Lambrechts, J Kroll, S Plaisance, M De Mol, FBono, S Kliche, G Fellbrich, K Ballmer-Hofer, DMaglione, U Mayr-Beyrle, M Dewerchin, S Dombrowski,D Stanimirovic, P Van Hummelen, C Dehio, DJ Hicklin,G Persico, JM Herbert, D Communi, M Shibuya, D Collen,EM Conway and P Carmeliet. (2003). Role of PlGF in theintra- and intermolecular cross talk between the VEGF re-ceptors Flt1 and Flk1. Nature Med 9:936–943.

40. Shore VH, TH Wang, CL Wang, RJ Torry, MR Caudle andDS Torry. (1997). Vascular endothelial growth factor, pla-centa growth factor and their receptors in isolated humantrophoblast. Placenta 18:657–665.

41. Bhattacharya B, T Miura, R Brandenberger, J Mejido, YLuo, AX Yang, BH Joshi, I Ginis, RS Thies, M Amit, ILyons, BG Condie, J Itskovitz-Eldor, MS Rao and RK Puri.

(2004). Gene expression in human embryonic stem celllines: unique molecular signature. Blood 103:2956–2964.

42. Ramalho-Santos M, S Yoon, Y Matsuzaki, RC Mulliganand DA Melton. (2002). “Stemness”: transcriptional pro-filing of embryonic and adult stem cells. Science298:597–600.

43. Bourne S, JM Polak, SP Hughes and LD Buttery. (2004).Osteogenic differentiation of mouse embryonic stem cells:differential gene expression analysis by cDNA microarrayand purification of osteoblasts by cadherin-11 magneticallyactivated cell sorting. Tissue Eng 10:796–806.

44. Calhoun JD, RR Rao, S Warrenfeltz, R Rekaya, S Dal-ton, J McDonald and SL Stice. (2004). Transcriptionalprofiling of initial differentiation events in human em-bryonic stem cells. Biochem Biophys Res Commun323:453–464.

45. Xu RH, X Chen, DS Li, R Li, GC Addicks, C Glennon,TP Zwaka and JA Thomson. (2002). BMP4 initiates hu-man embryonic stem cell differentiation to trophoblast. Na-ture Biotechnol 20:1261–1264.

Address reprint requests to:Steven L. Stice, Ph.D.

Director, Regenerative Bioscience CenterRhodes Animal Science Complex

425 River RoadThe University of Georgia

Athens, GA 30602

E-mail: sstice@uga.edu

Received December 17, 2005; accepted January 12, 2006.

TIBBITTS ET AL.

208