Neural stem cells from mouse forebrain are contained in a population distinct from the ‘side population’


Address correspondence and reprint requests to Dr Marc-André Mouthon, Laboratoire de Radiopathologie, CEA/DSV/DRR-IPSC, BP n°6, 92265 Fontenay-aux-Roses cedex, France.


Developing and adult forebrains contain neural stem cells (NSCs) but no marker is available to highly purify them. When analysed by flow cytometry, stem cells from various tissues are enriched in a ‘side population’ (SP) characterized by the exclusion of the fluorescent dye Hoechst 33342. Here, we characterize the SP in embryonic, neonatal and adult forebrains, as well as in neurosphere cultures and we have determined whether this SP could be a source of enriched NSCs. By using specific inhibitors, we found that the SP from embryonic forebrain results from the activity of the ABCG2 transporter, a characteristic of other stem cells, whereas the SP from adult forebrain probably results from the ABCB1 transporter. SP cells from embryonic and adult forebrains, however, expressed a range of cell surface markers more consistent with a haematopoietic/endothelial origin than with a neural origin; NSC markers were mostly expressed on cells outside the SP (in the main population, MP). Moreover, assays for NSC growth in vitro showed that SP cells from embryonic and adult forebrains did not generate NSC-derived colonies, whereas the MP did. We thus conclude that NSCs from developing and adult forebrains are not contained in the SP contrary to stem cells from other tissues.

Abbreviations used

7-aminoactinomycin D


ATP-binding cassette


bovine serum albumin


4′,6-Diamidino-2-phenylindole dihydrochloride


Dulbecco's modified Eagle's medium


epidermal growth factor


enhanced green fluorescent protein


fibroblast growth factor-2


Glial Fibrillary Acidic Protein


main population


neural colony forming cell


neural stem cell


neurosphere-initiating cells


phosphate-buffered saline


propidium iodide


lectin peanut agglutinin


side population


subventricular zone


von Willebrand factor

Neural stem cells (NSCs) are cells that can self-renew and differentiate into both neurons and glia. They have been isolated from many regions of the embryonic nervous system (Temple 2001). Between 10 and 20% of cells isolated from embryonic day 10 mouse forebrain are thought to be NSCs (Qian et al. 2000), but the proportion declines rapidly thereafter during development where the NSCs are diluted by the production of restricted progenitors and differentiated cells. In the adult forebrain, NSCs are still present, although in lower quantities, in the subgranular layer of the dentate gyrus and in the subventricular zone (SVZ) of the lateral ventricles (Alvarez-Buylla and Lim 2004).

NSCs are usually identified by their behaviour after isolation: in vitro, in the presence of epidermal growth factor (EGF) and fibroblast growth factor-2 (FGF-2), NSCs form clonally derived spheres called neurospheres (Reynolds and Weiss 1992). Neurospheres contain some NSCs, mostly progenitor cells and a few cells expressing differentiation markers of either neuronal or glial cell lineages. Cells from neurosphere cultures can be grown in culture for numerous passages and they differentiate into both neuronal and glial lineages when provided with an adequate medium, demonstrating the self-renewal and multipotency of NSCs. Hence, neurosphere cultures are widely used to study NSCs from adult and either embryonic human or murine brain tissues.

NSC biology suffers from a lack of specific membrane markers to unambiguously identify the cells. Nonetheless, a few studies have reported the use of membrane markers combined with cell sorting to isolate NSCs. Rietze and colleagues, for example, isolated NSCs from adult mouse brain by double-negative selection for the lectin peanut agglutinin (PNA) and CD24 (heat stable antigen) (Rietze et al. 2001), whereas Uchida et al. (2000) and Capela and Temple (2002) purified NSCs by positive selection of cells bearing the markers CD133 (prominin) or CD15 (LeX) from human fetal brain and adult mouse forebrain. By contrast, haematopoietic stem cells from adult bone marrow (Goodell et al. 1996), stem cells from various somatic tissues and germinal stem cells (Lassalle et al. 2004) can be enriched in a ‘side population’ (SP) of cells sorted by flow cytometry after exposure to the vital dye Hoechst 33342. This SP has lower fluorescence at two emission wavelengths (red and blue) than cells in the ‘main population’ (MP) as a result of the capacity of the stem cells to exclude the dye, which is pumped out specifically by ABCG2, a multidrug transporter of the ATP-binding cassette (ABC transporter) family (Zhou et al. 2001; Lassalle et al. 2004).

SP cells have been found in neurosphere cultures of embryonic neural precursor cells (Hulspas and Quesenberry 2000; Murayama et al. 2002), and highly proliferative multipotent cells in the SP prepared from embryonic and adult mouse brain neurospheres can be distinguished from less proliferative progenitor cells in the MP on the basis of their exclusion of Hoechst dye (Kim and Morshead 2003). Surprisingly, however, the stem cell marker ABCG2 mRNA is expressed in both the SP and the MP of embryonic neurosphere cultures, and not at all in the SP of neurosphere cultures prepared from adult brain (Kim and Morshead 2003). Furthermore, adult NSCs enriched by double-negative PNA and CD24 selection from freshly harvested SVZ, when analysed by flow cytometry, are not contained in the SP (Rietze et al. 2001). Hence, NSCs prepared from freshly harvested brain and from neurosphere cultures may be quite different.

