The role of multipotential progenitors and neural stem cells in the adult subventricular zone (SVZ) as cell-of-origin of glioblastoma has been suggested by studies on human tumors and transgenic mice. However, it is still unknown whether glial tumors are generated by all of the heterogeneous SVZ cell types or only by specific subpopulations of cells. It has been proposed that transformation could result from lack of apoptosis and increased self-renewal, but the definition of the properties leading to neoplastic transformation of SVZ cells are still elusive. This study addresses these questions in mice carrying the deletion of p53, a tumor-suppressor gene expressed in the SVZ. We show here that, although loss of p53 by itself is not sufficient for tumor formation, it provides a proliferative advantage to the slow- and fast-proliferating subventricular zone (SVZ) populations associated with their rapid differentiation. This results in areas of increased cell density that are distributed along the walls of the lateral ventricles and often associated with increased p53-independent apoptosis. Transformation occurs when loss of p53 is associated with a mutagenic stimulus and is characterized by dramatic changes in the properties of the quiescent adult SVZ cells, including enhanced self-renewal, recruitment to the fast-proliferating compartment, and impaired differentiation.
Together, these findings provide a cellular mechanism for how the slow-proliferating SVZ cells can give rise to glial tumors in the adult brain.
Relatively quiescent neural stem cells and fast-proliferating multipotential progenitors residing in the adult SVZ are characterized by the ability to self-renew and generate distinct cell types (Galli et al., 2003; Alvarez-Buylla and Lim, 2004), thus providing an important therapeutic tool (Picard-Riera et al., 2004). Given their ability to proliferate, self-renew, and migrate, these cells have also the potential to undergo neoplastic transformation (Recht et al., 2003), and a large body of evidence supports this concept, although the identity of the transformed cells is not well defined (Rao, 1999; Holland, 2000; Berger et al., 2004; Jang et al., 2004). Four distinct cell types reside in the adult SVZ (Doetsch et al., 1997, 1999a, b). The ependymal cells (type E cells) provide the lining of the lumen of the ventricles and are important for regulating the flow of the CSF. The astrocyte-like type B cells are characterized by the presence of intermediate filaments and nestin-positive (+)/GFAP+ immunoreactivity (Doetsch et al., 1997), and they have the property of self-renewal (Doetsch et al., 2002a) and proliferate very slowly (Morshead et al., 1994). Type C cells are large cells with invaginated nuclei and lack of intermediate filaments. These cells have also been called “transit amplifying progenitors, ” because they proliferate rapidly and give rise to migratory polysialylated neural cell adhesion molecule (PSA-NCAM)-positive neuroblasts (Lois and Alvarez-Buylla, 1994; Doetsch et al., 1997; Peretto et al., 1999) and oligodendrocyte progenitors (Nait-Oumesmar et al., 1999; Vitry et al., 2001).
The importance of SVZ cells in the genesis of glioblastomas (Recht et al., 2003) has been suggested by the isolation of neurosphere-forming cells from human glioblastomas (Ignatova et al., 2002) and by studies on transgenic mice (Holland, 2000) and on animals prenatally exposed to the mutagen N-ethyl-N-nitrosourea (ENU) (Oda et al., 1997; Leonard et al., 2001; Slikker et al., 2004; Savarese et al., 2005).
p53 is a tumor-suppressor gene expressed in SVZ cells (van Lookeren Campagne and Gill, 1998; Jori et al., 2003) and frequently deleted or mutated in adult (von Deimling et al., 1992; Watanabe et al., 1996; Hayashi et al., 2004) and pediatric glial tumors (Sung et al., 2000; Di Sapio et al., 2002). However, it is still not known whether loss of p53 can alter the behavior of adult SVZ cells and induce specific changes in specific subpopulations leading tumor formation. This study is aimed at testing this hypothesis.
