astrocytes negatively regulate neurogenesis through the jagged1-mediated notch pathway

TISSUE-SPECIFIC STEM CELLS

Astrocytes Negatively Regulate Neurogenesis Through the

Jagged1-Mediated Notch Pathway

ULRIKA WILHELMSSON,a

MARYAM FAIZ,a

YOLANDA DE PABLO,a

MARIKA SJOQVIST,c,d

DANIEL ANDERSSON,a

ASA WIDESTRAND,a MAJA POTOKAR,e,f MATJAZ STENOVEC,e,f PETER L. P. SMITH,a NORIKO SHINJYO,b TULEN PEKNY,a

ROBERT ZOREC,e,f,g ANDERS STAHLBERG,a MARCELA PEKNA,a CECILIA SAHLGREN,c,d MILOS PEKNYa

aCenter for Brain Repair and Rehabilitation, Department of Clinical Neuroscience and Rehabilitation, Institute of

Neuroscience and Physiology and bDepartment of Medical Chemistry and Cell Biology, Institute of Biomedicine,

Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden; cTurku Centre for Biotechnology,

University of Turku, Turku, Finland; dDepartment of Biosciences, Abo Akademi University, Turku, Finland;eInstitute of Pathophysiology, Faculty of Medicine, University of Ljubljana, Ljubljana, Slovenia; fCelica

Biomedical Center, Ljubljana, Slovenia; gIKERBASQUE, Basque Foundation for Science, Bilbao, Spain

Key Words. Astrocytes • Glial fibrillary acidic protein • Vimentin • Intermediate filaments • Neurogenesis

ABSTRACT

Adult neurogenesis is regulated by a number of cellular play-ers within the neurogenic niche. Astrocytes participateactively in brain development, regulation of the mature cen-tral nervous system (CNS), and brain plasticity. They are im-portant regulators of the local environment in adultneurogenic niches through the secretion of diffusible morpho-genic factors, such as Wnts. Astrocytes control the neurogenicniche also through membrane-associated factors, however,the identity of these factors and the mechanisms involved arelargely unknown. In this study, we sought to determine themechanisms underlying our earlier finding of increased neu-ronal differentiation of neural progenitor cells when cocul-tured with astrocytes lacking glial fibrillary acidic protein(GFAP) and vimentin (GFAP2/2Vim2/2). We used primaryastrocyte and neurosphere cocultures to demonstrate that

astrocytes inhibit neuronal differentiation through a cell–cellcontact. GFAP2/2Vim2/2 astrocytes showed reduced endocy-tosis of Notch ligand Jagged1, reduced Notch signaling, andincreased neuronal differentiation of neurosphere cultures.This effect of GFAP2/2Vim2/2

astrocytes was abrogated inthe presence of immobilized Jagged1 in a manner dependenton the activity of c-secretase. Finally, we used GFAP2/2

Vim2/2mice to show that in the absence of GFAP and vimen-

tin, hippocampal neurogenesis under basal conditions as wellas after injury is increased. We conclude that astrocytes nega-tively regulate neurogenesis through the Notch pathway, andendocytosis of Notch ligand Jagged1 in astrocytes and Notchsignaling from astrocytes to neural stem/progenitor cellsdepends on the intermediate filament proteins GFAP andvimentin. STEM CELLS 2012;30:2320–2329

Disclosure of potential conflicts of interest is found at the end of this article.

INTRODUCTION

Adult neurogenesis is restricted to two specific neurogenicniches: the subgranular zone (SGZ) of the hippocampus andthe subventricular zone (SVZ) of the lateral ventricles. Thecellular players within these niches are important for the regu-lation of neural stem/progenitor cell development and coordi-nation of cell genesis. Increasing evidence suggests an impor-tant role for astrocytes in the neurogenic niche: astrocytes

share certain properties with neural stem cells [1–3] andcreate an environment conducive to neurogenesis [4]. Astro-cytes regulate neurogenesis by the secretion of factors such asWnt3 [5], interleukin-1b, interleukin-6 [6], and thrombospon-din-1 [7]. Astrocytes control the neurogenic niche alsothrough membrane-associated factors [4], however, the iden-tity of these factors and the mechanisms involved are largelyunknown.

The Notch receptor family is involved in a large numberof cell fate decisions during development and their activation

Author contributions: U.W. and M.F.: conception and design, collection and/or assembly of data, data analysis and interpretation, andmanuscript writing; Y.d.P and D.A.: conception and design, collection and/or assembly of data, and data analysis and interpretation, M.Sj€oqvist, A.W., M. Potokar, M. Stenovec, P.L.P.S., and T.P.: collection and/or assembly of data and data analysis and interpretation;N.S.: provision of study material or patients; R.Z. and C.S.: conception and design, financial support, and data analysis andinterpretation; A.S.: conception and design; M. Pekna: conception and design, data analysis and interpretation, and manuscript writing;M. Pekny: conception and design, financial support, data analysis and interpretation, and manuscript writing. U.W. and M.F. contributedequally to this article. Y.d.P. and M. Sj€oqvist, contributed equally to this article.

Correspondence: Milos Pekny, Ph.D., M.D., Center for Brain Repair and Rehabilitation, Department of Clinical Neuroscience andRehabilitation, Institute of Neuroscience and Physiology, Sahlgrenska Academy at the University of Gothenburg, Box 440, SE-405 30Gothenburg, Sweden. Telephone: þ46 31 786 3269; Fax: þ46 31 416 108; e-mail: [email protected] Received January 11,2012; accepted for publication July 15, 2012; first published online in STEM CELLS EXPRESS August 7, 2012. VC AlphaMed Press1066-5099/2012/$30.00/0 doi: 10.1002/stem.1196

STEM CELLS 2012;30:2320–2329 www.StemCells.com

can promote an undifferentiated, precursor cell state. Notch1and its cognate ligand Jagged1 are expressed in the neuro-genic niches in the adult mammalian brain [8]. Using trans-genic Hes5-GFP mice, which enable the visualization of cellswith active canonical Notch signaling, it was shown thatcanonical Notch signaling in the SGZ is restricted to type 1neural stem cells with radial or nonradial morphology [9, 10].Activated Notch in postnatal SVZ cells suppresses neuronaldifferentiation and decreases proliferation [11], and activationof Notch signaling in neural stem cells leads to increasedastrogliogenesis in vitro [12]. Notch receptor activation pro-motes the survival of neural stem cells in vitro, and transientadministration of Notch ligands to the brain of adult ratsincreased the numbers of newly generated precursor cells[13]. A series of recent publications supports the notion thatNotch1 and Notch canonical signaling repressing pro-neuronalgenes are required for the maintenance of a pool of proliferat-ing undifferentiated cells in the adult SGZ and SVZ [9, 10,14–16]. Thus, despite controversy regarding the actual mecha-nisms and effects on neuronal cell fate, Notch signaling playsan essential role in controlling adult neurogenesis. In SVZ,astrocytes were shown to express Jagged1 [12, 17], whereasin the adult hippocampus, Jagged1 mRNA expression wasreported to be present in the SGZ [15] and in the granule celllayer (GCL) [8].

We and others have shown that ablation of the intermediatefilament proteins glial fibrillary acidic protein (GFAP) andvimentin in mice [18, 19] creates an environment more permis-sive to transplantation of neural grafts or neural stem cells [20,21] and increases axonal and synaptic regeneration [22–24].Neuronal differentiation of neural progenitor cells wasincreased when cocultured with astrocytes lacking GFAP andvimentin (GFAP�/�Vim�/�) [21]. Although the altered cellulardistribution of Wnt3 in the GFAP�/�Vim�/� astrocytes couldbe associated with changed secretion of this pro-neurogenicfactor and thus explain this finding, a direct cell–cell signalingfrom astrocytes to neural stem/progenitor cells and its involve-ment in neurogenesis remains as an attractive alternative. Inthis study, we investigated the role of astrocyte membrane-asso-ciated factors in the regulation of neurogenesis.

MATERIALS AND METHODS

Mice

Mice carrying a null mutation in the GFAP and/or vimentin genesand wild-type controls [25, 26] were on a C57Bl6/129Sv/129Olamixed genetic background. All mice were housed in a barrierfacility, and experiments were conducted according to protocolsapproved by the Ethics Committee of the University ofGothenburg.