In this paper, we have characterized SP cells from freshly isolated forebrain to determine whether or not they are true NSCs. We show that the vast majority of cells in the SP from both embryonic and adult forebrains are of endothelial and/or haematopoietic origin. We find that the SP from embryonic and adult forebrains does not contain NSC activity, whereas the SP from neurosphere cultures does. We conclude that isolating the SP of cells from forebrain cannot be used as a purification step for NSCs.

Materials and methods

Isolation of embryonic and adult forebrain cells

All animal procedures were carried out in accordance with French government regulations (Services vétérinaires de la santé et de la production animale, Ministère de l'Agriculture). Forebrains were obtained from either the embryos of pregnant C57BL6 mice or enhanced green fluorescent protein (EGFP) transgenic mice [C57BL6-TgN(beta-act-EGFP)01Osb; Okabe et al. 1997]. The meninges were removed from the telencephalon in phosphate-buffered saline (PBS) with 0.6% glucose and dissected under a binocular microscope. Forebrains were triturated and incubated in collagenase I (100 U/mL; Invitrogen, Carlsbad, CA, USA)/DNAse I (100 U/mL; Sigma, St Louis, MO, USA) in PBS with 0.6% glucose for 10–20 min at 37°C. They were then dissociated into single-cell suspensions by flushing through a p1000 micropipette tip and washed twice in Dulbecco's modified Eagle's medium (DMEM):F12 supplemented with 2% B27 minus antioxidant (Invitrogen). Cells from the SVZ of neonatal C57BL6 mice (2 days after birth) were prepared flowing the same procedure as for embryonic cells.

Either C57BL6 or p21Waf (B6; 129S2-Cdkn1atm1Tyj/J)-deficient mice (2–6 months old) were killed by cervical dislocation and their brains were immediately removed. A 2-mm-thick coronal slice was cut between +0.5 and −1.5 mm, relative to bregma. The slice was placed caudal side up and the lateral walls of the lateral ventricles were dissected under a binocular microscope. Dissected tissues were digested with 1 mg/mL papain (Worthington, Lakewood, NJ, USA) in Earl's Balanced Salt Solution (Invitrogen) containing 0.2 mg/mL l-cysteine (Sigma) and 0.2 mg/mL EDTA (Sigma), and then incubated for 30 min at 37°C on a rocking platform as previously described (Gritti et al. 2002). Tissues were then transferred to DMEM:F12 medium (Invitrogen) containing 0.7 mg/mL ovomucoid (Sigma) and the cells were dissociated mechanically by flushing through a p1000 micropipette tip. Cells were resuspended in DMEM:F12 medium supplemented with 2% B27 and filtered through a 70 µm cell sieve (Filcons®; BD Biosciences, Franklin Lakes, NJ, USA).

Neurosphere cultures

For embryonic and postnatal neurosphere cultures, cells were plated at a density of 105 cells/25-cm2 flask in 5 mL of Euromed medium (formerly known as NS-A; Euroclone, Pero, Italy) supplemented with a hormone mix (6 g/L glucose, 60 mmol/L putrescine, 20 nmol/L progesterone, 30 nmol/L sodium selenite, 25 µg/mL insulin and 100 µg/mL apo-transferrin) and human recombinant EGF and FGF-2 (Sigma), both at a final concentration of 20 ng/mL, as previously described (Mathieu et al. 2006). For adult SVZ neurosphere cultures, the medium was NeurobasalA supplemented with N2 (Invitrogen), 2 µg/mL heparin and 20 ng/mL EGF and 10 ng/mL FGF-2. Cultures were incubated at 37°C in a humidified atmosphere containing 5% CO2. Seven days after plating, neurospheres were mechanically dissociated and subcultured.

Staining with Hoechst 33342 and immunophenotyping of SP cells

Dissociated cells were incubated for 1 h 30 min at 37°C in an incubator containing 5% CO2 with 5 µg/mL Hoechst 33342 (Sigma) in DMEM:F12 supplemented with 2% B27. Cells were then washed twice in PBS containing 0.15% bovine serum albumin (BSA).

Immunophenotyping of SP cells was performed by incubation with monoclonal antibodies for 30 min on ice in PBS with 0.15% BSA after Hoechst staining. Primary antibodies CD15 (MMA), CD24 (30F1), CD31-PE (MEC13.3), CD45-PE-Cy5 (RA3-6B2), CD49f-PE (GoH3) and AA4-FITC were purchased from BD Biosciences; CD133-FITC (13A4) was purchased from eBiosciences (San Diego, CA, USA), and A2B5 from R & D (Abington, UK). Matched isotopic controls were used to set the gate of positivity. The secondary antibodies were goat anti-mouse IgG-Alexa Fluor 488, goat anti-rabbit-Alexa Fluor 488 (Molecular Probes, Eugene, OR, USA) and goat anti-mouse IgM–phycoerythrin (BD Biosciences).