Materials and Methods
All of the experiments were performed in 8- to 12-week-old mice obtained from Trp53Tm1Tyj C57BL/6J heterozygous (i.e., p53+/−) breeding pairs. Mouse genotypes were confirmed by tail clipping and PCR using the primers 5′-ACAGCGTGGTGGTACCTTAT-3 (ImRo36), 5′-TATACTCAGAGCCGGCCT-3′ (ImRo37), and 5′-TCCTCGTGCTTTACGGTATC-3′ (neo), yielding a fragment of 375 bp for p53+/+ and 525 bp in p53−/− mice. p53flox/flox mice (a gift from Dr. Anton Berns, Netherlands Cancer Institute, Amsterdam, The Netherlands) were genotyped using the primers 5′-CACAAAAACAGGTTAAACCCAG-3′ and 5′-AGCACATAGGAGGCAGAGAC-3′, yielding a fragment of 370 bp for p53flox/flox and 288 bp for wild-type mice.
A total dose of 2.21 Gy was used for brain irradiation using a linear accelerator with 6 MV of nominal photon energy (Chow et al., 2000). Anesthetized mice were placed in a prone position on an expanded polystyrene bed at a distance of 99.5 cm from the source. The dose variation within the target volume was estimated to be ±5%. Four hours after irradiation, the animals were perfused with 10% neutral buffered formalin.
Timed-pregnant p53+/− females were injected with ENU (25 mg/kg body weight) as described previously (Leonard et al., 2001). The offspring were allowed to reach 6–8 weeks of age and were then killed.
Protein extracts and Western blot analysis.
The SVZ was dissected and homogenized in cold lysis buffer (50 mm Tris-HCl, pH 7.4, containing 150 mm NaCl, 1% Triton X-100, 0.25% sodium deoxycholate, 1 mm EDTA, 0.1% SDS, 1 mm PMSF, and 5 μg/ml of the protease inhibitors aprotinin and leupeptin). Tissue extracts were sonicated and centrifuged at 14, 000 × g for 15 min at 4°C. Proteins were measured using a detergent-compatible assay (Bio-Rad, Hercules, CA) with bovine serum albumin as protein standard, separated by SDS-PAGE, and transferred to nitrocellulose (Hybond ECL; Amersham Biosciences, Buckinghamshire, UK). Immunoblotting was performed using a rabbit anti-p53 polyclonal antibody (FL-393; Santa Cruz Biotechnology, Santa Cruz, CA) at a 1:500 dilution. Mouse anti-actin monoclonal antibody (Sigma, St. Louis, MO) was used at a 1:1000 dilution. After incubation of secondary antibodies, immunoreactive bands were visualized using ECL plus (Amersham Biosciences).
RNA extraction and reverse transcription-PCR.
Freshly dissected SVZ regions were homogenized in Trizol reagent. RNA was isolated using the RNeasy Mini kit (Qiagen, Hilden, Germany). Total RNA at 1.5 μg was reverse transcribed and then amplified using SuperScript reverse transcription (RT)-PCR kit (Invitrogen, Carlsbad, CA). The following primers were used: 5′-CATCACCTCACTGCATGGAC-3′ and 5′-CTTGTGTACGGCGGTCTCTC-3′ to amplify p53; and 5′-TGGAATCCTGTGGCATCC-3′ and 5′-TCGTACTCCTGCTTGCTG-3 to amplify actin.
Bromodeoxyuridine (BrdU) (Sigma) was injected intraperitoneally (100 μg/g body weight). At the indicated times (either 1 h or 2 weeks later), mice were anesthetized and perfused with 4% paraformaldehyde (PFA). Brain tissues were sectioned and then processed for immunohistochemistry. BrdU+ cells were counted in at least four sections for each animal and expressed per squared millimeter. For data analysis, we used nonparametric statistics (Mann–Whitney U test) because we could not assume a normal distribution for our data. The previously described function of p53 as cell cycle inhibitor suggested that p53 loss-of-function would increase (and not decrease) the number of cells in S-phase. Thus, we performed planned one-tailed comparisons. Significance level for the rejection of the null hypothesis was set at α = 0.05.