Neurosphere Culture

Postnatal day 4 (P4) mouse forebrain was dissected in Leibovitzmedium (Invitrogen, Carlsbad, CA, http://www.invitrogen.com)and enzymatically (0.1% trypsin, 0.5 mM EDTA in Hank’s bal-anced salt solution [Invitrogen]) and mechanically dissociatedinto a single-cell suspension and plated in neurosphere medium(Neurobasal [Invitrogen] containing L-glutamine [2 mM, Invitro-gen], penicillin/streptavidin [100 U/0.1 mg/ml (1�), Invitrogen],B27 [1:50, Invitrogen], basic fibroblast growth factor [20 ng/ml,Invitrogen], epidermal growth factor [20 ng/ml, Stem Cell Tech-nologies, Vancouver, BC, Canada, http://www.stemcell.com],heparin [1 U/ml, Sigma-Aldrich, St. Louis, MS, http://www.sigmaaldrich.com], and fungizone [0.25 lg/ml, Bristol-MyersSquibb, New York City, NY, http://www.bms.com]). Neuro-spheres were dissociated with TrypLE (Invitrogen) into a single-

cell suspension and replated under the same conditions as primarycultures. Ablation of GFAP, vimentin, or both did not alter thepassaging capacity.

Neurosphere Differentiation

Seven-day-old P4 primary neurospheres were pipetted into 24-wellplates (15 neurospheres per well) with glass coverslips coated withpoly-L-ornithine (0.01 mg/ml, Sigma-Aldrich) and laminin (5 lg/ml, Invitrogen) and gently flooded with differentiation medium(neurosphere medium as described above without growth factorsand heparin added). Alternatively, primary neurospheres were dis-sociated with TrypLE and plated at equal densities for differentia-tion. One day after plating, 1% fetal bovine serum (FBS, Invitro-gen) was added. For experiments with conditioned medium,neurospheres were dissociated and plated in 90% neurosphere me-dium supplemented with 10% medium conditioned by astrocytesfor 3 days. Higher concentrations of conditioned medium resultedin neurosphere cell death. After 5 days of differentiation, cellswere fixed with 4% paraformaldehyde for immunocytochemistry.

Primary Astrocyte Culture

Primary astrocytes derived from forebrain of P2 mice werecultured in Dulbecco’s modified Eagle’s medium (DMEM,Invitrogen) with 10% FBS, as described [27]. For quantitativereal-time polymerase chain reaction (qRT-PCR) analysis, astro-cytes (Experiment 1) were prepared as described [28]. Alterna-tively, total RNA was extracted and purified with the RNeasyMicro Kit (Qiagen, Hilden, Germany, http://www.qiagen.com)(Experiments 2 and 3).

Cocultures

At confluence, primary astrocytes were trypsinized and plated in24-well plates or on coated glass coverslips. Twenty-four hourslater, the medium was changed to differentiation medium. Afteranother 24 hours, primary neurospheres were enzymatically disso-ciated with TrypLE for 5 minutes at 37�C and plated on astro-cytes. For the experiment in Figure 1D, neurospheres were stainedwith Hoechst 33342 dye (1:5,000, Invitrogen) for 10 minutes (todistinguish neurosphere cells from astrocytes) For experiments inFigure 1B and 1E, neurospheres were incubated with 5 lM 5-bromo-2-deoxyuridine (BrdU, Sigma-Aldrich) for 24 hours (to dis-tinguish neurosphere cells from astrocytes) before dissociation andreplating with astrocytes. For cultures with mixed GFAP�/�Vim�/�

and wild-type astrocytes, equal numbers of cells of respective geno-type were mixed and plated. One day after plating, 1% FBS wasadded. After 5 days of differentiation, cells were fixed with 4%paraformaldehyde for immunocytochemistry.

Immunocytochemistry and Cell Proliferation, CellDeath and Differentiation Assessment

To assess cell differentiation, cells were labeled with mouseanti-b-III-tubulin (1:2,000, Covance, Princeton, NJ, http://www.co-vance.com), rabbit anti-S100 (1:200, Dako, Glostrup, Denmark,http://www.dako.com), mouse anti-RIP (1:50, a kind gift from Dr.Hockfield, Yale University), and rat anti-BrdU (1:200, Nordic Bio-site, Stockholm, Sweden, http://www.biosite.se) antibodies followedby Alexa fluorochrome-conjugated secondary antibodies (1:1,000,Invitrogen), and 40-6-diamidino-2-phenylindole (DAPI, 1:10,000,Sigma-Aldrich). For detection of BrdU, cells were first treated with2N HCl. Cells were counted using an inverted fluorescence micro-scope (Leica) and Volocity software (PerkinElmer, Waltham, MA,http://www.perkinelmer.com). Neuronal differentiation of neuro-sphere cells was quantified by counting the total number of b-III-tubulinposHoechtpos or b-III-tubulinposBrdUpos cells per well.

To quantify cell proliferation, cell death, and Notch activationunder the differentiation conditions, neurospheres were dissoci-ated and plated in differentiation media. To assess proliferationafter 1 or 5 days, BrdU (5 lM) was added to the media for 6hours, and the cells were then fixed and immunostained againstBrdU. Alternatively, to assess cell death, propidium iodide

Wilhelmsson, Faiz, de Pablo et al. 2321

www.StemCells.com

(0.5 lg/ml) was added to the media for 15 minutes, and the cellswere then washed and fixed. To assess activation of Notch, cellswere incubated with goat anti-cleaved Notch1 (1:50, Santa CruzBiotechnology, Santa Cruz, CA, http://www.scbt.com). All cellswere counterstained with DAPI and the percentage of cells posi-tive for BrdU, propidium iodide, or cleaved Notch1 was assessedusing microscope and software as above.

Luciferase Assay to Measure Notch SignalingActivity

HEK-293 cells stably expressing the full-length Notch1 receptor(293HEK-FLN1) or neural stem/progenitor cell line derived fromadult mouse forebrain [29] were transfected with 12XCSL-lucif-erase Notch reporter and a cytomegalovirus (CMV)-b-galactosi-dase construct by electroporation [30, 31]. Transfected cells wereplated on monolayers of primary P2 astrocytes and cocultured for24 hours and then lysed and analyzed with Luciferase Assay andb-galactosidase Assay Kits (Promega, Madison, WI, http://www.promega. com), according to the manufacturer’s instruc-tions. b-Galactosidase activity was used to assess transfection ef-ficiency. Luciferase expression levels are given as light units rela-tive to b-galactosidase expression.

Fluorescent Activated Cell Sorting Analyses ofGeneral Endocytosis, Ligand Internalization, andMembrane Bound Jagged1

For quantification of general endocytosis, P2 primary astrocytecultures were incubated with fluorescence-tagged dextran-coatedbeads (Invitrogen) for 4 hours in 37�C. The cells were thendetached, centrifuged (450g, 5 minutes), and extracellular fluo-rescence was quenched with trypan blue in phosphate-bufferedsaline (PBS) for 5 minutes in room temperature. The cells werecentrifuged, excess trypan blue was removed, and the cells wereresuspended in PBS and analyzed by fluorescent activated cellsorting (FACS) as described [32] using FACSCalibur (BDPharmingen, San Diego, CA, http://www.bdbiosciences.com/).For ligand internalization, recombinant Notch1 extracellular do-main fused to Fc fragment of human IgG1 (rN1ECD, 1 lg/ml,R&D Systems, Minneapolis, MN, http://www.rndsystems.com)were preincubated in PBS with Alexa 488 goat anti-human anti-bodies (1:200, Invitrogen) in þ4�C for 1 hour. Primary astro-cytes cultures were blocked in DMEM containing 10% goat se-rum and 1% bovine serum albumin (BSA) for 45 minutes at37�C. The rN1ECD-Alexa 488 solution was diluted 1:5 inblocking solution and then added to the astrocytes for

Figure 1. Neuronal differentiation of stem/progenitor cells is determined by the environment and is cell–cell contact dependent. (A): GFAP�/�

Vim�/� (G�/�V�/�; n ¼ 8) neurospheres differentiate into more b-III-tubulin-expressing neurons (b-III-tubulinpos) than WT (n ¼ 11), GFAP�/�

(G�/�; n ¼ 6), or Vim�/� (V�/�; n ¼ 6) neurospheres. (B): Neuronal differentiation of P4 WT neurospheres cocultured with either WT, G�/�,V�/�, or G�/�V�/� primary P2 astrocytes (n ¼ 6, 3, 3, and 3, respectively). Images show 5-bromo-2-deoxyuridine (BrdU)-labeled dissociatedWT neurospheres plated on top of WT and G�/�V�/� astrocytes and allowed to differentiate for 5 days. Neurosphere cells are visualized by im-munostaining with antibodies against BrdU (red), neurons are visualized by antibodies against b-III-tubulin (green), and nuclei are labeled withDAPI. Arrows point to neurosphere cells differentiated into neurons (b-III-tubulinposBrdUpos), arrowheads point to neurosphere cells not differen-tiated into neurons (b-III-tubulinnegBrdUpos). (C): Neuronal differentiation of dissociated WT neurosphere cells in astrocyte-conditioned medium.No difference in the percentage of b-III-tubulinpos neurons was observed between WT neurosphere cells cultured under control conditions (noconditioned medium, n ¼ 4) and WT neurosphere cells cultured in the presence of medium conditioned by G�/�V�/� (n ¼ 9) or WT (n ¼ 7)astrocytes. (D): When neurosphere cells were exposed to conditioned medium from WT astrocytes while in direct contact with G�/�V�/� astro-cytes or vice versa, neuronal differentiation was increased by direct contact with G�/�V�/� astrocytes (n ¼ 4 in all groups). (E): The number ofb-III-tubulinpos neurons from WT neurospheres was higher in the presence of G�/�V�/� astrocytes than WT astrocytes or mixed G�/�V�/� andWT astrocytes (n ¼ 4 in all groups). N equals number of mice per experimental group. DAPI, nuclear staining; scale bars ¼ 75 lm (A) and 30lm (B); *, p < .05; **, p < .01; ***, p < .005; ****, p < .001 (A–C and E, ANOVA followed by Tukey honestly significant difference posthoc analysis; D, two-tailed t-test). Abbreviations: DAPI, 40-6-diamidino-2-phenylindole; WT, wild type.

2322 Astrocytes Regulate Neurogenesis

incubation at 37�C. The cells were then prepared for FACSanalyses as described above.

For detection of surface levels of Jagged1, detached astrocyteswere fixed in 4% paraformaldehyde, without further permeabilization,blocked with 3% BSA in PBS, and incubated with a fluorescent-con-jugated antibody recognizing the extracellular domain of Jagged1(R&D Systems) followed by FACS analyses as described above.

Quantification of Jagged1-Positive Vesicles

Live P2 primary astrocyte cultures were incubated with rabbitanti-Jagged1 antibodies (1:100, Abcam, Cambridge, U.K., http://www.abcam.com) and goat anti-rabbit Alexa Fluor 488 secondaryantibodies (1:600, Invitrogen). The cells were fixed in 2% para-formaldehyde, and images of single astrocytes were acquiredusing a laser-scanning confocal microscope (LSM 510, CarlZeiss, Oberkochen, Germany, http://www.zeiss.com) and were an-alyzed by ImageJ software (National Institute of Health, http://rsbweb.nih.gov/ij/) similarly as previously described [33].

Protein Extraction and Western Blot

For Western blot analysis, cells were lysed in RIPA buffer (0.15 MNaCl, 1% Nonidet P40, 0.5% sodium deoxycholate, 0.1% sodiumdodecyl sulfate, and 0.05 M Tris-HCl, pH 8) supplemented withprotease inhibitors (Roche, Basel, Switzerland, http://www.roche-applied-science.com) and dithiothreitol (1 mM, Sigma-Aldrich). Lysates were centrifuged for 15 minutes at 15,000g anddiluted with Laemmli sample buffer. Jagged1 was detected withgoat anti-Jagged1 antibody (C-20, Santa Cruz Biotechnology).

Jagged1 Activation of Notch Signaling

Plates were coated with Protein G (50 lg/ml, Invitrogen) overnightat room temperature, washed in PBS, and blocked with 1% BSA inPBS. The plates were washed with PBS and incubated with recombi-nant Jagged1-Fc chimera (R&D Systems) or a ChromPure IgG Fcfragment (Jackson Immunoresearch Laboratories, West Grove, PA,http://www.jacksonimmuno.com) at concentrations of 1, 2.5, or 5lg/ml in 0.1% BSA in PBS for 4 hours at room temperature. Seven-day-old P4 primary neurospheres were dissociated with TrypLE intoa single-cell suspension and immediately plated at equal densities onligand-coated plates. One day after plating, 1% FBS was added. Af-ter 5 days of differentiation, cells were fixed with 4% paraformalde-hyde for immunocytochemistry. No concentration-dependent differ-ences were seen between 1 and 5 lg/ml. For c-secretase inhibition,N-[N-(3,5-difluorophenylacetyl-l-alanyl)]-S-phenylglycine t-butyles-ter (DAPT; 5 lM) was added to the media daily.

Reverse Transcription and qRT-PCR

Reverse transcription was performed with the iScript cDNA Synthe-sis Kit (Bio-Rad Laboratories, Hercules, CA, http://www.bio-rad.com). qRT-PCR amplification analyses were carried out on a Light-Cycler 480 (Roche). Master mix was prepared with iQ SYBR GreenSupermix (Bio-Rad Laboratories) and primer sequences used were50-ATCGCATCGTACTGCCTTTC-30 and 50-GGCAATCCCTGTGTTTTGTT-30. PCR cycling condition was 3 minutes 95�C, fol-lowed by 45 cycles (20 seconds 95�C, 20 seconds 60�C, 20 seconds72�C) and melt curve from 60�C to 95�C to determine correct PCRproduct formation. Among six reference genes measured, ACTB andGAPDH were the most stably expressed as shown by NormFinder[34] and were used to normalize expression levels. Data from threeindependent experiments were pooled using mean centering.

BrdU Injections and Entorhinal Cortex Lesions

Three-month-old male mice were used in all in vivo experiments. Forbasal cell proliferation, mice received a single intraperitoneal injectionof BrdU (300 mg/kg) in sterile PBS and were killed 24 hours later.For cell fate determination experiments, 3-month-old males receivedBrdU twice daily (300 mg/kg) for 1 week and were killed 6 weeks af-ter the first injection. Unilateral entorhinal cortex lesions were per-formed by transection of perforant pathway with a retractable wireknife as previously described [35]. For cell fate determination after

lesion, mice received BrdU twice daily (200 mg/kg) during the firstweek after transection and were killed 2 weeks after the first injection.

Tissue Processing and Immunohistochemistry

Mice were anesthetized and perfused transcardially, and the dissectedbrains were postfixed with 4% paraformaldehyde overnight. Afterimmersion in 30% sucrose in PBS, 40 lm horizontal sections werecut. For BrdU immunohistochemistry, sections were treated with 2NHCl. Following primary antibodies were used: rat anti-BrdU (1:100,Nordic Biosite), mouse anti-BrdU and rabbit anti-S100 (both 1:200,Dako), biotinylated mouse anti-NeuN (1:100, Chemicon, Temecula,CA, http://www.chemicon.com), rabbit anti-Sox2 (1:200, Chemicon),rabbit anti Tbr2 (1:200, Abcam), goat anti-Jagged or anti-Sox2 (both1:50, Santa Cruz Biotechnology) antibodies. Secondary antibodiesused were streptavidin-conjugated Cy3 (1:300, Sigma-Aldrich), orAlexa fluorochrome-conjugated secondary antibodies and nucleimarker TOPRO-3 (both Invitrogen), and for light microscopy, biotin-ylated donkey anti-rat antibodies (Jackson ImmunoResearch) and Vec-tastain ABC kit (BioNordika, Stockholm, Sweden, http://www.bionordika.se). Sections were mounted with Mowiol and coverslipped.