Before commencing flow cytometry propidium iodide (PI) was added to a final concentration of 2 µg/mL in order to identify and exclude dead cells. Cells were analysed on either LSR I or FACStar Plus flow cytometers (BD Biosciences) as previously reported (Lassalle et al. 2004). SP cells exclude Hoechst 33342 and consequently have lower blue and red fluorescence emissions.

We investigated the drug transporters involved in Hoechst dye efflux by adding inhibitors at the beginning of Hoechst staining. Ko143, a specific inhibitor of ABCG2 (Allen et al. 2002), and verapamil, an inhibitor of ABC transporters with greatest specificity for ABCB1, were added at a final concentration of 200 nm and 50 µm, respectively.

After sorting, SP and MP cells were fixed for 15 min with 1% paraformaldeyde then cytospun at 115 g for 3 min to perform immunohistochemistry. After permeabilization with 0.1% Triton X-100 (Sigma) for 5 min, primary antibodies (either monoclonal anti-ABCB1 or P-Glycoprotein from Calbiochem, San Diego, CA, USA; and polyclonal anti-vWF, an endothelial cell marker from Dako, Carpinteria, CA, USA) were incubated for 1 h. The secondary antibodies were goat anti-mouse IgG–Alexa Fluor 488 and goat anti-rabbit-Alexa Fluor 546 (Molecular Probes). The nuclei were stained by incubating them for 5 min with 1 µg/mL 4′,6-Diamidino-2-phenylindole dihydrochloride (DAPI) (Sigma) and then the slides were mounted under Fluoromount (Southern Biotechnologies, Birmingham, AL, USA).

Cell sorting and analysis of NSC clonogenicity

We assessed the number of NSCs in the various cell populations of embryonic forebrains (SP, MP and total) by plating sorted cells at a limiting dilution using a modified version of a method described elsewhere (Tropepe et al. 1999). Viable cells were sorted on a FACStar Plus cytometer (BD Biosciences) equipped with a 360-nm UV laser and a 488-nm laser set at 50 mW. Sorted cells were plated using Clonecyt® at 15, 30, 60 and 120 cells/well in 96-well plates containing 150 µL of embryonic neurosphere medium. After 14 days of culture we scored the number of positive wells (i.e. wells containing at least one neurosphere). Clonogenicity, or the number of neurosphere-initiating cells (NS-IC), was calculated using limiting dilution analysis software (L-Calc®; StemCell Technologies, Grenoble, France).

Adult SVZ cells were plated using the mouse NeuroCult® neural colony forming cell assay (NCFC; StemCell Technologies) according to the manufacturer's instructions, except that cells were plated at a density of 500–17 000 cells per 35-mm culture dish. The cells were then incubated for 21 days in a humidified atmosphere with 5% CO2. EFG and FGF-2 were added every 7 days. After 21 days the colony diameters were measured using an eyepiece reticule on an inverted light microscope under phase contrast optics. Large clones (diameter = 1 mm) were considered to derive from NSC.

Determination of NSC self-renewal and differentiation

To determine whether the neurospheres formed were derived from NSCs, we assayed for self-renewal and differentiation. Single neurospheres from NS-IC tests of embryonic cells and single colonies from NCFC assays of adult cells were dissociated and plated in neurosphere culture conditions as described above. Secondary and tertiary neurosphere formation was tested. Thereafter, neurospheres were evaluated for their differentiation potential in Neurobasal medium in the presence of 2% B27 and 1% FCS on poly-d-lysine-coated culture slides (Labteck®; BD Biosciences).

After 1 week in differentiation medium, neurospheres were fixed in 1% paraformaldeyde for 15 min and permeabilized in 0.1% Triton X-100 (Sigma). Glial differentiation was assessed with anti-Glial Fibrillary Acidic Protein (GFAP) (G-A-5) and anti-O4 (clone 81) monoclonal antibodies purchased from Chemicon (Temecula, CA, USA). Neuronal differentiation was determined with a polyclonal anti-βIII tubulin antibody (Covance, Berkeley, CA, USA). Secondary antibodies were goat anti-mouse IgG1-Alexa Fluor 350 (Molecular Probes), goat anti-mouse IgM–phycoerythrin (BD Biosciences) and goat anti-rabbit-Alexa Fluor 488. The nuclei were stained by incubating them for 5 min with 1 µg/mL of either 7-aminoactinomycin D (7AAD) or DAPI (Sigma). Slides were mounted under Fluoromount (Southern Biotechnologies).

Analysis of NSC quiescence

NSC quiescence was determined by dual RNA/DNA staining with pyronin-Y (Sigma) and Hoechst 33342, as previously reported for haematopoietic stem cells (Gothot et al. 1997). After 45 min of incubation in Hoechst 33342, an equal volume of 2 µg/mL pyronin-Y was added and the cells were incubated for 45 min at 37°C. Cells were washed twice in 0.15% BSA in PBS and PI (2 µg/mL) was added to exclude dead cells from the analysis. The red fluorescence of pyronin-Y was measured with a 575/26 filter, after excitation with a 488-nm argon laser.


