Lineage-Restricted Neural Precursors Can Be Isolated from Both the Mouse Neural Tube and Cultured ES Cells

Developmental Biology 214, 113–127 (1999) Article ID dbio.1999.9418, available online at http://www.idealibrary.com on View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector Lineage-Restricted Neural Precursors Can Be Isolated from Both the Mouse Neural Tube and Cultured ES Cells

T. Mujtaba,* D. R. Piper,‡ A. Kalyani,* A. K. Groves,† M. T. Lucero,‡ and M. S. Rao*,1 *Department of Neurobiology and Anatomy, University of Utah School of Medicine, 50 North Medical Drive, Salt Lake City, Utah 84132; †Division of Biology, 139-74, Caltech, Pasadena, California 91125; and ‡Department of Physiology, University of Utah School of Medicine, Salt Lake City, Utah 84108

We have previously identiﬁed multipotent neuroepithelial (NEP) stem cells and lineage-restricted, self-renewing precursor cells termed NRPs (neuron-restricted precursors) and GRPs (glial-restricted precursors) present in the developing rat spinal cord (A. Kalyani, K. Hobson, and M. S. Rao, 1997, Dev. Biol. 186, 202–223; M. S. Rao and M. Mayer-Proschel, 1997, Dev. Biol. 188, 48–63; M. Mayer-Proschel, A. J. Kalyani, T. Mujtaba, and M. S. Rao, 1997, Neuron 19, 773–785). We now show that cells identical to rat NEPs, NRPs, and GRPs are present in mouse neural tubes and that immunoselection against cell surface markers E-NCAM and A2B5 can be used to isolate NRPs and GRPs, respectively. Restricted precursors similar to NRPs and GRPs can also be isolated from mouse embryonic stem cells (ES cells). ES cell-derived NRPs are E-NCAM immunoreactive, undergo self-renewal in deﬁned medium, and differentiate into multiple neuronal phenotypes in mass culture. ES cells also generate A2B5-immunoreactive cells that are similar to E9 NEP-cell-derived GRPs and can differentiate into oligodendrocytes and astrocytes. Thus, lineage restricted precursors can be generated in vitro from cultured ES cells and these restricted precursors resemble those derived from mouse neural tubes. These results demonstrate the utility of using ES cells as a source of late embryonic precursor cells. © 1999 Academic Press Key Words: ES cells; A2B5; E-NCAM; self-renewal; neuroblast; glioblast; development.

INTRODUCTION E-NCAM-immunoreactive neuronal-restricted precursor termed an NRP, that can generate multiple kinds of neu- Pluripotent stem cells in the central nervous system rons but not oligodendrocytes or astrocytes (Mayer- (CNS) generate differentiated neurons and glia either di- Proschel et al., 1997; Kalyani et al., 1998), and an A2B5- rectly or through the generation of lineage restricted inter- immunoreactive glial-restricted precursor termed a mediate precursors (Kalyani and Rao, 1999; Kilpatrick and glial-restricted precursors (GRP) that can generate oligoden- Bartlett, 1993, 1995; Price et al., 1988, 1992; Reynolds et al., drocytes and astrocytes, but cannot generate neurons (Rao 1992; Reynolds and Weiss, 1996; Temple and Davis, 1994; and Mayer-Proschel, 1997; Mayer-Proschel et al., 1997; Rao Williams, 1995). We have shown multipotent neuroepithe- et al., 1998). The multipotent NEP cell present at E10.5 is lial (NEP) cells to be present in the spinal cord of E10.5 rats. lineally related to the neuron-restricted precursors (NRPs) NEP cells are self-renewing cells that can differentiate into and GRPs and a transition from NEP cells to intermediate neurons, oligodendrocytes, astrocytes (Kalyani et al., 1997), restricted precursors has been documented in both mass and peripheral nervous system (PNS) derivatives (Mujtaba and clonal cultures (Mayer-Proschel et al., 1997; Rao and et al., 1998). Differentiation of NEP cells occurs via the Mayer-Proschel, 1997; Rao et al., 1998). generation of more restricted precursors. At E13.5, two Cells similar to NEPs, NRPs, and GRPs have not yet been major types of lineage restricted precursors are present: an described in mouse or chick embryos. Multipotent stem cells exist both in the mouse cortex (Reynolds et al., 1992; 1 To whom correspondence should be addressed. E-mail:rao@ Hulspas et al., 1997; reviewed in McKay, 1997) and in the msscc.med.utah.edu. spinal cord (Kilpatrick and Bartlett, 1993, 1995; Weiss et al.,

1996a,b) but these differ significantly from rat NEP cells in immunoreactive cells could subsequently be induced to growth factor requirements, culture conditions, and differ- generate multiple classes of neurons (Okabe et al., 1996). entiation potential (reviewed in Kalyani and Rao, 1999). It Equally importantly, the sequential development of pluri- remains to be determined whether NEP-like stem cells potent cells into specialized CNS derivatives appears to exist in the developing mouse spinal cord. Like NEP cells, involve many of the same cytokines and transcription GRP-like cells have not been isolated from either the factors identified as being important in normal develop- developing mouse cortex or the mouse embryonic neural ment (Gajovic et al., 1997; Nye et al., 1994). These obser- tube although O-2A-like glial progenitors have been iden- vations raise the possibility that it may be possible to tified in the optic nerve and cortex at late embryonic ages isolate NRPs and GRPs from ES cells maintained in culture, (reviewed in Collarini et al., 1991). Likewise, NRPs have rather than from developing embryos. To utilize such a not been described in the mouse spinal cord although system, however, it is necessary to identify neural- and neuronally restricted precursor cells have been described in glial-restricted precursors in mice similar to those we have the subventricular zone and hippocampus by clonal analy- previously characterized from the developing rat spinal sis in vivo and in vitro (Young and Levison,1996; Luskin et cord. These experiments have not been attempted previ- al., 1988; Williams and Price, 1995; Williams, 1995). In ously, because well-characterized cell surface markers that chick embryos, the existence of spinal cord NRPs has been distinguish neuroblasts and glioblasts from other differen- demonstrated by retroviral tracing (Leber et al., 1990; Leber tiating cells were unavailable. and Sanes, 1995) suggesting that such cells exist during Our recent results have identified highly polysialated normal development. In addition, NRP cell lines that are NCAM (E-NCAM) and A2B5 as cell surface markers that E-NCAM immunoreactive and phenotypically similar to may be used to select neuroblasts and glioblasts, respec- rat NRPs have been generated from human spinal cord tively. In this study, we have used both markers to show cultures (Li et al., 1998), further suggesting that such that it is possible to use immunoselection to selectively lineage-restricted precursors are not unique to the rat spinal isolate cells that are phenotypically identical to E13.5 rat cord and that precursors similar to NRPs and GRPs may neural tube-derived NRPs and GRPs. We show that exist in mouse tissue as well. E-NCAM- and A2B5-immunoreactive cells undergo self- Although lineage-restricted precursors may exist in the renewal and (in clonal culture) differentiate into neurons or mouse spinal cord, their antigenic characteristics, growth glial cells, respectively. Furthermore ES cells can be in- profile, and growth factor response may be different. For duced to generate NRP and GRP cells similar to those present in the developing spinal cord and that these NRPs example, antibodies against A2B5 in rat cultures appear to and GRPs can differentiate into functionally mature cells. label glial precursors (Rao and Mayer-Proschel, 1997), while We discuss the potential therapeutic application of our in mice, A2B5 recognizes an epitope also expressed by results. neuronal precursors (Eisenbarth et al., 1979v; our unpublished results). Culture conditions and proliferative potential may also differ. For example, human cortical stem cells MATERIALS AND METHODS are more difficult to maintain in culture than mouse precursor cells (Svendsen et al., 1996). The fibroblast Substrate preparation. Fibronectin (Sigma) was diluted to a growth factor (FGF) sensitivity of rat and mouse cells may concentration of 20 ␮g/ml in distilled H2O (Sigma). Fibronectin also be different (Rao and Anderson, 1997). Thus, charac- solution was applied to tissue culture dishes for a minimum of 4 h. terizing the proliferative potential, phenotypic markers, Laminin (Biomedical Technologies Inc.), used at a concentration of and differentiation properties will be critical for future 20 ␮g/ml, was dissolved in Dulbecco’s phosphate-buffered saline analyses of mouse precursor cell cultures. (DPBS ,Gibco-BRL). To prepare fibronectin–laminin double-coated A specific incentive for describing growth and differen- dishes, laminin was applied to fibronectin-coated dishes and plates tiation potential of restricted neural precursors in mice is were incubated overnight at 4°C. Excess laminin was withdrawn and the plates were rinsed with medium. In some case plates were the ability to analyze the effects of single gene defects in precoated with pLL (30–70 kDa) (Sigma Inc.). pLLysine was dis- cultures of cells from transgenic and gene knockout mice. solved in distilled water (13.3 ␮g/ml) and applied to tissue culture Similar analysis remains difficult in most other organisms plates for an hour. Excess pLL was withdrawn and the plates were as embryonic stem (ES) cell lines which represent the allowed to air dry. Plates were rinsed with water and then allowed earliest totipotent cell have until recently been generated to dry again. pDL plates were then coated with FN solution or only from mice (reviewed in Thompson and Marshall, laminin as described above. 1998). ES cells have been shown to recapitulate normal Mouse cell cultures. Mouse C57Bl6 embryos were removed at differentiation in vitro (Fraichard et al., 1995; Okabe et al., embryonic day 9.0 and placed in a petri dish containing DPBS 1996; Li et al., 1998). Postmitotic neurons and glia gener- (Gibco-BRL). The trunk segments of embryos were enzymatically treated with collagenase (Worthington; 10 mg/ml) and dispase ated from ES cells appear phenotypically normal and inte- (Boehringer Mannheim; 20 mg/ml) for 10 min at room temperature. grate and function normally after transplantation (Thomp- The enzyme solution was replaced with fresh medium with 10% son et al., 1998). For example, McKay and colleagues chick embryo extract (CEE). The trunk segments were gently showed that ES-cell-derived neural stem cells could be triturated to release the spinal cords from surrounding tissue. The maintained in culture and that these nestin- spinal cords were then incubated in 0.05% trypsin solution for 5