Thymidine labeling and autoradiography.
p53+/+ and p53−/− littermates were injected with 50 μl of 6.7 mCi titrated [3H]thymidine (Thy) (Amersham Biosciences) as described previously (Doetsch et al., 2002b). After 1 h or 4 weeks after injection, mice were perfused with 0.9% saline, followed by Karnovsky’s fixative (2% paraformaldehyde and 2.5% glutaraldehyde). Heads were removed and postfixed in the same fixative overnight, and the brains were dissected, washed in 0.1 m phosphate buffer, and cut into 200 μm vibratome sections. Sections were postfixed in 2% osmium and embedded in Araldite (Durcupan; Fluka, Buchs, Switzerland). Semithin sections (1.5 μm) were processed for titrated autoradiography (Doetsch et al., 2002b). [3H]Thy+ cells were counted in five sections, one every 7.5 μm, and the means were calculated and extrapolated per squared millimeter. Statistical analysis was performed as described for BrdU.
Terminal deoxynucleotidyl transferase-mediated biotinylated UTP nick end labeling in vivo and in vitro.
p53+/+ and p53−/− mice were perfused with 4% PFA. Brains were removed, cryopreserved in 30% sucrose, and sectioned. Cells or sections were treated with ethanol acetic (2:1) for 10 min at −20°C, washed three times in PBS, and processed with Apoptag Plus Fluorescein In Situ Detection kit (Chemicon, Temecula, CA) according to the manufacturer.
Immunohistochemistry and immunocytochemistry.
For O4 staining in cells, cultures were gently rinsed in PBS (10 mm sodium phosphate, pH 7.4, and 150 mm NaCl) and incubated with hybridoma supernatant (1:10; a gift from Dr. Bansal, University of Farmington, Farmington, CT) for 30 min at 37°C before fixation with 4% PFA for 20 min at room temperature. For immunocytochemical and immunohistochemical procedures, the following primary antibodies were used: anti-β-tubulin III (1:500, clone Tuj1; Covance, Berkeley, CA), anti-GFAP [mouse, 1:1000 (Sternberg Monoclonals, Lutherville, MD) or rabbit, 1:1000 (Dako, Glostrup, Denmark)], CC1 antibody (mouse, 1:50; Oncogene, San Diego, CA), anti-PSA-NCAM (mouse, 1:400; AbCys, Paris, France), anti-BrdU (mouse, 1:200; Dako), anti-p53 (rabbit, 1:500 CM5; Novocastra, New Castle upon Tyne, UK), anti-nestin (rat-401, 1:1000; Developmental Studies Hybridoma Bank, University of Iowa, Iowa City, IA), and anti-Cre (rabbit, 1:3000; Covance). Incubation of all of the primary antibodies was followed by incubation with the appropriate secondary antibodies conjugated with fluorophores (Jackson ImmunoResearch, West Grove, PA) or with biotin (Vector Laboratories, Burlingame, CA). Immunoreactive cells were analyzed using a fluorescence microscope (Leica, Heerbrugg, Switzerland), and images were captured using a Hamamatsu (Hamamatsu City, Japan) CCD camera interfaced with a G4 computer.
EM cell counts.
Ultrathin (60 nm) sections were cut with a diamond knife, stained with lead citrate, and examined under a Jeol (Tokyo, Japan) 1010 TEM electron microscope. Type A cells, were identified by the small size, scanty cytoplasm, and fusiform appearance; type B cells were identified as large, less electron-dense cells, rich of intermediate filaments; type C cells were characterized by the presence of characteristically large nuclei and lack of intermediate filaments. The different cell types in the adult SVZ were counted in an oblique reverse orientation and expressed as percentage of total cell counts and also per unit length, as described previously (Doetsch et al., 1997). No difference was observed in the calculated mean length of the ventricular wall of SVZ of p53−/− and p53+/+ mice. Graphical exploration of our data indicated that we could safely assume homogeneity of variance and normal distribution. Thus, we conducted a Student’s t test to analyze the differences in cell number in the SVZ. All reported probabilities were two tailed. Significance level for the rejection of the null hypothesis was set at α = 0.05
SVZ cells were collected from p53+/+ and p53−/− mice in Petri dishes containing PIPES buffer (in mm: 20 PIPES, 25 glucose, 120 NaCl, and 0.5 KCl, pH 7.4). After digestion with papain, cells were dissociated and resuspended in Neurobasal medium (Invitrogen) supplemented with B27 (Invitrogen), 2 mm l-glutamine (Invitrogen), antibiotic/antimycotic (Invitrogen), and 2 μg/ml heparin (Sigma). Equal number of cells were plated and kept in medium containing 20 ng/ml epidermal growth factor (EGF) and 10 ng/ml basic FGF. For the experiments performed at high density, the amount of plated cells was 10, 000 cells/cm2. For the experiments performed at low density, the amount of plated cells was 1000–2000 cells/cm2. The number of neurospheres was counted 7 d after plating. For the differentiation studies, the neurospheres were dissociated, and the cells were then plated in poly-ornithine-coated chamber slides without mitogens in the presence (or absence) of 2% FBS. The amplification rate was defined as x = (number of viable cells in passage N + 1/number of plated cells). Self renewal rate was calculated as x = (number of neurospheres in passage N + 1/number of plated cells) × 100. Clonal self-renewal was calculated by transferring individual neurospheres of similar size into individual wells of a 48-well plate and counting the total number of neurospheres generated after 5 d.