Neurosphere cultures contain SP cells able to self-renew

We wanted to compare the SP cells from neurospheres with those from freshly prepared forebrains. To do so, we first produced neurospheres by culturing embryonic and neonatal mouse forebrains in the presence of EGF and FGF-2 and we characterized their SP cells (Fig. 1). Embryonic neurosphere cultures contained an SP (Fig. 1ai), which disappeared upon the addition of Ko143, an ABCG2-specific inhibitor (Allen et al. 2002) that blocks Hoechst dye efflux (Fig. 1aii). After 4 days in culture, only a small proportion of cells from the embryonic forebrain neurospheres were in the SP (Fig. 1b), whereas after 7 days significantly more were found in the SP (p = 0.05). The proportion of cells in the SP from neonatal forebrain neurospheres after 7 days in culture was not statistically different from that observed for embryonic cells. Incubation of the cells with Ko143 strongly blocked Hoechst dye efflux (by 72% in embryonic cells and 89% in postnatal cells), demonstrating that the SP phenotype in neurosphere cultures is a result of the activity of ABCG2 (Fig. 1b).

Figure 1.

 Side population (SP) cells from neurosphere cultures can self-renew. Neurosphere cultures were established either from embryonic (embryonic day 14.5; E14.5) or from postnatal (2-days old; PN2) forebrains. The SP was present in embryonic neurosphere cultures (ai) and was strongly inhibited by Ko143 (aii). The percentage of total neurosphere cells found in the SP were determined after 4 (D4) and 7 (D7) days in culture (b). Total, SP and main population (MP) cells were sorted by flow cytometry and their potential to generate neurospheres was determined in limited dilution conditions (c). The percentage of clonogenicity for total neurosphere cells, MP and SP sorted cells is expressed as the mean of 24 replicate wells for four dilutions of seeded cells. The majority of neurospheres obtained in clonogenicity assays gave secondary neurospheres (d, phase contrast) that were composed of nestin-positive cells (e, nestin in blue and DNA counterstaining with 7-aminoactinomycin D, 7AAD). Neurospheres were transferred into differentiation medium for 5 days then simultaneously analysed for expression of differentiation markers (photos in fi and fii are in the same field: βIIItubulin in green; O4 in red; GFAP in blue; DNA counterstaining with DAPI). The scale bars in panels (d), (e) and (fi) represent 100 µm.

We next analysed the NSC content of the SP from neurosphere cultures based on their self-renewal capacity, as previously reported (Kim and Morshead 2003). SP and MP cells from embryonic neurospheres were sorted by flow cytometry and plated at a limiting dilution in neurosphere medium containing EGF and FGF-2. Both SP and MP cells formed neurospheres; the SP cells formed about twice as many neurospheres as the total cell population and the MP (Fig. 1c). The majority of neurospheres (78%) had characteristics of NSCs as they generated secondary neurospheres, and all of them differentiated into neuronal and glial lineages (Figs 1d, e and f). These data suggest that the SP of neurosphere cultures contain more NSCs than the MP, although self-renewal activity was observed in both.

Freshly harvested developing and adult forebrains contain SP cells

To determine whether freshly harvested forebrain contains a similar SP to that in neurospheres, we carried out a similar flow cytometric analysis on cells prepared from forebrains at various stages of development as well as from adult forebrains (Table 1 and Fig. 2). In embryonic forebrain, the profile of Hoechst 33342 staining was similar to that obtained for bone marrow (compare Figs 2ai and di; we verified that the bone marrow SP cells contained haematopoietic stem cells, data not shown). We found that freshly harvested embryonic forebrain cells contained a small SP that did not increase significantly between embryonic day 13.5 and 15.5 (from 0.20% to 0.28% of total cells; Table 1 and Fig. 2ai), whereas among the cells from the SVZ of newborn mice, 0.8% were SP cells (Table 1). The proportion of cells in the SP from embryonic forebrain was profoundly decreased by incubation with the specific ABCG2 inhibitor Ko143 (Fig. 2bi). By contrast, the addition of 50 µm verapamil, which inhibits primarily ABCB1 at this concentration, had no effect on the SP (Fig. 2ci).

Table 1.   Side population in the developing forebrain
 Developmental stage
  1. The proportion of cells in the side population (%SP) was determined by flow cytometry of total cells from either embryonic forebrains at various developmental stages (embryonic day 13.5–15.5) or from the subventricular zone of postnatal (2-day-old) mice. n, number of samples analysed.

% SP0.20 ± 0.070.24 ± 0.060.28 ± 0.050.80 ± 0.39
n =2442
Figure 2.

 The side population (SP) in freshly harvested embryonic and adult forebrains. The SP profile upon flow cytometry is shown for embryonic forebrain cells (ai–ci) and adult subventricular zone (SVZ) cells (aii–cii). The effects of the specific ABCG2 transporter inhibitor Ko143 (bi and bii) and the ABCB1-selective inhibitor verapamil (ci and cii) are shown. The classical SP profile obtained with bone marrow cells is shown in (di) and the inhibitory effect of Ko143 is shown in (dii).

