B R A I N R E S E A R C H 1 0 7 1 ( 2 0 0 6 ) 2 4 – 3 3

ava i l ab l e a t www.sc i enced i rec t . com

www.e l sev i e r. com/ loca te /b ra in res

Research Report

De novo and salvage pathways of DNA synthesis in primarycultured neurall stem cells

Kenichi Satoa,b, Junko Kannoa, Teiji Tominagab, Yoichi Matsubaraa,c, Shigeo Kurea,c,⁎aDepartment of Medical Genetics, Tohoku University School of Medicine, 1-1 Seiryo-machi, Aobaku, Sendai 980-8574, JapanbDepartment of Neurosurgery, Tohoku University School of Medicine, Sendai, JapancTohoku University 21st Century COE Program “Comprehensive Research and Education Center for Planning of Drug Development andClinical Evaluation,” Sendai, Japan

A R T I C L E I N F O

⁎ Corresponding author. Department of Medic8574, Japan. Fax: +81 22 717 8142.

E-mail address: skure@mail.tains.tohoku.

0006-8993/$ – see front matter © 2005 Elsevidoi:10.1016/j.brainres.2005.11.039

A B S T R A C T

Article history:Accepted 6 November 2005Available online 10 January 2006

We studied the de novo and salvage pathways of DNA synthesis in sphere-forming neuralstem cells obtained from mouse embryos by a neurosphere method. The former pathwayneeds folic acid (FA) for nucleotide biosynthesis, while the latter requiresdeoxyribonucleosides (dNS). We examined the proliferative activity of sphere-formingcells in E14.5 embryos by counting the number of spheres formed in media that lacked FAand/or dNS. Proliferation failure and apoptosis occurred in a deficient medium lacking ofboth FA and dNS. Spheres formed in the deficient medium supplemented with dNS, withoutFA, did not produce neuron, but rather only seem to generate astrocytes andoligodendrocytes when plated under differentiation condition in culture. On the otherhand, a subpopulation of cultured cells formed spheres in the deficient mediumsupplemented with FA alone in an appropriate concentration, and did possess the self-renewing and multipotential characteristics of neural stem cells. Spheres formed in themedia containing low dose Azathioprine and methotrexate, inhibitors of de novo DNAsynthesis, were selectively prevented from producing neurons even in the presence of FA.These results suggested that activating de novo DNA synthesis was needed for neural stemcells to proliferate with multipotentiality.

© 2005 Elsevier B.V. All rights reserved.

Keywords:Neural stem cellNeurosphereProliferationDNA synthesis pathwayFolic acid

al Genetics, Tohoku University School of Medicine, 1-1 Seiryo-machi, Aobaku, Sendai 980-

ac.jp (S. Kure).

er B.V. All rights reserved.

25B R A I N R E S E A R C H 1 0 7 1 ( 2 0 0 6 ) 2 4 – 3 3

Abbreviations:FA, folic aciddNS, deoxyribonucleosidesdNTPs, deoxynucleosidetriphosphatesdAdo, 2′-deoxyadenosinedCyd, 2′-deoxycytidinedGuo, 2′-deoxyguanosineThy, thymidineAZP, azathioprineMTX, methotrexateNTD, neural tube defectDIV, days in vitroGFAP, glial fibrillary acidic protein

Fig. 1 – De novo and salvage biosynthesisand salvage pathway (right panel). AbbrevdCyd, 2′-deoxycytidine; dGuo, 2′-deoxyguN5,N10-methylenetetrahydrofolate.

1. Introduction

Normal CNS development involves the sequential differenti-ation of multipotent stem cells. Alteration in the number ofstem cells or their proliferative capability has major effectson the appropriate development of the nervous system(Sommer and Rao, 2002). One of the essential steps for theproliferation of stem cells is DNA synthesis. DNA precursors,i.e., nucleotides, are synthesized through two metabolicpathways referred to as de novo pathway and salvagepathway (Fig. 1). In the de novo pathway, deoxynucleosidetriphosphates (dNTPs) are generated in multiple steps from5-phosphoribosyl-1-pyrophosphate, glutamine, glycine, as-partate, and folic acid (FA). Since animal cells cannotsynthesize FA, FA supplementation into their culture mediais required for activating de novo DNA synthesis (Murray et

of DNA. DNA is synthesiations: IMP, inosinic acanosine; Thy, thymidi

al., 1999). In the salvage pathway, dNTPs are produced insingle steps from exogenous and endogenous deoxyribonu-cleosides (dNS). The de novo pathway is predominant innormal differentiated tissues, whereas the salvage pathwayis activated in rapidly proliferating tissues, including fetaland regenerative tissues and cancer cells (Hatse et al., 1999).Previous studies functionally evaluated the DNA synthesis inerythroblasts using culture media deficient of FA and/or dNS,and that indicated that erythroblasts underwent apoptosisduring S phase in the lack of both FA and dNS (James et al.,1994; Koury and Horne, 1994; Koury et al., 2000). To date, theDNA synthesis pathway employed in neural stem cellsremains uninvestigated.

