C.Y.B. and T.L.L. contributed equally to this work.
Sox2 expression defines a heterogeneous population of neurosphere-forming cells in the adult murine brain
Version of Record online: 6 JUL 2005
Volume 4, Issue 4, pages 197–207, August 2005
How to Cite
Brazel, C. Y., Limke, T. L., Osborne, J. K., Miura, T., Cai, J., Pevny, L. and Rao, M. S. (2005), Sox2 expression defines a heterogeneous population of neurosphere-forming cells in the adult murine brain. Aging Cell, 4: 197–207. doi: 10.1111/j.1474-9726.2005.00158.x
- Issue online: 6 JUL 2005
- Version of Record online: 6 JUL 2005
- Accepted for publication 16 May 2005
- neural stem cell;
- subgranular layer;
- subventricular zone
The identification of neural stem cells (NSCs) in situ has been prevented by the inability to identify a marker consistently expressed in all adult NSCs and is thus generally accomplished using the in vitro neurosphere-forming assay. The high-mobility group transcription factor Sox2 is expressed in embryonic neural epithelial stem cells; because these cells are thought to give rise to the adult NSC population, we hypothesized that Sox2 may continue to be expressed in adult NSCs. Using Sox2:EGFP transgenic mice, we show that Sox2 is expressed in neurogenic regions along the rostral–caudal axis of the central nervous system throughout life. Furthermore, all neurospheres derived from these neurogenic regions express Sox2, suggesting that Sox2 is indeed expressed in adult NSCs. We demonstrate that NSCs are heterogeneous within the adult brain, with differing capacities for cell production. In vitro, all neurospheres express Sox2, but the expression of markers common to early progenitor cells within individual neurospheres varies; this heterogeneity of NSCs is mirrored in vivo. For example, both glial fibrillary acidic protein and NG2 are expressed within individual neurospheres, but their expression is mutually exclusive; likewise, these two markers show distinct staining patterns within the Sox2+ regions of the brain's neurogenic regions. Thus, we propose that the expression of Sox2 is a unifying characteristic of NSCs in the adult brain, but that not all NSCs maintain the ability to form all neural cell types in vivo.
The identification of neural stem cells (NSCs) in situ has been difficult due to the lack of markers that can reliably identify these cells. Therefore, NSCs are frequently identified in vitro as those cells that are capable of forming multipotent self-renewing aggregates of cells termed neurospheres. Nestin has been suggested to label NSCs; however, this protein is also present in immature neurons, and when green fluorescent protein (GFP)-labeled nestin+ cells are sorted and tested for neurosphere formation, only 0.4% of the cells go on to form neurospheres, suggesting that only a small fraction of the nestin+ cells in vivo are actually NSCs (Mignone et al., 2004). Furthermore, more and more regions of the adult brain are being shown to harbor multipotent cells, including the subventricular zone (SVZ) surrounding the anterior lateral ventricle, the subgranular layer of the hippocampus (SGL) and polysialated neural cell adhesion molecule (PSA-NCAM)-positive cells within the parenchyma (reviewed in Pevny & Rao, 2003). Likewise, progenitor cell populations that previously had been thought to produce a single cell type in vivo acquire stem-cell characteristics when placed in permissive conditions in vitro (Kondo & Raff, 2000; Laywell et al., 2000; Gabay et al., 2003; Nunes et al., 2003). Thus, a unifying characteristic of neural stem cells from various regions of the brain has yet to be identified.
Recent studies suggest that the high-mobility group transcription factor Sox2 is present in NSCs (Avilion et al., 2003; D’Amour & Gage, 2003; Graham et al., 2003). During early development, Sox2 is seen in cells of the neural tube (Cai et al., 2002); as development proceeds, Sox2+ cells become largely restricted to the VZ and delaminate toward the mantle zone following their final mitosis. Sox2 is necessary for the maintenance of NSCs, as inhibition of Sox2 signaling promotes an early exit of neural precursors from the cell cycle, while constitutive expression inhibits neuronal differentiation (Graham et al., 2003). Additionally, all Sox2-expressing neuroepithelial (NEP) cells co-express nestin, and Sox2 expression is lost upon neuronal differentiation, although expression is retained in early glial restricted precursors. In the adult mouse, all neurospheres derived from the SVZ express Sox2 (Ellis et al., 2004), further suggesting that the expression of Sox2 may be common among different neural stem-cell populations.
In the present study, we hypothesize that Sox2 can be used as a marker for NSCs throughout lifespan. We show that Sox2 expression is maintained in neurogenic regions of the central nervous system (CNS) throughout adulthood, and that all neurospheres formed in vitro from these neurogenic regions express Sox2. We also show that, although all neurospheres express Sox2, other markers common for early progenitor cells are differentially expressed among the neurospheres. This pattern of differential expression is preserved in vivo, suggesting that neural stem cells isolated from different regions of the brain may be intrinsically dissimilar to each other.