Hippocampal Cell Counts

Stereological counting of cells was performed using an epifluorescencemicroscope (Eclipse 80i, Nikon) and by using a modified optical dis-sector method as described previously [36]. For absolute counts ofBrdUpos cells in the SGZ and GCL of the dentate gyrus at 24 hour timepoint and 2 weeks after lesion time point, every sixth horizontal section40 lm thick covering a depth of 1,680 lm (seven sections) was used.For the 6 week time point, we used every fourth horizontal section 40lm thick covering a depth of 960 lm (six sections). On each section,all cells positive for the respective marker or the combination thereofin the specified brain region were counted. The average cell numberper section was multiplied by a factor to obtain the estimated total cellnumber for a depth of 1,680 lm through the hippocampus in the dorso-ventral direction. Colocalization of BrdU with NeuN, S100, Sox2, andTbr2, respectively, was quantified in three to six sections per mouse bylaser-scanning confocal microscopy (Leica Microsystems, Wetzlar,Germany, http://www.leica-microsystems.com/). Absolute numbers ofdouble-positive cells were calculated by multiplying the ratio of dou-ble-positive cells by the total number of BrdUpos cells. For the assess-ment of absolute number of Sox2posS100neg neural stem/progenitorcells, stacks of optical sections of the dentate gyrus covering 6 lm (ba-sal condition) and 12 lm (lesioned mice) on every sixth horizontal tis-sue section 40 lm thick (three sections) were acquired with a laser-scanning confocal microscope (Leica Microsystems).

To compare the area of GCL in GFAP�/�Vim�/� and wild-typemice, every fourth horizontal section 40 lm thick covering a depth of960 lm (a total of six sections), with GCL visualized by NeuN antibod-ies, was photographed with an epifluorescence microscope (NikonInstruments, Amstelveen, the Netherlands, http://www.nikoninstruments.com) and the GCL area was measured with ImageJ software (ver-sion 1.30, National Institutes of Health). The area of the GCL did notdiffer between GFAP�/�Vim�/� and wild-type mice (data not shown).

Statistical Analysis

Data were analyzed in SPSS (version 16.0, IBM, Armonk, NY, http://www.ibm.com) or Prism (version 5.0, GraphPad Software, La Jolla,CA, http://www.graphpad.com). One-factor or two-factor ANOVAwas used followed by post hoc analysis (Bonferroni correction orTukey honestly significant difference test). Two-tailed t-tests wereused for comparisons between two groups. Differences were consid-ered significant at p < .05. Values are presented as mean6 SEM.

RESULTS

Neuronal Differentiation of Stem/Progenitor Cells IsDetermined by the Environment

To assess the mechanisms by which astrocytes regulate neuro-nal differentiation of neural stem/progenitor cells, we first


www.StemCells.com

determined the effects of GFAP and vimentin deficiency ondifferentiation of neurosphere cultures derived from P4 mice.There were no differences in neuronal differentiation betweenwild-type, GFAP�/�, or Vim�/� neurospheres (Fig. 1A). How-ever, several fold more b-III-tubulinpos neurons were gener-ated from GFAP�/�Vim�/� neurospheres than wild-type neu-rospheres (p < .001, Fig. 1A). This experiment was repeatedwith neurospheres from wild-type and GFAP�/�Vim�/� micethat were dissociated prior differentiation. Again, there weremore b-III-tubulinpos neurons generated from GFAP�/�Vim�/�

than from wild-type neurospheres (1.9% 6 0.4% and 0.6% 60.1%, respectively, p < .001). Twofold more RIPpos oligoden-drocytes were generated from GFAP�/�Vim�/� compared towild-type neurospheres (0.4% 6 0.07% vs. 0.2% 6 0.02%,p < .0001). The neurosphere genotype had no effect on astro-genesis (95% 6 0.5% and 96% 6 0.4% S100pos cells forGFAP�/�Vim�/� and wild-type neurospheres, respectively), cellproliferation (28% 6 2.5% and 29% 6 2.5% BrdUpos cells at24 hours and 10% 6 0.8% and 10% 6 0.9% BrdUpos cells at 5days of differentiation of GFAP�/�Vim�/� and wild-type neuro-spheres, respectively), or cell death (17% 6 1.0% and 15% 61.1% propidium iodidepos cells at 24 hours and 4% 6 0.6% and4% 6 0.4% propidium iodidepos cells at 5 days of differentiationof GFAP�/�Vim�/� and wild-type neurospheres, respectively).

To assess the effects of the astrocyte environment on neuro-nal differentiation, we used an astrocyte-neurosphere coculturesystem in which P4 neurospheres were prelabeled with BrdUfor 24 hours before coculturing with P2 astrocytes to excludethe contribution of neural stem/progenitor cells from the astro-cyte cultures (Fig. 1B). Coculture of dissociated wild-type neu-rosphere cells on top of wild-type astrocytes increased neuronaldifferentiation 2.4-fold when compared to neurosphere cellsplated directly on laminin-coated surface (3.2% 6 0.4%, n ¼ 6and 1.3%6 0.4%, n ¼ 11, respectively, p < .05). Also, the totalnumber of neurosphere-derived cells was higher when neuro-spheres were cultured in the presence of astrocytes (data notshown). These results confirm that the presence of astrocytes isbeneficial for the proliferation, survival, and neuronal differen-tiation of neurosphere cells as previously shown for adult hippo-campus-derived neural progenitor cells [4, 37]. Comparable neu-ronal differentiation was seen when dissociated wild-typeneurospheres were plated on top of wild-type, GFAP�/�, orVim�/� astrocytes. However, when plated on top of GFAP�/�

Vim�/� astrocytes, dissociated wild-type neurospheres generated2.1-fold more neurons compared to plating on top of wild-type,GFAP�/�, or Vim�/� astrocytes (Fig. 1B; p < .01). These find-ings provide evidence for enhanced neuronal differentiation ofneurosphere cells cocultured with GFAP�/�Vim�/� astrocytes.As the total number of neurosphere-derived cells did not differbetween neurospheres cocultured with GFAP�/�Vim�/� andwild-type astrocytes (82,988 6 4,005 and 88,424 6 12,665 cellsper coverslip), the absence of GFAP and vimentin in astrocytesdoes not appear to have any measurable effect on neurospherecell survival and/or proliferation in this coculture system.

Effect of GFAP2/2Vim2/2 Astrocytes Is Mediatedby Cell–Cell Contact

Next, we determined whether the increase in neuronal differ-entiation was due to secretion of neurogenic factors byGFAP�/�Vim�/� astrocytes or due to signaling dependent oncell–cell contact. Neuronal differentiation of wild-type neuro-sphere cells was similar in medium conditioned by GFAP�/�

Vim�/� or wild-type astrocytes (Fig. 1C). In contrast, whenprelabeled neurosphere cells were exposed to conditioned me-dium from wild-type astrocytes while in direct contact withGFAP�/�Vim�/� astrocytes or vice versa, neuronal differen-

tiation was increased by direct contact with GFAP�/�Vim�/�

astrocytes (p < .05, Fig. 1D). To further analyze the effect ofcell–cell contact, we cocultured prelabeled dissociated neuro-sphere cells on wild-type, GFAP�/�Vim�/�, and mixed wild-type and GFAP�/�Vim�/� astrocytes. The presence of wild-type astrocytes abolished the increase in neuronal differentia-tion mediated by GFAP�/�Vim�/� astrocytes (p < .01, Fig.1E), suggesting the presence of an inhibitory signaling fromwild-type astrocytes to neurosphere cells.

GFAP2/2Vim2/2 Astrocytes Show Reduced NotchSignaling

The increased neuronal differentiation of neurosphere cellscocultured with GFAP�/�Vim�/� astrocytes indicated a cell–cell contact-dependent signaling mechanism that is suppressedby contact with wild-type astrocytes. Since Notch signalingsuppresses neuronal differentiation through cell–cell contact[38], we investigated the possible involvement of Notch sig-naling in GFAP�/�Vim�/� astrocyte-mediated neuronal differ-entiation by coculturing P2 wild-type and GFAP�/�Vim�/�

astrocytes with Notch reporter cells [30, 31]. There was lessNotch signaling between GFAP�/�Vim�/� astrocytes and re-porter cells than between wild-type astrocytes and reportercells (p < .005; Fig. 2A).

Next, we assessed the expression level of Jagged1, theprincipal Notch ligand, in GFAP�/�Vim�/� and wild-typeastrocytes by qRT-PCR in three independent experiments.Expression of Jagged1 was downregulated by 40% in GFAP�/�

Vim�/� astrocytes (p < .001; pooled data from three independ-ent experiments, Fig. 2B). Western blot analysis confirmed thatGFAP�/�Vim�/� astrocytes contained less Jagged1 protein (p <.001, Fig. 2C). FACS analyses of Jagged1pos astrocytes showedcomparable amount of cell membrane bound Jagged1 on wild-type and GFAP�/�Vim�/� astrocytes (46.9 6 2.4 and 40.9 64.1 mean fluorescence intensity, respectively). Thus, althoughthe total amount of Jagged1 in GFAP�/�Vim�/� astrocytes isreduced, the membrane-associated fraction is not altered.