Reynolds and Weiss (1996) reported that neural stem cellscould be established from rodent brains by the neurosphereculture method. Using the neurosphere method, single

ized through two pathways: the de novo pathway (left panel)id; UMP, uridine monophosphate; dAdo, 2′-deoxyadenosine;ne; H2-FA, dihydrofolate; H4-FA, tetrahydrofolate; CH2H4-FA,

26 B R A I N R E S E A R C H 1 0 7 1 ( 2 0 0 6 ) 2 4 – 3 3

neural stem cells can be selectively expanded in a chemi-cally defined medium containing epidermal growth factor(EGF), fibroblast growth factor 2 (FGF-2), or both, giving riseto floating spheroid cell aggregates called neurospheres.Since D-MEM/F-12, the commonly used medium for theneurosphere culture, contains both FA and two kinds of dNS,namely, thymidine (Thy) and hypoxanthine, the pathway ofDNA synthesis that operates in the proliferation of neuralstem cells remains unknown.

In this report, we performed the neurosphere method byusing a custom-prepared D-MEM/F-12 medium that lackedFA and/or dNS for the functional evaluation of DNAsynthesis pathways in embryonic sphere-forming neuralstem cells. We examined the efficiency of the sphereformation and characterized the multipotency of the neuralstem cells by in vitro differentiation analysis. Specificinhibitors of de novo DNA synthesis were used for thefurther characterization of sphere-forming cells. Theseapproaches enabled us, for the first time, to perform thefunctional evaluation of DNA synthesis in neural stem cells.

Fig. 2 – (A and B) Dose–response curves of sphere formationcultured for 7 DIV in the deficient medium supplementedwith FA (A) or dNS (B). During this developmental period,spheres can be formed in the presence of FA or dNS in anappropriate concentration. (C) The number of spherescultured for 7 DIV in the deficient medium supplementedwith 16 different combinations of deoxyribonucleosides,each at a concentration of 10 μM. Graphs show mean ± SEMnumber of spheres per 1 × 104 cells prepared from the E14.5mouse striata. Abbreviations: A, 2′-deoxyadenosine; C,2′-deoxycytidine; G, 2′-deoxyguanosine; T, thymidine.

2. Results

2.1. Sphere formation derived from the E14.5 mousestriata in FA+/dNS− or FA−/dNS+ media

To determine the DNA synthesis pathway utilized byembryonic neural stem cells for proliferation, we culturedneurospheres derived from the E14.5 mouse striata in D-MEM/F-12 medium lacking the precursors of DNA synthesis,defined as a deficient medium. We supplemented thedeficient medium with varying amounts of FA, as a precursorfor the de novo pathway, or dNS (equimolar addition of dAdo,dGuo, dCyd, and Thy), as precursors for the salvage pathway.In the deficient medium, designated as FA−/dNS− medium,no sphere was formed, and the condensation of cultured cellsand the fragmentation of nuclei, which are characteristic ofapoptosis, were observed. Terminal deoxynucleotidyl trans-ferase-mediated dUTP-biotin nick end labeling (TUNEL)showed increased TUNEL-positive cells (mean ± SEM,56.7% ± 1.5%; n = 4) in the deficient medium when comparedwith those in the deficient medium containing 10 μM FA and50 μM dNS (29.2% ± 1.1%, n = 5, P b 0.01) (data not shown).These results suggested that cultured cells did not proliferateand underwent apoptosis in the medium without theprecursors of DNA synthesis. In the medium containingonly FA, maximum number of spheres was formed with 10μM FA (Fig. 2A). In the absence of FA, peak sphere formationefficiency with each dNS was observed at a concentration of50 μM (Fig. 2B). We defined the culture condition of 10 μM FAwithout dNS as FA+/dNS−. Similarly, the condition in whichthe culture was supplemented with 50 μM of each dNSwithout FA was designated as FA−/dNS+.

2.2. Sphere-forming activity in the presence of variouscombinations of dNS

Eukaryotes generally have converting enzymes among purineand pyrimidine deoxyribonucleosides (Murray et al., 1999). To

determine the precursors of the salvage pathway that arerequired for sphere formation, we tested 16 different combi-nations of 10 μM dNS. As shown in Fig. 2C, the combination ofdAdo with Thy yielded maximum sphere-forming efficiencyand additional supplementation with other dNSs had littleeffect on the efficiency.