min at 37°C. The cells were dissociated and plated in 35-mm at room temperature. Cells were allowed to bind to the dish for 1 h fibronectin-coated dishes as mass cultures (5000 cells/dish).Cells at room temperature. The supernatant was discarded and the dish were maintained at 37°C in 5% CO2/95% air in NEP basal medium was washed 8ϫ with DPBS. The bound cells were mechanically (see below) and 10% CEE. scraped off and plated on fibronectin/laminin-coated dishes in 1 ml Mouse C57B6 embryos were removed at embryonic day 12.0 and of NEP basal medium either as mass (5000 cells/dish) or clonal (100 placed in a petri dish containing DPBS. Spinal cords were mechani- cells/dish) cultures. Growth factors were added every other day. In cally dissected from surrounding connective tissue using sharpened all cases, an aliquot of cells was tested for panning efficiency by No. 5 forceps. Isolated spinal cords were incubated in 0.05% immunocytochemistry. In general, the panning efficiency was trypsin solution for 30 min. The trypsin solution was replaced with greater than 90%. For E-NCAM-panned or NRP cultures, the NEP fresh medium. The spinal cords were gently triturated with a basal medium was supplemented with the following growth fac- Pasteur pipette to dissociate the cells. The dissociated cells were tors: NGF, PDGF, brain-derived nerve growth factor (BDNF), and plated on fibronectin/laminin-coated dishes. Cells were main- FGF (all at 10ng/ml). For A2B5-panned or GRP cultures, the NEP tained at 37°C in 5% CO2/95% air in NEP complete medium (see basal medium was supplemented with PDGF (10 ng/ml) and FGF below). (10 ng/ml). Differentiation into neural crest was promoted by adding bone Immunocytochemistry. Staining procedures were as described morphogenetic protein-4 (BMP-4, 10 ng/ml) to the mouse NEP cells previously (Rao and Mayer-Proschel,1997). To stain spinal cord growing on fibronectin for a period of 4 days. The cells were then sections embryos were harvested, and either fresh frozen or fixed replated on fibronectin/laminin-coated dishes in neural crest me- embryos (4% paraformaldehyde overnight) were sectioned on a dium [NEP basal medium supplemented with bFGF (10 ng/ml), cryostat to obtain 20-␮m sections. These sections were processed nerve growth factor (NGF, 50 ng/ml), epidermal growth factor for immunocytochemistry as described below. Staining for the cell (EGF, 100 ng/ml)]. All growth factors were obtained from Upstate surface markers such as p75, E-NCAM, and galactocerebroside Biotechnolgy (Lake Placid, NY) unless otherwise stated. To pro- (GalC) was carried out in cultures of living cells without perme- mote maturation into Schwann cells, dibutryl cAMP (5 ␮m/ml, abilization. To stain cells with antibodies against cytoplasmic Sigma) was added for an additional 7 days. To promote smooth antigens, cultures were fixed with 2% formaldehyde for 20 min at muscle differentiation, crest cells were grown in neural crest room temperature. medium supplemented with 10% fetal bovine serum (FBS, Hy- In general, cultured cells or sections were incubated with pri- clone, Logan, UT). mary antibody for 1 h followed by incubation with an appropriate The NEP basal medium used in all experiments was a chemi- secondary antibody for 30 min. Double labeling experiments were cally defined medium modified from that described by (Stemple performed by simultaneously incubating cells in appropriate com- and Anderson, 1992; Kalyani et al., 1997). The medium consisted of binations of primary antibodies followed by noncrossreactive sec- DMEM-F12 (Gibco-BRL) supplemented with additives described by ondary antibodies. Bisbenzimide (DAPI) histochemistry was per- Bottenstein and Sato (1979), and bFGF (25 ng/ml, Upstate Biotech- formed as described previously (Kalyani et al., 1997). Bisbenzimide nology). The NEP complete medium is NEP basal medium with (Sigma) staining was generally done after labeling had been com- 10% CEE. pleted. p75, E-NCAM, and GalC antibodies were hybridoma super- ES cell cultures. Undifferentiated ES-D3 cells (ATCC) were natants obtained from DSHB. The A2B5 antibody was obtained grown as aggregates in suspension dishes (Nunc) in DMEM-F12 from ATCC. Microtubule-associated protein (MAP2) and ␤-III with 10% fetal calf serum (FCS) and leukemia inhibitory factor tubulin (Sigma) antibodies, which stain neurons, and nestin (mono- (LIF, 10 ng/ml, Gibco-BRL) for 4 days. The medium was then clonal from DSHB; polyclonal, a kind gift from Dr. Keith Cauley at changed to NEP basal medium and the cells were plated on Signal Pharmaceuticals), a marker for undifferentiated stem cells fibronectin. The medium was changed every 2 days. Differentiation (Zimmerman et al., 1994; Lendahl et al., 1990; Dahlstrand et al., into NRPs and GRPs was induced by plating the cells on poly-L- 1995), were used as described previously (Rao and Mayer-Proschel, lysine/laminin double-coated dishes. For long-term neuronal dif- 1997). Antibodies to glutamate and glycine were obtained from ferentiation, FGF was withdrawn and retinoic acid (RA, 100 ␮M, Signature Immunologicals and used as per the manufacturer’s Sigma) was added to the medium. Similarly for long-term glial recommendations. Antibodies to glial fibrillary acidic protein induction, cells were maintained in platelet-derived growth factor (GFAP) and glutamate decarboxylase (GAD) were obtained from (PDGF, 10 ng/ml, Upstate Biotechnology) and tri-iodothyronine Chemicon and used at 1:500 dilution. hormone (T3, 30 nM, Sigma) with the withdrawal of FGF. 5-Bromo-2؅-deoxyuridine (BRDU) incorporation. To assess the ,Immunopanning of E-NCAM؉ and A2B5؉ cells. E-NCAMϩ proliferation of neuronal and oligodendrocyte precursor cells and A2B5ϩ cells were purified from dissociated E12.0 neural tube BRDU (Sigma) at a concentration of 10 ␮M was added to the cells cells using a specific antibody capture assay (Wysocki and Sato, for 24 h. The cells were then fixed with 2% paraformaldehyde for 1978) with minor modifications. In brief, the cells were trypsinized 15 min at room temperature followed by 95% methanol for 30 min and 10 ml of cell suspension (NEP basal medium and 20% FCS) was at Ϫ20°C. Cells were then washed 3ϫ with PBS and 5% goat serum plated on a 100-mm petri dish and incubated at 37°C in a 5% CO2 and permeabilized with 2 N HCl for 10 min. Acid was removed by humidified atmosphere for 2–3 h for the maximum attachment of 3ϫ washes with PBS and 5% goat serum and the residual HCl was flat cells. The supernatant, containing an enriched population of neutralized with sodium borate (Sigma) for 10 min. After rinsing neuronal and oligodendrocyte precursor cells, was plated on dishes with PBS, cells were incubated with anti-BRDU antibody (1:100, coated with either E-NCAM antibody (5A5, Developmental Stud- Sigma) for 30 min at room temperature in buffer containing 0.5% ies Hybridoma Bank (DSHB, 1:1 dilution)) or A2B5 antibody Triton X-100. The cells were then incubated with goat anti-mouse (ATCC, 1:1 dilution) to allow binding of all E-NCAMϩ and A2B5ϩ IgG1 (1:100, Jackson Immunologicals) for 30 min. After three cells. E-NCAM and A2B5 dishes were prepared by sequentially washes with PBS, the cells were observed with a Zeiss Fluorescence coating petri dishes with an unlabeled anti-mouse IgM antibody (10 microscope. ␮g/ml) overnight and rinsing 3ϫ with DPBS, followed by incuba- RNA extraction, cDNA synthesis, and PCR reactions. Total tion with E-NCAM or A2B5 hybridoma supernatants for 1–3 hours RNA was isolated from cells or neural tubes by a modification of

TABLE 1

Gene Product (bp) Primers sense (top); antisense (bottom)

p75 329 5Ј GCA CAT ACT CAG ACG AAG CCA 3Ј 5Ј AGC AGC CAA GAT GGA GCA ATA GAC 3Ј ChAT 377 5Ј CTG AAT ACT GGC TGA ATG ACA TG 3Ј 5Ј AAA TTA ATG ACA ACA TCC AAG AC 3Ј Isl-1 350 5Ј GCA GCA TAG GCT TCA GCA AG 3Ј 5Ј GTA GCA GGT CCG CAA GGT G 3Ј

GAD65 327 5Ј GAA TCT TTT CTC CTG GTG GTG 3Ј 5Ј GAT CAA AAG CCC CGT ACA CAG 3Ј Calbindin28 276 5Ј GCA GAA TCC CAC CTG CAG - 3Ј 5Ј GTT GCT GGC ATC GAA AGA G 3Ј Glutaminase 560 5Ј GCA CAG ACA TGG TTG GGA TAC TAG 3Ј 5Ј GCA GGG CTG TTC TGG AGT CG 3Ј Cyclophilin 302 5Ј CCA CCG TGT TCT TCG ACA TC 3Ј 5Ј GGT CCA GCA TTT GCC ATG G 3Ј EGF-R 205 5Ј GCT GGG GAA GAG GAG AGG AGA 3Ј 5Ј ACG AGT GGT GGG CAG GTG TCT T 3Ј FGFR-1 764 5Ј TGG GAG CAT CAA CCA CAC CTA CC 3Ј 5Ј GCC CGA AGC AGC CCT CGC C 3Ј FGFR-4 672 5Ј ATC GGA GGC ATT CGG CTG CG 3Ј 5Ј AGA ACT GCC GGG CCA AAG GG 3Ј PLP/DM20 505/400 5Ј GAC ATG AAG CTC TCA CTG GCA C 3Ј 5Ј CAT ACA TTC TGG CAT CAG CGC 3Ј MAP2 404 5Ј GAA GGA AAG GCA CCA CAC TG 3Ј 5Ј GCT GGC GAT GGT GGT GGG 3Ј NF-M 186 5Ј GCC GAG CAG ACC AAG GAG GCC ATT 3Ј 5Ј CTG GAT GGT GTC CTG GTA GCT GCT 3Ј GFAP 346 5Ј TTG CAG ACC TCA CAG ACG CTG CGT 3Ј 5Ј CGG TTT TCT TCG CCC TCC AGC AAT 3Ј

