Proliferation of Progenitor Cells inthe Adult Rat Brain Correlates Withthe Presence of Vimentin-Expressing

AstrocytesGERARD ALONSO*

CNRS-UMR 5101, CCIPE, Montpellier, France

KEY WORDS glial fibrillary acidic protein; hippocampus; subventricular zone oflateral ventricle; aging; glucocorticoids

ABSTRACT It is well established that proliferation of progenitor cells persistswithin the hippocampal dentate gyrus (DG) and the subventricular zone of the lateralventricle (SVZ) in the adult brain. The aim of the present study was to determinewhether the rate of cell proliferation within these germinative zones could be correlatedto the occurrence of a particular glial environment. The cell proliferation marker bro-modeoxyuridine (BrdU) was administrated to rats under different physiological andexperimental conditions known to modify the rate of progenitor cell proliferation. Withinboth germinative zones, BrdU-labeled nuclei were associated with cell bodies immuno-stained for the neuronal marker polysialylated neural cell adhesion molecule, but not forthe glial markers glial fibrillary acidic protein (GFAP) or vimentin (VIM). In all the ratsexamined, however, proliferating (BrdU-labeled) cells always exhibited close relation-ships with immature-like astrocytes that expressed both GFAP and VIM. There was adramatic decrease of cell proliferation in the DG from both the aged rats and thecorticosterone-treated adult rats that was correlated with a decreased expression ofvimentin by the astrocytes present in this region. In contrast, both cell proliferation andvimentin expression were only slightly affected in the SVZ from these two treatmentgroups. Conversely, after either adrenalectomy or a surgical lesion through the lateralhippocampus, the increase in cell proliferation observed in the DG was correlated to theoccurrence of an increased number of GFAP and VIM double immunostained structuresin these regions. All together, these data suggest that immature-like astrocytes presentin the germinative zones may provide a microenvironment involved in sustaining theproliferation of progenitor cells. GLIA 34:253–266, 2001. © 2001 Wiley-Liss, Inc.

INTRODUCTION

During many years, it has been considered that theneurons and glial cells that compose the central ner-vous system (CNS) arise from progenitor cells of theventricular zone, which actively proliferate during lateembryonic and early postnatal period and then stopdividing (Cowan, 1979). Recently, however, evidencehas been provided that in adult rodents as well as inmost mammals including man, active neurogenesispersists within specific germinative zones. These zonesmainly include the subventricular zone of the lateralventricle (SVZ) (Lois and Alvarez-Buylla, 1993; Mors-

head et al., 1994; Goldman, 1995; Hauke et al., 1995;Alvarez-Buylla et al., 2000) and the hippocampal den-tate gyrus (DG) (Altman and Das, 1965; Kuhn et al.,1996; Gould et al., 2000; Kempermann and Gage,2000). To date, it is not well understood why prolifer-ation of progenitor cells decreases after brain matura-tion and why neurogenesis is maintained in specific

*Correspondence to: Gerard Alonso, UMR 5101, Biologie des Neurones Endo-crines, CCIPE, 141 rue de la Cardonille, 34094 Montpellier cedex, France.E-mail: alonso@bacchus.montp.inserm.fr

Received 13 November 2001; Accepted 15 March 2001

Published online 00 Month 2001.

GLIA 34:253–266 (2001)

© 2001 Wiley-Liss, Inc.

regions of the adult brain. It is possible that in theadult brain precursor cell proliferation is restricted bythe local production of specific proliferating factorsthat, in contrast, are present in large quantitiesthroughout the periventricular regions in the develop-ing brain. Interestingly, the postnatal period duringwhich the proliferation of progenitor cells progressivelydecreases throughout the ventricular zones corre-sponds to the period of maturation of astrocytes, whenthe so-called radial glial cells migrate outside from theperiventricular area and undergo major modificationsof their morphology and phenotype (Schmechel andRakic, 1979; Pixley and De Vellis, 1984; Sancho-Telloet al., 1995). Although the roles played by radial glialcells in the developing brain are not yet fully under-stood, there is evidence that they play an importantrole in the different steps of neurogenesis, includingthe survival, migration, and differentiation of neuronalprecursors (Rakic, 1981, 1991), and that these func-tions are lost by mature astrocytes (Vernadakis, 1988).

Interestingly, the two germinative zones in which pro-liferation of neuronal progenitors continues throughoutadulthood (i.e., the SVZ and the hippocampal DG) havebeen shown to contain specific astrocytes that exhibitmorphological and phenotypic patterns of immatureastrocytes. Indeed, in contrast with the surroundingforebrain regions, both the SVZ and hippocampal DGregions contain elongated astrocytes (i.e., the morphol-ogy of radial glial cells) that continue to express highlevels of vimentin, the main intermediate filamentcomponent of immature astrocytes (Doetsch et al.,1997; Peretto et al., 1997; Seki and Arai, 1999). Arecent in vitro study has shown that the proliferation ofadult subventricular zone progenitors is stimulated bydirect cell-cell contact between these progenitors andthe cultured astrocytes (Lim and Alvarez-Buylla,1999). A possible role for the immature-like astrocytespresent in the germinative zones of the adult braincould, therefore, be to provide specific factors locallythat support the proliferation of progenitor cells. Totest this hypothesis, the relationships between prolif-erating progenitor cells and astrocytes were examinedin different germinative zones of the adult rat brainunder various physiological and experimental condi-tions known to modify this cell proliferation.

MATERIALS AND METHODSAnimals

Male Sprague-Dawley rats (Iffa-Credo, l’Arbresle,France) were divided into three groups, including adult(2–3 month old; n 5 6) and aged (12–14 month old; n 55) untreated animals; adult animals given treatmentsto modify their circulating corticosteroid levels or con-trol treatments (n 5 15); and adult animals given alateral hippocampus lesion and daily injections of ei-ther corticosterone or oil vehicle (n 5 10). All the ex-periments were performed according to the principles

of laboratory animal care published by the French Eth-ical Committee.

Experimental Modifications of CirculatingCorticosteroid Levels

Three groups of animals were used: adult controlrats receiving a daily subcutaneous injection of sesameoil (0.2 ml; n 5 5); rats receiving a daily subcutaneousinjection of corticosterone (10 mg/kg in 0.2 ml sesameoil; n 5 5); and adrenalectomized rats receiving a dailysubcutaneous injection of sesame oil (0.2 ml; n 5 5).Injections were performed every day at 18:00–19:00p.m. for 15 days. Adrenalectomized rats received drink-ing water containing 0.1% NaCl.

Surgical Lesion

Two groups of rats were used, including untreatedadult rats (2–3 month old; n 5 5) and corticosterone-treated adult rats (n 5 5). As described above, theyreceived a daily injection of either sesame oil (untreat-ed) or of corticosterone (10 mg/kg) during the 7 dayspreceding and the 7 days following the lesion. Afterdeep anesthesia with Equithesin (3 ml/kg), animalswere fixed in a stereotaxic device. A metallic rectangu-lar knife 2 mm wide and 18 mm long was loweredunilaterally into the forebrain according to the stereo-taxic atlas of Paxinos and Watson (1982): the knifeblade was fixed parallel to the midline and placed 4 mmlateral to it, 3 mm posterior to bregma, and lowered to6 mm below the surface of the skull. This protocol wasfound to produce a vertical lesion that extends throughthe cortical layers and the lateral portions (CA2 andCA3 regions) of the hippocampus (Fig. 1C). Animalswere sacrificed 7 days after the lesion.