2.3. Developmental-stage specific changes ofsphere-forming activities under the FA+/dNS− and FA−/dNS+conditions

To examine the developmental changes of sphere-formingactivities under the FA+/dNS− and FA−/dNS+ conditions, weperformed the neurosphere cultures derived from the E12.5and E16.5 mice striata in the deficient medium supplemen-ted with varying amounts of FA or dNS (Fig. 3). In thecultures derived from the E16.5 striata, nearly the samenumber of spheres was formed with appropriate concentra-tions of FA and dNS. However, in the culture derived from

Fig. 4 – FA−/dNS+ spheres contain astrocytes and oligoden-drocytes, but not neurons. (A and B) Phase-contrast images oflive floating spheres under the FA+/dNS condition (A) andFA−/dNS+ condition (B). (C–J) Immunofluorescence ofchemical markers for neural progenitors (nestin, red; C andD), neurons (tubulin β III, red; E and F), astrocytes (GFAP,green; G and H), oligodendrocytes (O4, green; I and J), andnucleus (DAPI, blue; C–J) derived from the FA+/dNS− spheres(C, E, G, and I) and FA−/dNS+ spheres (D, F, H, and J). Whitearrowheads indicate tubulin β III-positive neurons. Inset inpanel E is high magnification of a white box. Scale bars: A–J,200 μm; Inset, 20 μm. Abbreviations: B.F., bright field; GFAP,glial fibrillary acidic protein.

Fig. 3 – Developmental change of dose–response curve ofsphere formation, cultured for 7 DIV in the deficient mediumsupplemented with FA (A and C) or dNS (B and D). Graphsshow mean ± SEM number of spheres per 1 × 104 cellsprepared from E12.5 (A and B) and E16.5 (C and D) micestriata.

27B R A I N R E S E A R C H 1 0 7 1 ( 2 0 0 6 ) 2 4 – 3 3

E12.5, only a few spheres were formed under the FA−/dNS+condition.

2.4. Morphological and immunochemicalcharacterizations of the formed spheres in theFA+/dNS− or FA−/dNS+ media

Spheres formed under the FA−/dNS+ condition morphologi-cally differed from those formed under the FA+/dNS− condi-tion (Figs. 4A, B). After 7 DIV, the FA+/dNS− spheres were large,smooth, and spherical in shape, while the FA−/dNS+ sphereswere small and irregular in shape. The FA+/dNS− sphereswere77.0 ± 3.8 μm (n = 66) in diameter, larger than the FA−/dNS+spheres (49.8 ± 2.5 μm, n = 78, P b 0.01). It is unlikely that bothspheres were derived from ciliated cells by their morpholog-ical characteristics (Chiasson et al., 1999; Laywell et al., 2000).Immunochemical characterization showed that both sphereswere immunopositive for Nestin (a neural progenitor marker),GFAP (an astrocyte marker), and O4 (an oligodendrocytemarker) (Figs. 4C, D, G–J). However, tubulin β III (a neuron

marker) positive cells were observed only in the FA+/dNS−spheres (Fig. 4E), and not in the FA−/dNS+ spheres (Fig. 4F),suggesting that cells in the FA−/dNS+ spheres barely differ-entiated into neurons.

Fig. 5 – Tri-colored immunofluorescence of markers for neurons (tubulin β III, red), astrocytes (GFAP, blue), andoligodendrocytes (O4, green) shows that the FA+/dNS− clonal spheres gave rise to different cells expressing markers of all threeneural cell types, as depicted in a triple-labeled survey (A), and the details of individual cells of each type (C–F). The FA−/dNS+spheres gave rise to astrocytes and oligodendrocytes, but not neurons, as shown in a triple-labeled survey (B), which alsodepicted the details of individual cells (G–J). Scale bars: A, B, 200 μm; C–J, 20 μm. (K) Experimental protocol. The 7-DIV primaryculture in the FA−/dNS+ (a) and FA+/dNS− (b) conditionwas placed under the differentiation condition for 4 DIV (total 11 DIV). (c),The 7-DIV primary culture in the FA+/dNS−mediumwas placed under the FA+/dNS− condition for 7 DIV (total 14 DIV) and, then,the cultured cells were placed under the differentiation condition for 4 DIV (total 18 DIV). (d) Likewise, the FA−/dNS+ primaryculturewas placed under the FA−/dNS+ condition and, then, under the differentiation condition. (L) Graph showingmean ± SEMpercentage of tubulin β III-immunopositive neurons of the total number of cells per field derived from each protocol.**Significant difference between the two groups (P b 0.01).