Notch ligand and receptor availability is known to be regu-lated by endocytosis and membrane trafficking [39], and wehave previously shown that intermediate filaments are impor-tant for astrocyte vesicle trafficking dynamics and interferon-cinduced mobility of major histocompatibility complex (MHC)class II compartment [28, 40, 41]. Thus, we investigated boththe general endocytosis as well as the endocytosis of Jagged1,which is important for eliciting a Notch signal [42]. InGFAP�/�Vim�/� astrocytes, we observed a general reductionin endocytosis as shown by FACS analysis of the uptake ofdextran-coated beads (p < .01, Fig. 2D) and a decrease in theNotch ligand-mediated internalization of the Notch extracellu-lar domain (p < .001, Fig. 2E). Also, the number of Jagged1pos

vesicles was reduced in GFAP�/�Vim�/� astrocytes (p < .05,Fig. 2F), suggesting that reduced endocytosis of Jagged1 in in-termediate filament-deficient astrocytes might be the cause ofdecreased Notch signaling from GFAP�/�

Vim�/� astrocytes to neural stem/progenitor cells.To determine the efficiency of Notch signaling from

GFAP�/�Vim�/� astrocytes specifically to neural stem cells,we transfected adult mouse neural stem cells [29] with aNotch reporter [30, 31]. We found that the Notch signalingactivity in neural stem cells cocultured with GFAP�/�Vim�/�

compared to wild-type P2 astrocytes was reduced by 78%(p < .001, Fig. 2G). Next, we assessed Notch signaling in P4neurospheres, which contain both astrocytes and stem/progeni-tor cells, by quantifying the expression of the Notch intracel-lular domain (NICD) using an antibody specific for cleaved/active Notch. After 5 days of differentiation, GFAP�/�Vim�/�


neurospheres showed a 19% decrease in NICDpos cells com-pared to wild-type controls (62% 6 2.4% vs. 76% 6 0.1%NICDpos cells/well; p < .05).

Jagged1 Reverses the Increase in NeuronalDifferentiation Mediated by GFAP2/2Vim2/2

Astrocytes

To determine whether adding Jagged1 to the system would abro-gate the effect of GFAP�/�Vim�/� astrocytes on neuronal differ-entiation, we allowed P4 neurosphere cells to differentiate in thepresence of immobilized recombinant Jagged1-Fc or a controlprotein Fc [43]. Under control conditions, neuronal differentia-tion was greater in GFAP�/�Vim�/� neurosphere cells (p < .01,Fig. 3A). In the presence of Jagged1, however, neuronal differen-tiation of GFAP�/�Vim�/� neurospheres decreased to the levelcomparable to wild-type neurospheres (p < .05, Fig. 3A).

To investigate if this decrease was specific to Notch-medi-ated signaling, we added DAPT, a c-secretase inhibitor thatprevents cleavage and activation of the Notch receptor [44],to GFAP�/�Vim�/� neurosphere cells differentiating in thepresence of Jagged1. The addition of DAPT abrogated theJagged1-mediated decrease in neuronal differentiation ofGFAP�/�Vim�/� neurospheres (p < .05, Fig. 3B).

Increased Neurogenesis in the Hippocampal DentateGyrus of GFAP2/2Vim2/2 Mice

Since our in vitro experiments were performed with astrocytesobtained from the whole brain and the neurogenic as well asother properties differ between astrocytes from different brainregions, we determined the effects of GFAP and vimentinablation specifically on hippocampal neurogenesis. Astrocytesidentified by S100 expression in the dentate gyrus of the

hippocampus of wild-type and GFAP�/�Vim�/� mice were posi-tive for Jagged1 (Fig. 4A). To assess endogenous stem/progenitorcell division in the hippocampal dentate gyrus, we administereda single pulse of BrdU to GFAP�/�Vim�/� and wild-type mice.No differences were seen in the number of proliferating cells(BrdUpos) in the SGZ and GCL between GFAP�/�Vim�/� andwild-type mice 24 hours after BrdU administration (Fig. 4B).To assess different subpopulations of neural stem/progenitorcells, we combined BrdU labeling with antibodies against tran-scription factors Sox2 and Tbr2. Sox2 is present in rarely divid-ing neural stem cells with radial glia-like morphology andactive, horizontally aligned neural stem cells (both correspond-ing to type 1 cells), and early intermediate progenitor cells (type2a cells) [9, 45]. As Sox2 is also present in non-neurogenicastrocytes, we combined Sox2 antibodies with astrocyte markerS100 to be able to exclude the Sox2-positive non-neurogenicastrocytes. Tbr2 expression is found in early to late intermediatestage progenitor cells (type 2a and 2b cells) [46]. Twenty fourhours after a single BrdU injection, there was no difference inthe fraction of Sox2posS100neg cells or Tbr2pos cells amongBrdU labeled cells in the SGZ and GCL of GFAP�/�Vim�/�

and wild-type mice (Fig. 4B). In addition, the total cell numbersof Sox2posS100neg cells and Tbr2pos cells were similar betweenGFAP�/�Vim�/� and wild-type mice (Fig. 4C). This suggeststhat reduced Jagged1-mediated Notch signaling from GFAP�/�

Vim�/� astrocytes in the adult hippocampus does not affect neu-ral stem cell pool maintenance, proliferation, and lineage pro-gress into early intermediate progenitor cells.

To determine how the absence of GFAP and vimentinaffects the differentiation and survival of newly formed astro-cytes and neurons, we administered BrdU twice daily for aweek and assessed the numbers of the newly formed cells 6weeks after the first injection. We found that the two groups

Figure 2. GFAP�/�Vim�/� astrocytes show reduced Notch signaling. (A): In coculture with Notch reporter cells, GFAP�/�Vim�/� (G�/�V�/�)astrocytes showed less Notch signaling than wild-type (WT) astrocytes (n ¼ 4 in both groups). (B): Quantitative real-time polymerase chain reactionanalysis of expression levels of Jagged1 in G�/�V�/� astrocytes relative to those in WT astrocytes showed downregulation of Jagged1 (data from threeindependent experiments with n ¼ 3–5 mice per group in each experiment). (C): Western blot analysis showed less Jagged1 protein in G�/�V�/�

astrocytes than in WT astrocytes (three independent experiments). (D): G�/�V�/� (n ¼ 3) astrocytes showed reduced general endocytosis compared toWT (n ¼ 5) astrocytes. (E): G�/�V�/� astrocytes showed a major reduction in Notch ligand-mediated endocytosis of the NEC, a prerequisite forNotch signaling, compared to WT (n ¼ 4 in both groups). (F): Number of Jagged1pos vesicles was reduced in G�/�V�/� astrocytes compared to WTastrocytes (n ¼ 35 astrocytes from two mice in each group). (G): Adult mouse neural stem cells transfected with a Notch reporter showed reducedNotch signaling activity when cocultured with G�/�V�/� compared to WT astrocytes (n ¼ 8 per group). *, p < .05; **, p < .01; ***, p < .005; ****,p < .001 (two-tailed t-test). Abbreviations: MFI, mean fluorescence intensity; NEC, Notch extracellular domain; WT, wild type.


www.StemCells.com

of mice had similar numbers of newly formed astrocytes(BrdUposS100pos). However, GFAP�/�Vim�/� mice had 40%more BrdUpos cells, which implies an enhanced survival ofnewly formed cells in the dentate gyrus and 74% more newlyborn neurons (BrdUposNeuNpos) in the GCL (p < .05, Fig.4D–4F). Thus, in the absence of GFAP and vimentin, neuro-genesis in the hippocampal dentate gyrus is increased.