Tissue Fixation

After deep anesthesia with Equithesin, animals wereperfused through the ascending aorta with phosphate-buffered saline (PBS), pH 7.4, followed by 500 ml offixative composed of 4% paraformaldehyde 0.1 M phos-phate buffer, pH 7.4. The forebrain was dissected andfixed by immersion in the same fixative for 2 to 4 days.It was then cut frontally with a vibratome into 40–50mm thick sections. These were carefully rinsed in PBSand subsequently treated for double fluorescence im-munostaining.

Immunocytochemistry

Vibratome sections were incubated with two primaryantibodies, including a rabbit IgG polyclonal antibodyagainst glial fibrillary acidic protein (GFAP; diluted1:1,000; Dakko, Denmark) and a mouse IgG monoclo-

254 ALONSO

Fig. 1. Immunostaining of bromodeoxyuridine (BrdU) and glialmarkers in the hippocampal dentate gyrus of adult (A–C), aged (D–F),corticosterone-treated (G–I), and adrenalectomized (J–L) rats. Singleperoxidase immunostaining of BrdU shows that, as compared withthe adult rat (A), the number of BrdU-IS nuclei detected along thegranular layer of the hippocampal DG is dramatically decreased inthe aged (D) and corticosterone-treated (G) rats, whereas it is in-creased in the adrenalectomized rat (J). Double fluorescence immu-nostaining of glial fibrillary acidic protein (GFAP; B, E, H, K) andvimentin (VIM; C, F, I, L) shows that GFAP and VIM double immu-

nostained structures are present throughout the granular layer of thehippocampal DG of adult (B and C) and to a higher extent adrenalec-tomized rats (K and L) but are absent from this region in both aged (Eand F) and corticosterone-treated rats (H and I). Note that theseGFAP and VIM double immunostained astrocytes exhibit elongatedprocesses crossing through the granular layer of the hippocampal DG.The dotted lines in A–C indicate the limits of the granular layer of thedentate gyrus. Scale bars 5 150 mm (in J and L apply for peroxidaseand fluorescence immunostaining, respectively).

255PROLIFERATION OF PROGENITOR CELLS

nal antibody against vimentin (VIM; diluted 1:1,000;Sigma). The vibratome sections were incubated for 48 hat 4°C with the two primary antibodies diluted in PBScontaining 1% normal goat serum and 0.1% TritonX-100. After rinsing in PBS, they were incubated for2 h at 4°C with the corresponding secondary antibod-ies, including an antimouse IgG conjugated with Cy3and an antirabbit IgG or an antimouse IgM conjugatedwith fluorescein isothiocyanate (FITC; Jackson labora-tories). The secondary antibodies were diluted 1:500 inPBS containing 1% normal goat serum and 0.1% TritonX-100.

After careful rinsing, sections were mounted inMowiol (Calbiochem, La Jolla, CA) and observed undera Biorad MRC 1024 confocal laser scanning microscopeequipped with a krypton/argon mixed gas laser. Twolaser lines emitting at 488 nm and 568 nm were usedfor exciting the fluorescein- and Cy3-conjugated sec-ondary antibodies, respectively. The background noiseof each confocal image was reduced by averaging fiveimage inputs. The organization of the immunostainedstructures was studied on single confocal images of 1 to2 mm thick. Unaltered digitalized images were trans-ferred to a PC-type computer and Powerpoint (Mi-crosoft) was used to prepare and print final figures.

Some sections were treated for the quantification ofthe colocalization between GFAP and VIM immuno-staining within specific forebrain regions. For this, 1mm thick confocal images were obtained by collectingsequentially the green (GFAP) and red (VIM) imagesthrough superficial layers of vibratome sections (inwhich the penetration of both antibodies is optimum).Such paired confocal images were collected from atleast three different sections per rat for at least threerats of the different groups examined. Colocalizationanalysis was performed by using the Lasersharp soft-ware (Biorad) within a fixed area of 100 3 100 mm sidecentered on the different forebrain regions considered.For each area analyzed, a colocalization coefficient wasgenerated between 0 (no colocalization) and 1 (com-plete colocalization). Data were statistically comparedusing the nonparametric test of Mann and Whitney.

BrdU Labeling and Detection

BrdU (Sigma) was administered intraperitoneallytwice a day during the 2 days preceding their perfusion(i.e., days 13 and 14 postadrenalectomy or the start ofthe corticosterone treatment, or days 6 and 7 postsur-gical lesion). Six hours after the last BrdU administra-tion, animals were perfused as described above with4% paraformaldehyde. Brains were postfixed in 4%paraformaldehyde for 12 h and cut frontally with avibratome into 40 to 50 mm thick sections. Sectionswere then treated for either single peroxidase or doublefluorescence immunostaining.

Peroxidase immunostaining

Vibratome sections were incubated with 2 N HCL for30 min at room temperature, carefully rinsed, and in-cubated for 48 h at 4°C with an anti-BrdU antibody(monoclonal mouse IgG; Novocastra Lab, Newcastle,U.K.; diluted 1:100); for 12 h at 4°C with a peroxidase-labeled Fab fragment of goat IgG antimouse IgG (Bio-sys, Compiegne, France; diluted 1:1,000); and with0.1% 3,39-diaminobenzidine diluted in 0.05 M Trisbuffer, pH 7.3, in the presence of 0.2% H2O2. The pri-mary and secondary antibodies were diluted in PBScontaining 0.1% Triton X100, 1% BSA, and 1% normalgoat serum. Immunostained sections were mounted inpermount and observed under a light microscope.

For each of the different groups of rats, series ofsections passing through the SVZ (0.2–0.8 mm caudalto bregma) or the hippocampus (2–4 mm caudal tobregma) were selected and treated for the semiquanti-tative analysis of BrdU-immunostained nuclei. In or-der to avoid any bias, the identity of the sections wasmasked during the quantitative analysis. The numberof BrdU-labeled nuclei was counted within squarefields of 150 mm side (final magnification 3 450) cen-tered on the SVZ or the hippocampal DG. For eachstructure, two to six fields per section were analyzed inat least four sections per animal. Within each fieldanalyzed, the numerical density was expressed asnumber of BrdU-labeled nuclei/mm2.

Double fluorescence immunostaining

Vibratome sections were incubated with a rabbit IgGpolyclonal antibody against GFAP (diluted 1:1,000;Dakko) or VIM (diluted 1:1,000; kindly provided by Dr.A.M. Hill, Paris, France) or a mouse IgM monoclonalantibody against polysialylated neural cell adhesionmolecule (PSA-NCAM; diluted 1:1,000; kindly providedby Dr. T. Seki, Tokyo, Japan). The sections were thentreated with 2 N HCL for 30 min at room temperature,carefully rinsed, and incubated for 48 h at 4°C with themouse monoclonal antibody anti-BrdU. The two pri-mary antibodies (anti-BrdU and anti-GFAP, -VIM, orPSA-NCAM) were then visualized with correspondingsecondary antibodies conjugated with Cy3 and FITCand the double-labeled sections were examined underthe confocal microscope.