28 B R A I N R E S E A R C H 1 0 7 1 ( 2 0 0 6 ) 2 4 – 3 3

29B R A I N R E S E A R C H 1 0 7 1 ( 2 0 0 6 ) 2 4 – 3 3

2.5. In vitro differentiation analysis of sphere-initiatingcells in the FA+/dNS− or FA−/dNS+ media

To examine the multipotency of spheres under the FA+/dNS− and FA−/dNS+ conditions, we transferred and placedisolated spheres under the differentiation condition. After 4DIV of differentiation, triple-labeled immunocytochemistryfor tubulin β III, GFAP, and O4 revealed that cells with themorphological and antigenic characteristics of neurons,astrocytes, and oligodendrocytes, respectively, were gener-ated from the FA+/dNS− spheres (Figs. 5A, C–F). In contrast,the FA−/dNS+ spheres yielded astrocytes and oligodendro-cytes, but neither tubulin β III nor MAP2 positive neurons(Figs. 5B, G–I). The percentage of tubulin β III positiveneurons that differentiated from the FA−/dNS+ primaryspheres was 0.067% ± 0.009% (n = 4) significantly lower thanthat from the FA+/dNS− primary spheres (6.618% ± 0.188%,n = 5, P b 0.01), as well as the percentage of tubulin β IIIpositive neurons from secondary spheres (FA−/dNS+,0.035% ± 0.004% vs. FA+/dNS−, 9.81% ± 1.516%, n = 4,P b 0.01) (Fig. 5J). According to previous studies, thepercentage of neurons derived from the neurosphereculture under the regular condition was 5%–10% (Laeng et al.,2004; Palmer et al., 1999). These results suggested that spheresformed under the FA+/dNS− condition, but not under the FA−/dNS+ condition, coincided with the definition of a neuro-sphere, self-renewal capability, and multipotency (Seabergand van der Kooy, 2002).

2.6. Effects of inhibitors of de novo DNA synthesis on invitro differentiation

We hypothesized that the disturbance of de novo DNAsynthesis in the sphere-forming neural stem cells resulted inthe decrease of neurons. To test the hypothesis, we addedde novo synthesis inhibitors to the FA+/dNS+ medium. AZPblocks the conversion of inosinate to adenyl succinate,which is the key reaction in the formation of purineadenylate; additionally, it also blocks an important step inthe production of the purine base guanylate (Belgi andFriedmann, 2002) (Fig. 1). MTX is a competitive inhibitor ofdihydrofolate reductase (EC 1.5.1.3), which generates tetra-hydrofolate from dihydrofolate. As a result, MTX interfereswith the conversion of deoxyuridylate to thymidylate in thesynthesis of DNA (Olsen, 1991) (Fig. 1). Both AZP and MTXinhibited sphere formation in a dose-dependent manner(Figs. 6A–C). The inhibition was most evident under the FA+/NS− condition. These results suggested that lower doses ofAZP and MTX (e.g., b1 μM) selectively inhibited the de novoDNA synthesis in sphere-forming cells. We transferred thespheres obtained under these conditions and placed themunder the differentiation condition (Fig. 6D). After 4 DIV, thepercentage of tubulin β III positive neurons derived from theFA+/dNS+ spheres significantly (n = 4, P b 0.01) decreased inthe presence of AZP or MTX (Fig. 6E). Figs. 6F–I showrepresentative fields of the culture under the FA+/dNS+condition (Fig. 6F), FA−/dNS+ condition (Fig. 6G), FA+/dNS+with 1 μM AZP condition (Fig. 6H), and FA+/dNS+ with 0.1 μMMTX condition (Fig. 6I). The number of astrocytes did notsignificantly differ among these conditions (data not shown).

3. Discussion

We performed the neurosphere method by using modifiedmedia that lacked FA or dNS and added specific inhibitors ofDNA synthesis. Cultured cells barely proliferated and under-went apoptosis under the FA−/dNS− condition. Althoughsphere formation was observed, the spheres formed underthe FA−/dNS+ condition poorly produced neurons. Sinceextracellular nucleosides exert various effects on cell differ-entiation, apoptosis, mitogenesis, and stimulators of cytokinerelease in the nervous system (Neary et al., 1996), the de-creased percentage of neurons derived from the FA−/dNS+spheres could be attributed to the effects of extracellularnucleosides. However, there was no significant difference(P N 0.05) between the percentage of neurons derived fromFA+/dNS+ spheres (7.662 ± 1.672%, n = 4) (Fig. 6E) andFA+/NS− spheres (6.618% ± 0.188%, n = 5) (Fig. 5J). Addition ofinhibitors of de novo DNA synthesis, namely, AZP and MTX,resulted in poor neuronal differentiation even under the FA+/dNS+ condition. Taken together, low percentage of neuronsderived from spheres formed under the FA−/dNS+ conditionis not due to the extracellular effects of dNS, but due to theproliferation failure of sphere-forming neural stem cellsinduced by inactivation of the de novo DNA synthesispathway.