Sox2:EGFP expression in the adult mouse brain
The high-mobility group transcription factor Sox2 is first expressed during embryogenesis in neuroepithelial stem cells; its expression in neurogenic regions of the brain – such as the SVZ lining the lateral ventricle and the SGL within the hippocampus – is maintained throughout adulthood (Ellis et al., 2004; Ferri et al., 2004). Further to determine Sox2 expression within the adult mouse brain, we utilized the Sox2:EGFP transgenic mouse, where enhanced green fluorescent protein (EGFP) is expressed from the Sox2 promoter. Thus, EGFP expression serves as a surrogate marker for Sox2 expression. In these animals, EGFP expression in situ mirrors Sox2 expression as detected by immunohistochemistry (data not shown). Ten-micron-thick frozen sections were analysed from young adult (4–8 week) and middle-aged (1 year) adult mice. EGFP was expressed throughout the entire rostrocadal axis in both young adult and middle-aged mice (summarized in Table 1). EGFP was expressed in the core of the olfactory bulb of both young adult and middle-aged animals, although expression was less pronounced in the older animals (Fig. 1A,B). Highest EGFP expression was found in regions enriched in NSCs, the SVZ and the SGL, in both young and adolescent mice. Along the anterior lateral ventricle (aLV), EGFP expression was seen in more than 50% of the cells comprising the ependymal layer and the immediately subjacent SVZ (Fig. 1C,D). EGFP was also pronounced within the hippocampus, localizing to the hilus, the SGL of the dentate gyrus (DG), and the ependymal and subependymal layers of the fimbria (see Fig. 2B,E). Scattered cells within the cortex in both young adult and middle-aged animals were EGFP+ (Fig. 1E,F), as were small cells proximal to the Purkinje cells of the cerebellum (Fig. 1G,H). In addition, EGFP was expressed in ependymal cells lining the central canal of the spinal cord (Fig. 1I,J). Middle-aged animals contained fewer Sox2+ cells in all regions examined as assessed by counting nuclei in sections of the hippocampus and the rostral migrating stream. Numbers decline even further in older animals (data not shown), although no detailed quantitative assessment was made.
|Brain region||Subregion||Sox2:EGFP expression|
|Rostral migratory stream||+++||++|
|Anterior lateral ventricle||Ependyma||+++||++|
|Cerebellum||Small cells adjacent to Purkinje cells||++||+|
|Scattered cells in white and grey matter||+||+|
Sox2:EGFP is expressed in neurosphere-forming cells, and expression is lost upon differentiation
Previous studies indicate that Sox2+ NEP cells are multipotent and form neurospheres in vitro (Graham et al., 2003). To determine whether neurosphere-forming cells from the adult mouse brain express Sox2, we generated neurospheres from the SVZ, hippocampus, olfactory bulb and cortex of young adult (4–8 weeks) Sox2:EGFP mice. Sox2 expression was assessed visually with EGFP and on the molecular level using the reverse transcription-polymerase chain reaction (RT-PCR). The analysis was limited to passage 1–3 neurospheres initially. At 7 days in vitro (div), more than 70% of the spheres arising from the SVZ and the hippocampus expressed EGFP (Fig. 2A,B). Likewise, the majority of spheres arising from the olfactory bulb and the cortex also expressed EGFP (Fig. 2C,D). More neurospheres were isolated from the SVZ and were larger than spheres arising from the other three regions (data not shown). Interestingly, even in early-passage neurospheres at 7 div, not all of the cells within the sphere appeared to be EGFP+, suggesting that neurospheres arising from young adult mouse brain comprised a heterogeneous population of cells. The number of EGFP-expressing cells in a neurospehre declined rapidly and by the tenth passage fewer than 1% of cells were green.
To determine whether EGFP expression is maintained following passaging, primary spheres from the SVZ were enzymatically dissociated into a single cell suspension and re-plated at clonal density in the presence of EGF and fibroblast growth factor-2 (FGF-2) for 7 div. EGFP was expressed in a substantial fraction of the resultant secondary neurospheres (Fig. 2E). We next determined whether differentiated cells continue to express Sox2. Primary SVZ spheres were adhered to poly-l-lysine/laminin-coated six-well plates in DMEM:F12 containing N2 but no EGF or FGF-2. These conditions allow for the attachment of neurospheres and their spontaneous differentiation into neurons, astrocytes and oligodendrocytes. On the day of plating (day 0), EGFP was highly expressed in the primary neurosphere (Fig. 2F). By 2 div (day 2), the sphere had attached to the substrate, and the majority of the cells had crawled out of the main body of the sphere (Fig. 2G). At this stage, it is clear that some cells no longer express EGFP (Fig. 2G). By 5 div (day 5), the majority of the cells no longer express EGFP (Fig. 2H). Thus, the majority of neurosphere-forming cells isolated from the adult brain express EGFP when the cells are maintained in EFG/FGF-2; EGFP expression is lost as the cells begin to differentiate.
We next determined whether Sox2 was characteristic of fetal human cortical spheres. Cortical neurospheres were cultured from the E16 Sox2:EGFP mouse or human fetus. At 7 div, spheres were frozen for Sox2 immunohistochemistry on 10-µm sections. Sox2 expression (in green) was seen in both murine (Fig. 2I) and human (Fig. 2J) cortical neurospheres. RT-PCR confirmed the presence of Sox2 mRNA in these cultures (Fig. 2K), as well as in neurospheres cultured from the adult murine SVZ and hippocampus, suggesting that Sox2 expression is characteristic of neurosphere cultures.