For the untreated and adrenalectomized adult rats,series of sections passing through the hippocampal DGand the SVZ were analyzed to determine the phenotypeof the BrdU-labeled cells. The 3 40 objective was usedto collect confocal images of sections double immuno-stained for BrdU on the one hand, and for GFAP, VIM,or PSA-NCAM on the other hand.

The specificity of the antibody against PSA-NCAMhas been previously described and documented (Sekiand Arai, 1993) and the specificity of commercial anti-bodies has been guaranteed by the manufacturers. Ad-ditional controls consisted of omitting the primary an-

256 ALONSO

tibodies and applying the secondary antibodies alone;applying each primary antibody sequentially and thenreacting them with an inappropriate secondary anti-body; and exciting each fluorochrome by the inappro-priate laser line. This allowed us to confirm that thetwo secondary antibodies used in double immunostain-ing experiments did not induce artifactual fluorescentlabeling and that there was no overlap of the emissionspectra of the two fluorochromes.

RESULTS

Single peroxidase immunostaining of BrdU was firstused to evaluate the rate of cell proliferation within thehippocampal dentate gyrus and the subventricularzone of the lateral ventricle under different physiolog-ical and experimental conditions. Under these differentconditions, double fluorescence was then used to char-acterize the coexpression of GFAP and VIM in astro-cytes present within these forebrain regions. Finally,double immunostaining for both BrdU and either neu-ronal or glial markers was used to characterize thenature of proliferative cells and their anatomical rela-tionships with astrocytes within these regions.

Single Immunostaining for BrdU

In all the sections examined, BrdU immunostainingwas always associated with the nucleus of cells distrib-uted through different forebrain regions. In agreementwith previous descriptions, different regions of theadult rat forebrain could be categorized by the varyingdensities of BrdU-labeled nuclei. These regions in-cluded regions exhibiting high numerical densities, in-cluding the subventricular zone of the lateral ventricle,and the rostral migratory stream (RMS) extendingfrontally from the SVZ toward the olfactory bulb; re-gions exhibiting moderate numerical densities, includ-ing the subgranular zone of the hippocampal dentategyrus and most white matter regions such as the cor-pus callosum and the fimbria; and regions exhibitinglow numerical densities, including all the other graymatter regions. In this study, the observations werefocused on the two germinative zones in which theproduction of new neurons is well documented, i.e., theSVZ and hippocampal DG (Alvarez-Buylla et al., 2000;Kempermann and Gage, 2000).

Untreated rats of different ages

The densities of BrdU-labeled nuclei detected theSVZ and hippocampal DG was found to decrease inaged rats compared to the adult rats. In all the animalsexamined, however, the importance of such a decreasewas found to vary markedly with the region considered:it was important within the hippocampal DG, but only

slight within the SVZ and the RMS (Fig. 1A and D; Fig.2A and D).

Adrenalectomized and corticosterone-treatedrats

As compared with controls, the number of BrdU-labeled nuclei detected within the hippocampal DGwas found to increase in adrenalectomized rats,whereas it was markedly decreased in corticosterone-treated rats (Fig. 1G and J). By contrast, the number ofBrdU-labeled nuclei detected in the SVZ was found toremain unchanged in both groups of rats (Fig. 2G).

Lesioned rats

In the lesioned animals, the observations were re-stricted to the hippocampus. On the side contralateralto the lesion, the number of BdrU-labeled nuclei andtheir distribution throughout the hippocampus wassimilar to that observed in unlesioned rats of the sameage. On the lesioned side, numerous BrdU-labeled nu-clei were located in the area closely surrounding thesurgical cut. Moreover, as compared to the oppositeintact side, an increased number of BrdU-labeled nu-clei was detected along the DG granular layer adjacentto and at a distance from the lesion (Fig. 3A and B).

Double Immunostaining for Glial Markers

Untreated rats of different ages

In the two groups of age considered, intense GFAPimmunostaining was associated with the processesand, to a lesser extent, the cell bodies of a large numberof astrocytes located throughout the different forebrainregions examined. By contrast, VIM immunostainingwas associated with a limited number of structures,including the ependymocytes bordering the ventriclesand some astrocyte-like structures located in specificregions. In adult rats, GFAP and VIM double immuno-stained structures were essentially detected within thesuperficial layers of the brain; the circumventricularorgans, including the subfornical organ, and the hypo-thalamic median eminence; the white matter regions(including the corpus callosum and the fimbria); andthe germinative zones, including the hippocampal DGand the SVZ and RMS (Fig. 1B and C; Fig 2B and C).Within all these regions, such double immunostainedstructures frequently exhibited an elongated morphol-ogy characteristic of immature astrocytes (i.e. a cellbody connected to a restricted number of elongatedprocesses). All the GFAP-immunostained (IS) struc-tures detected throughout the other forebrain regionsalways appeared negative for VIM and presented astellate morphology characteristic of mature astrocytes(i.e., a cell body connected to several radiating processes).

257PROLIFERATION OF PROGENITOR CELLS

Within the different forebrain regions, the extent ofcolocalization between GFAP and VIM was found todecrease in the aged as compared with the adult rats.The importance of such a decrease was, however, foundto differ markedly between the two germinative zonesconsidered. First, in the hippocampal dentate gyrus,the GFAP- and VIM-IS elongated processes that werefound to cross through the granular layer in the adultrats were not detected in the aged rats (Fig. 1E and F).Secondly, the extent of colocalization as well as mor-phological organization of the GFAP and VIM doubleimmunostained structures in the SVZ and the RMSwas found to remain grossly unchanged (Fig. 2E and

F). Similarly, the organization of the astrocytes doubleimmunostained for GFAP and VIM detected in thesuperficial brain layers and in the circumventricularorgans of aged rats was not modified as compared toadult rats.

Adrenalectomized and corticosterone-treatedrats

Throughout the different forebrain regions of adre-nalectomized or corticosterone-treated adult rats, the

Fig. 2. Immunostaining of bromodeoxyuridine (BrdU) and glialmarkers in the subventricular zone of the lateral ventricle of adult(A–C), aged (D–F), and corticosterone-treated (G–I) rats. Single per-oxidase immunostaining of BrdU shows that no marked variation inthe numerical density of BrdU-labeled nuclei present within the SVZis observed between the adult (A), aged (D), and corticosterone-treated rats (G). Similarly, double fluorescence immunostaining of

glial fibrillary acidic protein (GFAP; B, E, H) and vimentin (VIM; C, F,I) shows that, as for the adult rat (B and C), numerous GFAP and VIMdouble immunostained structures are localized to the SVZ of aged (Eand F) and corticosterone-treated (H and I) rats. CC: corpus callosum;LV: lateral ventricle; St: striatum. Scale bars 5 150 mm (in G and Iapply for peroxidase and fluorescence immunostaining, respectively).