The neurosphere cultures established from the E12.5mouse proliferated to a lesser extent than those obtainedfrom the E14.5 or E16.5 mouse under the FA−/dNS+ condition,while they proliferated to an equivalent extent under theFA+/dNS− condition. As shown in the differentiation assay,FA−/dNS+ spheres were likely to consist of cells committedto the glial lineage (Seaberg and van der Kooy, 2002). Thisresult may be in line with the fact that neurogenesisprecedes gliogenesis (Qian et al., 2000). To clarify themechanism responsible for the developmental change inthe DNA synthesis pathway, we compared the mRNAexpression of thymidine kinase (EC. 2.7.1.21), deoxycytidinekinase (EC. 2.7.1.74), and adenosine kinase (EC. 2.7.1.45)—key enzymes for the salvage pathway (Karbownik et al.,2003)—and that of thymidylate synthetase (EC. 2.1.1.45)—akey enzyme for the de novo pathway (Chu et al., 2003)—inthe E11.5 and E16.5 mouse neuroepithelium. There was nodetectable change in semi-quantitative RT-PCR (data notshown). The activities of several key enzymes involved inDNA synthesis were previously reported using developmen-tal brain tissues (Hyndman and Zamenhof, 1978; Suleimanand Spector, 1982; Sung, 1971). These reports suggested thatthe specific activities of both the DNA synthesis pathwaysequally decreased in proportion with maturation. Sincethere is a discrepancy between the mRNA levels andenzymatic activities, the activities of the key enzymesinvolved in the DNA synthesis pathways may also beregulated translationally and/or posttranslationally (Ayu-sawa et al., 1986; Sherley and Kelly, 1988).

FA is a cofactor in one-carbon metabolism, and itpromotes the remethylation of homocysteine. Folate defi-ciency allows the accumulation of intracellular homocys-teine, a potentially neurotoxic amino acid that can induceDNA strand break, thereby triggering apoptosis (Mattson

30 B R A I N R E S E A R C H 1 0 7 1 ( 2 0 0 6 ) 2 4 – 3 3

and Shea, 2003). FA deficiency and elevated homocysteinelevels endanger postmitotic neurons in neurodegenerativedisease model mice (Duan et al., 2002; Kruman et al., 2002).To test the effect of homocysteine on sphere-forming neuralstem cells, we performed a neurosphere assay in the FA+/dNS+ medium supplemented with 0.1–1000 μM homocys-teine. However, no toxic effect was observed with anyconcentration of homocysteine. Additionally, no inhibitoryeffect of homocysteine on neuronal differentiation wasobserved in the in vitro differentiation assay (unpublisheddata). These results suggest that the high concentration ofhomocysteine due to FA deficiency cannot explain theproliferation failure of neural stem cells observed in ourstudy.

De novo DNA synthesis inhibitors, such as MTX and AZP,are reported to induce selective malformation of the rhom-bencephalic and telencephalic brain regions in rat embryocultures (Schmid, 1984) and neural tube defects (NTD) inrabbits (Lloyd et al., 1999). In humans, NTD can be preventedby FA supplementation in the periconceptual period (Czeizeland Dudas, 1992; Daly et al., 1995). Although FA deficiencydoes not cause NTDs in normal mice (Heid et al., 1992) or incultured rat embryos (Cockroft, 1991), Fleming and Copp (1998)identified a mouse model of NTD, namely, homozygousSplotch (Pax3) mouse embryos, in which exogenous FA andthymidine prevent NTDs in culture. This observation sug-gested that the disturbance of DNA synthesis was involved inthe pathogenesis of NTDs.

Our observation may explain the vulnerability of embry-onic brain to FA deficiency. Craciunescu et al. (2004) reportedthat FA deficiency decreased progenitor cell proliferationand increased apoptosis in the fetal mouse brain (Craciu-nescu et al., 2004). In this study, we showed that the activityof de novo DNA synthesis is necessary for primary culturedneural stem cells to proliferate with multipotency. Glial cellscan proliferate under the activation of either de novo orsalvage DNA synthesis pathway whereas neuronal cellscannot be generated under the activation of salvagepathway alone. Neural tube defect can be prevented byadministration of FA during the periconceptual period,which may result from activation of de novo DNA synthesisand differentiation to neurons from neural stem cells.Further studies are necessary to elucidate a role of denovo DNA synthesis pathway in differentiation from neuralstem cells into neurons in both embryonic and adult centralnervous system.