Retroviral infection of adult SVZ cells.
Neurospheres were mechanically dissociated and incubated with MSCV-NRasV12-IRES-GFP (a gift from Dr. James Downing, St Jude Children’s Research Hospital, Memphis, TN) or MSCV-IRES-GFP empty vector ecotropic retroviruses in the presence of 8 μg/ml polybrene. Infection was allowed to proceed at 37°C under continuous shaking during the first 30 min. The infected cells were then transferred to culture dishes and placed in the incubator for an additional 2 h. Infection was terminated by adding fresh growth medium to the cultures. On the following day, cells were centrifuged and replated onto eight-well chamber slides in differentiation medium to assess their ability to differentiate.
Adenoviral infection of SVZ cells.
After dissociation of primary neurospheres, an equal number of cells (5 × 104) was resuspended in serum-free medium containing mitogens and 106 viral particles of the adenovirus-CMV-Cre recombinase (Cre) (from Gene Transfer Vector Core, University of Iowa, Iowa City, IA) corresponding to a multiplicity of infection of 35. The infection was performed in cells kept in suspension in a 15 ml conical tube (Falcon; BD Biosciences, Bedford, MA) for 1 h at 37°C in the incubator. After collection by centrifugation, the supernatant containing the virus was discarded, and the cell pellet was resuspended in fresh medium. After a 2 d recovery period, the majority of the cells were grown in medium containing mitogens to assess neurosphere formation, whereas some cells were plated in slide chambers and cultured for 7 d in differentiation medium. After fixation, cultured cells were stained with antibodies against Cre or Tuj1 or were processed for terminal deoxynucleotidyl transferase-mediated biotinylated UTP nick end labeling (TUNEL) assay. The gene-transfer efficiency was 98 ± 1.8%, as calculated by dividing the number of 4′, 6′-diamidino-2-phenylindole (DAPI)+/Cre+ cells by the total number of DAPI+ cells.
In physiological conditions, the levels of p53 detected in the SVZ were very low (Fig. 1A), but they increased after brain irradiation, especially along the lateral walls of the ventricles (Fig. 1B). Loss of p53 induced detectable changes in this region, as a randomly distributed thickening of the cellular layer bordering the ventricles (Fig. 1C). Areas of cell density comparable with that of p53+/+ littermates were observed within regions of hyperplasia (Fig. 1C). This overall effect on cell density was confirmed by cell counts that revealed 343.47 ± 13.26 cells/mm in p53−/− mice (C57BL/6 Trp53; n = 5) compared with 255.24 ± 5.63 cells/mm in the SVZ of p53+/+ littermates (C57BL/6; n = 5). Because the SVZ is composed of distinct cellular subpopulations, we asked whether the effect of loss of p53 was generalized or specific for distinct cell types. Quantification of the distinct cell types along the ventricular wall (Fig. 1D) revealed a statistically significant increase of type A (91.48 ± 7.95) and type B (94.2 ± 9.6) cells in p53−/− mice (n = 5) compared with p53+/+ littermates (type A, 65 ± 3 cells/mm; type B, 70.8 ± 7.8 cells/mm; n = 5). In contrast, the number of ependymal cells lining the lumen of the ventricles (type E) and type C precursors was relatively constant (p53−/−, 34.35 ± 7.96 cells/mm; p53+/+, 30.17 ± 0.83 cells/mm) (Fig. 1D). In general, we observed an increase of B cells contacting the ventricle (Bv), a feature of B cell activation and lineage progression toward the type C, but their low number precluded any accurate statistical analysis. The selective increase of these cell types in the SVZ of p53−/− mice was confirmed by immunohistochemistry, using antibodies against PSA-NCAM to identify A cells (see Fig. 5A) and against GFAP to identify B cells (data not shown). The excess A cells were detected along the walls of the lateral ventricle and especially at the anterolateral corner (Fig. 2A) (see Fig. 5), whereas the B cells