258 ALONSO

Fig. 3. Immunostaining of bromodeoxyuridine (BrdU) and glialmarkers in the hippocampal dentate gyrus of a lesioned rat treatedwith corticosterone. Single peroxidase immunostaining of BrdUshows that on the side ipsilateral to the lesion (A), numerous BrdU-labeled nuclei are detected throughout the area surrounding the sur-gical lesion (arrow) and all along the granular layer of dentate gyrus,whereas very few are detected in this region on the contralateral side

to the lesion (B). Double fluorescence immunostaining of glial fibril-lary acidic protein (GFAP; C, E) and vimentin (VIM; D, F) shows thaton the side ipsilateral to the lesion, structures double immunostainedfor GFAP and VIM are detected within both the granular layer of thedentate gyrus adjacent to the lesion (C, D) and along the surgicallesion (E, F). Scale bars 5 150 mm (in B and F apply for peroxidaseand fluorescence immunostaining, respectively).

259PROLIFERATION OF PROGENITOR CELLS

Fig. 4. Quantitative evaluation of therate of cell proliferation and of the extentof colocalization between GFAP and VIMin the germinative zones of rats underdifferent physiological or experimentalconditions. Drawings are schematic rep-resentations of frontal sections throughthe forebrain of intact or lesioned rats,illustrating the location of the areas inwhich both parameters were quantified(open squares). The vertical bar in thelesioned brain illustrates the position ofthe surgical cut made through the lateralhippocampus. Within the SVZ and thehippocampal DG, the numerical densityof BrdU-labeled nuclei (black bars) andthe extent of colocalization betweenGFAP and VIM (gray bars) show parallelvariations under the different conditionsconsidered. First, in the SVZ (A), bothparameters only slightly vary with age orwith modified levels of circulating cortico-sterone. Secondly, in the hippocampal DGof intact rats (B), they decrease with in-creasing age and after corticosteronetreatment, whereas they increase afteradrenalectomy. Thirdly, in lesioned rats(C) either untreated or corticosterone-treated, they increase in the hippocampalDG ipsilateral to the lesion as comparedwith the contralateral DG. Left and righty-axis values, respectively, represent thenumber of BrdU-labeled nuclei/mm2 andthe coefficient of colocalization 3 100 ofGFAP vs. VIM in the areas considered.Note that for reasons of legibility, thescales differ on right and left y-axes fromone diagram to the other. Values repre-sent means 6 SEM of data. Single aster-isk: P , 0.05; double asterisk: P , 0.01,Mann-Whitney test: statistically differentfrom values obtained in adult rats in Aand B, and DG contralateral to the lesionin C. Adult: 2- to 3-month-old untreatedrats; Adx: adrenalectomized adult rats.Aged: 12- to 14-month-old untreated rats;Cort: corticosterone-treated adult rats;DG: dentate gyrus of the hippocampus;SVZ: subventricular zone of the lateralventricle; contra: contralateral to the le-sion; ipsi: ipsilateral to the lesion.

260 ALONSO

organization of GFAP-IS profiles was found to begrossly similar to that detected in control adult rats.The examination of double immunostained sections,however, showed that marked modifications occurredconcerning the organization of those GFAP and VIMdouble immunostained structures present in the ger-minative zones. In corticosterone-treated rats, GFAPand VIM double immunostained profiles were almostundetectable within the hippocampal dentate (Fig. 1Hand I) gyrus as well as within white matter regions (notshown). Conversely, the organization of such doubleimmunostained structures was not changed through-out the SVZ and RMS (Fig. 2H and I), as well as thesuperficial brain layers and the circumventricular or-gans, including the subfornical organ and the medianeminence (not shown). In adrenalectomized rats, thenumber of GFAP and VIM double immunostainedstructures was found to increase throughout the gran-ular layer of the hippocampal DG (Fig. 1K and L),whereas it was grossly unchanged within the SVZ andthe other brain regions.

Lesioned rats

On the side contralateral to the lesion, the organiza-tion of the GFAP and VIM double immunostained pro-files detected throughout the different forebrain re-gions was comparable to that observed in untreated orcorticosterone-treated intact rats. As described above,a limited number of such double immunostained pro-files were found to cross through the granular layer ofthe hippocampal DG of untreated rats, whereas theywere undetectable in corticosterone-treated rats. Onthe lesioned side, an increased number of GFAP andVIM double immunostained structures was detectedthroughout the granular layer of the hippocampal DGof both untreated and corticosterone-treated rats (Fig.3C and D), while these tightly packed structures werenumerous throughout the areas closely surroundingthe surgical cut (Fig. 3E and F).

Double Immunostaining for BrdU and Neuronalor Glial Markers

Unlesioned rats

Previous studies have clearly established that themembranous protein PSA-NCAM is essentially associ-ated with proliferative neuronal progenitors in boththe SVZ and the hippocampal DG (Seki and Arai, 1993;Doetsch et al., 1997). In agreement with these studies,the present observations indicated that in all the SVZand hippocampal DG examined, BrdU-labeled nucleiwere frequently associated with PSA-NCAM-IS cellbodies (Fig. 5D and J). By contrast, although closeanatomical relationships were constantly observed inthese regions between the BrdU-labeled cells and theGFAP- or VIM-IS structures, labeled nuclei were never

associated with any GFAP- or VIM-IS cell bodies (Fig.5A–I, Table 1). The observation of sections double im-munostained for BrdU and VIM further indicated thatthroughout the forebrain, VIM-IS profiles were alwayslocalized to the germinative areas exhibiting the high-est densities in BrdU-IS nuclei. Such preferential an-atomical relationships were particularly evidentwithin the SVZ and along the connected RMS whereeach of the numerous BrdU-labeled cell was closelysurrounded by VIM-IS processes (Fig. 5A and B). Al-though less evident, such preferential relationshipswere also observed throughout the hippocampal DG inboth intact adult and adrenalectomized rats. In theseanimals, z-series examination of double immuno-stained sections at high magnification further indi-cated that most of the proliferating (BrdU-labeled) cellswere in close apposition with one or several VIM-ISprocesses (Fig. 5F–H). In all sections examined, how-ever, a number of BrdU-IS nuclei located in the hip-pocampal DG appeared at a distance from any VIM-ISstructures.

Lesioned rats

In both control and corticosterone-treated rats, theareas closely surrounding the surgical cut exhibitinghigh concentrations of BrdU-IS nuclei always con-tained a high number of cell bodies and processes im-munostained for GFAP or VIM. Within these areas, anumber of BrdU-IS nuclei was found to be associatedwith cell bodies immunostained for PSA-NCAM,GFAP, or VIM (Fig. 5O and P). Throughout the sub-granular layer of the hippocampal DG distant from thelesion, a majority of the BrdU-IS nuclei were associatedwith PSA-NCAM-IS cell bodies, but never with GFAP-or VIM-IS cell bodies (Table 1). As in the unlesionedhippocampus, z-series examination indicated thatalong the DG of the lesioned hippocampus, BrdU-la-beled cell bodies frequently exhibited close appositionwith VIM-IS processes (Fig. 5K–N).