the guanidinium–isothiocyanate extraction method (TRIzol, pluronic F127 (Sigma, 80 ␮g/ml) in rat Ringer’s solution (RR, see Gibco-BRL). cDNA was synthesized using 10,000 cells or one to below) at 23°C in the dark. Following the 20-min incubation, cells five neural tubes. The amount of first-stand cDNA synthesized was were washed three times in RR and allowed to deesterify for 30 estimated using a Bioquant fluorometer. Superscript II (Gibco- min. Relative changes in intracellular Ca2ϩ concentrations were BRL), a modified Moloney murine leukemia virus reverse transcrip- measured from the background-corrected ratio of fluorescence tase. and oligo(dT)12–18 primers were used, and the Gibco-BRL intensity by excitation at 340/380 nm. A response was defined as a protocol was followed. Equal amounts of cDNA (in general equiva- minimum rise of 10% of the ratioed baseline value. A Zeiss- lent to 1/20 of the above reaction) were used in a 50-␮l reaction Attofluor imaging system and software (Atto Instruments Inc.) volume. PCR amplification was performed using Elongase poly- were used to acquire and analyze the data. Data points were merase (Gibco-BRL). Primer sequences used for PCR amplification sampled at 1 Hz. Neurotransmitters (500 ␮M concentration) were of receptors are shown in the Table 1 below. The reactions were run made fresh in RR and delivered by bath exchange using a small for 35 cycles and a 10-min incubation at 72°C was added at the end volume loop injector (200 ␮l). RR consisted of (in mM): 140 NaCl, to ensure complete extension. Several controls were run to assure 3 KCl, 1 MgCl2, 2 CaCl2, 10 Hepes, and 10 glucose. A concentration the fidelity of the reaction. No RT, no primer controls, and positive of 500 ␮M ascorbic acid was added to dopamine solutions to controls with whole embryo extracts were routinely run alongside prevent oxidation. Control application of 500 ␮M ascorbic acid had test samples. In cases where levels needed to be estimated samples no effect (Figs. 5A and 5C). The pH of all solutions was adjusted to were quantified by running a portion of the amplified product on an 7.4 with NaOH. High K RR was made by substituting 50 mM Kϩ ethidium bromide gel and quantifying levels using a Kodak gel for Naϩ in the normal RR. documentation and analysis system. To normalize samples and test for the integrity of cDNA synthesized GAPDH/actin were amplified in the same tube (or in sister tubes) and levels were RESULTS normalized to the amount of GAPDH/actin. In all cases tests were run to ensure that samples harvested were from the log phase of amplification. Stem cells present in the embryonic mouse neural tube Intracellular Ca2؉ measurements. Calcium imaging experi- resemble FGF-dependent NEP stem cells isolated from rat ments were performed on clonal cultures of E-NCAM-panned cells embryos at E10.5. Two kinds of stem cells have been obtained from mouse E12.0 embryos as described above. Cells were described in the CNS of developing embryos: an EGF- loaded with 5 ␮M fura-2 AM (Sigma, Grynkiewics et al., 1985) plus dependent stem cell present in more cranial portions of the

neural tube (Santa-Ollala and Covarrubias, 1995; Reynolds et al., 1992; Weiss et al., 1996a) that grows in suspension culture and a FGF and/or FGF/EGF-dependent stem cell that is present in more caudal neural tube (Kalyani et al., 1997; Weiss et al., 1996a,b) as well as in the cortex (Palmer et al., 1995; McKay, 1997). Both stem cells have been shown to be present at later developmental stages, but the properties of the stem cells present at early stages of mouse neurogenesis remain to be determined. To characterize the earliest mouse NEP precursor cells, we examined the antigenic properties, growth characteristics, and differentiation potential of stem cells present at the time of neural tube closure. E9 mouse neural tubes were isolated as previously described (Rao and Anderson, 1997; see also Materials and Methods) and cells were plated at low density to test the growth characteristics of NEP stem cells. NEP cells appeared homogenous and were nestin immunoreactive. Many cells (Ͼ70%) incorporated BRDU over a 6-h period indicating that they were actively dividing cells when maintained in 25 ng/ml of FGF (Fig. 1A). NEP cells did not express any markers of early as tested by either immunocytochemistry or RT-PCR differentiation (Figs. 1D and 2A, lanes 2, 4, and 6). In particular, NEP cells did not express E-NCAM, ␤-III tubulin, or A2B5 immunoreactivity (Fig. 1D) indicating that they were undifferentiated precursor cells. Nestin-immunoreactive cells persisted around the ventricular zone at E10 (Fig. 1C) while ␤-III tubulin- immunoreactive neurons appeared to differentiate at the ventricular margins (Fig. 1C). To determine the growth factor dependence of E9 NEP cells, several growth factors were tested at a variety of different concentrations (Fig. 1E). FGF was sufficient to maintain NEP cell proliferation. In particular, NGF, PDGF, and EGF were ineffective at any dose tested. None of the growth factors tested (Fig. 1F) acted synergistically with FGF to alter cell division rates (Fig. 1F) suggesting that EGF and PDGF were neither survival or proliferation factors for NEP cells. The failure to see a response to EGF indicated that these cells were distinct from both the EGF-dependent neurospheres and the FGF/EGF-dependent cells isolated from later stages in the spinal cord (Reynolds et al., 1992; Reynolds and Weiss, 1996; Weiss et al., 1996a,b). To confirm the similarity between E9.0 mouse NEP cells and previously characterized rat neuroepithelial stem cells, we FIG. 1. Mouse spinal cord NEP cells are nestin-immunoreactive examined EGF-R expression in our cultures. As can be seen FGF-dependent cells. E10 mouse neural tubes were sectioned (C in Fig. 2 we failed to detect EGF-R in mouse NEP cells by and D) or E9 neural tube cells were dissociated and plated onto 35-mm fibronectin-coated dishes at low density in NEP basal medium with bFGF alone (A, AЈ,B,BЈ) or in combination with different growth factors (E and F). Cells were allowed to grow overnight, pulsed with BRDU for 6 h, and processed for nestin zone. D, summarizes the antigenic profile of nestin- expression (B and BЈ), BRDU incorporation (A, AЈ, and F), and immunoreactive NEP cells in vitro and in vivo (section staining). markers of differentiation (D). As can be seen in A (red) and AЈ Mouse NEP cells at this stage express nestin and are immunonega- (phase) cells have divided in culture. Panel B (green) and B’ (phase) tive for all lineage markers tested. E, shows the concentration range show that all cells express nestin immunoreactivity. C, a trans- of various growth factors tested. Dose–response curves in F show verse section of an E10 mouse embryo neural tube double labeled that bFGF alone (at 10 ng/ml) is sufficient to induce a maximal with antibodies against mouse monoclonal nestin (green) and proliferative effect while PDGF, EGF, and NGF have no additive mouse monoclonal ␤-III tubulin (red). Note that nestin positive effect. All experiments were run in duplicate and repeated three cells are ␤-III tubulin negative and are present in the ventricular times with five random fields counted in each dish.

neurospheres as well as FGF/EGF-dependent mouse stem cells isolated at later stages of development. E9.0 mouse NEP cells could be maintained in NEP basal medium with the addition of FGF for several passages (at least 10) and could then be induced to differentiate by replating the cells on laminin or poly-L-lysine and reducing FGF concentrations. NEP cells under these conditions readily differentiated into neurons, astrocytes and oligodendrocytes (Figs. 3A–3D). Clonal analysis of mouse NEP cells (N ϭ 65) confirmed that individual cells (N ϭ 40 or 61%) were multipotent and that a single cell could generate neurons, astrocytes, and oligodendrocytes. Thus, like rat NEP cells, mouse NEP cells are FGF dependent, grow in adherent culture, and can generate CNS derivatives. To further confirm the similarities between mouse and rat NEP cells we asked if mouse NEP cells could generate neural crest derivatives. We have shown previously that rat NEP cells can generate neural crest stem cells either sto- chastically or in response to the addition of BMP-2 or BMP-4 (Mujtaba et al., 1998). As can be seen in Fig. 3, mouse NEP cells generate p75-immunoreactive crest cells, FIG. 2. Mouse spinal cord NEP cells are undifferentiated cells smooth muscle actin (SMA)-immunoreactive smooth that express FGFRs but do not express EGFRs. RNA isolated from mouse neural tubes at E9, dissociated cells at E12, and rat NEP cells muscle cells that coexpress desmin immunoreactivity but at E10.5 was reverse transcribed as described under Materials and do not express GFAP immunoreactivity and peripheral glia Methods to obtain cDNA. Equal concentrations of the resulting (Schwann cells) and neurons (Figs. 3E and 3G and data not cDNA (10 ng/reaction) were used for PCR analysis using primers shown). PCR analysis also confirms an increase in the listed in Table 1. PCR products were resolved on a 2% agarose gel expression of p75 (Fig. 3F) as described for rat NEP cells and single bands of appropriate sizes were detected. A, shows that (Mujtaba et al., 1997), when NEP cells are induced to E9 mouse NEP cells do not express MAP-2 (lane 2) or GFAP (lane differentiate into neural crest by the addition of 10 ng/ml of 4) and express a single band of DM-20 (lane 6). In contrast these BMP-4. Neural crest and its derivatives have not been markers are readily detected in differentiated cells (lanes 3, 5, and generated from EGF-dependent stem cells (reviewed in 7, respectively). Note that both PLP and DM-20 are expressed by differentiated cells (compare lane 6 and lane 7). Lane 1 shows that Kalyani and Rao, 1999). Thus, the developing mouse neural no amplification is seen when primers are omitted from the PCR tube contains nestin-immunoreactive, FGF-dependent plu- reaction and lane 8 shows that cyclophilin can be readily amplified ripotent cells that resemble rat NEP cells, rather than from E9 cDNA indicating the integrity of the cDNA. Thus PCR EGF-dependent neurosphere stem cells. amplification confirms the immunocytochemical analysis demon- Lineage-restricted neuronal precursors exist in the de- strating that NEP cells do not express any neuronal or glial veloping mouse spinal cord. We have previously shown markers. Controls for the PCR analysis are described under Mate- that differentiation in rat neural tubes involves progressive rials and Methods. E9 neural tube cDNA was also used to test for restriction of cell fate (reviewed in Kalyani and Rao, 1999). expression of EGF and FGF receptors (B). FGFR1 and FGFR4 were readily detected in mouse NEP cells (lane 5 and lane 7). In contrast NEP cells generate more restricted (but still multipotent) no EGF-R expression was seen (lane 4). EGF-R expression, however, precursors both in vitro and in vivo (Mayer-Proschel et al., was readily detected in rat brain (lane 1) and mouse glial cells (lane 1997; Rao et al., 1998). Two such restricted multipotent 3) which were run in parallel demonstrating the sensitivity of the precursors have been described: an E-NCAM- amplification reaction. Comparison of mouse NEP cell cDNA with immunoreactive neuronal precursor (Mayer-Proschel et al., rat NEP cell cDNA shows that the profile of EGF and FGF receptor 1997; Kalyani et al., 1998) and an A2B5-immunoreactive expression is similar (compare lanes 4, 5, and 7 with lanes 2, 6, and glial precursor (Rao and Mayer-Proschel, 1997; Rao et al., 8, respectively). Thus mouse NEP cells like rat NEP cells express 1998). To determine if similar neuronal precursors could be FGFRs but not EGFRs. isolated from mouse neural tubes, we analyzed E12.0 mouse neural tubes for the presence of E-NCAM- immunoreactive cells. A significant proportion of E12.0 PCR. In contrast, EGF-R was readily detected in whole neural tubes cells were E-NCAM immunoreactive (ranging brain (Fig. 2B) as well as on glial precursors, neurons, and from 70 to 80% in three independent experiments). Immu- ϩ astrocytes (Fig. 2B and data not shown). Identical results noselected E-NCAM cells divided in culture (Fig. 4A) and were obtained using an antibody specific to mouse EGF-R expressed neuronal but not glial markers (table in Fig. 4). (data not shown). Thus, mouse NEP cells appeared more E-NCAMϩ cells could be maintained as a dividing popula- similar to rat NEP cells as described previously by Kalyani tion of cells for a period of a month (data not shown) when et al. (1997) and were distinct from EGF-dependent mouse grown in FGF (10 ng/ml), NGF (10 ng/ml), PDGF (10 ng/ml),