Newly formed neurospheres from adolescent young adult are a heterogeneous population
The above data suggest that whereas both embryonic and adult neurospheres abundantly express Sox2, neurospheres derived from various brain regions comprise a heterogeneous population of cells after even limited culture, as reported by multiple groups (reviewed in Pevny & Rao, 2003). To test this hypothesis further, neurospheres were isolated from the young adult murine SVZ and analysed for markers indicative of undifferentiated and differentiating cells. All neurospheres tested were positive for nestin (Fig. 3A), Sox1 (Fig. 3B) and epidermal growth factor receptor expression (EGFR, Fig. 3C), all markers of immature undifferentiated cells. Likewise, neurospheres were negative for MAP2 expression (for mature neurons, Fig. 3D), PSA-NCAM (a dividing neuronal precursor, data not shown) and A2B5 (glial restricted precursor, data not shown), suggesting that the neurospheres comprised predominantly undifferentiated cells.
Varying subclasses of neural stem cells have been described within the SVZ in situ; each subclass retains the capacity for neurosphere formation and gives rise to neurons, oligodendroctyes and astrocytes in vitro. These subclasses include the type B cell that expresses glial fibrillary acidic protein (GFAP) (Doetsch et al., 1999; Laywell et al., 2000), NG2+ and/or Olig2+ cells (Takebayashi et al., 2000; Belachew et al., 2003), and ependymal cells (Johansson et al., 1999). However, it has not been precisely determined whether there is any overlap within these cell populations. Thus, we next stained young adult SVZ neurospheres for markers indicative of the above cell types. S100β, a calcium binding protein most abundantly expressed in astrocytes, was expressed in nearly 100% of the neurospheres observed (Fig. 3K). Likewise, the transcription factor Olig2 is expressed in nearly 100% of the neurospheres examined (Fig. 3H). Importantly, Olig2 was expressed evenly throughout the entire neurosphere, in sharp contrast to what is seen in neurospheres isolated from embryonic animals. In primary neurosphere cultures derived from the embryonic day 16 (E16) Sox2:EGFP mouse, Olig2 is expressed in only a few cells at the periphery of the sphere (Fig. 3I). Furthermore, Olig2 expression is lost when embryonic neurospheres are passaged and embryonic neurospheres do not express S100β (data not shown).
Labeling with GFAP and NG2 (Fig. 3E–G) allowed us to further divide adult neurospheres into two distinct classes: those that were highly GFAP+, and those which were highly NG2+. Sixty per cent of the neurospheres tested contained GFAP+ cells (Fig. 3E; 66/107) half of which expressed GFAP in the majority of the cells within the neurosphere (31/107), while the other half expressed GFAP in only a few cells (35/107). The remaining 40% (41/107) did not express GFAP. By contrast, 15% of neurospheres tested comprised mainly of NG2+ cells (5/38), while 50% of the neurospheres contained no NG2+ cells (20/38). Co-expression of GFAP and NG2 within the same neurosphere was rare, with very few observed instances of co-expression in the same cell (Fig. 3G). These data suggest that neurospheres having a high content of either GFAP+ or NG2+ cells comprise non-overlapping populations. Double-labeling experiments reveal that GFAP+ spheres also contain some olig2- and S-100β-positive cells (Fig. 3J–L), although several Olig2+/GFAP– neurospheres were seen (Fig. 3J).
Labeling with CD24 failed to identify any CD24+ neurosphere population in SVZ-derived neurosphere culture, although CD24/Sox2 co-expression was seen in vivo (see below). RT-PCR indicated CD24 expression within the SVZ neurosphere population (data not shown), but the extent of this expression could not be verified by antibody labeling.
Adult neurosphere heterogeneity in vitro is reflected by Sox2:EGFP+ cells in vivo
Based on the above observations that early-passage Sox2+ neurospheres are heterogeneous in nature with at least two distinct neurosphere populations present in vitro, we hypothesized that Sox2+ cells in vivo could likewise be categorized into at least two different populations. To test this hypothesis, we labeled 10-µm cryostat sections from young adult Sox2:EGFP mice with antibodies against GFAP, NG2, S100-β and Olig2. Representative examples of staining are shown in Figs 4 and 5.
As seen in adult SVZ neurospheres, Sox2:EGFP and S100β are co-expressed in situ, primarily within the ependymal cell layer surrounding the lateral ventricles and in a significant number of subependymal cells (Fig. 4A). However, there are also many S100β(–)/Sox2:EGFP+ cells present in this region (Fig. 4A, inset, arrow). S100β is also strongly co-expressed with Sox2:EGFP in the SGL and hilus of the hippocampus (Fig. 4B), and in the fimbria (Fig. 4C). Again, not all Sox2:EGFP+ cells were positive for S100β, and vice-versa.
Similarly, a small number of Olig2+ cells are located in the SVZ surrounding the aLV, where confocal analysis confirmed that all Olig2+ cells in this area are Sox2:EGFP+ (Fig. 4D). There are also a small number of Olig2+ cells in the hilus and SGZ of the hippocampus, all of which co-express Sox2:EGFP (Fig. 4E). Intriguingly, there are a large number of Olig2+/Sox2:EGFP+ cells in the subependymal region of the fimbria (Fig. 4F), which is reportedly the primary source of multipotent NSCs in adult neurosphere preparation from the hippocampus (Seaberg & van der Kooy, 2002). Although Olig2 expression is observed in white matter, Sox2:EGFP is excluded from these areas (data not shown).