TABLE 1. Phenotype of BrdU-labeled cells*

BrdU PSA-NCAM GFAP VIM

Intact rats SVZ adult 100 60 (60) 0 0SVZ adx 100 65 (65) 0 0DG adult 65 22 (34) 0 0DG adx 120 66 (55) 0 0

Lesionedrats

Lesion 100 10 (10) 7 (7) 10 (10)DG 70 20 (28) 0 0

*The number of BrdU-labeled cells was pooled among forebrain double immu-nostained sections from three rats per group analyzed. Because of the largenumber of BrdU-labeled cells in the SVZ, a limit of 100 sampled cells wasimposed. In lesioned rats, the phenotype of BrdU-labeled cells was determinedboth in the area closely surrounding the lesion (lesion) and in the hippocampalDG adjacent to the lesion. Numbers in parentheses indicate the percentage ofBrdU-labeled nuclei associated with cell bodies immunostained for PSA-NCAM,GFAP, or VIM. Adult: 2- to 3-month-old untreated adult rat; adx: adrenalecto-mized adult rat; DG: hippocampal dentate gyrus; SVZ: subventricular zone of thelateral ventricle; GFAP: glial fibrillary acidic protein; PSA-NCAM: polysialy-lated neural cell adhesion protein; VIM: vimentin.

261PROLIFERATION OF PROGENITOR CELLS

Figure 5.

262 ALONSO

DISCUSSION

A recent in vitro study demonstrated that the prolif-eration of SVZ precursors was stimulated by their in-teraction with astrocytes (Lim and Alvarez-Buylla,2000). In the present study, we attempted to providesupport to the hypothesis that glial cells present in situwithin the germinative zones of the adult rat braincould provide a specific microenvironment to sustainthe proliferation of progenitor cells. Our observationsindicate that the germinative zones of the adult ratbrain exhibiting high to moderate rate of cell prolifer-ation always contain particular, nonproliferating, imma-ture-like astrocytes expressing both GFAP and vimentin,and that aging and circulating levels of corticosteronethat affect the rate of cell proliferation within the hip-pocampal DG correlatively affect the expression of vimen-tin within the astrocytes of these regions. Finally, incontrast with this germinative zone, aging and alter-ations in circulating glucocorticoid levels have only slighteffects on both cell proliferation and astrocytic expressionof VIM within the SVZ and the RMS.

Germinative Zones of Adult Rat Brain ContainImmature-Like Astrocytes Expressing Vimentin

In the course of maturation, the expression of inter-mediate filament proteins in astrocytes shows a se-quential change: vimentin is initially expressed andlater replaced by GFAP, both proteins being simulta-

neously expressed during the transition period (Pixleyand De Vellis, 1984; Sancho-Tello et al., 1995). Theexpression of Vimentin and/or GFAP has thus gener-ally been considered as markers of glial differentiation.Another modification that affect astrocytes during CNSmaturation concerns their morphology: immature as-trocytes exhibit a radial organization with a smallnumber of elongated processes (Levitt and Rakic,1980), whereas mature astrocyte exhibit a typical stel-late organization with a large number of ramified pro-cesses. A consistent finding of the present study is thatthe two germinative zones of the adult rat exhibitingmoderate to high rates of cell proliferation always con-tain particular astrocytes with features of immatureastrocytes, i.e., a radial-like, elongated morphology,and the expression of both GFAP and VIM. Previousstudies have provided evidence that the adult SVZcontains neural stem cells that have the capacity toself-renew and differentiate into neurons and glia(Morshead et al., 1994; Gitti et al., 1999). In two recentstudies in the mouse, such neural stem cells have beenidentified as being either ependymocytes (Johansson etal., 1999) or subventricular astrocytes (Doetsch et al.,1999). Our data, however, indicate that the cell prolif-eration marker BrdU never labeled GFAP-IS astro-cytes and VIM-IS ependymocytes present in the SVZ ofthe adult rat. Besides possible differences related tothe different species considered (rat versus mouse), twopossibilities may explain such discrepant results. First,putative stem cells exhibit particularly low mitotic ac-tivity (they could only be BrdU-labeled after 2 weeks ofcontinuous BrdU administration), as compared withthe high rate of proliferation of their progeny. It is thusvery likely that only the highly proliferating progenitorcells issued from the stem cells (including the PSA-NCAM–labeled neuronal progenitors) could be labeledunder the present conditions of BrdU administration(i.e., two times a day over a 3-day period). Secondly,proliferative stem cells express very low levels of GFAPor VIM. Indeed, after a single administration of the cellproliferation marker 3H-thymidine, proliferative SVZastrocytes were identified by means of the ultrastruc-tural characteristics of labeled cells and not by theirimmunocytochemical staining for GFAP or VIM(Doetsch et al., 1997).

Therefore, although it appears unlikely that thoseparticular astrocytes intensely immunostained for bothGFAP and VIM present within the germinative zonesof the adult rat brain correspond to stem cells, it cannotbe excluded that they correspond to postmitotic newlyformed astrocytes issued from proliferating progeni-tors. In this case, it is not surprising that the fates ofboth cell types are intimately linked. However, anotherpossibility is that these particular astrocytes present inthe germinative zones correspond to mature residentastrocytes that maintain an immature phenotype. In-deed, such astrocytes intensely immunostained forGFAP and VIM are present within nongerminativezones of the adult brain such as the brain surface or thecircumventricular organs.

Fig. 5. Double fluorescence immunostaining of BrdU (red) and neu-ronal or glial markers (green) in germinative zones of adult rats. A–D:Sections through the SVZ (A, C, D) and the RMS (B) of an untreatedadult rat immunostained for BrdU and for either VIM (A and B),GFAP (C), or PSA-NCAM (D). Within the SVZ (A) and the connectedRMS (B), BrdU-labeled nuclei are always closely surrounded by elon-gated VIM-labeled processes localized to these areas. Within thisregion, BrdU-labeled nuclei are never associated with VIM- (A and B)or GFAP-IS cell bodies (C), whereas they are frequently associatedwith PSA-NCAM-IS cell bodies (arrows in D). Note that BrdU-labelednuclei are never associated with any of the VIM-IS ependymocytesbordering the lateral ventricle (arrowheads in A) or its recess extend-ing along the RMS (arrowheads in B). E–J: Sections through thehippocampus of an adrenalectomized rat immunostained for BrdUand for either VIM (E-H), GFAP (I), or PSA-NCAM (J). A portion ofthe dentate gyrus that contains high concentrations of BrdU-labelednuclei also contains numerous VIM-IS processes (E). Higher magni-fication confocal images show that a majority of the BrdU-labeled cellsare closely apposed onto VIM- (arrows in F, G, H) or GFAP-IS pro-cesses (arrows in I), and BrdU-labeled nuclei of this region are fre-quently associated with PSA-NCAM-IS cell bodies (arrows in J), butnever with VIM- (F-H) or GFAP-IS (I) cell bodies. K–P: Sectionsthrough the hippocampal DG of a lesioned rat treated with cortico-sterone, immunostained for BrdU and either VIM (K–O) or GFAP (P).Low magnification shows that BrdU-labeled nuclei and VIM-IS struc-tures are localized to the same areas, including the region surround-ing the surgical cut (large arrow) and the adjacent subgranular layerof the dentate gyrus (small arrows). Higher magnifications show thatin the DG at a distance from the lesion, most of the BrdU-labeled cellsare closely apposed to VIM-IS structures (arrows in L, M, N), and inthe area closely surrounding the lesion, a number of BrdU-labelednuclei is associated with cell bodies immunostained for VIM (arrow-heads in O) or GFAP (arrowheads in P). Figures F–H and I–N corre-spond to z-series images of the same field through the thickness of thevibratome sections. LV: lateral ventricle; V: recess of the lateralventricle; Scale bars 5 50 mm (in P apply for B–D, F–J, and I–P).