and BDNF (10 ng/ml). Thus, based on their antigenic profile, E-NCAM-immunoreactive cells appeared to be dividing neuronal precursor cells (Fig. 4A). To test the differentiation potential of E-NCA-immunoreactive cells, purified populations of E-NCAMϩ cells were analyzed in mass and clonal culture. E-NCAMϩ cells, when allowed to differentiate by addition of retinoic acid and reduction of FGF concentration, readily differentiated into multiple kinds of neurons as assessed by RT-PCR (Fig. 4D) and immunocytochemistry (data not shown). However, E-NCAM- immunoreactive cells failed to differentiate into either oligodendrocytes or astrocytes when grown under glial promoting conditions, indicating that these cells are limited in their differentiation potential. Glial promoting conditions tested included and addition of T3 and PDGF or 10% fetal calf serum, conditions under which A2B5 immunoselected cells or NEP cells readily generated glial derivatives (Figs. 3 and 6 and data not shown). To examine the ability of differentiated E-NCAMϩ cells to respond to neurotransmitters and elevated Kϩ, we used fura-2 Ca2ϩ imaging techniques. E12.0 E-NCAMϩ cells were grown in culture for 10 days and allowed to differentiate. They were then loaded with fura-2 (see Materials and Methods) and the changes in internal Ca2ϩ concentrations in response to stimulus application were monitored. Figure 5A shows a bar graph of the number of cells (sum of cells from two independent experiments) responding to application of the indicated neurotransmitter. Most cells (Ͼ95%) responded to external application of glutamate or acetylcholine. Smaller fractions of cells responded to dopamine, GABA, or glycine. A substantial fraction of the cells re- FIG. 3. Mouse NEP cells can differentiate into neurons, astro- sponded to more than one neurotransmitter (summarized in Fig. 5B) demonstrating heterogeneity in the neuronal popu- cytes, and oligodendrocytes. E9 mouse NEP cells grown on fi- ϩ bronectin for 5 days were harvested by trypsinization and replated lation derived from E-NCAM precursors. Representative 2ϩ onto fibronectin/laminin-coated dishes in NEP medium without traces of internal Ca changes plotted over time are shown CEE for a period of 5–10 days. Some culture dishes were then for two different cells from the same clonally derived processed for labeling with ␤-III tubulin (red, A and B), A2B5 (green, culture (Fig. 5C). While both cells responded to glutamate, B), GalC (red, C), and GFAP (green, D). Other culture dishes were acetylcholine, and high Kϩ, the cell in the lower trace also exposed to BMP-4 (10 ng/ml) for 4 days. The cells were then responded to dopamine, indicating that the receptor expres- replated on fibronectin/laminin-coated dishes in neural crest me- sion profile of cells in the same culture varies. Thus dium (see Materials and Methods) and were allowed to differentiate E-NCAM-immunoreactive cells are neuronal-restricted for an additional period of 5 (E and F) or 10 days (G and H). Cells were then processed for the expression of specific markers by either precursor cells that can differentiate into a heterogeneous RT-PCR (H) or immunocytochemistry for expression of p75 (red, E), population of neurons and closely resemble their rat coun- SMA/GFAP (red/green, respectively, F) and SMA/desmin (red/ terparts in antigenic characteristics, differentiation proper- green, respectively, G and GЈ). NEP cells could differentiate into ties, and growth potential. ␤-III-immunoreactive neurons (A), A2B5-immunoreactive/␤-III Lineage-restricted glial precursors exist in the develop- tubulin-negative glial precursors (B), GalC-immunoreactive oligo- ing mouse spinal cord. To determine if A2B5 immunore- dendrocytes (C), and GFAP-immunoreactive astrocytes (D). Double activity could be used to isolate glial precursors at early labeling with A2B5 and ␤-III tubulin (B) shows that a small subset stages of embryonic development, we isolated A2B5- ␤ ϩ ϩ of -III tubulin are also A2B5 cells (B, arrowhead). These cells immunoreactive cells by immunopanning and mass cul- were excluded from subsequent analysis (see Results for a detailed discussion). NEP cells can also generate neural crest cells in culture when exposed to BMP (E and F). Note the large number of p75-immunoreactive cells (E) and the increased expression of p75 relative to cyclophilin when NEP cells are exposed to BMP-2 (H). p75-immunoreactive cells did not express neuronal or glial markers green, respectively, G and GЈ). NEP cells differentiated into SMA/ at this stage (unpublished results). NEP cells cultured for 10 days in desmin-immunoreactive smooth muscle cells (G and GЈ) that did neural crest medium supplemented with 10% FBS were stained for not express GFAP (F). Thus, NEP cell cells can differentiate into SMA/GFAP (red/green, respectively, F) and SMA/desmin (red/ both CNS and PNS cells.

FIG. 5. Differentiated E-NCAMϩ cells show heterogeneous responses to neurotransmitters. Ca2ϩ imaging experiments using fura-2 showed that cell populations from two different mouse E-NCAMϩ clones express different percentages of neurotransmitter receptors. The bar graph (A) shows the percentage of cells responding to 500 ␮M GABA, glycine (Gly), dopamine (DA), ascorbic acid (AA), L-glutamate (Glu), acetylcholine (ACh), betaine (Bet), and 50 mM Kϩ (high K). The red bars represent 95 cells from clone 1 and the blue bars represent 99 cells from clone 2. Pie charts (B) show the FIG. 4. E-NCAM-immunoreactive cells are neuron-restricted pre- percentage of cells that responded to one or specific combinations cursor (NRP) cells. E12.0 mouse neural tube cells were immunos- of neurotransmitters. The left and right graphs correspond to cells elected for E-NCAM expression. The immunoselected cell population from clones 1 and 2, respectively. Cells from clone 1 showed was plated on fibronectin/laminin-coated dishes at medium density more heterogeneity than clone 2. The ratio of intensity from in NEP medium supplemented with FGF (10 ng/ml), PDGF (10 excitation at 340 and 380 nm plotted over time shows that Ca2ϩ ng/ml), BDNF (10 ng/ml), and NT-3 (10 ng/ml) which were added to responses to neurotransmitters differ in two different cells from the cultured cells at day 2 for a period of 24 h. Cells were then stained for same clone (C). E-NCAM expression (green) and BRDU incorporation (red). Immunos- elected cells were E-NCAMϩ and a subset of cells had divided and incorporated BRDU within 24 h (A). B, lists the antigenic profile of acutely dissociated E-NCAM-immunoreactive cells. Note that ϩ tures of cells were tested for cell division by BRDU incor- E-NCAM cells express some neuronal markers and do not express poration. A2B5-imunoreactive cells (Fig. 6A) had divided in glial markers. An example of coexpression of a neuronal marker ϩ culture, indicating that these are a mitotic population. We MAP-2 (CЈ) with E-NCAM (C) is shown. Most E-NCAM cells are ϩ further examined the antigenic properties and differentia- MAP2 . Some of the cultures were allowed to mature for an additional 10 days and then processed for PCR analysis. Controls for the tion potential of these cells and noted one important PCR analysis are described under Materials and Methods. D, shows difference between rat and mouse A2B5-immunoreactive the presence of mature neuronal markers upon differentiation. These cell cultures: a small subset of the mouse A2B5- markers are not seen in acutely dissociated E-NCAMϩ cells (B). immunoreactive cells (Ͻ5%) were ␤-III tubulin immunore-

immunonegative population was repanned to isolate the A2B5-immunoreactive cells. Pooled A2B5ϩ/E-NCAMϪ cells were analyzed by PCR and immunocytochemistry; the results are summarized in Fig. 6. A2B5ϩ/E-NCAMϪ cells were nestin immunoreactive (Fig. 6) and did not express neuronal markers or markers of differentiated glial cells (Figs. 6B and 6C). A2B5ϩ/E- NCAMϪ cells however could differentiate into oligodendrocytes and two kinds of astrocytes (Figs. 6D–6F) but failed to differentiate into neurons when grown under neuron promoting conditions, indicating that these cells are limited in their differentiation potential. Thus glial-restricted and neuronal-restricted precursors are present in the developing mouse spinal cord and these precursor cells can be distinguished by E-NCAM and A2B5 immunoreactivity. ES-cell-derived E-NCAM cells are neuron-restricted precursor cells similar to E12.0 NRPs. ES cells have been