Occasional Sox2:EGFP+/GFAP+ cells were found scattered throughout the brain in cells with significant arborizations that could not be distinguished morphologically from Sox2:EGFP(–)/GFAP+ cells in situ. In the area around the aLV, GFAP was co-expressed with Sox2:EGFP in subependymal cells (Fig. 5A), particularly in cells lying in the SVZ directly adjacent to the ependymal cells (Fig. 5B). In the cortex, the majority of the scattered Sox2:EGFP+ cells co-expressed GFAP (Fig. 5C), and a large number of Sox2:EGFP+ cells in both the hippocampal SGL and hilus were GFAP+ (Fig. 5D). By contrast, the majority of the Sox2:EGFP+ cells in the fimbria were GFAP-negative (Fig. 5E), where a large number of Olig2+/Sox2:EGFP+ cells were observed (Fig. 4F). It should be noted that in both the SVZ and the hippocampus, there were a number of non-ependymal, GFAP–/Sox2:EGFP+ cells (yellow arrows, Fig. 5A,D). Furthermore, the vast majority of GFAP cells throughout the adult brain were both Sox2– and Olig2– (data not shown), contrasting with the significant overlap between GFAP and Olig2 in adult neurospheres (Fig. 3J).
Sox2:EGFP+/NG2+ cells were present in the SVZ surrounding the aLV (Fig. 5F,G), as well as in the SGL of the hippocampus (Fig. 5H) and in the subependymal region of the fimbria (Fig. 5I). However, NG2+/Sox2:EGFP+ cells constitute a minority of the Sox2+ cells in all regions examined, consistent with the low percentage of highly NG2+ neurospheres formed in vitro. Combined, these data indicate that the Sox2-expressing population of cells in situ is a heterogeneous population in which subpopulations can be identified on the basis of marker expression.
Telomerase activity is more abundant in embryonic neurospheres than in neurospehres from young adult animals
To determine whether telomerase activity in Sox2+ cells declines with age, we propagated neurospheres from E16 or from the SVZ of young adult (4 weeks) Sox2:EGFP mice and assayed for telomerase activity utilizing the Telomeric Repeat Amplification Protocol (TRAP). Negative control samples for each group were prepared by heat-treating the lysates (Fig. 6A). The sum of the relative telomerase activities were calculated for each lane containing 50 ng protein, and compared with the E16 neurosphere sample #1 (Fig. 6A, lane 1). As shown in Fig. 6(B), the spheres derived from young adult animals contained approximately 80% less telomerase activity than E16 spheres, suggesting that perhaps one mechanism by which neural stem cells decrease in number in the adult brain is through the loss of telomerase activity. Telomerase levels were undetectable in the neurospheres obtained from older animals and from passaged fetal neurospheres and those derived from adult human tissue (data not shown).
The high-mobility group transcription factor Sox2 is first expressed during early embryonic development in neuroepithelial stem cells, and is thought to remain active in multipotent neural cells throughout adulthood (Graham et al., 2003; Ferri et al., 2004). Here we show that Sox2+ cells are, in fact, present in neurogenic regions of the mouse brain at up to 1 year of age. These Sox2+ cells give rise to neurospheres in vitro, and the total number of neurospheres derived from each brain region can be correlated to the number of Sox2+ cells present in that region in vivo. Thus, a larger number of neurospheres form from adult SVZ and hippocampal dissections than from dissections of the olfactory bulb or cortex. Additionally, individual Sox2:EGFP+ neurospheres variably express GFAP, NG2, S-100β and Olig2, an observation that is supported by the differential expression of these same markers with Sox2+ cells in vivo. Finally, Sox2 expression is lost upon neurosphere differentiation, suggesting that the expression of this transcription factor may be unique to multipotent stem and progenitor cells in neurogenic regions of the brain.
The idea that multiple populations of neural stem/progenitor cells (NSPs) exist in vivo is not a new one (Zappone et al., 2000; Hitoshi et al., 2002; and references below). In fact, many laboratories have described neural stem/progenitor cells of different competencies in both rodents and humans, largely dependent upon the region from which these cells were isolated (for example, Sheen et al., 1999; Nunes et al., 2003; Pevny & Rao, 2003). The data presented here support and extend this hypothesis, suggesting not only that various populations of NSPs exist within the adult murine brain, but also that one unifying characteristic of these early progenitors is the expression of Sox2. When grown in media allowing for NSP proliferation, individual Sox2+ neurospheres express either GFAP or NG2 with little overlap (Fig. 3). Additionally, in vivo, GFAP and NG2 show distinct staining patterns within the Sox2+ regions of the SVZ, SGL and hippocampal fimbria (Fig. 5). Thus, at least two populations of neural stem/progenitor cells exist within these regions.
Previous studies have shown that both GFAP+ and GFAP– progenitors reside within the adult brain (Palmer et al., 1999; Doetsch et al., 2002; Seaberg & van der Kooy, 2002; Mignone et al., 2004) and clonal analysis of neurospheres has suggested the presence of GFAP+ and GFAP– neurospheres (Suslov et al., 2002; Steiner et al., 2004). Our results support these findings, as Sox2+ neurospheres were not always positive for GFAP. Other results have suggested positional heterogeneity of stem cells (Zappone et al., 2000; Hitoshi et al., 2002) and are consistent with our data showing regional variability in antigen expression of Sox2 cells. Interestingly, Dlx2+ progenitor cells give rise to 70% of the neurospheres seen in vitro, all of which are GFAP– (Doetsch et al., 2002; Seri et al., 2001, 2004), supporting the hypothesis that GFAP– cells can give rise to neurospheres. Likewise, in our experiments, 70% of the neurospheres derived from the SVZ were GFAP– (Fig. 3). It is likely that these GFAP– neurospheres arose from Dlx2+ cells, while the remaining 30% arose from Dlx2– cells, although further work will need to be performed to answer this question definitively. These results contrast with previous findings showing the elimination of neurosphere formation upon ablation of GFAP-expressing cells (Imura et al., 2003; Morshead et al., 2003). Further experiments will be required to reconcile these results.