We then analysed Hoechst dye efflux from the cells of freshly harvested adult SVZ, which is known to contain NSCs. The SP represented 1.0 ± 0.4% of SVZ cells (Fig. 2aii). In addition, we also observed a distinct population of cells, referred hereafter as ‘low SP’ that intensively excluded Hoechst dye. This low SP represented 7.1 ± 2.7% of adult SVZ cells. Kol43 reduced the low SP by less than 10% (Fig. 2bii), whereas verapamil almost completely removed it (Fig. 2cii). Unfortunately, the effect of the inhibitors of ABC transporters on the adult SP cells could not be analysed because verapamil and Ko143 might falsely increase the number of SP cells as a result of the shift of some low SP cells within the gate of SP (Fig. 2).

We have further characterized the expression of ABCB1 transporter by immunofluorescence in low SP cells after sorting with flow cytometry. We observed that ABCB1 was expressed in the great majority of low SP cells (Figs 3ei and eii) consistently with the inhibition of Hoechst efflux by verapamil.

Figure 3.

 The majority of side population (SP) cells are of either haematopoietic or endothelial origin. Expression of various membrane markers was determined by flow cytometry on SP and main population (MP) cells from embryonic (a) and adult (b) forebrains. MP cells (ci and cii) and low-SP cells (di–eii) were sorted from adult SVZ, and the expression of CD15, nestin (ci and cii), vWF (di and dii), ABCB1 (ei and eii) was examined by immunofluorescence. A negative control was performed with an irrelevant mouse IgG (ei).

We conclude that developing forebrain contains SP cells, and that the capacity of these cells to efflux Hoechst 33342 involves ABCG2, as we observed in embryonic neurosphere cultures, whereas adult SP cells contain a distinct population, the low SP, which probably involves ABCB1.

Brain SP contains mainly haematopoietic/endothelial cells

We next characterized the cells in the SP from freshly harvested embryonic forebrains by using antibodies against membrane markers (Fig. 3). The majority of embryonic SP cells lacked CD24, a characteristic of NSCs (Rietze et al. 2001), whereas the majority of MP cells expressed CD24 (Fig. 3a). CD15/LeX and CD133/prominin, two markers of NSCs, were expressed, respectively, on 7.1 ± 3.3% and 10.9 ± 6.4% of SP cells from freshly harvested embryonic forebrains, whereas the astroglial cell marker A2B5 was expressed on approximately half of SP cells (Fig. 3a). The haematopoietic/endothelial marker CD31 was present on 68 ± 12% of the embryonic SP cells (Fig. 3a) and on 79 ± 3% of neonatal SP cells (data not shown). Three additional markers of endothelial cells and haematopoietic stem cells, Tie2, AA4 and c-kit (data not shown), were also expressed on the majority of SP cells from embryonic forebrain. However, CD45, a pan-haematopoietic marker, was rarely expressed on embryonic forebrain cells (0.05%), and no CD45-positive cells were found within the SP (data not shown). These data suggest that most of cells in the embryonic forebrain SP were of endothelial origin and that few cells were NSCs.

As described above, adult SVZ cells could be divided into three populations based on Hoechst dye staining: an MP, an SP and a low SP. The NSC marker CD15/LeX was present on only a few SP cells (1%) and almost entirely absent from low SP, whereas it was expressed on around 5% of MP cells (Fig. 3b). By contrast, the NSC marker CD133 was expressed, respectively, on 31% and 55% of SP cells and low SP cells. Double CD133/CD31 staining showed that nearly all CD133+ cells within the SP and low SP also expressed CD31 (data not shown), suggesting that these cells were of either haematopoietic or endothelial origin. Candidate NSCs (CD133+CD31) were present in the MP but nearly absent from the SP. The A2B5 astroglial marker was present on 25% of adult SP cells but almost absent from low SP cells (Fig. 3b).

As in embryonic cells, CD31 was expressed on the majority of adult SP cells and on the low SP cells (57 ± 4% and 83 ± 13%, respectively; Fig. 3b). The haematopoietic marker CD45 was present on adult MP cells but absent from SP and low SP cells (data not shown). CD11b was present on both SP and low SP cells (data not shown), indicating the presence of microglial cells. CD49f, a marker expressed on NSCs (Campos et al. 2004) but also on haematopoietic/endothelial cells, was expressed on 22 ± 16% and 83 ± 23 of adult SP and low SP, respectively (Fig. 3b).

Expression of NSC and endothelial markers was further characterized by immunofluorescence after the sorting of MP and low SP cells from SVZ (Figs 3c–eii). CD15 expression was almost exclusively expressed in MP cells and was associated with nestin expression (Figs 3ci and cii). Expression of von Willebrand factor (vWF), a specific endothelial marker, was detected in less than 2% of MP cells, whereas a strong staining for vWF was observed in 76% of low SP cells (Figs 3di and dii).

These data suggest that most cells in the SP from freshly harvested adult and embryonic forebrains were probably of either endothelial or haematopoietic origin, whereas only a few SP cells expressed markers of NSCs.