Next, we asked whether the absence of GFAP and vimentinwould affect the neurogenic response to injury. Entorhinal cortexlesion leads to hippocampal injury and promotes neurogenesis[47]. First, we assessed the total population of Sox2pos neuralstem/progenitor cells by counting Sox2pos cells negative for theastrocyte marker S100. We found that 4 days after entorhinalcortex lesion, GFAP�/�Vim�/� (n ¼ 3) and wild-type (n ¼ 6)mice had comparable number of Sox2posS100neg neural stem/progenitor cells in the SGZ of the denervated hippocampus (69.66 5.8 vs. 74.76 4.0 cells per section, respectively). To examinethe effect of entorhinal cortex lesion on cell proliferation, weadministered BrdU twice daily during the first week after lesion.Two weeks later, GFAP�/�Vim�/� mice had lower number ofBrdUpos cells in the SGZ and GCL on the lesioned side com-pared to wild-type mice (Fig. 4G; p < .05). To assess the fate ofthe dividing cells, we determined the percentage of astrocytesand neurons formed from BrdUpos cells. The total number ofnewly born neurons and astrocytes was comparable in GFAP�/�

Vim�/�and wild-type mice (Fig. 4G), however, the percentage ofnewly born neurons (BrdUposNeuNpos) was higher in GFAP�/�

Vim�/� than in wild-type mice (p < .05, Fig. 4H). Thus, whilethe lesion-triggered proliferative response in the hippocampuswas lower, the cell fate was more directed toward neuronal line-age in GFAP�/�Vim�/� compared to wild-type mice.

DISCUSSION

Astrocytes Regulate Neurogenesis Through Cell–CellContact with Neural Stem/Progenitor Cells

Astrocytes play an active role in adult neurogenesis through thesecretion of factors, of which several have been characterized

[4–7], while the astrocyte membrane-associated factors involvedin the regulation of neurogenesis have been far less studied [4].Although primary astrocytes and especially hippocampus-derived primary astrocytes direct neural stem/progenitor cell dif-ferentiation toward a neuronal fate in vitro [4], astrocytes in theneurogenic niches may exert several distinct and possibly evenopposing regulatory actions that affect neurogenesis. Thus, thenet effect of astrocytes on neurogenesis should reflect the inte-gration of all these effects. While factors secreted by astrocytesregulate neurogenesis positively [5–7], here we provide evi-dence that the negative control of neuronal differentiation ofneural stem/progenitor cells by astrocytes is mediated through acell–cell contact and that intermediate filament proteins GFAPand vimentin play an important role in this process. First, weshowed that compared to wild-type neurospheres, neurogenesisfrom GFAP�/�Vim�/� neurospheres is highly increased. To dis-criminate between the effects of GFAP and vimentin ablationon the intrinsic properties of neural stem/progenitor cells andthe effects on the niche, we cocultured neurosphere cells withprelabeled astrocytes. In this system, we found that simultane-ous ablation of GFAP and vimentin in astrocytes increased neu-ronal differentiation of neurosphere cells. Furthermore, weshowed that to exert this inhibitory effect on neurosphere differ-entiation, astrocytes need to be in direct contact with the neuro-sphere cells. These findings support the notion that wild-typeastrocytes inhibit neuronal differentiation of stem/progenitorcells through cell–cell contact and the ablation of GFAP andvimentin lifts this inhibition. This conclusion is also supportedby our previous report showing that GFAP�/�Vim�/� astrocytessupport neuronal differentiation from adult rat hippocampus-derived neural stem/progenitor cells [21]. This does not excludethe possibility that the absence of GFAP and vimentin alsodirectly affects the properties of neural stem/progenitor cells,which were shown to express vimentin and transiently alsoGFAP [48, 49]. This question merits further investigation.

Astrocytes Regulate Neurogenesis Through theNotch Pathway

Notch signaling is an evolutionarily conserved signaling path-way that controls many aspects of neuronal fate choice during

Figure 3. Immobilized Jagged1 reverses the increase in neuronal differentiation mediated by GFAP�/�Vim�/� astrocytes. (A): Differentiating WT (n¼ 4) neurospheres cultured in the presence of immobilized Jagged1 or a control protein (Fc) showed no differences in the percentage of b-III-tubulin-expressing (b-III-tubulinpos) neurons. In contrast, differentiating GFAP�/�Vim�/� (G�/�V�/�; n ¼ 4) neurospheres cultured in the presence of Fcshowed greater neuronal differentiation than G�/�V�/� neurospheres cultured with Jagged1 or WT neurospheres cultured with Jagged1 or Fc. Thus, im-mobilized Jagged1 abrogated the proneurogenic effect of G�/�V�/� astrocytes on neurosphere cells. (B): Addition of DAPT, a c-secretase inhibitor thatprevents cleavage of the Notch receptor, to differentiating G�/�V�/� neurospheres, abrogated the Jagged1-mediated decrease in neuronal differentiationof G�/�V�/� (n ¼ 4) neurospheres. N equals number of mice per experimental group. *, p < .05; **, p < .01 (ANOVA followed by Tukey honestly sig-nificant difference post hoc analysis). Abbreviations: DAPT, N-[N-(3,5-difluorophenylacetyl-l-alanyl)]-S-phenylglycine t-butylester; WT, wild type.


development and plays an essential role in controlling adultneurogenesis [50]. Astrocytes in SVZ were previously shownto express Jagged1, however, reports concerning the effects ofNotch signaling in SVZ on neurogenesis are controversial[12, 51]. Although cortical injury has been shown to be asso-ciated with increased Jagged1 expression in the ipsilateralSVZ [12], a direct link between altered Jagged1 expression orfunction in astrocytes and neurogenesis is lacking.

Our results show that in the absence of GFAP and vimen-tin, Jagged1 mRNA and protein levels as well as Notch sig-naling activity from astrocytes to reporter cells weredecreased. Importantly, adult mouse neural stem cells [29]

transfected with a Notch reporter showed reduced Notch sig-naling activity when cocultured with GFAP�/�Vim�/� astro-cytes compared to wild type. Together, these findings point tothe involvement of Notch pathway in the control of neuralstem/progenitor cell differentiation. Additional support forthis hypothesis comes from our data showing that comparedto wild-type, Notch ligand endocytosis, a prerequisite for bothNotch receptor and ligand activation [39, 42], was reduced by71% in GFAP�/�Vim�/� astrocytes. Furthermore, the GFAP�/�

Vim�/� astrocytes contained 36% fewer Jagged1pos vesiclesthan wild type, consistent with our previous reports that inter-mediate filaments have a role in astrocyte vesicle trafficking

Figure 4. Increased neurogenesis in the hippocampal dentate gyrus of GFAP�/�Vim�/� mice under basal conditions and after injury. (A):

S100pos astrocytes in the SGZ (arrows) and in the molecular cell layer (arrowheads) of the dentate gyrus show immunoreactivity for Jagged1.(B): After 24 hours of a single BrdU injection, GFAP�/�Vim�/� mice (G�/�V�/�; n ¼ 4–7) and wild-type mice (WT; n ¼ 4–6) had comparablenumber of dividing cells (BrdUpos) and subpopulations of dividing cells (BrdUposSox2posS100neg cells and BrdUposTbr2pos cells) in SGZ andGCL (SGZþGCL). (C): The total numbers of Sox2pos and Tbr2pos cells were comparable between genotypes (n ¼ 4–5 per group). (D): At 6weeks, G�/�V�/� mice had more BrdUpos cells in GCL and more BrdUposNeuNpos cells in SGZþGCL and in the GCL alone than WT mice (n¼ 12 per group). There was no difference in the number of BrdUposS100pos cells. (E): SGZ and GCL in the dentate gyrus visualized with anti-bodies against BrdU, NeuN, and S100 (arrowhead depicts BrdUpos cell). (F): Images of BrdUposNeuNpos and BrdUposS100pos cells in the dentategyrus of the hippocampus. (G): After 2 weeks of entorhinal cortex lesion, G�/�V�/� mice showed lower number of BrdUpos cells compared toWT mice (n ¼ 7 per group), but there was no difference between genotypes in the number of BrdUposNeuNpos cells (n ¼ 4 per group) or BrdU-posS100pos cells (n ¼ 5 per group). (H): G�/�V�/� mice showed a higher percentage of BrdUposNeuNpos cells in SGZþGCL than WT mice (n ¼4 per group). There was no difference in the percentage of BrdUposS100pos cells (n ¼ 5 per group). Scale bars ¼ 50 lm (A and E), 10 lm (F).*, p < .05; **, p < .01 (two-tailed t-test). Abbreviations: BrdU, 5-bromo-2-deoxyuridine; GCL, granule cell layer; SGZ, subgranular zone; WT,wild type.


www.StemCells.com

dynamics [28, 40]. Thus, it is conceivable that the activity ofNotch ligands in GFAP�/�Vim�/� astrocytes is reduced, eventhough the amount of Jagged1 found at the cell membrane ofastrocytes was not altered. Immobilized Jagged1, which is usedto activate Notch signaling [52], abolished the increased neuro-nal differentiation mediated by GFAP�/�Vim�/� astrocytes. Incontrast, the addition of a c-secretase inhibitor DAPT, whichprevents cleavage of the Notch receptor and inhibits Notch sig-naling, abrogated the effect of immobilized Jagged1 on neuronaldifferentiation of neurosphere cells. Together, these data implythat astrocytes exert their control of Notch signaling in the sig-nal receiving cells through the endocytosis of Jagged1 and thisprocess is dependent on the normal function of cytoplasmic in-termediate filaments in astrocytes. Our results add support to theclaims that Notch signaling inhibits neuronal differentiation ofneural stem/progenitor cells.