As with a subset of GFAP-expressing cells, a subset of NG2-expressing cells appears to have stem-cell properties. NG2+ cells in the adult brain, particularly the hippocampus, were recently identified as being multipotential both in vitro and in vivo (Belachew et al., 2003). NG2-expressing cells have previously been classified as a large, slowly proliferating pool of oligodendrocyte progenitors in the postnatal brain (reviewed in Dawson et al., 2000; Levine et al., 2001). Kondo & Raff (2000) have suggested that NG2+ oligodendrocyte progenitor cells could be ‘reprogrammed’ into multipotent NSCs through sequential exposure to a variety of growth factors. Our data showing co-expression of Sox2 and NG2 in a subset of cells both in vitro and in vivo suggest that exposure to different growth factors may not ‘reprogram’ oligodendrocyte-committed progenitor cells, but rather may promote proliferation of a subset of Sox2+ cells that are also NG2+. Note that a recent study has shown that Olig2+ cells are multipotent in culture (Gabay et al., 2003).
It is important to note that fetal-derived neurospheres differ from adult neurospheres. For instance, we show here that fetally derived neurospheres are highly enriched in telomerase activity, and by postnatal week 4 this activity has dropped by 80% (Fig. 6). This result is supported by earlier data describing a decline in telomerase activity in both the SVZ and the olfactory bulb by 60% over the first 7 days of life (Caporaso et al., 2003). Likewise, neither fetal neurospheres nor neuroepithelial stem cells (NEP) express GFAP, NG2, S-100β or Olig2. However, NEPs do express Sox2, can self-renew and are multipotent, as are adult NSCs (reviewed in Pevny & Rao, 2003). Therefore, NSCs may undergo maturation during development while retaining varying degrees of multipotentiality. This maturation is reminiscent of the maturation of ventricular zone-derived neuroepithelial stem cells to SVZ-type neurosphere-forming stem-cell pools, which differ in their cytokine dependence and relative proportion of neurons and astrocytes formed. We suggest that the phenotypically distinct populations of Sox2+ neurosphere-forming cells present in the adult may likewise differ in the kinds of neurons or proportion of neurons and glia generated. Confirmation will require sorting individual populations of spheres, which is currently technically challenging. The more recent availability of NG2-GFP mice and Olig2-GFP mice may make this feasible in the near future.
Our data also suggest that cells within neurospheres acquire additional markers in vitro. In particular, the number of Olig2-expressing neurospheres far exceeds the fraction of Olig2+ cells in vivo. Additionally, triple-labeled GFAP+/Olig2+/Sox2+ cells and NG2+/Olig2+/Sox2+ cells are present in the neurospheres. This raises the possibility that either Olig2+/Sox2– cells present in vivo de-differentiate into stem cells when placed in vitro (Gabay et al., 2003), or that Olig2 expression is rapidly induced as cells are maintained in culture. We cannot at present distinguish between these possibilities, although we tend to favor the latter as it has been difficult to de-differentiate Olig2+ glial progenitors into multipotent stem/progenitor cells (our unpublished data), and it has been difficult to induce Sox2 expression in differentiated cells (our unpublished data). In either case, cells maintained in culture acquire phenotypes or properties that have no correlating phenotype in vivo, supporting the notion that adult neurospheres may consist of, or form from, artificially de-differentiated cells. Indeed, previous work shows that transit-amplifying neural precursors can be converted into multipotent cells following EGF exposure (Doetsch et al., 2002). Likewise, when Olig2+ cells isolated from the spinal cord are exposed to FGF-2 in vitro, these otherwise bi-potent cells are converted into cells which produce neurons, astrocytes and oligodendrocytes upon differentiation (Gabay et al., 2003). Our results suggest that even this de-differentiated cell will require the acquisition of Sox2 to function as a true neural stem cell.
From this study alone, it is difficult to determine whether Sox2 identifies primarily neural stem cells, or both NSCs and rapidly proliferating multipotent progenitors. Furthermore, GFP is likely down-regulated at a slower rate than Sox2 and may persist in cells that have recently lost their stem-cell characteristics. However, our data showing a reduction in the number of Sox2+ cells in the aging brain – and by correlation a reduction in the total number of stem cells – is supported by previous studies in which the G1 stage-specific replication factor Mcm2 was used to identify NSCs in young and aging mice (Maslov et al., 2004). The authors demonstrated an approximate two-fold reduction in the number of NSCs in the SVZ of aging mice. As these two technically different studies show agreement in results, we argue that Sox2 can be used as a reliable marker for NSCs in vivo. In vivo, Sox2:EGFP+ cells are clearly distinguished from migrating doublecortin-expressing cells in the dentate gyrus, suggesting that Sox2 expression is restricted to pre-migratory cells (data not shown). By contrast, Sox2 is expressed in doublecortin-positive cells within the rostral migratory stream, indicating that Sox2 is differentially down-regulated in different regions of the brain. In our in vitro studies, primary neurospheres comprised mainly Sox2:EGFP-expressing cells, suggesting that Sox2 may be expressed in both NSCs and rapidly proliferating progenitors. Alternatively, Sox2 may be present in only the NSC population, but EGFP expression persists due to a lag-time in protein breakdown.