Brain SP cells have no clonogenic capacity

To determine whether the SP from freshly prepared embryonic forebrain contained NSC activity, we assayed the clonogenicity of SP cells in the presence of EGF and FGF-2. We first verified that Hoechst 33342 staining and exposure to the laser beam during cell sorting in our conditions did not prevent neurosphere formation, that the majority of neurospheres formed in the clonogenic assay gave rise to secondary and tertiary neurospheres, and that most of them gave rise to neurons and glial cells when transferred into differentiation medium (data not shown).

In embryonic day 15.5 forebrains, the proportion of SP cells that formed neurospheres was 4.3–9.3-fold lower than the proportion of total cells that formed neurospheres, and the MP was slightly enriched in clonogenic cells when compared with the total (Table 2). By comparison, CD133+CD31 cells, which contained no SP cells, were 4.0-fold enriched in neurosphere-initiating cells compared with total cells (Table 2).

Table 2.   Neural stem cells are enriched in the main population of embryonic forebrain cells and not in the side population
  1. Clonogenicity, or number of neurosphere-initiating cells (NS-ICs), was determined in side population (SP) and main population (MP) cells from embryonic forebrains (embryonic day 15.5). Data represent the percentage of clonogenicity (NS-IC) and the number of sorted cells (× 103) obtained in two independent experiments (Exp1 and Exp2). The gate for SP selection was more stringent in Exp2 (SP = 0.35%) than in Exp1 (SP = 0.66%) to minimize contamination by MP cells.

Exp1NS-IC (%)
× 103
Exp2NS-IC (%)
× 103
 0.429 10.80.046

An autocrine factor has been reported to favour the formation of neurospheres (Taupin et al. 2000). We hypothesized that this factor may be necessary for the formation of neurospheres by SP cells. To test this hypothesis, we sorted SP cells from the forebrains of embryonic mice expressing EGFP and mixed these fluorescent cells with total C57BL6 mouse forebrain cells to allow any autocrine factor to be produced. No fluorescent SP cells grew in culture, although the normal mouse cells produced neurospheres (data not shown). Similarly, SP cells plated at high density (1000 SP cells per well of a 96-well plate) did not produce neurospheres. Therefore, the SP from embryonic forebrain did not contain the clonogenic activity associated with NSCs.

We next investigated whether the SP from adult SVZ contained NSC activity by measuring colony formation in semisolid medium in the presence of EGF and FGF-2. In this assay, colonies initiated by NSCs grow up to 21 days after plating, yielding large colonies, whereas restricted progenitors produce only small colonies (Fig. 4a). Thirty-nine percent of the large colonies (diameter ≥ 1 mm) that grew up were initiated by NSCs, as they gave rise to secondary and tertiary neurospheres, and most of them gave rise to neurons and glial cells when transferred into differentiation medium (data not shown). The number of colonies was directly correlated with the number of plated cells even with low quantities of cells (Fig. 4b).

Figure 4.

 Adult subventricular zone (SVZ) cells form colonies in neural colony-forming cell (NCFC) assay (a). Large colonies (1 mm in diameter) are noted by arrowheads. Increasing numbers of SVZ cells (× 103) were plated in NCFC assay and the total number of colonies was scored 3 weeks after plating (b).

Adult SVZ cells containing both SP and low-SP cells (called here ‘SP’ for simplicity) were compared with MP cells. We observed that these SP cells generated only a few small colonies (Fig. 5). The sorting of SP cells combined with positive selection for CD49f did not improve the efficiency of neurosphere formation. To rule out the possibility that contaminating non-neural cells within the SP might interfere with colony formation, we further selected the SP with an NSC marker; the rare SP cells that expressed CD15 (0.6%) did not give rise to colonies. By contrast, MP cells contained almost all the capacity to generate colonies and gave rise to large clones (≥1 mm in diameter) containing NSCs (Fig. 5).

Figure 5.

 The main population (MP) cells from adult forebrain form colonies, whereas the side population (SP) cells do not. SP and MP cells sorted by flow cytometry were assayed for neural colony-forming cells (NCFC) to quantify their neural stem cells (NSCs). Three weeks after plating, colonies were scored according to their diameter: white shading, colonies of less than 0.5 mm; grey shading, colonies of 0.5–1.0 mm; black shading, colonies of more than 1 mm, which were considered to derive from NSCs. NCFC (‰), or clonogenicity, is represented as the mean of four independent experiments with the total number of sorted cells (× 103) indicated below the x-axis.

Finally, we cultured SP and MP cells of adult SVZ EGFP-expressing mice in the presence of endothelial cells, which have been shown to increase the proliferation of NSCs (Shen et al. 2004; Ramirez-Castillejo et al. 2006). We observed that indeed SVZ cells generated more colonies in the presence of endothelial cells than they did on their own (Fig. 6). However, these colonies all grew from the MP cells; no SP cells formed colonies in either the presence or the absence of endothelial cells (Fig. 6).

Figure 6.

 Side population (SP) cells do not generate colonies even in the presence of endothelial cells. SP cells (5 × 103) and main population (MP) cells (10 × 103) from the subventricular zone (SVZ) of adult enhanced green fluorescent protein (EGFP)-expressing mice were analysed by neural colony forming cell (NCFC) assay in either the absence (SVZ) or the presence (+ endothelial cells) of 15 × 103 endothelial cells. Three weeks after plating fluorescent EGFP colonies were scored. Large clones (diameter = 1 mm; darkest shading) were considered to derive from NSCs.