ϩ Ϫ shown to be capable of differentiating into neurons, astro- FIG. 6. A2B5 /E-NCAM cells express DM-20 and can differentiate into oligodendrocytes and two types of astrocytes. E12.0 cytes, and oligodendrocytes. More recently Okabe et al. mouse neural tube cells were sequentially immunopanned for (1996) and Li et al. (1998) have shown that NEP-like cells E-NCAM and A2B5 to obtain A2B5ϩ/E-NCAMϪ cells. A2B5ϩ/E- can be harvested from ES cells, bypassing the need to NCAMϪ cells were plated on fibronectin/laminin-coated dishes in harvest stem cells from fetal tissue. To determine if more NEP basal medium and their mitotic ability (A), antigenic proper- restricted neural precursors that appear at developmentally ties (B and C), and ability to differentiate (D–F) were tested. BRDU later embryonic stages could be harvested similarly, we was added to some cultures on day 2 for a period of 24 h and cells analyzed ES cell cultures. The ES-D3 cell line which has were processed for BRDU incorporation (A, red). Note that immu- ϩ been shown to be able to contribute to all cell lineages was nopanned cells are A2B5 and a subset has divided within 24 h (A). Sister plates of cells were either fixed and processed for immuno- obtained from ATCC and grown in nondifferentiating (as cytochemistry (B) or RNA harvested for RT-PCR analysis (C). The aggregates in DMEM/F12, 10% FCS, and LIF 10 ng/ml) and chart (B) summarizes the antigenic profile of undifferentiated differentiating conditions (as adherent cultures on A2B5-immunoreactive cells which are A2B5ϩ and nestinϩ at this polylysine/laminin substrate in NEP basal medium). Undif- stage and immunonegative for all neuronal and astrocytic markers ferentiated ES cells did not express E-NCAM immunoreac- tested. RT-PCR analysis (C) confirms the immunocytochemistry tivity (Fig. 7A) but upon aggregation and treatment with RA ϩ Ϫ results and shows that A2B5 /E-NCAM cells do not express NF, (1␮M), a subset (approx. 5%) of cells began to express GFAP, or PLP but do express DM-20 (see Results for a detailed E-NCAM immunoreactivity (Fig. 7B). E-NCAMϩ cells were discussion). To test the ability of A2B5ϩ/E-NCAMϪ to differentiate ϩ Ϫ a mitotic population as assessed by BRDU incorporation into oligodendrocytes and astrocytes cells A2B5 /E-NCAM cells were grown in medium containing ciliary neurotrophic factor (Figs. 8A and 8B). E-NCAM cells coexpressed MAP2, a ␤ (CNTF, 10 ng/ml), 10% FCS, or PDGF (10 ng/ml) for a period of marker for neurons (DeCamilli et al., 1984) and -III tubu- 7–14 days (D–F, respectively) and labeled for A2B5 expression lin immunoreactivity, but did not express GFAP, GalC, or (green, D and E), GFAP expression (red, D and E), and GalC O4 immunoreactivity (Figs. 7C, 7CЈ, 7D, 7DЈ, and 7F and expression (red, F). 50% of the A2B5ϩ/E-NCAMϪ-derived cells data not shown). A subset of the E-NCAM cells were nestin ϩ grown in CNTF differentiated into A2B5 process-bearing astro- immunoreactive (Fig. 7E) but none of the cells expressed cytes (D), while the others were A2B5Ϫ fibroblast-looking astro- Ϫ GFAP, a glial marker (Fig. 7F). The coexpression of neuronal cytes. The cells exposed to 10% FCS differentiated into A2B5 / markers with E-NCAM and the absence of glial markers GFAPϩ astrocytes (E). Exposure to PDGF/T3 resulted in mature ϩ suggest that ES-cell-derived E-NCAM-immunoreactive looking oligodendrocytes most of which were GalC (F). Thus, A2B5ϩ/E-NCAMϪ-immunoselected cells can generate oligodendro- cells are similar to E12.0 mouse neuronal-restricted precur- cytes and two kinds of astrocytes. sor cells. To further confirm the neuronal differentiation of ES- cell-derived E-NCAM cells, ES cells were induced to differentiate and E-NCAM-immunoreactive cells were selected active (Fig. 3B). This population was also E-NCAM immu- by immunopanning. Cells were grown in mass (Fig. 8) or noreactive as determined by immunopanning experiments clonal culture (data not shown) in medium supplemented (data not shown). Thus unlike rat spinal cord cultures, the with FGF and NT-3. When mass cultures of E-NCAM cells A2B5 epitope was not uniquely present on glial cells. To were induced to differentiate by withdrawal of FGF and the determine if the E-NCAM/␤-III tubulin-negative subset of addition of RA, cells became postmitotic (data not shown), A2B5 cells represented glial precursors, we isolated this elaborated extensive processes, and synthesized multiple population by sequential panning. E-NCAM-immunoreac- neurotransmitters. Immunocytochemistry and PCR analy- tive cells were isolated and discarded and the E-NCAM- sis show the presence of various excitatory, inhibitory, and

FIG. 8. ES-cell-derived E-NCAM-immunoreactive cells express mature neuronal markers upon differentiation. ES-D3-derived E-NCAMϩ cells were isolated by immunopanning, plated on fibronectin-coated dishes, and pulsed with BRDU for 24 h. Note that within 24 h, a significant proportion of the E-NCAMϩ-panned, ␤-III tubulin-immunoreactive cells (B) had incorporated BRDU (A). Cultured cells were allowed to differentiate for 10 days and were then analyzed by RT-PCR or by immunocytochemistry. Differen- tiated E-NCAMϩ cells were fixed and stained with antibodies to glutamate (C and CЈ), GAD (D and DЈ), and glycine (E and EЈ). Phase (CЈ,DЈ, and EЈ) and bright-field (C–E) images of representative fields FIG. 7. ES cells express early neuronal markers upon differen- ϩ tiation. ES-D3 cells were grown as aggregates for 4 days and then show staining with each antibody. Subsets of E-NCAM neurons plated on fibronectin-coated dishes in NEP basal medium. To expressed each marker, the proportions of which differed between allow differentiation, the cells were replated on polylysine/ antibodies [note both the presence (straight arrow) and the absence laminin and were grown in NEP basal medium. The cells were (curved arrow) of staining in different cells of the same population]. The PCR panel shows the expression of various mature neuronal processed for DAPI histochemistry (blue, A and B) and E-NCAM ϩ expression before and after differentiation. A, shows the absence markers tested. Thus, ES-cell-derived, E-NCAM cells, which do of E-NCAM immunoreactivity in undifferentiated ES cells. After not express mature neuronal markers (data not shown) when differentiation, the cells expressed E-NCAM immunoreactivity undifferentiated, begin to express neurotransmitter-synthesizing (B). The E-NCAM-labeled cells (red, C and D) were double enzymes or mature phenotypic markers upon differentiation in labeled with ␤-III tubulin and MAP2 (green, CЈ and DЈ, respec- culture. tively). Note that all the E-NCAMϩ cells are also ␤-III tubulinϩ and MAP2ϩ. Differentiated cell cultures were also stained with antibodies to nestin (green) and GFAP (green, E and F, respec- ϩ cholinergic neurotransmitters after differentiation (Fig. tively). Staining showed that a subset of E-NCAM cells express nestin immunoreactivity (E) but these cells do not coexpress 8C–8E and PCR profile). E-NCAM-immunoreactive cells GFAP immunoreactivity (F). Thus ES-cell-derived E-NCAMϩ failed to differentiate into either oligodendrocytes or astro- cells resemble neuron-restricted precursors isolated from mouse cytes when grown under glial promoting conditions, indi- neural tube cultures. cating that these cells are limited in their differentiation

potential. Thus ES cells can be used as a source of neuronal- restricted precursor cells. ES-cell-derived A2B5؉/E-NCAM؊ cells are self-renewing glial-restricted stem cells. To test the differentiation potential of A2B5ϩ/E-NCAMϪ cells, ES cells were induced to differentiate and E-NCAM-immunoreactive cells were de- pleted by immunopanning. E-NCAM-immunonegative cells were reselected for A2B5 immunoreactivity. This double selection procedure was utilized as described above because in mice, A2B5 immunoreactivity has been detected on glial precursors, as well as on subsets of neurons (Eisen- barth et al., 1979). A2B5-immunopositive cells were grown in mass or clonal culture in medium supplemented with FGF and PDGF. These two factors were used because preliminary results had shown that both molecules are required to maintain A2B5-immunoreactive cells in a proliferative state. A2B5-immunoreactive cells expressed nestin but did not express neuronal markers such as ␤-III tubulin or glial markers such as GFAP or GalC. When mass cultures of A2B5ϩ/E-NCAMϪ cells were induced to differentiate by withdrawal of FGF, cells differentiated into different glial populations. Two kinds of astrocytes could be identified as early as 7 days after induction of differentiation: an A2B5Ϫ/GFAPϩ flat astrocyte which appeared similar to astrocytes previously characterized as Type 1 astrocytes and a A2B5ϩ/GFAPϩ astrocyte which appeared similar to type 2 astrocytes (Figs. 9E and 9F). In addition, if cells were allowed to differentiate for longer time periods (10 days), oligodendrocyte differentiation could be detected (Fig. 9G) as assessed by GalC immunocytochemistry. A2B5- immunoreactive cells failed to differentiate into neurons (data not shown) when grown under neuron promoting FIG. 9. ES-cell-derived A2B5-immunoreactive cells differentiate conditions, indicating that these cells are limited in their into oligodendrocytes and two types of astrocytes. ES-D3 cells were differentiation potential to glial lineages. The ability of plated on fibronectin/laminin-coated dishes in NEP basal medium ES-cell-derived A2B5-immunoreactive cells to differentiate and were allowed to differentiate for 5 days. The cells were then into astrocytes and oligodendrocytes confirms their resem- labeled with A2B5. Note the absence of A2B5 immunoreactivity in blance to E12.0 A2B5-derived GRP cells and suggests that ϩ undifferentiated cells (A) and the presence of A2B5 cells in culture like neuron-restricted precursors, glial-restricted precursors ϩ after differentiation (red, B). The A2B5 cells were immunopanned can be isolated directly from differentiating ES cell cultures. and a purified population isolated (C and CЈ). Note that all DAPI-stained nuclei, representing the total number of cells present Ј ϩ ϩ in the field (C ) are A2B5 (C). A2B5 cells were either fixed and DISCUSSION processed for immunocytochemistry (D) or allowed to differentiate (E–H). Undifferentiated cells expressed nestin immunoreactivity but did not show detectable levels of GFAP, ␤-III tubulin, neuro- We have shown that NEP cells, neuroblasts, and glio- filament, or GalC immunoreactivity and thus resembled mouse blasts are present in the developing mouse spinal cord. The neural tube-derived glial precursors. To promote differentiation characteristics of mouse CNS precursor cells are similar to ES-cell-derived A2B5ϩ cells were plated on polylysine/laminin- those described for rat precursors and differ from previously coated dishes and were grown in NEP basal medium supplemented described neurosphere cells. E-NCAM and A2B5 immuno- with 10% FCS (E), CNTF (10 ng/ml, F), or PDGF (10 ng/ml, G and reactivity can be used to isolate restricted neuronal and H) for 7–14 days and then labeled with antibodies to A2B5 (green, glial precursors both from the mouse spinal cord and from E and F), GFAP (red, E and F), and GalC (red, G) antibodies. The cells exposed to FCS differentiated almost entirely into A2B5Ϫ astrocytes with a fibroblast-looking morphology (E). 50% of the GRP cells exposed to CNTF differentiated into A2B5ϩ process- bearing astrocytes, while the others were A2B5Ϫ astrocytes (F). Exposure to PDGF resulted in mature looking oligodendrocytes analyzed by PCR. H, shows the PCR profile of differentiated ϩ which were immunoreactive for GalC (G). The A2B5ϩ ES cells were ES-cell-derived A2B5 cells. Note the strong expression of PLP/ grown on polylysine/laminin in NEP basal medium supplemented DM-20 (lane 2) and GFAP (lane 4, an astrocytic marker) and the with CNTF (10 ng/ml) for 15 days and were then harvested and absence of neurofilament (lane 3, a neuronal marker).