Previously we showed that ablation of GFAP and vimen-tin attenuates reactive gliosis and creates a more favorableenvironment for the survival of neural grafts and neural stemcells as well as promotes axonal and synaptic regeneration[20, 21, 24, 53]. Here, we demonstrate that ablation of GFAPand vimentin increases hippocampal neurogenesis in unchal-lenged mice and neuronal cell fate determination after injury.However, we did not detect any changes in the population oftype 1 and 2 neural stem/progenitor cells as determined bythe number of cells expressing Tbr2 or Sox2 alone or in com-bination with BrdU incorporation. As judged from immuno-staining with antibodies against Jagged1 (Fig. 4A), S100pos

astrocytes are not the only cell type expressing Jagged1 in theSGZ. Jagged1 may also be present in neural stem cells or in-termediate progenitor cells, and thus affect Notch signaling inthe niche in an astrocyte-independent manner. Notch receptoractivation promotes the survival of neural stem cells in vitroand transient administration of Notch ligands to the brain ofadult rats increased the numbers of newly generated precursorcells [13]. After inducible conditional inactivation of Notchsignaling in the brain of adult mice, neural stem cells differ-entiated into transit-amplifying precursors resulting in anincreased number of immature neurons 1–3 weeks after Notchinactivation and followed by a dramatic loss of neurogenesis2–3 months later [10, 16]. In contrast, in the GFAP�/�Vim�/�

mice, hippocampal neurogenesis is still detectable andincreased even in old age (18 months) [54]. In addition,GFAP�/�Vim�/� mice and wild-type mice used in this studyshowed comparable numbers of Sox2posS100neg neural stem/progenitor cells in the hippocampus under both basal condi-tions and after entorhinal cortex lesion, suggesting that thehippocampal neural stem cell pool is not depleted in GFAP�/�

Vim�/� mice. Thus, the reduced inhibition of Notch signalingin the hippocampal neurogenic niche of the GFAP�/�Vim�/�

mice has much milder effects than its complete inactivation.Notably, the choice of gene promoter (Gfap and Glast vs. Nes-tin) to induce the Notch gene inactivation in the neurogenic

niche seems to affect the dynamics in neurogenesis differen-tially [10, 14, 15]. These distinct effects of Notch inactivationpoint to the complexity of the interplay between cell types andtheir role in Notch signaling in the hippocampus. In contrast toconditional Notch1 inactivation, which essentially abolisheslong-term neurogenesis, the compensatory mechanisms inresponse to constitutive absence of GFAP and vimentin mayaccount for the milder effects on astrocyte-mediated Notch sig-naling and neurogenesis observed in GFAP�/�Vim�/� mice.Notch signaling is also present in newly born neurons in GCLof the adult hippocampus where it regulates dendritic morphol-ogy [15, 55]. This may play a role in the integration of the newneurons into the neuronal networks and ultimately affect theirsurvival. In addition, the ablation of GFAP and vimentin con-ceivably has multiple effects on the neurogenic niche, includingthe endogenous properties of neural stem cells and altered distri-bution or secretion of soluble neurogenic factors such as Wnt3[21] that may promote neuronal differentiation or survival.

SUMMARY

We conclude that astrocytes inhibit neuronal differentiation ofneural stem/progenitor cells through a cell–cell contact. Endo-cytosis of Notch ligand Jagged1 by astrocytes plays an essen-tial role in this process and is dependent on normal functionof the intermediate filament proteins GFAP and vimentin inthese cells.

ACKNOWLEDGMENTS

This work was supported by the Swedish Medical Research Coun-cil (project 11548), AFA Research Foundation, ALF G€oteborg(project 11392), Sten A. Olsson Foundation for Research and Cul-ture, S€oderberg Foundations, Hj€arnfonden, the Swedish StrokeFoundation, the Swedish Society for Medical Research, the FreeMason Foundation, Aml€ov’s Foundation, E. Jacobson’s DonationFund, NanoNet COST Action (BM1002), the EU FP 7 ProgramEduGlia (237956 to M.P. and R.Z.), the EU FP 7 Program Target-BraIn (279017 to M.P.), Trygg-Hansa, Academy of Finland(C.S.), Turku Graduate School for Biomedical Sciences and Acad-emy of Finland (M. Sj.), and by the Research Agency of Slovenia(P30310, J30031, J30133, J34051, J34146, J33632 to R.Z).

DISCLOSURE OF POTENTIAL

CONFLICTS OF INTEREST

The authors indicate no potential conflicts of interest.

REFERENCES

1 Buffo A, Rite I, Tripathi P et al. Origin and progeny of reactive glio-sis: A source of multipotent cells in the injured brain. Proc Natl AcadSci USA 2008;105:3581–3586.

2 Laywell ED, Rakic P, Kukekov VG et al. Identification of a multipo-tent astrocytic stem cell in the immature and adult mouse brain. ProcNatl Acad Sci USA 2000;97:13883–13888.

3 Seri B, Garcia-Verdugo JM, McEwen BS et al. Astrocytes give rise tonew neurons in the adult mammalian hippocampus. J Neurosci 2001;21:7153–7160.

4 Song H, Stevens CF, Gage FH. Astroglia induce neurogenesis fromadult neural stem cells. Nature 2002;417:39–44.

5 Lie DC, Colamarino SA, Song HJ et al. Wnt signalling regulates adulthippocampal neurogenesis. Nature 2005;437:1370–1375.

6 Barkho BZ, Song H, Aimone JB et al. Identification of astrocyte-expressed factors that modulate neural stem/progenitor cell differentia-tion. Stem Cells Dev 2006;15:407–421.

7 Lu Z, Kipnis J. Thrombospondin 1—A key astrocyte-derived neuro-genic factor. FASEB J 2010;24:1925–1934.

8 Stump G, Durrer A, Klein AL et al. Notch1 and its ligands Delta-likeand Jagged are expressed and active in distinct cell populations in thepostnatal mouse brain. Mech Dev 2002;114:153–159.

9 Lugert S, Basak O, Knuckles P et al. Quiescent and active hippocam-pal neural stem cells with distinct morphologies respond selectively tophysiological and pathological stimuli and aging. Cell Stem cell 2010;6:445–456.


10 Ehm O, Goritz C, Covic M et al. RBPJkappa-dependent signaling isessential for long-term maintenance of neural stem cells in the adulthippocampus. J Neurosci 2010;30:13794–13807.

11 Chambers CB, Peng Y, Nguyen H et al. Spatiotemporal selectivity ofresponse to Notch1 signals in mammalian forebrain precursors. Devel-opment 2001;128:689–702.

12 Givogri MI, de Planell M, Galbiati F et al. Notch signaling in astro-cytes and neuroblasts of the adult subventricular zone in health and af-ter cortical injury. Dev Neurosci 2006;28:81–91.

13 Androutsellis-Theotokis A, Leker RR, Soldner F et al. Notch signal-ling regulates stem cell numbers in vitro and in vivo. Nature 2006;442:823–826.

14 Ables JL, Decarolis NA, Johnson MA et al. Notch1 is required formaintenance of the reservoir of adult hippocampal stem cells. J Neu-rosci 2010;30:10484–10492.