were homogeneously distributed to the inferior border of the SVZ. The distribution of these areas of cellular hyperplasia corresponding to specific cell types was further confirmed by a camera lucida serial reconstruction. A diagram of the SVZ of p53+/+ and p53−/− mice was drawn based on toluidine blue-stained semithin sections. Each cell type was identified in the corresponding ultrathin section, and its topographic location along the lateral wall of the ventricles was accurately recorded on the diagram, using a different color for each cell type (Fig. 2A). This pseudocolor map confirmed the distribution of the distinct cell types. Variability on the length of the SVZ was observed in distinct mice. However, a systematic measurement of the length of the lateral wall of the ventricles in six p53+/+ and p53−/− mice did not reveal statistically significant differences between the two groups, because the length in p53+/+ mice was 1.66 ± 0.12 and in p53−/− mice was 1.56 ± 0.04.
Because the number of SVZ cells in each subpopulation is driven by a steady-state equilibrium between their birth and their death attributable to apoptosis or lineage progression, we systematically investigated the effects of p53 loss-of-function on these properties in vivo and in vitro. Proliferation of SVZ cells was analyzed by assessing the number of cells in S-phase after 1 h in vivo labeling with BrdU followed by immunohistochemistry or titrated [3H]Thy followed by autoradiography (Fig. 3). The number of immunoreactive BrdU+ cells in the anterior SVZ of p53−/− mice (x = 127 ± 3.66; n = 3) was 62% higher than the number of BrdU+ cells detected in equivalent regions of the SVZ of p53+/+ animals (x = 77 ± 1; n = 3), and represented a statistically significant increase (p < 0.005) of cells in S-phase in p53−/− mice (Fig. 3B). The total number of [3H]Thy+ cells was counted in semithin sections after autoradiography (Fig. 3C) and confirmed the proliferative advantage conferred by the loss of p53 (x = 20.64 ± 0.39 cells/mm; n = 3) to SVZ cells (Fig. 3D) compared with p53+/+ littermates (x = 12.92 ± 1.68 cells/mm; p < 0.05). To define whether loss of p53 equally affected proliferation of distinct cell types, the identity of the proliferating cells was defined by serial ultrastructural analysis of [3H]Thy+ sections. The number of [3H]Thy+ cells in each SVZ subpopulation was detected per unit of surface area and then normalized to the total number of cells in that population, to obtain the relative proportion of proliferating cells in each cell type (Fig. 3E). This analysis revealed a statistically significant increase of fast-proliferating C cells in S-phase in p53−/− mice compared with p53+/+, whereas no difference was observed in proliferating A and B cells (Fig. 3E). These results suggest that the increase in specific SVZ subpopulations detected in p53−/− mice could not be simply explained in terms of growth advantage provided by p53 loss-of-function, but revealed a much more complex effect on the population dynamics.
Because astrocyte-like B cells are known to be relatively quiescent (Morshead et al., 1994), we asked whether p53 loss-of-function affected this slow-dividing cell population by performing long-term labeling experiments. Mice of both genotypes received a single injection of BrdU or [3H]Thy+ and were analyzed 2–4 weeks later. Of all of the cells incorporating the label at the time of injection, only the relatively quiescent ones retained the label after several weeks, whereas those fast-proliferating lost the label by dilution after several rounds of division. The detection of label-retaining cells in the SVZ was consistently higher in p53−/− mice compared with p53+/+ (Fig. 3F), although the low number precluded accurate statistical analysis. Thus, loss of p53 increased the division of the relatively quiescent astrocyte-like B cells, without significantly modifying their slow rate of proliferation.