Fig. 6 – Inhibitors of de novo DNA synthesis decrease tubulin βExperimental protocol. Neurospheres were cultured under the FAwith AZP or MTX for 7 DIV. (B and C) Dose–response curves of spsupplemented with AZP (B) and MTX (C). Among the three condiinhibitors. Graphs showmean ± SEM number of spheres per 1 × 1conditions, as described above, were placed under the common dof de novo inhibitors. Graph showsmean ± SEMpercentage of tubper field derived from the spheres under the FA+/dNS+ conditiondifference compared with the FA+/dNS+ condition (P b 0.01). (F–I)and astrocytes (GFAP, green). Representative fields derived fromFA+/dNS+ with 1 μM AZP (H), and FA+/dNS+ with 0.1 μM MTX (IIII-positive neurons. Scar bars: F–I, 50 μM.

4. Experimental procedures

4.1. Neurosphere culture

Neurosphere cultures were prepared as described previously(Reynolds and Weiss, 1996). In brief, striata were removed fromE12.5, E14.5, and E16.5 C57BL/6 mice (Japan SLC, Hamamatsu,Japan) embryos and were mechanically dissociated. In place ofthe complete D-MEM/F-12 medium (Invitrogen, San Diego, CA),which contains 6 μM FA, 1.5 μM Thy, and 1.5 μM hypoxanthine,the cells were cultured in a custom-prepared deficient D-MEM/F-12 medium lacking FA, Thy, and hypoxanthine in order to stressthe availability of precursors for both the de novo pathway(folate derivatives) and salvage pathway (Thy, hypoxanthine) ofDNA synthesis. The cells were resuspended in the deficient D-MEM/F-12 medium supplemented with N-2 supplement (Invitro-gen) (progesterone, 20 nM; apo-transferrine, 100 μM; sodiumselenite, 30 nM; insulin, 25 μg/ml), 20 ng/ml human recombinant(hr−) basic FGF-2 (R&D Systems, Minneapolis, MN), 20 ng/ml hr-EGF (Upstate Biotechnology, Lake Placid, NY), and variousconcentrations of FA and dNS as described. FA was purchasedfrom Takeda Pharmaceutical Co. Ltd. (Tokyo, Japan) andprepared as a 15 mg/ml stock solution before it was added tothe culture at a final concentration of 1.0–400 μM. 2′-deoxyade-nosine (dAdo), 2′-deoxycytidine (dCyd), 2′-deoxyguanosine(dGuo), Thy, azathioprine (AZP), and methotrexate (MTX) werepurchased from Sigma (St. Louis, MO) and prepared as a 20 mMstock solution in water. Viable single cells at a density of1.0 × 105 cells/ml were seeded into 75-cm2 tissue culture flasksand incubated at 37 °C in an atmosphere of 5% CO2 and treatedwith 20 ng/ml basic fibroblast growth factor (bFGF) every otherday. Secondary neurosphere cultures were prepared by mechan-ical dissociation of 7 days in vitro (DIV) primary neurospheresinto single cells and resuspended at a density of 2.0 × 104 cells/ml in a fresh culture medium. For neurosphere counting, viablesingle cells at a density of 1.0 × 104 cells/ml were seeded intouncoated 96-well microplates. It has been demonstrated previ-ously that culturing cells at this density will result in clonalneurosphere colonies, as form in single-cell cultures, and thatneurospheres do not arise as a result of cell aggregation at thecell culture densities used here (Seaberg and van der Kooy, 2002;Tropepe et al., 2000). To assess their differentiation potential, 7-DIV spheres were plated onto poly-L-ornithine-coated chamberslides (Nunc, Naperville, IL) in D-MEM/F-12 supplemented withN-2 supplement and 10% fetal bovine serum (differentiationcondition) and cultured for 4 DIV before fixation for immuno-cytochemistry. For neuron counting, 7 DIV spheres weremechanically dissociated, and viable cells were resuspended ata density of 5.0 × 104 cells/ml and plated onto the precoatedchamber slides under the differentiation condition. Cell numbersand viability were assessed by trypan blue dye exclusion byusing a hemocytometer.

III-immunopositive neurons in the neurosphere culture. (A)+/dNS−, FA−/dNS+, and FA+/dNS+ conditions supplementedhere formation, cultured for 7 DIV under each conditiontions, the FA+/dNS− condition was the most sensitive to the04 cells. (D) Experimental protocol. Spheres formed under theifferentiation condition. (E) Quantitative analysis of the effectsulinβ III-immunopositive neurons of the total number of cellsor under the FA+/dNS+ condition with inhibitors. **SignificantDouble immunocytochemistry for neurons (tubulin β III, red)the spheres under the FA+/dNS+ conditions (F), FA−/dNS+ (G),) conditions are shown. White arrowheads indicate tubulin β