263PROLIFERATION OF PROGENITOR CELLS

Rate of Cell Proliferation Varies CorrelativelyWith Extent of Vimentin Expression

Within Astrocytes

The existence of a preferential anatomical associa-tion between neuronal precursors and particular astro-cytes has already been reported in both the hippocam-pal DG and the SVZ of the adult brain. In thehippocampus, radial-like glial cells expressing bothGFAP and VIM have been described to extend through-out the DG granular layer (Cameron and Gould, 1994;Seki and Arai, 1999), and similar elongated astrocytesexpressing both GFAP and VIM have also been de-scribed in the SVZ and the connected RMS (Jankowskiand Sotelo, 1996; Thomas et al., 1996; Peretto et al.,1997). Although largely speculative, some roles havebeen proposed for those specific astrocytes present inthe germinative zones of the adult brain. In the hip-pocampal DG, the astrocytes were proposed to play arole in the dendritic growth of the newly formed hip-pocampal granule cells (Seki and Arai, 1999). Such arole appears unlikely in the SVZ and the RMS, how-ever, since these regions only contain migrating neu-roblasts or newly formed neurons that develop theirdendritic arborization on their site of destination, i.e.,the olfactory bulb. Those specific astrocytes associatedwith the SVZ and the connected RMS were proposed tobe involved in the guidance of the neuronal precursorsalong their long migratory pathway to the olfactorybulb, forming glial tubes preventing their dispersionfrom their migratory pathway (Jankowski and Sotelo,1996; Thomas et al., 1996; Peretto et al., 1997). Al-though cells issued from the proliferation of precursorcells in the hippocampal DG (Gould and Cameron,1996) do migrate along short distances within the gran-ular layer, such a migration appears limited as com-pared with that of neuroblasts generated in the SVZ.Since evidence has been recently provided that theproliferation of SVZ progenitor cells was stimulated byastrocytes (Lim and Alvarez-Buylla, 1999), an alternativerole for these particular immature-like astrocytes presentin the germinative regions of adult brain could thus be tosustain the proliferation of the progenitor cells.

The present data clearly indicate that, under variousphysiological or experimental conditions, the rate ofcell proliferation within the germinative zones is al-ways strongly correlated to the expression of VIM bythe astrocytes present within these areas. Such a cor-relation is clearly illustrated in the hippocampal DGwhere the numerical density of proliferating cells andthe number of GFAP and VIM double immunostainedstructures was found to decrease dramatically withincreasing age or after chronic corticosterone treat-ment, which increases corticosterone circulating levels,and to increase after adrenalectomy, which decreasescorticosterone circulating levels. Conversely, our dataclearly show that in the SVZ and the RMS, whichexhibited numerous GFAP and VIM double immuno-stained structures whatever the age or the circulatingcorticosterone levels, the rate of cell proliferation re-

mained grossly unchanged. The role played by inter-mediate filaments (IF) within astrocytes remains un-clear. As mentioned above, vimentin is the main IFprotein in radial glial cells that plays a major role indifferent steps of the neurogenesis in the developingCNS (Rakic, 1981, 1991). Whereas vimentin is not ex-pressed in the large majority of the mature astrocytespresent in the adult brain (Pixley and De Vellis, 1984;Sancho-Tello et al., 1995), it is expressed by the so-called reactive astrocytes that appear in the CNS aftertrauma, tumor growth, or neurodegenerative diseases(Eddleston and Mucke, 1993). Our present data onlesioned rats show that numerous of such GFAP- andVIM-expressing reactive astrocytes are detected bothalong the surgical cut extending through different fore-brain regions and at a distance from the lesion, alongthe hippocampal dentate gyrus. The appearance ofthese reactive astrocytes at a distance from the surgi-cal cut may result from the degeneration of neuronalstructures of the DG that have been lesioned by thesurgical cut crossing through the lateral hippocampus.The present data clearly show that the occurrence ofsuch vimentin-expressing reactive astrocytes through-out the DG is always associated with an increasedproliferation of nonastrocytic (i.e., GFAP- and VIM-negative) cells within this region. This strongly sug-gests that VIM-IS reactive astrocytes play the samerole than the VIM-IS astrocytes present in the intacthippocampal DG as regard to the proliferation of pro-genitor cells. As previously demonstrated (Miyake etal., 1988; Janeczko, 1993; Norton, 1999), the presentobservations in lesioned rats also show that a dramaticincrease of cell proliferation was associated with theareas containing a large number of GFAP and VIMdouble immunostained reactive astrocytes closely sur-rounding the surgical cut. Although in these areassome of these BrdU-IS nuclei were associated withVIM-IS astrocytes, the majority was associated withNG2-IS cell bodies, illustrating the increased prolifer-ation of the oligodendrocyte progenitor cells present inthe lesioned area (data not shown). It can thus beassumed that the increase of cell proliferation that isclassically detected in the vicinity of mechanical injuryto the adult brain at least partly corresponds to astimulation of the proliferation of residing progenitorcells by the VIM-IS reactive astrocytes. Whereas onemay admit that the differential expression of GFAP orVIM by astrocytes will play a role in their morpholog-ical organization (Eliasson et al., 1999), it appears un-likely that the differential expression of IF proteinsmay directly influence surrounding cells. It can be as-sumed, however, that the expression of vimentinwithin astrocytes depends on the functional status ofthe cell and is associated with the expression of otherastrocytic molecules, some of which may play a role inthe control of cell proliferation. Several astrocytic fac-tors expressed by immature and/or reactive astrocytesare known to stimulate the proliferation of progenitorcells. This is the case for basic fibroblast growth factor(bFGF) (Flott-Rahmel et al., 1992; Gonzales et al.,

264 ALONSO

1995; Riva et al., 1998), epidermal growth factor (EGF)(Santa-Olalla and Covarrubias, 1995), transforminggrowth factor (TGFa) (Santa-Olalla and Covarrubias,1995; Weickert and Blum, 1995), and insulin growthfactor-1 (IGF-1) (O’Kusky et al., 2000). Future studiesshould determine whether these factors are expressedby the particular astrocytes located in the differentgerminative areas of the adult brain and if their ex-pression is affected by aging and/or by glucocorticoids.

A recent in vitro study reported that the proliferationof adult progenitor cells by cultured astrocytes resultedfrom direct cell-cell contacts (Lim et al., 1999). Here,we show that in both the SVZ and the hippocampal DG,most proliferating (BrdU-IS) cells were closely sur-rounded by VIM-IS structures. This strongly suggeststhat the control of cell proliferation by astrocytespresent in the germinative zones at least partly in-volves specific membranous proteins and/or proteinsbound to the pericellular matrix.