Copyright © 1999 by Academic Press. All rights of reproduction in any form reserved. 124 Mujtaba et al. mouse ES cell cultures. E-NCAM-immunoreactive cells Mouse E-NCAM cells appear similar to the rat-restricted present in populations of differentiated ES cells have lost neuronal precursors (Gensburger et al., 1987; Ray et al., 1993; the ability to generate astrocytes or oligodendrocytes, but Ray and Gage, 1994; Mayer-Proschel et al., 1997). Both mouse retain the ability to generate multiple neuronal phenotypes. and rat cells are E-NCAM immunoreactive, proliferate in ES-cell-derived, A2B5-immunoreactive GRP cells do not response to FGF-1 and –2, and can differentiate into multiple differentiate into neurons, but instead differentiate into classes of neurons. We have however noted some small oligodendrocytes and astrocytes. differences in proliferation and differentiation potential. Rat An important observation was that the properties of NRP cells appear more robust and survive over multiple mouse NEP cells resembled those of rat NEP cells rather passages, while mouse NRPs appear more fragile and we have than other classes of stem cells previously described (Kil- been unable to maintain them for more than 10–12 passages patrick and Bartlett, 1995; Gritti et al., 1996; Johe et al., (data not shown). Rat cells appear to be more heterogeneous as 1996; Weiss et al., 1996a,b). Mouse NEP cells grew in assessed by Ca2ϩ imaging (Kalyani et al., 1998), while mouse adherent culture, did not express EGF-R immunoreactivity, cells show less heterogeneity (present results) and appear to could not survive or proliferate in EGF, and absolutely mature more slowly. It remains to be determined whether required FGF for both proliferation and survival. PDGF, differences in culture requirements underlie these differences which has been shown to be an instructive molecule for or whether fundamental differences exist. It is interesting to cortical stem cells (Williams et al., 1997; Lachyankar et al., note that while mouse ES cells have been readily obtained, rat 1997), had no observable effect either alone or in combina- ES cells have not been developed successfully. tion with FGF. Our results are consistent with recent Our results show that A2B5 immunoreactivity is not reports showing that FGF2 injections in vivo will increase exclusive to glial precursors at this stage (E12.0). A small the progenitor pool and that the ventricular zone is reduced proportion of neuronal cells (Ͻ5%) express A2B5 immuno- in mice lacking FGF2 (Vaccarino et al., 1999) and with reactivity. We have not analyzed this subpopulation al- earlier reports showing expression of FGFRs in the ventric- though morphologically they do not appear to be distinct ular zone (see for example Orr-Urteger et al., 1991). Our from other neuronal cells. When the A2B5ϩ/E-NCAMϪ results also show that EGF is not a mitogen for NEP cells population of cells is excluded by prepanning for E-NCAM- and that EGF-R is not expressed or expressed at low levels immunoreactive cells, the remaining population appears to early in development. These results are consistent with be restricted to glial differentiation. This A2B5- reports of EGF-R expression in vivo (Adamson and Meek, immunoreactive glial precursor is distinct from the more 1984; Lazaar and Blum, 1992; Burrows et al., 1997) and the restricted A2B5-immunoreactive O-2A precursor (oligoden- functional effects seen in EGF-R knockouts. Transgenic drocyte type 2 astrocyte precursors, reviewed in Mayer- animals lacking EGF-R show no observable deficit in ven- Proschel et al., 1997) and appears similar to rat GRP cells. tricular zone stem cells but do show deficits in neuronal Like rat GRP cells, mouse GRPs generate oligodendrocytes migration (Miettinen et al., 1995; Threadgill et al., 1995). and two kinds of astrocytes (A2B5ϩ/GFAPϩ and A2B5Ϫ/ Thus, our data suggest that at least during early time GFAPϩ). Like rat GRPs and unlike O-2A cells, mouse GRPs periods of development, only FGF-dependent pluripotent respond to CNTF by differentiating into astrocytes rather stem cells are present in the developing spinal cord. Taken than oligodendrocytes (Rao et al., 1998; data not shown). together with evidence that at later stages an EGF- or Thus, mouse GRPs resemble their rat counterparts and are EGF/FGF-dependent cell is present (see for example Gritti present at corresponding times of development (see results et al., 1999), we suggest that at least two classes of stem and unpublished observations). cells exist. Recent reports suggest that the FGF-dependent To determine if NRP and GRP cells can be isolated stem cell may be the precursor of the EGF-dependent stem directly from ES cells differentiated in vitro, we tested a cell (Ciccolini and Svendsen, 1998; Tropepe et al., 1999). variety of differentiation protocols (data not shown). We These reports, together with our observation that the early have described an experimental paradigm in which totipo- FGF-dependent stem cell does not express EGF-R immuno- tent embryonic stem cells generate restricted neuronal and reactivity, identify EGF-R as a marker that can be used to glial precursors and have shown that E-NCAM and A2B5 select between these two populations of stem cells. immunoreactivity is sufficient to purify ES-cell-derived Our results also show that E-NCAM and A2B5 immunore- neuroblasts and glioblasts. ES-cell-derived NRPs and GRPs activity can be used to isolate lineage restricted precursor are morphologically and phenotypically identical to the cells. E-NCAM appears to be expressed exclusively by neuro- neuroblasts and glioblasts that we have described as being nal precursors at this stage (E12.0) in development. Neither present in E12.0 mouse embryos. Like E12.0 precursor cells, NEP cells nor glial precursors express detectable levels of the ES-cell-derived NRPs are dividing cells that are limited in E-NCAM epitope recognized by the 5A5 antibody. E-NCAM- their differentiation into neurons. Multiple kinds of neu- immunoreactive cells appear capable of generating multiple rons are generated. In particular, glutamatergic, GABAergic, kinds of neurons as demonstrated by immunocytochemistry, and cholinergic neurons could be identified by PCR and PCR, and Ca2ϩ imaging. The Ca2ϩ imaging experiments immunocytochemistry (see results). Likewise, ES-cell- clearly indicate that cells are functionally heterogeneous and derived A2B5ϩ glial precursors resembled E12.0 GRPs can respond to multiple classes of neurotransmitter signals. rather than O-2A or other glial precursors. Our data suggest