31B R A I N R E S E A R C H 1 0 7 1 ( 2 0 0 6 ) 2 4 – 3 3

32 B R A I N R E S E A R C H 1 0 7 1 ( 2 0 0 6 ) 2 4 – 3 3

4.2. Immunocytochemistry

Indirect immunocytochemistry was carried out either immedi-ately after plating (for neurosphere staining) or after 4 DIV (fortriple-labeling and for neuronal counting). The cells plated on theprecoated chamber slides were fixed in 4% paraformaldehyde for30 min, followed by permeabilization with 0.3% Triton X-100 for 5min, and stained by immunofluorescence with the followingprimary antibodies: mouse anti-tubulin β III (1:200; Chemicon,Temecula, CA), mouse anti-MAP2 (1:500; Chemicon), mouse anti-nestin (1:200; Chemicon), rabbit anti-GFAP (1:400; Dako, Carpin-teria, CA), andmouse anti-O4 (1:20, Chemicon). Primary antibodieswere visualized with Cy3- (red), AMCA- (blue), and FITC- (green)conjugated secondary antibodies (Jackson Immuno-Research,West Grove, PA). 4′,6-Diamidino-2-phenylindole (DAPI) was usedas a fluorescent nuclear counterstain. Stained cultures wereexamined and photographed by fluorescence microscopy (Leica,Nussloch, Germany).

4.3. Statistical analysis

Statistical analyses were carried out using ANOVA or Student's ttest. A P value b0.05 was considered significant.

R E F E R E N C E S

Ayusawa, D., Shimizu, K., Koyama, H., Kaneda, S., Takeishi, K.,Seno, T., 1986. Cell-cycle-directed regulation of thymidylatesynthase messenger RNA in human diploid fibroblastsstimulated to proliferate. J. Mol. Biol. 190, 559–567.

Belgi, G., Friedmann, P.S., 2002. Traditional therapies:glucocorticoids, azathioprine, methotrexate, hydroxyurea.Clin. Exp. Dermatol. 27, 546–554.

Chiasson, B.J., Tropepe, V., Morshead, C.M., van der Kooy, D., 1999.Adult mammalian forebrain ependymal and subependymalcells demonstrate proliferative potential, but onlysubependymal cells have neural stem cell characteristics.J. Neurosci. 19, 4462–4471.

Chu, E., Callender, M.A., Farrell, M.P., Schmitz, J.C., 2003.Thymidylate synthase inhibitors as anticancer agents: formbench to bedside. Cancer Chemother. Pharmacol. 52,S80–S89.

Cockroft, D.L., 1991. Vitamin deficiencies and neural-tube defects:human and animal studies. Hum. Reprod. 6, 148–157.

Craciunescu, C.N., Brown, E.C., Mar, M.H., Albright, C.D.,Nadeau, M.R., Zeisel, S.H., 2004. Folic acid deficiency duringlate gestation decreases progenitor cell proliferation andincreases apoptosis in fetal mouse brain. J. Nutr. 134,162–166.

Czeizel, A.E., Dudas, I., 1992. Prevention of the first occurrence ofneural-tube defects by periconceptional vitaminsupplementation. N. Engl. J. Med. 327, 1832–1835.

Daly, L.E., Kirke, P.N., Molloy, A., Weir, D.G., Scott, J.M., 1995. Folatelevels and neural tube defects. Implications for prevention.J. Am. Med. Assoc. 27, 1698–1702.

Duan, W., Ladenheim, B., Cutler, R.G., Kruman, I.I., Cadet, J.L.,Mattson, M.P., 2002. Dietary folate deficiency and elevatedhomocysteine levels endanger dopaminergic neurons inmodels of Parkinson's disease. J. Neurochem. 80, 101–110.

Fleming, A., Copp, A.J., 1998. Embryonic folate metabolism andmouse neural tube defects. Science 280, 2107–2109.

Hatse, S., De Clercq, E., Balzarini, J., 1999. Role of antimetabolites ofpurine and pyrimidine nucleotide metabolism in tumor celldifferentiation. Biochem. Pharmacol. 58, 539–555.

Heid, M.K., Bills, N.D., Hinrichs, S.H., Clifford, A.J., 1992. Folatedeficiency alone does not produce neural tube defects in mice.J. Nutr. 122, 888–894.

Hyndman, A.G., Zamenhof, S., 1978. Thymidine phosphorylase,thymidine kinase and thymidylate synthetase activities incerebral hemispheres of developing chick embryos.J. Neurochem. 31, 577–580.

James, S.J., Basnakian, A.G., Miller, B.J., 1994. In vitro folatedeficiency induces deoxynucleotide pool imbalance, apoptosis,and mutagenesis in Chinese hamster ovary cells. Cancer Res.54, 5075–5080.