Cell Proliferation and Astrocytic Phenotype AreDifferentially Affected in SVZ vs. Other

Germinative Zones

The present observations that aging and corticoste-rone treatment similarly affect the proliferation of pro-genitor cells within the hippocampal DG strongly rein-force the idea that the age-related decrease inhippocampal neurogenesis results from an increasedsecretion of glucocorticoids (Porter and Landfield,1988; Cameron and McKay, 1999). Moreover, thepresent data indicate that within the SVZ and RMS,neither cell proliferation nor astrocytic phenotypes areaffected by aging or corticosterone treatment.

Although the glucocorticoid receptors are expressedby a large number of neurons and glial cells throughoutthe adult brain (Ovadia et al., 1984; Jung-Testas et al.,1992; Cintra et al., 1994), it is believed that the effectsof glucocorticoids on the proliferation of the hippocam-pal neuronal precursor do not involve direct effects ofthe steroid hormones via these receptors. Indeed, it hasbeen clearly established that the effects of glucocorti-coids on cell proliferation are mediated by N-methyl-D-aspartate (NMDA) receptors (Gould et al., 1994;Cameron et al., 1995; Bernabeu and Sharp, 2000) andacute or chronic administration of glucocorticoids in-duces an increased release of glutamate throughout theCNS (Reagan and McEwen, 1997; Vereno and Borrell,1999). Since neuronal precursors of the hippocampalDG express NMDA receptors (Gould and Tanapat,1997), it is possible that the increased stimulation ofthese receptors directly affects their rate of prolifera-tion. The modest effects of either age or corticosteronetreatment on progenitor cell proliferation within theSVZ and RMS regions could thus be explained by eitherthe absence of glutamate receptors on neuronal precur-sors and/or astrocytes present within these regions, ora very scarce glutamatergic innervation of these re-gions as compared with the hippocampus. Interest-

ingly, apart from the astrocytes present in the hip-pocampal DG, aging and corticosterone were found todecrease dramatically the expression of VIM in theso-called fibrous astrocytes present in the white matterregions (data not shown). This provides support to analternative possibility: aging and corticosterone act oncell proliferation by modifying the phenotype of theimmature-like astrocytes present in the germinativeareas, possibly via a stimulation of their NMDA recep-tors, then modifying their supporting role on the pro-liferation of progenitor cells. Our data indicate thataging and corticosterone have no effect on the expres-sion of VIM by those astrocytes located within specificbrain regions that have direct access to the circulatingmedium and/or the cerebroventricular fluid, for exam-ple, regions located outside from the blood-brain bar-rier (BBB; such as the circumventricular organs), orlocated on the external brain border or within the ven-tricular lumen (such as the glia limitans and mostcircumventricular organs, respectively). The astrocytespresent in the SVZ, like those astrocytes present in theRMS (see Fig. 5B), are located on the border of theventricular lumen, thus facilitating their access to theintraventricular cerebrospinal fluid. Moreover, ourdata also indicate that corticosterone treatment doesnot modify the expression of VIM by the reactive as-trocytes located in the area closely surrounding a sur-gical cut, i.e., a region in which the BBB has beendisrupted. Such a differential access to the extracere-bral milieu or the cerebrospinal fluid may provide analternative explanation for the differential responses toaging and glucocorticoids of the astrocytes present inthe SVZ and RMS versus those present in the DG andwhite matter regions. An unidentified factor or factorspresent in the circulating medium and/or the cerebro-spinal fluid may induce these astrocytes to counteractthe effects of aging or corticosterone treatment, thuspreserving their immature-like phenotype and theircapacity to sustain the proliferation of progenitor cells.

Taken together, these data suggest that specific as-trocytes expressing vimentin provide a particular mi-croenvironment that sustains the proliferation of theprogenitor cells present in their close vicinity. Futurestudies should focus on the characterization of the as-trocyte factor(s) involved in such positive effects onprogenitor cells proliferation, and the circulatingand/or intraventricular factors responsible for themaintained expression of an immature phenotype byparticular astrocytes located in contact with the extra-cerebral medium and/or the cerebrospinal fluid.

ACKNOWLEDGMENTS

Confocal microscopy has been realized using the fa-cilities at CRIC, Montpellier, France. The authorthanks Florence Duchamp for her excellent technicalassistance and Vicky Tobin for reading the article.

265PROLIFERATION OF PROGENITOR CELLS

REFERENCES

Alvarez-Buylla A, Herrera DG, Wichterle H. 2000. The subventricularzone: source of neuronal precursors for brain repair. Prog Brain Res127:1–11.

Altman J, Das GD. 1965. Autoradiographic and histological evidenceof postnatal hippocampal neurogenesis in rats. J Comp Neurol124:319–335.

Bernabeu R, Sharp FR. 2000. NMDA and AMPA/kainate glutamatereceptors modulate dentate neurogenesis and Ca3 synapsin-I innormal and ischemic hippocampus. Cereb Blood Flow Metab 20:1669–1680.

Cameron HA, Gould E. 1994. Adult neurogenesis is regulated byadrenal steroids in the dentate gyrus. Neuroscience 61:203–209.

Cameron HA, McEwen BS, Gould E. 1995. Regulation of adult neu-rogenesis by excitatory input and NMDA receptor activation in thedentate gyrus. J Neurosci 15:4687–4692.

Cameron HA, McKay RDG. 1999. Restoring production of hippocam-pal neurons in old age. Nat Neurosci 2:894–897.

Cintra A, Zoli M, Rosen L, Agnati LF, Okret S, Wikstrom AC, Gustafs-son JA, Fuxe K. 1994. Mapping and computer assisted morphome-try and microdensitometry of glucocorticoid receptor immunoreac-tive neurons and glial cells in the rat central nervous system.Neuroscience 62:843–897.

Cowan WM. 1979. The development of the brain. Sci Am 241:112–133.Doetsch F, Garcia-Verdugo JM, Alvarez-Buylla A. 1997. Cellular com-

position and three-dimensional organization of the subventriculargerminal zone in the adult mammalian brain. J Neurosci 17:5046–5061.

Doetsch F, Caille I, Lim DA, Garcia-Verdugo JM, Alvarez-Buylla A.1999. Subventricular zone astrocytes are neural stem cells in theadult mammalian brain. Cell 97:703–717.

Eddleston M, Mucke L. 1993. Molecular prolfile of reactive astrocytes-implications for their role in neurologic disease. Neuroscience 54:15–36.

Eliasson C, Sahlgren C, Berthold CH, Stakeberg J, Celis JE, Bet-sholtz C, Eriksson JE, Pekny M. 1999. Intermediate filament pro-tein partnership in astrocytes. J Biol Chem 274:23996–24006.

Flott-Rahmel B, Gerdes W, Pechan PA, Brysch W, SchlingensiegenKH, Seifert W. 1992. bFGF induces its own gene expression inastrocytic and hippocampal cell cultures. Neuroreport 3:1077–1080.

Goldman JE. 1995. Lineage, migration and fate determination ofpost-natal subventricular zone cells in the mammalian CNS. J Neu-rooncol 24:61–64.

Gonzales AM, Berry M, Maher PA, Logan A, Baird A. 1995. A com-prehensve analysis of the distribution of FGF-2 and FGFR1 in therat brain. Brain Res 701:201–226.