The level of neurogenesis in the brain is known to decrease with age. Likewise, recent data suggest that neural stem cells also age (Bailey et al., 2004). Furthermore, our results documenting the decrease in telomerase levels suggest that the self-renewal capabilities of adult neurospheres may be much less than that of younger stem cells and could account for the decreased neurogenesis seen. The previously documented decrease in the number of NSCs within neurogenic regions, a decrease in the number of Sox2+ cells in vivo and a reduction in the number of neurospheres obtained in vitro all suggest that stem cells are altered as the animal ages, and that changes in stem-cell properties could account for the reduction of neurogenesis in aged animals (Maslov et al., 2004). Additionally, based on the data presented here showing that not all NSCs are similarly competent, it is plausible that as animals age, the proportion of NSCs capable of producing neurons declines, resulting in a concomitant increase in the proportion of glia- producing NSCs. Future experiments are required to test the validity of these hypotheses.
Neurospheres and tissue sections were obtained from young adult (4–8 weeks) Sox2:EGFP mice. Sox2:EGFP mice were generated as described previously (Ellis et al., 2004). All Sox2:EGFP mice were heterozygous for the replacement of the Sox2 gene with EGFP, as the homozygous replacement is embryonic lethal. Genotyping of Sox2:EGFP mice was confirmed using previously described primer sequences. All animals were housed and bred following NIH animal use guidelines.
Young adult (4–8 weeks) and middle-aged (1 year) brains from Sox2:EGFP mice were fixed, frozen and sectioned as described previously (Limke et al., 2003). Sections were cut at 8–12 µm thickness and mounted on laminin-coated coverslips for immunohistochemistry. Blocking buffer, primary antibodies and secondary antibodies were applied as described previously (Limke et al., 2003). Primary antibodies included: βIII tubulin, MAP2 and s100 (Sigma); CD24 and GFAP (BD Pharmingen); doublecortin, GFP, Nestin and Sox2 (Chemicon); EGFR (Santa Cruz Biotech); NG2 (kind gift from Dr W. Stallcup); Olig2 (kind gift from Dr D. Rowitch); and Sox1 and Sox2 (kind gifts from Dr L. Pevny). Appropriate secondary antibodies conjugated to Alexafluor 568 or Alexafluor 488 (Molecular Probes, Eugene, OR, USA); FITC- or TRITC-conjugated anti-mouse (Southern Biotechnology, Birmingham, AL, USA) and FITC- or TRITC- conjugated anti-rat, or FITC- or Cy3-conjugated anti-mouse (Jackson ImmunoResearch, West Grove, PA, USA) were applied. Nuclei were visualized with DAPI (1 : 1000, Sigma, St Louis, MO, USA). Control experiments were performed simultaneously using sections exposed to blocking buffer without primary antibody, followed by secondary antibody. Antibody labeling was performed in at least two separate experiments using sections from different, non-littermate animals. Slides were rinsed in distilled water, preserved with Fluorsave (Calbiochem, San Diego, CA, USA) and cover-slipped. Images were captured with a digital camera coupled to an Olympus microscope and processed using Magnafire and Adobe Photoshop software. For detailed analysis of antigen co-expression, z-stack images were obtained using an Olympus confocal microscope at a frequency of 0.5 µm to permit visualization of cells throughout the section.
Adult mouse neurosphere cultures
Neurospheres were derived from young adult (4–5 weeks) wild-type C57Bl/6 and Sox2:EGFP mice following previously established protocols (Gritti et al., 1996; Tropepe et al., 1999). Briefly, the region surrounding the aLV, including the ependymal and subependymal zones, was isolated and enzymatically dissociated to produce forebrain neurosphere cultures, while the hippocampus was isolated and enzymatically dissociated to produce hippocampal neurospheres. Single cells were suspended in uncoated 24-well culture plates in growth medium consisting of DMEM/F12 (Gibco, Grand Island, NY, USA) supplemented with N2 (Gibco), 20 ng mL−1 EGF (Peprotech, Rocky Hill, NJ, USA), 20 ng mL−1 FGF-2 (Peprotech), 2 µg mL−1 heparin (Sigma) and penicillin-streptomycin (Sigma) at a density of 2 × 104 cells mL−1. Cells were fed every 2–4 days by supplementation with FGF-2 and EGF. Cell passaging was accomplished by incubating primary spheres in Trypsin/EDTA for 10 min followed by mechanical dissociation of the spheres into a single cell suspension. Cells were replated at 2 × 104 cells mL−1 in growth medium to allow secondary neurosphere formation.
Embyronic mouse and human neurosphere cultures
Timed-pregnant Sox2:EGFP or C57Bl/6 mice were anesthetized and the pups removed and decapitated in sterile PBS on E16. Cortices were dissected and placed into ice-cold PBS. The tissue was minced and enzymatically dissociated using trypsin-EDTA (Sigma) for 10 min. Trypsin was inhibited with 10% FBS for 1 min. FBS was removed and the cells were mechanically dissociated. Cells were washed three times with Neurocult media (Stem Cell Technologies, Vancouver, BC, Canada) before being plated at a density of 4 × 104 cells mL−1 in Neurocult media supplemented with murine growth supplement (Neurocult) and 20 ng mL−1 EGF. Neurospheres were harvested at 7 div for the isolation of RNA or freezing for immunocytochemistry.
Human cortical neurospheres derived from late first trimester (∼110 gestational days) human cortex were a kind gift from Dr Norm Haughey, Johns Hopkins University.