Because a characteristic feature of SVZ cells is self-renewal, we asked whether the increased number of B cells in p53−/− mice could be attributed to an enhancement of this property. It is well accepted that the property of self-renewal can be assessed in vitro (Reynolds and Weiss, 1992; Morshead et al., 1994; Parker et al., 2005; Reynolds and Rietze, 2005) and is measured by the ability to form neurospheres (Fig. 4A). However, this assay cannot be used as the only criterion of analysis of stem cell behavior (Parker et al., 2005; Reynolds and Rietze, 2005). Thus, SVZ cells were isolated from p53−/− and p53+/+ mice plated at high or low density and the total number of neurospheres counted after 7 d (Fig. 4B). The effect of p53 loss-of-function on the number of neurospheres was dependent on cell density. At high density, the number of neurospheres was similar in the two groups, whereas at low density, the p53−/− cultures had 70% more neurospheres than p53+/+ cultures (Fig. 4B). Besides the increased number of spheres, we consistently noticed that the size of the p53−/− neurospheres was significantly larger than p53+/+ (Fig. 4A). The increase in size was associated with a large increase in amplification rate (i.e., the number of viable cells detected after 7 d in culture) (Fig. 4C). To address whether the observed effects of p53 loss-of-function were cell autonomous or consequent to a compensatory mechanism, we repeated the experiments on cells isolated from the SVZ of p53flox/flox (Marino et al., 2000) and p53+/+ mice. The excision of p53 was achieved by infecting the p53flox/flox cultures with adenoviral vectors expressing Cre (Fig. 4D, E). Wild-type cultures also received Cre infection as an additional control. After a 2 d recovery period, the cells were fixed, and infection efficiency was calculated to be 98 ± 1.8% (data not shown). Given the high level of transfection efficiency, we performed additional experiments addressing the effect of p53 excision by Cre on self-renewal and proliferation. Cre infection of p53+/+ cultures behaved as wild type not transduced control, generating a similar number of neurospheres as Cre-infected p53+/+ cultures (Fig. 4D). We also observed the predicted increase in the number of neurospheres generated from Cre-infected p53flox/flox cultures compared with controls (Fig. 4D). Thus, we concluded that the effect of p53 on self-renewal is attributable to a direct effect of p53 on the adult SVZ cells rather than to a compensatory effect. Because p53 loss-of-function also increased the amplification rate compared with p53+/+ mice (Fig. 4C), we asked whether a similar effect could be observed in Cre-infected p53flox/flox cultures compared with controls. For this reason, we compared the total number of viable cells in Cre-infected p53+/+ and p53flox/flox cultures (Fig. 4E). The results were remarkably similar to those reported previously for the p53−/− cultures. Therefore, these results strongly argue in favor of a direct, rather than compensatory, effect of p53 loss-of-function on the regulation of self-renewal and proliferation of adult SVZ pluripotent progenitors.
It remained to be explained, however, why the increase of slow-proliferating B cells and fast-proliferating C cells detected in p53−/− mice was not associated with the expansion of the C cell compartment. A possible explanation for the lack of accumulation of C cells in vivo was their rapid differentiation toward the generation of A cells or oligodendrocytes. This hypothesis was supported by the significant increase of A cells detected in p53−/− mice (Fig. 5A) and the increased number of CC1+ oligodendrocytes observed in vivo (data not shown). Thus, in the absence of p53, SVZ cells proliferated more but retained the ability to differentiate along distinct lineages. Experiments conducted in vitro, in neurospheres generated from p53−/− and p53+/+ mice and allowed to spontaneously differentiate into neurons or glial cells, supported this interpretation (Fig. 5B, C). In p53+/+ mice, 82.13 ± 10.29% of cells were GFAP+, 3.22 ± 1.28% of cells were TuJ1+, and 2.66 ± 0.42% were O4+. In contrast, in p53−/− mice, 60.39 ± 7.64% of cells were GFAP+, 35.85 ± 6.61% of cells were TuJ1+, and 5.63 ± 1.51% were O4+ (Fig. 5D). Thus, the significantly higher number of neuroblasts (11-fold) and oligodendrocytes (2-fold) generated by p53−/− neurospheres SVZ cells in vitro was consistent with the increased number of differentiated cells detected in p53−/− mice. Higher number of neurons were also observed in cultures derived from Cre-infected p53flox/flox mice compared with uninfected p53flox/flox and Cre-infected p53+/+ cultures, thus indicating also that the effect of p53 on neurogenesis was direct and not consequent to the activation of compensatory mechanisms.