Gould E, Cameron HA, McEwen BS. 1994. Blockade of NMDA recep-tor increases cell death and birth in the developping dentate gyrus.J Comp Neurol 340:551–565.

Gould E, Cameron HA. 1996. Regulation of neuronal birth, migrationand death in the rat dentate gyrus. Dev Neurosci 18:22–35.

Gould E, Tanapat P. 1997. Lesion-induced proliferation of neuronalprogenitors in the dentate gyrus of the adult rat. Neuroscience80:427–36.

Gould E, Tanapat P, Rydel T, Hastings N. 2000. Regulation of hip-pocampal neurogenesis in adulthood. Biol Psychiatry 48:715–720.

Gitti A, Frolichsthal-Schoeler P, Gali R, Parati EA, Cova L, PaganoSF, Bjornson CR, Vescovi AL. 1999. Epidermal and fibroblastgrowth factors behave as mitotic regulators for a single multipotentstem cell-like population from the subventricula region of the adultmouse forebrain. J Neursci 19:3287–3297.

Hauke C, Ackermann I, Korr H. 1995. Cell proliferation in the sub-ependymal layer of the adult mouse in vivo and in vitro. Cell Prolif28:595–607.

Janeczko K. 1993. Co-expression of GFAP and vimentin in astrocytesproliferating in response to injury in the mous cerebral hemisphere:a combined autoradiographic and double immunocytochemicalstudy. Int J Dev Neurosci 11:139–147.

Jankowski A, Sotelo C. 1996. Subventricular zone-olfactory bulb mi-grattory pathway in the adult mouse: cellular composition andspecificity as determined by heterochronic and heterotopic trans-plantation. J Comp Neurol 371:376–396.

Johansson CB, Momma S, Clarke DL, Risling M, Lendahl U, Frisen J.1999. Identification of a neural stem cell in the adult mammaliancentral nervous system. Cell 96:25–34.

Jung-Testas I, Renoir M, Bugnard H, Greene GL, Beaulieu EE. 1992.Demonstration of steroid hormone receptors and steroid action in

primary cultures of rat glial cells. J Steroid Biochem Molec Biol41:621–631.

Kempermann G, Gage FH. 2000. Neurogenesis in the adult hippocam-pus. Norvatis Found Symp 8:220–235.

Kuhn HG, Dickinson-Anson H, Gage FH. 1996. Neurogenesis in thedentate gyrus of the adult rat: age-related decrease of neuronalprogenitor proliferation. J Neurosci 16:2027–2033.

Levitt P, Rakic P. 1980. Immunoperoxidase localization of glial fibril-lary acidic protein in radial glial cells and astrocytes of the devel-oping Rhesus monkey brain. J Comp Neurol 193:815–840.

Lim DA, Alvarez-Buylla A. 1999. Interaction between astrocytes andadult subventricular zone precursors stimulate neurogenesis. ProcNatl Acad Sci USA 96:7526–7531.

Lois C, Alvarez-Buylla A. 1993. Proliferating subventricular zone cellsin the adult mammalian forebrain can differentiate into neuronsand glia. Proc Natl Acad Sci USA. 90:2074–2077.

Miyake T, Hattori T, Fukuda M, Kitamura T Fujita S. 1988. Quanti-tative studies on proliferative changes of astrocytes in mouse cere-bral cortex. Brain Res 451:133–138.

Morshead CM, Reynolds BA, Craig CG, McBurney MW, Staines WA,Morassutti D, Weiss S, van der Kooy D. 1994. Neural stem cells inthe adult mammalian forebrain: a relatively quiescent subpopula-tion of subependymal cells. Neuron 13 1071–1082.

Norton WT. 1999. Cell reactions following acute brain injury: a re-view. Neurochem Res 24:213–218.

O’Kurski JR, Ye RP, D’Ercole AJ. Insulin-like growth factor-I pro-motes neurogenesis and synaptogenesis in the hippocampal dentategyrus during postnatal development. J Neurosci 20:8435–8442.

Ovadia H, Vlodavsky I, Abramsky O, Weidenfeld J. 1984. Binding ofhormonal steroids to isolated oligodendria and astroglia grown invitro an a naturally produced extracellular matrix. Clin Neurop-harmacol 7:307–311.

Paxinos G, Watson C. 1982. A rat stereotoxic coordinates. London:Academic Press.

Peretto P, Merighi A, Fasolo A, Bonfanti L. 1997. Glial tubes in therostral migratory stream of the adult rat. Brain Res Bull 42:9–21.

Pixley SK De Vellis J. 1984. Transition between radial glia andmature astrocytes studied with a monoclonal antibody to vimentin.Brain Res 15:201–209.

Porter NM, Landfield LW. 1998. Stress hormones and brain aging:adding injury to insult? Nat Neurosci 1:3–4.

Rakic P. 1981. Neuron-glial interaction during brain development.Trends Neurosci 4:184–187.

Rakic P. 1991. Glial cells in development. In vivo and in vitro ap-proaches. Ann NY Acad Sci 633:96–99.

Reagan LP, McEwen BS. 1997. Controversies surrounding glucocor-ticoid-mediated cell death in the hippocampus. J Chem Neuroanat13:149–167.

Riva MA, Molteni R, Racagni G. 1998. Differential regulation ofFGF-2 and FGFR-1 in rat cortical astrocytes by dexamethason andisoprotenerol. Mol Brain Res 57:38–45.

Sancho-Tello M, Valles S, Montoliu C, Renau-Piqueras J Guerri C.1995. Developmental pattern of GFAP and vimentin expression inrat brain and radial glial cultures. Glia 15:157–166.

Santa-Olalla J, Covarrubias L. 1995. Epidernal growth factor (EGF)transforming growth factor alpha (TGF-alpha) and basic fibroblastgrowth factor (bFGF) differentially influence neural precursor cellsof mouse embryonic mesencephalon. J Neurosci 42:172–183.

Schmechel DE, Rakic P. 1979. A golgi study of radial glial cells indevelopping monkey telencephalon: morphogenesis and transfor-mation into astrocytes. Anat Embryol 156:115–152.

Seki T, Arai Y. 1993. Highly polysialylated neural cell adhesion mol-ecule (NCAM-H) is expressed by newly generated granule cells inthe dentate gyrus of the adult rat. J Neurosci 13:2351–2358.

Seki T, Arai Y. 1999. Temporal and spacial relationships betweenPSA-NCAM–expressing, newly generated granule cells and radialglia-like cells in the adult dentate gyrus. J Comp Neurol 410:503–513.

Thomas LB, Gates MA, Steindler DA. 1996. Young neurons from theadult subependymal zone proliferate and migrate along an astro-cyte, extracellular matrix-rich pathway. Glia 17:1–14.

Vereno C, Borrell J. 1999. Rapid glucocorticoid effects on excitatoryamino acid levels in the hippocampus: a microdialysis study infreely moving rats. Eur J Neurosci 11:2465–2473.

Vernadakis A. 1988. Neuron-glia interrelations. Int Rev Neurobiol30:149–224.

Weickert CS, Blum M. 1995. Striatal TGF-alpha: postnatal develop-mental expression and evidence for a role in the proliferation ofsubependymal cells. Dev Brain Res 86:203–216.

266 ALONSO