Mouse neuroepithelial (NEP) cell cultures
NEP cells were cultured from E8 mice, as described previously (Cai et al., 2002). Briefly, the trunk segments of the embryos were dissected and incubated in an enzyme solution containing Collagenase type I (1 mg mL−1; Worthington Biochemical) and Dispase II (2 mg mL−1; Roche) in Ca2+/Mg2+-free Hank's balanced salt solution (Gibco/BRL), at room temperature for approximately 10 min. The enzyme solution was then replaced by NEP basal medium (Kalyani et al., 1997) with 10% chicken embryo extract (CEE). The segments were gently triturated with a Pasteur pipette (Fisher) to release neural tubes from surrounding somites and connective tissue. Isolated neural tubes were dissociated by Trypsin–EDTA, and NEP cells were grown in NEP basal medium with 10% CEE.
Tissue culture dishes were coated with fibronectin (Sigma), which was diluted to a concentration of 20 µg mL−1 in distilled H2O (Sigma), for a minimum of 4 h. Poly-l-lysine (PLL)/laminin double-coated dishes were prepared by first coating with PLL (30–70 kDa; Sigma; dissolved in distilled water at a concentration of 13.3 µg mL−1) for 1 h, followed by incubation with laminin (Biomedical Technologies Inc.; dissolved in distilled water at a concentration of 15 µg mL−1) overnight at 4 °C. Excess laminin was discarded, and the dishes were rinsed with medium just prior to use. Acutely dissected or sorted NEP cells were plated on a fibronectin-coated 60-mm dish (Corning) and maintained at 37 °C, 5% CO2 in NEP-basal medium with 10% CEE and basic fibroblast growth factor (bFGF, 30 ng mL−1; Peprotech). Acutely dissected or sorted E14.5 neural tube cells were plated on PLL/laminin-coated dishes and maintained at 37 °C, 5% CO2 in NEP-basal medium with bFGF (20 ng mL−1).
Immunocytochemistry on frozen neurospheres
Neurospheres were washed twice with PBS then fixed in a solution of 4% paraformaldehyde in PBS for 30 min. Neurospheres were then frozen in OCT embedding solution and cut into 8-µm sections for mounting on lysine-coated glass slides. This approach removed the question of whether antibodies were able to penetrate into the inner cells of large neurospheres. Blocking buffer, primary antibodies and secondary antibodies were applied as described under Immunohistochemistry (above). Antibody labeling was repeated in at least two separate neurosphere isolations per experiment. Images were captured using a digital camera coupled to an Olympus fluorescence microscope, and analysed using Magnafire and Adobe Photoshop software. Alternatively, images were captured using an Olympus scanning confocal microscope. Co-expression of markers was confirmed using z-stacks of images captured at intervals of 0.5 µm. x- and y-axis scans are shown along the bottom and left margins of each photograph, respectively.
cDNA synthesis and RT-PCR analysis
Total RNA was isolated from neurospheres at 7 div using a standard Trizol (Gibco/BRL) extraction method. cDNA was synthesized and PCR was performed following the protocol outlined in Kalyani et al. (1999). Primers included murine Sox2 (5′GTGGAAACTTTTGTCCGAGAC3′/5′TGGAGTGGGAGGAAGAGGTAAC3′) human Sox2 (5′TGGAGTGGGAGGA-AGAGGTAAC3′/5′GTGGAAACTTTTGTCCGAGAC3′) murine G3PDH (5′TGATGGGTGTGAACCACGAG3′/5′CTCCTGTTGTT-ATGGGGTCTG3′) and human G3PDH (5′GCTCAGACACCATGGG-GAAGGT3′/5′GTGGTGCAGGAGGCATTGCTGA3′).
Cell extracts were prepared according to protocols provided by the manufacturer. Telomerase activity was measured using the TRAPeze telomerase detection kit (Chemicon). For TRAPeze, telomerase elongation conditions were 30 °C for 30 min and 94 °C for 4 min. PCR cycling condtions were for 33 cycles (94 °C for 30 s, 59 °C for 30 s, 72 °C for 60 s). The TRAP ladders were run on a 10% non-denaturing polyacrylamide gel and visualized by SYBR green. All telomerase assays were performed in triplicate and were repeated at least twice. In order to estimate telomerase activity, we used FluorChem™ 8900 (Alpha Innotech, San Leandro, CA, USA) to measure the total signal intensity of the bands corresponding to all TRAP products. Negative controls were performed by heat-treating the cell lysates for 10 min at 85 °C.
We thank members of our laboratories for critical review of this manuscript. We also thank Dr Norm Haughey for the kind gift of human neurospheres. Dr Ollivier Milhavet is thanked for advice on siRNA experiments. T.L.L. is supported by a PRAT fellowship from NIGMS (NIH).