Copyright © 1999 by Academic Press. All rights of reproduction in any form reserved. Neuroblasts and Glioblasts from Mouse Embryos 125 that ES cells can serve as a source of more downstream, Our experimental design did not allow us to determine if more restricted precursor cells. Since ES cells can be readily ES cells first generate neuroepithelial cells that subse- maintained in vitro, these cells serve as a ready source of quently generate intermediate lineage restricted precursors differentiated cells, and A2B5 and E-NCAM immunoreac- or that ES cells directly generate neuroblasts and glioblasts. tivity can be used to isolate purified fractions of restricted These experiments are difficult to perform as no cell surface precursor cells. markers that will unambiguously distinguish NEP cells E-NCAM expression has been described on a variety of from ES and other cells exist. Thus, cell sorting and clonal cell types including mesodermal derivatives (Husmann et analysis remain difficult to perform. Nevertheless recent al., 1989; Brunet et al., 1989; Gerety and Watanbe, 1997; results (Okabe et al., 1996; Li et al., 1998) suggest that Dubois et al., 1994), epidermal derivatives (see for example pluripotent nestin-immunoreactive cells may also be gen- Muller-Rover et al., 1998), and astrocytes and oligodendro- erated from ES cell cultures under virtually identical cul- cytes (Seki and Arai, 1993; Bhat et al., 1988; Minana et al., ture conditions to those described in our present studies. 1998). It was therefore somewhat surprising that a large Thus it is reasonable to assume that at least some neuro- proportion of the E-NCAM-immunoreactive cells present blasts or glioblasts arose from ES cells that went through a in differentiating ES cell cultures are neuroblasts. A pos- NEP-like stem cell phase. Direct demonstration of this sible explanation may lie in the fact that ES cell differen- transition would provide compelling evidence that in vitro tiation was biased to neural versus epidermal or mesoder- differentiation recapitulates in vivo development and allow mal differentiation by aggregation and retinoic acid. Within a detailed analysis of neural differentiation that occurs over the developing neural tube E-NCAM immunoreactivity 13 days of embryonic development in a simple in vitro appears limited to early developing neurons and their pre- model. cursors at early stages of development (Mayer-Proschel et In summary our results provide a culture model to study al., 1997). Thus, the large proportion of E-NCAM- the process of differentiation from totipotent ES cells to immunoreactive cells that are neuroblasts may reflect not multipotent NEP stem cells to more restricted neuronal just the specificity of the antibody, but also the specificity and glial precursors. Further, our results show that of the culture conditions that we utilized to generate E-NCAM and A2B5 antibodies can be used for selectively neurons. Consistent with this possibility is our observation sorting NRPs and GRPs and amplifying them through serial that nonneuronal E-NCAM-immunoreactive cells are passages for gene expression studies and therapeutic use. present when ES cells are allowed to differentiate without aggregation and the addition of retinoic acid (data not ACKNOWLEDGMENTS shown). Under conditions in which neuronal E-NCAM cells were present, our cultures also contained other neural This work was supported by an NIH FIRST award, a MDA grant, ϩ precursors such as A2B5-immunoreactive cells. A2B5 cells and MOD award to M.S.R. A part of this research was supported by appeared in a relatively higher proportion than E-NCAM- funds from Accorda Therapeutics. A.K. was supported by a gradu- immunoreactive cells (10 versus 5%). Thus, while signifi- ate student fellowship. M.T.L. was supported by NIH NIDCD cant numbers of neurons and glial cells can be obtained, it DC02994-03. We thank Dr. Mayer-Proschel for her advice and input and Sehba Kudiya for technical support. We thank Dr. Ann must be emphasized that the actual efficiency of the pro- Greig, Jeff Lee, and all members of our laboratories for constant cess is rather low. Despite this low efficiency, however, we stimulating discussions. M.S.R. acknowledges the support of Dr. S. would argue that ES cells represent the most likely source Rao through all phases of this project. of large numbers of cells for transplant tissue. ES cells can be grown in culture indefinitely and indeed have been grown in this fashion for years. Thus, it is possible to obtain REFERENCES hundreds of millions of cells easily. Other laboratories have shown that other classes of intermediate precursors can be Adamson, E. D., and Meek, J. (1984). The ontogeny of epidermal growth factor receptors during mouse development. Dev. Biol.. isolated from differentiating ES cells. For example, hema- 103, 62–70. topoietic stem cells that are restricted to differentiation Bhat, S., and Silberberg, D. H. (1988). Developmental expression of into bone marrow stem cell derivatives have been isolated neural cell adhesion molecules of oligodendrocytes in vivo and in from differentiating ES cells (reviewed in Weiss, 1997; see culture. J. Neurochem. 50, 1830–1838. also Suwabe et al., 1998). Similarly, a mesodermal stem cell Bottenstein, J. E., and Sato, G. (1979). Growth of a rat neuroblas- that can differentiate into cartilage, bone, and connective toma cell line in serum free supplemented medium. Proc. Natl. tissue cells has been isolated from ES cell cultures (Katoka Acad. Sci. USA 76, 514–517. et al., 1997). Our demonstration that two additional classes Brunet, J. F., Hirsch, M. R., Naquet, P., Uberla, K., Diamantstein, of intermediate precursors (NRPs and GRPs) that can be T., Lipinski, M., and Goridis, C. (1989). Developmentally regulated expression of the neural cell adhesion molecule (NCAM) by distinguished from the previously characterized precursors mouse thymocytes. Eur. J. Immunol. 5, 837–841. extends the range of possible precursors, harvested from ES Burrows, R. C., Wancio, D., Levitt, P., and Lillien, L. (1997). cells, and suggests that ES cell cultures are a useful source Response diversity and the timing of progenitor cell maturation of dividing intermediate precursors that normally develop are regulated by developmental changes in EGFR expression in at different stages in vivo. the cortex. Neuron 19(2), 251–267.

Ciccolini, F., and Svendsen, C. N. (1998). Fibroblast growth factor Kalyani, A., Hobson, K., and Rao, M. S. (1997). Neuroepithelial 2 (FGF-2) promotes acquisition of epidermal growth factor (EGF) stem cells from the embryonic spinal cord: Isolation, character- responsiveness in mouse striatal precursor cells: Identification of ization and clonal analysis. Dev. Biol. 186, 202–223. neural precursors responding to both EGF and FGF-2. J. Neurosci. Kalyani, A., Piper, D., Mujtaba, T., Lucero, M. T., and Rao, M. S. 18, 7869–7880. (1998a). Spinal cord neuronal precursors generate multiple neu- Collarini, E. J., Pringle, N., Mudhar, H., Stevens, G., Kuhn, R., ronal phenotypes in culture. J. Neurosci. 18, 7856–7868. Monuki, E. S., Lemke, G., and Richardson, W. D. (1991). Growth Kalyani, A., Mujtaba, T., and Rao, M. S. (1999). Expression of EGF factors and transcription factors in oligodendrocyte develop- receptor and FGF receptor isoforms during neuroepithelial stem ment. J. Cell Sci. Suppl. 15, 117–123. cell differentiation. J. Neurobiol. 38, 207–224. Dahlstrand, J., Lardelli, M., and Lendahl, U. (1995). Nestin mRNA Kalyani, A., and Rao, M. S. (1999). Lineage determination in the expression correlates with the central nervous system progenitor developing spinal cord. Cell Biol. Biochem. 76, 1–17. state in many, but not all regions of the developing nervous Kilpatrick, T. J., and Bartlett, B. F. (1993). Cloning and growth of system. Dev. Brain Res. 84, 109–129. multipotential neural precursors: Requirements for proliferation DeCamilli, P., Miller, P. E., Navone, F., Theurkauf, W. E., and and differentiation. Neuron 10, 255–265. Vallee, R. B. (1984). Distribution of microtubule-associated pro- Kilpatrick, T. J., and Bartlett, B. F. (1995). Cloned multipotential tein 2 in the nervous system of the rat studied by immunof- precursors from the mouse cerebrum require FGF-2 whereas glial lourescence. Neuroscience 11, 817–846. restricted precursors are stimulated by either FGF-2 or EGF. Dubois, C., Figarella-Branger, D., Pastoret, C., Rampini, C., Kar- J. Neurosci. 15, 3653–3661. pati, G., and Rougon, G. (1994). Expression of NCAM and its Kataoka, H., Takakura, N., Nishikawa, S., Tsuchida, K., Kodama, polysialylated isoforms during mdx mouse muscle regeneration H., Kunisada, T., Risau, W., Kita, T., and Nishikawa, S. I. (1997). and in vitro myogenesis. Neuromuscul. Disord. 4, 171–182. Expressions of PDGF receptor alpha, c-Kit and Flk1 genes clus- Eisenbarth, G. S., Walsh, F. S., and Nirenberg, M. (1979). Monoclo- tering in mouse chromosome 5 define distinct subsets of nascent Dev. Growth Differ. nal antibody to plasma membrane antigen of neurons. Proc. Natl. mesodermal cells. 39, 729–740. Lachyankar, M. B., Condon, P. J., Quesenberry, P. J., Litofsky, N. S., Acad. Sci.USA 76, 4913–4917. Recht, L. D., and Ross, A. H. (1997). Embryonic precursor cells Fraichard, A., Chassande, O., Bilbaut, G., Dehay, C., Savatier, P., that express Trk receptors: induction of different cell fates by and Samarut, J. (1995). In vitro differentiation of embryonic stem NGF, BDNF, NT-3, and CNTF. Exp. Neurol. 144, 350–360. cells into glial cells and functional neurons. J. Cell Sci. 108, Lazaar, L. M., and Blum, M. (1992). Regional distribution and 3181–3188. developmental expression of epidermal growth factor and trans- Gajovic, S., St-Onge, L., Yokota, Y., and Gruss, P. (1997). Retinoic forming growth factor-a mRNA in mouse brain by quantitative acid mediates Pax6 expression during in vitro differentiation of nuclease protection assay. J. Neurosci. 12, 1688–1697. embryonic stem cells. Differentiation 62(4), 187–192. Leber, S. M., Breedlove, S. M., and Sanes, J. R. (1990). Lineage, Gensburger, C., Labourdette, G., and Sensenbrenner, M. (1987). arrangement, and death of clonally related motoneurons in chick Brain basic fibroblast growth factor stimulates the proliferation spinal cord. J. Neurosci. 10, 1688–1697. of rat neuronal precursor cells in vitro. FEBS Lett. 217, 1–5. Leber, S. M., and Sanes, J. R. (1995). Migratory paths of neurons and glia Gerety, M., and Watanabe, M. (1997). Polysialylated NCAM ex- in the embryonic chick spinal cord. J. Neurosci. 15, 1236–1248. pression on endocardial cells of the chick primary atrial septum. Lendahl, U., Zimmerman, L. B., and McKay, R. D. (1990). CNS Anat. Rec. 7, 71–84. stem cells express a new class of intermediate filament protein. Gritti, A., Parati, E. A., Cova, L., Frolichsthal, P., Galli, R., Wanke, Cell 60, 585–595. E., Faravelli, L., Morassutti, D. J., Roisen, F., Nickel, D. D., and Li, M., Pevny, L., Lovell-Badge, R., and Smith, A. (1998). Genera- Vescovi, A. (1996). Multipotential stem cells from the adult tion of purified neural precursors from embryonic stem cells by mouse brain proliferate and self-renew in response to basic lineage selection. Curr. Biol. 8, 971–974. fibroblast growth factor. J. Neurosci. 16, 1091–1100. Li et al. (1998) Soc. Neurosci. abst 605.11, pg. 1529. Gritti, A., Frolichsthal-Schoeller, P., Galli, R., Parati, E. A., Cova, Luskin, M. B., Pearlman, A. L., and Sanes, J. R. (1988). Cell lineage L., Pagano, S. F., Bjornson, C. R., and Vescovi, A. L. (1999). in the cerebral cortex of the mouse studied in vivo and in vitro Epidermal and fibroblast growth factors behave as mitogenic with a recombinant retrovirus. Neuron 1, 635–647. regulators for a single multipotent stem cell-like population from Luskin, M. B., Parnavelas, J. G., and Barfield, J. A. (1993). Neurons, the subventricular region of the adult mouse forebrain. J. Neu- astrocytes, and oligodendrocytes of the rat cerebral cortex origi- rosci. 19(9), 3287–3297. nate from separate progenitor cells: an ultrastructural analysis of Grynkiewics, D., Neufeld, G., and Schweigerer, L. (1985). A new clonally related cells. J. Neurosci. 13, 1730–1750. generation of calcium indicators with greatly improved floures- Mayer-Proschel, M., Kalyani, A. J., Mujtaba, T., and Rao, M. S. cence properties. J. Biol. Chem. 260, 3440–3450. (1997). Isolation of lineage- restricted precursors from multipo- Hulspas, R., Tiarks, C., Reilly, J., Hsieh, C. C., Recht, L., and tent neuroepithelial stem cells. Neuron 19, 773–785. Quesenberry, P. J. (1997). In vitro cell density-dependent clonal McKay, R. (1997). Stem cells in the central nervous system. Science growth of EGF-responsive murine neural progenitor cells under 276, 66–71. serum-free conditions. Exp. Neurol. 148, 147–156. Miettinen, P. J., Berger, J. E., Meneses, J., Phung, Y., Pedersen, R. A., Husmann, M., Pietsch, T., Fleischer, B., Weisgerber, C., and Bitter- Werb, Z., and Derynck, R. (1995). Epithelial immaturity and Suermann, D. (1989). Embryonic neural cell adhesion molecules multiorgan failure in mice lacking epidermal growth factor on human natural killer cells. Eur. J. Immunol. 19, 1761–1763. receptor. Nature 376, 337–341. Johe, K. K., Hazel, T. G., Muller, T., Dugich-Djordjevic, M. M., and Minana, R., Sancho-Tello, M., Climent, E., Segui, J. M., Renau- McKay, R. D. (1996). Single factors direct the differentiation of Piqueras, J., and Guerri, C. (1998). Intracellular location, tempo- stem cells from the fetal and adult central nervous system. Genes ral expression, and polysialylation of neural cell adhesion mole- Dev. 10, 3129–3140. cule in astrocytes in primary culture. Glia. 24, 415–427.