Karbownik, M., Brzezianska, E., Zasada, K., Lewinski, A., 2003.Expression of genes for certain enzymes of pyrimidine andpurine salvage pathway in peripheral blood leukocytescollected from patients with Graves' or Hashimoto's disease.J. Cell. Biochem. 89, 550–555.

Koury, M.J., Horne, D.W., 1994. Apoptosis mediates thymidineprevents erythroblast destruction in folate deficiency anemia.Proc. Natl. Acad. Sci. U. S. A. 99, 4067–4071.

Koury, M.J., Price, J.O., Hicks, G.G., 2000. Apoptosis inmegaloblastic anemia occurs during DNA synthesis by ap53-independent, nucleoside-reversible mechanism. Blood96, 3249–3255.

Kruman, I.I., Kumaravel, T.S., Lohani, A., Pedersen, W.A., Cutler,R.G., Kruman, Y., Haughey, N., Lee, J., Evans, M., Mattson, M.P.,2002. Folic acid deficiency and homocysteine impair DNArepair in hippocampal neurons and sensitize them to amyloidtoxicity in experimental models of Alzheimer's disease.J. Neurosci. 22, 1752–1762.

Laeng, P., Pitts, R.L., Lemire, A.L., Drabik, C.E., Weiner, A., Tang, H.,Thyagarajan, R., Mallon, B.S., Altar, C.A., 2004. The moodstabilizer valproic acid stimulate GABA neurogenesis from ratforebrain stem cells. J. Neurochem. 91, 238–251.

Laywell, E.D., Rakic, P., Kukekov, V.G., Holland, E.C., Steindler, D.A.,2000. Identification of a multipotent astrocytic stem cell in theimmature and adult mouse brain. Proc. Natl. Acad. Sci. U. S. A.97, 13883–13888.

Lloyd, M.E., Carr, M., McElhatton, P., Hall, G.M., Hughes, R.A., 1999.The effects of methotrexate on pregnancy, fertility andlactation. QJM 92, 551–563.

Mattson, M.P., Shea, T.B., 2003. Folate and homocysteinemetabolism in neural plasticity and neurodegenerativedisorders. Trends Neurosci. 26, 137–146.

Murray, R.K., Granner, D.K. (Eds.), 1999. Harper's Biochemistry,25th edition. Appleton and Lange.

Neary, J.T., Rathbone, M.P., Cattabeni, F., Abbracchio, M.P.,Burnstock, G., 1996. Trophic actions of extracellularnucleotides and nucleosides on glial and neuronal cells.Trends Neurosci. 19, 13–18.

Olsen, E.A., 1991. The pharmacology of methotrexate. J. Am. Acad.Dermatol. 25, 306–318.

Palmer, T.D., Markakis, E.A., Willhoite, A.R., Safar, F., Gage, F.H.,1999. Fibroblast growth factor-2 activates a latent neurogenicprogram in neural stem cells from diverse regions of the adultCNS. J. Neurosci. 19, 8487–8497.

Qian, X., Shen, Q., Goderie, S.K., He, W., Capela, A., Davis, A.A.,Temple, S., 2000. Timing of CNS cell generation: a programmedsequence of neuron and glial cell production from isolatedmurine cortical stem cells. Neuron 28, 69–80.

Reynolds, B.A., Weiss, S., 1996. Clonal and population analysesdemonstrate that an EGF-responsive mammalian embryonicCNS precursor is a stem cell. Dev. Biol. 175, 1–13.

Schmid, B.P., 1984. Monitoring of organ formation in ratembryos after in vitro exposure to azathioprine,mercaptopurine, methotrexate or cyclosporin A. Toxicology31, 9–21.

Seaberg, R.M., van der Kooy, D., 2002. Adult rodent neurogenicregions: the ventricular subependyma contains neural stemcells, but the dentate gyrus contains restricted progenitors.J. Neurosci. 22, 1784–1793.

Sherley, J.L., Kelly, T.J., 1988. Regulation of human thymidinekinase during the cell cycle. J. Biol. Chem. 263, 8350–8358.

33B R A I N R E S E A R C H 1 0 7 1 ( 2 0 0 6 ) 2 4 – 3 3

Sommer, L., Rao, M., 2002. Neural stem cells and regulation ofcell number. Prog. Neurobiol. 66, 1–18.

Suleiman, S.A., Spector, R., 1982. Identification, development,and regional distribution of thymidylate synthetase in adultrabbit brain. J. Neurochem. 38, 392–396.

Sung, S.C., 1971. Thymidine kinase in the developing rat brain.Brain Res. 35, 268–271.

Tropepe, V., Coles, B.L.K., Chiasson, B.J., Horsford, D.J., Elia, A.J.,McInnes, R.R., van der Kooy, D., 2000. Retinal stem cells in theadult mammalian eye. Science 287, 2032–2036.