Altogether, these data indicate that the SP prepared from freshly harvested embryonic and adult forebrains contains few, if any, NSCs.

SP contains quiescent cells but not NSC

We recently reported that the SP prepared from neurosphere cultures contained fewer cycling cells than the MP (Mathieu et al. 2006), suggesting that SP cells are quiescent as previously shown for haematopoietic stem cells (Arai et al. 2004). Consistent with this, we found by using a fluorescent stain for RNA, pyronin-Y, which labels quiescent cells weakly compared with cycling cells, that 65 ± 21% of adult SP cells and 48 ± 15% of embryonic forebrain SP cells were quiescent (Fig. 7). One might speculate therefore that any NSCs in the SP did not generate colonies in vitro because they were quiescent. NSCs from p21waf-deficient mice (p21waf–/–) cannot enter into quiescence, and consequently generate more neurospheres than NSCs from their wild-type littermates (Kippin et al. 2005). We showed that the profile of Hoechst dye efflux in adult SVZ cells isolated from p21waf–/– mice was similar to that of wild-type mice (data not shown). The percentage of CD133+CD31 cells was 1.5-fold higher in p21waf–/– mice, probably as a consequence of the more intense NSC proliferation previously reported (Kippin et al. 2005). In addition, total SVZ cells from p21waf–/– mice generated more colonies than wild-type mice (Fig. 8). We thus took advantage of the fact that quiescence is down-regulated in SVZ cells from p21waf–/– mice to determine whether the SP cells from the forebrains of these mice had the potential to generate colonies in vitro. As shown in Fig. 8, SP cells from p21waf–/– mice did not generate more colonies than wild-type SP cells.

Figure 7.

 The side population (SP) contains quiescent cells. Embryonic and adult forebrain cells were stained with Hoechst 33342 and pyronin-Y and then analysed by flow cytometry to detect quiescent cells, i.e. those with a low level of fluorescent staining with pyronin-Y (PYlow). The percentage of SP cells that are PYlow is indicated.

Figure 8.

 Side population (SP) cells from p21waf-deficient mice do not generate colonies. Total cells (14–25 × 103 cells, n = 3) and SP cells (10 × 103) from the subventricular zone (SVZ) of adult either wild-type (p21waf+/+) or p21-deficient (p21waf–/–) mice were assayed for neural colony forming cells (NCFC). Three weeks after plating, the colonies were measured and counted. White shading, colonies of less than 0.5 mm diameter; grey shading, colonies of 0.5–1.0 mm diameter, black shading, colonies of more than 1 mm diameter, which were considered to derive from NSCs.


In this study, we report that the NSCs harvested freshly from developing and adult mouse forebrains are not found in the SP that is defined by its capacity to exclude Hoechst 33342 dye. This discovery contrasts with the situation in neurosphere cultures in which the SP apparently contains NSCs (Kim and Morshead 2003). We demonstrate that the SP from forebrains contains mainly cells of haematopoietic and endothelial origin, and that these cells have no capacity to self-renew. We exclude the possibility that the SP from forebrain contains quiescent NSCs.

Flow cytometric analysis of Hoechst 33342-stained cells showed the presence of an SP in neurosphere cultures, as previously described, which is a result of the exclusion of the dye by a small proportion of the neurosphere cells (Hulspas and Quesenberry 2000; Murayama et al. 2002; Kim and Morshead 2003). We have extended this analysis by showing that the SP also exists in cells from freshly harvested developing and adult forebrains. The SP from developing forebrain was almost entirely abolished by a specific inhibitor of ABCG2 (Ko143), a typical characteristic of stem cells, whereas verapamil, an inhibitor of ABCB1, had a moderate effect. In adult SVZ, two populations that excluded Hoechst dye could be distinguished: a typical SP, similar to that in embryonic forebrain, and a low SP that excluded the dye very effectively. The low SP was very sensitive to verapamil but only weakly inhibited by Ko143. Hoechst dye efflux mediated by the ABCG2 transporter is a characteristic of various types of stem cells (Zhou et al. 2001; Lassalle et al. 2004), which suggests that the SP of embryonic forebrain contains stem cells, but that the low SP of adult forebrain does not. Hoechst dye efflux by the low SP is probably caused by the activity of the verapamil-sensitive ABCB1 transporter.