Muller-Rover, S., Peters, E. J., Botchkarev, V. A., Panteleyev, A., growth factor-responsive precursor cells following grafting into and Paus, R. (1998). Distinct patterns of NCAM expression are the lesioned adult central nervous system. Exp. Neurol. 137, associated with defined stages of murine hair follicle morpho- 376–388. genesis and regression. J. Histochem. Cytochem. 46, 1401–1410. Suwabe, N., Takahashi, S., Nakano, T., and Yamamoto, M. (1998). Mujtaba, T., Mayer-Proschel, M., and Rao, M. S. (1998). A common GATA-1 regulates growth and differentiation of definitive ery- neural progenitor for the CNS and PNS. Dev. Biol. 200, 1–15. throid lineage cells during in vitro ES cell differentiation. Blood Nye, J. S., Kopan, R., and Axel, R. (1994). An activated Notch 92, 4108–4118. suppresses neurogenesis and myogenesis but not gliogenesis in Temple, S., and Davis, A. (1994). Isolated rat cortical progenitor mammalian cells. Development 120, 2421–2430. cells are maintained in division in vitro by membrane associated Okabe, S., Nilsson, K. F., Spiro, A. C., Segal, M., and McKay, factors. Development 120, 999–1008. R. D. G. (1996). Development of neuronal precursor cells and Thomson, J. A., and Marshall, V. S. (1998). Primate embryonic stem functional postmitotic neurons from embryonic stem cells in cells. Curr. Top. Dev. Biol. 38, 133–165. vitro. Mech. Dev. 59, 89–102. Thomson, J. A., Marshall, V. S., and Trojanowski, J. Q. (1998). Orr-Urtreger, A., Givol, D., Yayon, A., Yarden, Y., and Lonai, P. Neural differentiation of rhesus embryonic stem cells. APMIS (1991). Developmental expression of two murine fibroblast growth Jan. 106, 149–156; discussion 156–157. factor receptors, flg and bek. Development 113(4), 1419–1434. Threadgill, D. W., Dlugosz, A. A., Hansen, L. A., Tennenbaum, T., Palmer, T. D., Ray, J., and Gage, F. H. (1995). FGF-2-responsive Lichti, U., Yee, D., LaMantia, C., Mourton, T., Herrup, K., Harris, neuronal progenitors reside in proliferative andquiescent regions R. C., Barnard, J. A., Yuspa, S. H., Coffey, R. J., and Magnuson, T. of the adult rodent brain. Mol. Cell. Neurosci. 6, 474–486. (1995). Targeted disruption of mouse EGF receptor: Effect of Price, J., and Thurlow, L. (1988). Cell lineage in the rat cerebral genetic background on mutant phenotype. Science 269, 230–234. cortex: a study using retroviral mediated gene transfer. Develop- Tropepe, V., Sibilia, M., Ciruna, B. G., Rossant, J., Wagner, E. F., ment 104, 473–482. and van der Kooy, D. (1999). Distinct neural stem cells proliferate Price, J., Williams, B., and Grove, E. (1992). The generation of in response to EGF and FGF in the developing mouse telenceph- cellular diversity in the cerebral cortex. Brain Pathol. 2, 23–29. alon. Dev. Biol. 208(1), 166–188. Rao, M. S., and Anderson, D. J. (1997). Immortalization and Vaccarino, F. M., Schwartz, M. L., Raballo, R., Nilsen, J, Rhee, J., controlled in vitro differentiation of murine multipotent neural Zhou, M., Doetschman, T., Coffin, J. D., Wyland, J. J., and Hung, crest stem cells. J. Neurobiol. 32, 722–746. Y. T. (1999). Changes in cerebral cortex size are governed by Rao, M. S., and Mayer-Proschel, M. (1997). Glial-restricted precur- fibroblast growth factor during embryogenesis. Nature Neurosci. sors are derived from multipotent neuroepithelial stem cells. 2(3), 246–253. Dev. Biol. 188, 48–63. Weiss, M. J. (1997). Embryonic stem cells and hematopoietic stem Rao, M. S., Noble, M., and Mayer-Proschel, M. (1998). A tripoten- cell biology. Hematol. Oncol. Clin. North Am. 11, 1185–1198. tial glial precursor is present in the developing spinal cord. Proc. Weiss, S., Dunne, C., Hewson, J., Wohl, C., Wheatley, M., Peterson, Natl. Acad. Sci. USA 95, 3996–4001. A. C., and Reynolds, B. A. (1996a). Multipotent CNS stem cells Ray, J., Peterson, D. A., Schinstine, M., and Gage, F. H. (1993). are present in the adult mammalian spinal cord and ventricular Proliferation, differentiation, and long-term culture of primary neuroaxis. J. Neurosci. 16, 7599–7609. hippocampal neurons. Proc. Natl. Acad. Sci. USA 90, 3602–3606. Weiss, S., Reynolds, B. A., Vescovi, A. L., Morshead, C., Craig, Ray, J., and Gage, F. (1994). Spinal cord neuroblasts proliferate in response C. G., and van der Kooy, D. (1996b). Is there a neural stem cell in to basic fibroblast growth factor. J. Neurosci. 14, 3548–3564. the mammalian forebrain?. Trends Neurosci. 19, 387–393. Reynolds, B. A., Tetzlaff, W., and Weiss, S. (1992). A multipotent Williams, B. P., Park, J. K., Alberta, J. A., Muhlebach, S. G., Hwang, EGF-responsive striatal embryonic progenitor cell produces neu- G. Y., Roberts,T. M., and Stiles, C. D. (1997). A PDGF-regulated rons and astrocytes. J. Neurosci. 12, 4565–4574. immediate early gene response initiates neuronal differentiation Reynolds, B. A., and Weiss, S. (1996). Clonal and population in ventricular zone progenitor cells. Neuron 18, 553–562. analysis demonstrate that an EGF-responsive mammalian em- Williams, B. P. (1995). Precursor cell types in the germinal zone of bryonic CNS precursor is a stem cell. Dev. Biol. 175, 1–13. the cerebal cortex. Bioessays 17, 391–393. Sanes, J. R. (1989). Analysing cell lineage with a recombinant Williams, B. P., and Price, J. (1995). Evidence for multiple precursor retrovirus. TINS 12, 21–28. cell types in the embryonic rat cerebral cortex. Neuron 14, Santa-Ollala, J., and Covarrubias, L. (1995). Epidermal growth 1181–1188. factor (EGF), Transforming growth factor a (TGF-a) and basic Wysocki, L. J., and Sato,V. L. (1978) “Panning” for lymphocytes: A fibroblast growth factor (bFGF) differentially influence neural method for cell selection. Proc. Natl. Acad. Sci. USA 75, 2844–2848. precursor cells of mouse embryonic mesencephalon. J. Neurosci. Young, G. M., and Levison, S. W. (1996). Persistence of multipo- Res. 42, 172–183. tential progenitors in the juvenile rat subventricular zone. Dev. Seki, T., and Arai, Y. (1993). Distribution and possible roles of the Neurosci. 18, 255–265. highly polysialylated neural cell adhesion molecule (NCAM-H) Zimmerman, L., Parr, B., Lendahl, U., Cunningham, M., McKay, in the developing and adult central nervous system. Neurosci. R., Gavin, B., Mann, J., Vassileva, G., and McMahon, A. (1994). Res. 17, 265–290. Independent regulatory elements in the nestin gene direct trans- Stemple, D. L., and Anderson, D. J. (1992). Isolation of a stem cell gene expression to neural stem cells or muscle precursors. for neurons and glia from the mamalian neural crest. Cell 71, Neuron 12, 11–24. [published erratum appears in Neuron, 1994, 973–985. 12(6), following 1388]. Svendsen, C. N., and Rosser, A. E. (1995). Neurones from stem cells? Trends Neurosci. 18, 465–467. Received for publication May 19, 1999 Svendsen, C. N., Clarke, D. J., Rosser, A. E., and Dunnett, S. B. Revised July 21, 1999 (1996). Survival and differentiation of rat and human epidermal Accepted July 21, 1999