- 2003) Multipotent cell lineages in early mouse development depend on SOX2 function. Genes Dev. 17, 126–140. , , , , , (
- 2004) Accumulation of mutations and somatic selection in aging neural stem/progenitor cells. Aging Cell 3, 391–397. , , (
- 2003) Postnatal NG2 proteoglycan-expressing progenitor cells are intrinsically multipotent and generate functional neurons. J. Cell Biol. 161, 169–186. , , , , , , (
- 2002) Properties of a fetal multipotent neural stem cell (NEP cell). Dev. Biol. 251, 221–240. , , , , , , , (
- 2003) Telomerase activity in the subventricular zone of adult mice. Mol. Cell Neurosci. 23, 693–702. , , , (
- 2003) Genetic and functional differences between multipotent neural and pluripotent embryonic stem cells. Proc. Natl Acad. Sci. USA 100 (Suppl. 1), 11866–11872. , (
- 2000) NG2-expressing cells in the central nervous system: are they oligodendroglial progenitors? J. Neurosci. Res. 61, 471–479. , , (
- 1999) Regeneration of a germinal layer in the adult mammalian brain. Proc. Natl Acad. Sci. USA 96, 11619–11624. , , (
- 2002) EGF converts transit-amplifying neurogenic precursors in the adult brain into multipotent stem cells. Neuron 36, 1021–1034. , , , , (
- 2004) SOX2, a persistent marker for multipotential neural stem cells derived from embryonic stem cells, the embryo or the adult. Dev Neurosci. 26, 148–165. , , , , , , , , (
- 2004) Sox2 deficiency causes neurodegeneration and impaired neurogenesis in the adult mouse brain. Development 131, 3805–3819. , , , , , , , , , , (
- 2003) Deregulation of dorsoventral patterning by FGF confers trilineage differentiation capacity on CNS stem cells in vitro. Neuron 40, 485–499. , , , (
- 2003) SOX2 functions to maintain neural progenitor identity. Neuron 39, 749–765. , , , (
- 1996) Multipotential stem cells from the adult mouse brain proliferate and self-renew in response to basic fibroblast growth factor. J Neurosci. 16(3), 1091–100. , , , , , , , , , , (
- 2002) Neural stem cell lineages are regionally specified, but not committed, within distinct compartments of the developing brain. Development. 129, 233–244. , , , (
- 2003) The predominant neural stem cell isolated from postnatal and adult forebrain but not early embryonic forebrain expresses GFAP. J. Neurosci. 23, 2824–2832. , , (
- 1999) Identification of a neural stem cell in the adult mammalian central nervous system. Cell 96, 25–34. , , , , , (
- 1997) Neuroepithelial stem cells from the embryonic spinal cord: isolation, characterization, and clonal analysis. Dev. Biol. 186, 202–223. , , (
- 1999) Expression of EGF receptor and FGF receptor isoforms during neuroepithelial stem cell differentiation. J. Neurobiol. 38, 207–224. , , (
- 2000) Oligodendrocyte precursor cells reprogrammed to become multipotential CNS stem cells. Science 289, 1754–1757. , (
- 2000) Identification of a multipotent astrocytic stem cell in the immature and adult mouse brain. Proc. Natl Acad. Sci. USA 97, 13883–13888. , , , , (
- 2001) The oligodendrocyte precursor cell in health and disease. Trends Neurosci. 24, 39–47. , , (
- 2003) Distinguishing features of progenitor cells in the late embryonic and adult hippocampus. Dev. Neurosci. in press. , , , , (
- 2004) Neural stem cell detection, characterization, and age-related changes in the subventricular zone of mice. J. Neurosci. 24, 1726–1733. , , , (
- 2004) Neural stem and progenitor cells in nestin-GFP transgenic mice. J. Comp. Neurol. 469, 311–324. , , , , (
- 2003) The ablation of glial fibrillary acidic protein-positive cells from the adult central nervous system results in the loss of forebrain neural stem cells but not retinal stem cells. Eur. J. Neurosci. 18, 76–84. , , , (
- 2003) Identification and isolation of multipotential neural progenitor cells from the subcortical white matter of the adult human brain. Nat. Med. 9, 439–447. , , , , , , , , (
- 1999) Fibroblast growth factor-2 activates a latent neurogenic program in neural stem cells from diverse regions of the adult CNS. J. Neurosci. 19, 8487–8497. , , , , (
- 2003) The stem-cell menagerie. Trends Neurosci. 26, 351–359. , (
- 2002) Adult rodent neurogenic regions: the ventricular subependyma contains neural stem cells, but the dentate gyrus contains restricted progenitors. J. Neurosci. 22, 178–1793. , (
- 2004) Cell types, lineage, and architecture of the germinal zone in the adult dentate gyrus. J. Comp. Neurol. 478, 359–378. Erratum 480, 427. , , , , (
- 2001) Astrocytes give rise to new neurons in the adult mammalian hippocampus. J. Neurosci. 18, 7153–7160 . , , , (
- 1999) Neural precursor differentiation following transplantation into neocortex is dependent on intrinsic developmental state and receptor competence. Exp. Neurol. 158, 47–62. , , , (
- 2004) Differential regulation of gliogenesis in the context of adult hippocampal neurogenesis in mice. Glia. 46, 41–52. , , , , , (
- 2002) Neural stem cell heterogeneity demonstrated by molecular phenotyping of clonal neurospheres. Proc. Natl Acad. Sci. USA 99, 14506–14511. , , , (
- 2000) Dynamic expression of basic helix-loop-helix Olig family members: implication of Olig2 in neuron and oligodendrocyte differentiation and identification of a new member, Olig3. Mech. Dev. 99, 143–148. , , , , , , (
- 1999) Distinct neural stem cells proliferate in response to EGF and FGF in the developing mouse telencephalon. Dev Biol. 208(1), 166–88. , , , , , (
- 2000) Sox2 regulatory sequences direct expression of a (beta)-geo transgene to telencephalic neural stem cells and precursors of the mouse embryo, revealing regionalization of gene expression in CNS stem cells. Development. 127, 2367–2382. , , , , , , , , , , (