15 Breunig JJ, Silbereis J, Vaccarino FM et al. Notch regulates cell fateand dendrite morphology of newborn neurons in the postnatal dentategyrus. Proc Natl Acad Sci USA 2007;104:20558–20563.

16 Imayoshi I, Sakamoto M, Yamaguchi M et al. Essential roles of Notchsignaling in maintenance of neural stem cells in developing and adultbrains. J Neurosci 2010;30:3489–3498.

17 Wang X, Mao X, Xie L et al. Involvement of Notch1 signaling inneurogenesis in the subventricular zone of normal and ischemic ratbrain in vivo. J Cereb Blood Flow Metab 2009;29:1644–1654.

18 Eliasson C, Sahlgren C, Berthold CH et al. Intermediate filament pro-tein partnership in astrocytes. J Biol Chem 1999;274:23996–24006.

19 Pekny M, Johansson CB, Eliasson C et al. Abnormal reaction to cen-tral nervous system injury in mice lacking glial fibrillary acidic pro-tein and vimentin. J Cell Biol 1999;145:503–514.

20 Kinouchi R, Takeda M, Yang L et al. Robust neural integration fromretinal transplants in mice deficient in GFAP and vimentin. Nat Neu-rosci 2003;6:863–868.

21 Widestrand A, Faijerson J, Wilhelmsson U et al. Increased neurogene-sis and astrogenesis from neural progenitor cells grafted in the hippo-campus of GFAP�/� Vim�/� mice. Stem Cells 2007;25:2619–2627.

22 Cho KS, Yang L, Lu B et al. Re-establishing the regenerative poten-tial of central nervous system axons in postnatal mice. J Cell Sci2005;118:863–872.

23 Menet V, Prieto M, Privat A et al. Axonal plasticity and functional re-covery after spinal cord injury in mice deficient in both glial fibrillaryacidic protein and vimentin genes. Proc Natl Acad Sci USA 2003;100:8999–9004.

24 Wilhelmsson U, Li L, Pekna M et al. Absence of glial fibrillary acidicprotein and vimentin prevents hypertrophy of astrocytic processes andimproves post-traumatic regeneration. J Neurosci 2004;24:5016–5021.

25 Colucci-Guyon E, Portier MM, Dunia I et al. Mice lacking vimentindevelop and reproduce without an obvious phenotype. Cell 1994;79:679–694.

26 Pekny M, Leveen P, Pekna M et al. Mice lacking glial fibrillary acidicprotein display astrocytes devoid of intermediate filaments but developand reproduce normally. EMBO J 1995;14:1590–1598.

27 Pekny M, Eliasson C, Chien CL et al. GFAP-deficient astrocytes arecapable of stellation in vitro when cocultured with neurons and exhibita reduced amount of intermediate filaments and an increased cell satu-ration density. Exp Cell Res 1998;239:332–343.

28 Potokar M, Kreft M, Li L et al. Cytoskeleton and vesicle mobility inastrocytes. Traffic 2007;8:12–20.

29 Ray J, Gage FH. Differential properties of adult rat and mouse brain-derived neural stem/progenitor cells. Mol Cell Neurosci 2006;31:560–573.

30 Jarriault S, Brou C, Logeat F et al. Signalling downstream of activatedmammalian Notch. Nature 1995;377:355–358.

31 Kato H, Taniguchi Y, Kurooka H et al. Involvement of RBP-J in bio-logical functions of mouse Notch1 and its derivatives. Development1997;124:4133–4141.

32 Rosenholm JM, Meinander A, Peuhu E et al. Targeting of poroushybrid silica nanoparticles to cancer cells. Acs Nano 2009;3:197–206.

33 Stenovec M, Kreft M, Grilc S et al. Ca2þ-dependent mobility ofvesicles capturing anti-VGLUT1 antibodies. Exp Cell Res 2007;313:3809–3818.

34 Andersen CL, Jensen JL, Orntoft TF. Normalization of real-time quan-titative reverse transcription-PCR data: A model-based variance esti-mation approach to identify genes suited for normalization, applied tobladder and colon cancer data sets. Cancer Res 2004;64:5245–5250.

35 Wilhelmsson U, Bushong EA, Price DL et al. Redefining the conceptof reactive astrocytes as cells that remain within their unique domainsupon reaction to injury. Proc Natl Acad Sci USA 2006;103:17513–17518.

36 Kempermann G, Gast D, Kronenberg G et al. Early determination andlong-term persistence of adult-generated new neurons in the hippo-campus of mice. Development 2003;130:391–399.

37 Oh J, McCloskey MA, Blong CC et al. Astrocyte-derived interleukin-6 promotes specific neuronal differentiation of neural progenitor cellsfrom adult hippocampus. J Neurosci Res 2010;88:2798–2809.

38 Ninkovic J, Gotz M. Signaling in adult neurogenesis: From stem cellniche to neuronal networks. Curr Opin Neurobiol 2007;17:338–344.

39 Kopan R, Ilagan MX. The canonical Notch signaling pathway:Unfolding the activation mechanism. Cell 2009;137:216–233.

40 Potokar M, Stenovec M, Gabrijel M et al. Intermediate filamentsattenuate stimulation-dependent mobility of endosomes/lysosomes inastrocytes. Glia 2010;58:1208–1219.

41 Vardjan N, Gabrijel M, Potokar M et al. IFN-gamma-induced increasein the mobility of MHC class II compartments in astrocytes dependson intermediate filaments. J Neuroinflammation 2012;9:144.

42 Fortini ME, Bilder D. Endocytic regulation of Notch signaling. CurrOpin Genet Dev 2009;19:323–328.

43 Sahlgren C, Gustafsson MV, Jin S et al. Notch signaling mediates hy-poxia-induced tumor cell migration and invasion. Proc Natl Acad SciUSA 2008;105:6392–6397.

44 Dovey HF, John V, Anderson JP et al. Functional gamma-secretaseinhibitors reduce beta-amyloid peptide levels in brain. J Neurochem2001;76:173–181.

45 Steiner B, Klempin F, Wang L et al. Type-2 cells as link betweenglial and neuronal lineage in adult hippocampal neurogenesis. Glia2006;54:805–814.

46 Hodge RD, Kowalczyk TD, Wolf SA et al. Intermediate progenitorsin adult hippocampal neurogenesis: Tbr2 expression and coordinateregulation of neuronal output. J Neurosci 2008;28:3707–3717.

47 Cameron HA, McEwen BS, Gould E. Regulation of adult neurogene-sis by excitatory input and NMDA receptor activation in the dentategyrus. J Neurosci 1995;15:4687–4692.

48 Doetsch F, Caille I, Lim DA et al. Subventricular zone astrocytes areneural stem cells in the adult mammalian brain. Cell 1999;97:703–716.

49 Imura T, Kornblum HI, Sofroniew MV. The predominant neural stemcell isolated from postnatal and adult forebrain but not early embry-onic forebrain expresses GFAP. J Neurosci 2003;23:2824–2832.

50 Ables JL, Breunig JJ, Eisch AJ et al. Not(ch) just development: Notchsignalling in the adult brain. Nat Rev Neurosci 2011;12:269–283.

51 Wang L, Chopp M, Zhang RL et al. The Notch pathway mediatesexpansion of a progenitor pool and neuronal differentiation in adultneural progenitor cells after stroke. Neuroscience 2009;158:1356–1363.

52 Beckstead BL, Santosa DM, Giachelli CM. Mimicking cell-cell inter-actions at the biomaterial–cell interface for control of stem cell differ-entiation. J Biomed Mater Res A 2006;79:94–103.

53 Pekny M, Pekna M, Wilhelmsson U et al. Astroglia sitting at the con-trols? Trends Neurosci 2004;27:243–244.

54 Larsson A, Wilhelmsson U, Pekna M et al. Increased cell proliferationand neurogenesis in the hippocampal dentate gyrus of old GFAP�/�Vim�/� mice. Neurochem Res 2004;29:2069–2073.

55 Brandt MD, Maass A, Kempermann G et al. Physical exerciseincreases Notch activity, proliferation and cell cycle exit of type-3progenitor cells in adult hippocampal neurogenesis. Eur J Neurosci2010;32:1256–1264.


www.StemCells.com

astrocytes negatively regulate neurogenesis through the jagged1-mediated notch pathway

Documents