We observed that the majority of SP cells from freshly harvested embryonic and adult forebrains expressed markers of endothelial and haematopoietic stem cells (CD31, vWF, AA4, Tie-2 and c-kit), whereas a minority (< 10%) expressed CD15/LeX, which is reported to be enriched in NSCs (Capela and Temple 2002; Corti et al. 2005). On the other hand, CD133, which has been shown to be expressed on embryonic NSCs (Sawamoto et al. 2001), was present on SP cells from the adult SVZ, but almost all these CD133+ cells also expressed the haematopoietic/endothelial cell marker CD31, arguing strongly that these SP CD133+ cells are of either haematopoietic or endothelial origin. Brain endothelial cells, which are involved in the impermeability of the blood–brain barrier, have active ABCB1 and ABGC2 transporters (Cisternino et al. 2004; Mercier et al. 2004), and microglial cells, which are of haematopoietic origin, are also reported to express functional ABC transporters (Dallas et al. 2003). Furthermore, we observed that low SP from adult SVZ contained CD11b+ microglial cells (data not shown). Hence, the SP phenotype may correspond to endothelial cells in developing brain and involve ABCG2, whereas in the adult brain both microglial and endothelial cells may efflux Hoechst in an ABCB1-dependent manner.

Although the SP of embryonic and adult forebrains seems to comprise mainly endothelial and/or microglial cells, the presence of a small proportion of cells bearing the NSC-related marker CD15+ suggested that the SP might also contain some NSCs. Therefore, we used clonogenic assays to identify candidate NSCs in sorted populations by their capacity to generate neurospheres in vitro. We confirmed previous data obtained on fetal human brain (Uchida et al. 2000) that indicate the selection of CD133+CD31 cells enriched NSCs from embryonic forebrains. Under the same conditions, SP cells from embryonic forebrains gave rise to dramatically fewer neurospheres than did the MP cells from the same tissue. We ruled out the possibilities that poor neurosphere formation by SP cells was a result of the lack of an autocrine factor. Similar results were obtained with adult SVZ cells: SP cells were unable to produce colonies in an in vitro NSC assay, whereas MP cells produced NSC-derived colonies. Furthermore, even in the presence of endothelial cells, which are known to stimulate NSC proliferation (Shen et al. 2004; Ramirez-Castillejo et al. 2006), SP cells from adult SVZ did not produce colonies, whereas they improved colony formation by MP cells.

The absence of NSCs from the SP cannot be the result of a failure to sort NSCs with our Hoechst dye exclusion method because we observed that SP cells from neurosphere cultures were more clonogenic than MP cells from neurospheres. Moreover, we confirmed that, in our hands, Hoechst dye efflux enriched haematopoietic stem cells from bone marrow and germinal stem cells (Lassalle et al. 2004). Consistent with our findings, NSCs enriched from adult SVZ by double-negative selection for PNA and CD24 produced cells that mostly labelled intensely with Hoechst 33342 (Rietze et al. 2001).

To exclude the possibility that SP cells do not generate neurospheres in vitro because they are in a quiescent state, we used SVZ cells from p21waf-deficient mice that cannot enter quiescence (Kippin et al. 2005). We observed that SP cells from these mice do not form neurospheres, whereas total SVZ cells generated more neurospheres compared with those from wild-type mice. Altogether, our data demonstrate that NSCs from forebrains do not exclude Hoechst dye and that the SP from forebrain does not contain quiescent NSCs.

We observed that SP cells from neurosphere cultures had only a two-fold higher rate of clonogenicity than MP cells. By contrast, Kim and Morshead (2003) have reported that NSCs from neurospheres are almost all contained in the SP. This discrepancy probably results from different culture conditions and media, as our culture conditions gave rise to smaller neurospheres than those in Kim and Morshead's study (500–1000 cells/neurosphere instead of 15 000–20 000 cells/neurosphere).

It is surprising that there is such a qualitative difference between the NSCs in fresh brain tissues and those in neurosphere cultures. Our data indicate that the SP in neurospheres increases with time in culture, perhaps reflecting the size of the neurospheres. In addition, a lower percentage of SP cells were obtained in our culture conditions as compared with Morshead's group (< 1% instead of 3.6%), and this correlates with smaller neurosphere size. As the neurospheres grow, the cells inside them are exposed to a lower level of oxygen and this may induce the ABCG2 transporter in response to hypoxia, as recently reported for haematopoietic cells (Krishnamurthy et al. 2004). Consistent with this idea, we observed that the SP increased in the presence of cobalt chloride, a chemical inducer of hypoxic-like conditions (data not shown). Hypoxia may therefore regulate Hoechst-dye efflux in neurosphere cultures in cells that have characteristics of NSCs, but further experiments are needed to determine if it is such a case in vivo.

Altogether our data demonstrate that NSCs are not contained in SP from freshly harvested developing and adult forebrains. The SP is rather caused by the presence of endothelial and/or microglial cells. Nonetheless, we could not totally rule out that a Hoechst efflux could emerge in NSCs in response to several stresses such as hypoxia. In conclusion, although the SP in neurosphere cultures contained a fraction of NSCs, Hoechst efflux could not be used as a marker to purify NSCs in embryonic and adult forebrains.


We would like to thank Dr A. H. Schinkel (Institute of Molecular Chemistry, University of Amsterdam, Amsterdam, the Netherlands) for providing us with the ABCG2 inhibitor Ko143 and Dr M. Okabe (Research Institute for Microbial Diseases, Osaka University, Japan) for EGFP mice. We also thank V. Neuville, C. Chauveau and S. Leblay for their technical assistance in our animal facilities and Dr L. Gauthier for helpful discussions. This work was supported by a grant from Electricité de France.