An alternative explanation for the lack of C cell accumulation in p53−/− mice was an increase in p53-independent apoptosis. In contrast with the well established role of p53, we consistently observed an increased number of TUNEL+ cells in the SVZ of p53−/− mice localized to those areas of increased cell density (Fig. 6A) and possibly resulting from compensatory death. The greater number of spontaneously dying cells in p53−/− cultures (x = 5 ± 0.7%) compared with controls (x = 1.2 ± 0.6%) was also detected in vitro (Fig. 6B). Increased apoptosis, however, was not observed in Cre-infected p53flox/flox mice compared with uninfected p53flox/flox and Cre-infected p53+/+ cultures, thus supporting the interpretation of a compensatory p53-independent apoptosis in the p53−/− mice. Together, these data suggest a critical role for p53 in regulating the number of adult SVZ cells by directly modulating proliferation, differentiation, and self-renewal or survival, whereas the compensatory increase of p53-independent death is consistent with the lack of spontaneous glial tumors in p53−/− mice.
It is important to mention, however, that loss of p53 results in a 60% incidence of glioblastoma-like tumors in p53−/− mice, but not in p53+/− or in p53+/+ littermates, after prenatal exposure to the mutagen ENU (Oda et al., 1997; Leonard et al., 2001). These tumors were typically localized in periventricular locations, at the border between the SVZ and the corpus callosum (Fig. 7A, B) or between the SVZ and the striatum. The tumoral masses were often spherical and not capsulated with clear signs of parenchymal infiltration. Neovascularization was frequently detected and associated with microhemorrhages and small necrotic areas (Fig. 7C), consistent with the histological features of glioblastoma-like tumors. In semithin sections obtained from tumoral masses, several aberrant mitotic figures were detected together with atypical giant cells (Fig. 7D). At the immunohistochemical level, the tumor was highly heterogeneous, with scattered distribution of GFAP- and nestin-immunoreactive cells (Fig. 7E). Proliferating cells were observed within the tumoral mass and more prominently at its periphery (Fig. 7F). At the ultrastructural level, tumoral cells were characterized by electron-dense cytoplasm, the absence of intermediate filaments and microtubules, an irregularly dilated endoplasmic reticulum, enlarged Golgi apparatus, and mitochondria with vacuoles or vesicles (Fig. 7E).
To identify the changes in population dynamics leading to tumor formation, we further quantified [3H]Thy+ incorporation in semithin sections from ENU-treated p53−/− mice (that develop spontaneous glioblastoma-like tumors) and compared these changes with those occurring in untreated p53−/− mice and in ENU-treated p53+/− mice (that do not develop these tumors). The identity of the labeled cells was assessed and quantified in serial sections from mice of different genotypes. The relative proportion of [3H]Thy+ B cells in the SVZ was increased in ENU-treated p53−/− mice compared with untreated p53−/− mice, whereas the relative proportion of type A and C cells in S-phase was decreased (Fig. 8A). Because the number of [3H]Thy+ cells identified after a 1 h labeling pulse reflects the number of cells in S-phase within a relatively short time, the greater proportion of relatively quiescent [3H]Thy+ type B cells identified in the ENU-treated p53−/− mice suggested the recruitment of these cells into the fast proliferative compartment. This interpretation was supported by the detection of several cells with “immature” features, a transition stage between type B and C cells (Fig. 8B). Together, these data indicate that the neoplastic transformation of SVZ cells in adult p53−/− mice after prenatal exposure to ENU is preceded by important changes of the SVZ population dynamics, including the recruitment of B cells to the fast-proliferating compartment (Fig. 8C) and the accumulation of undifferentiated cells with intermediate characteristics between types B and C (Fig. 8D), which cannot progress toward the neuronal or glial cell lineage.
To verify the cell-autonomous nature of these effects and define the changes leading to neoplastic transformation, we isolated cells from the SVZ of ENU-treated p53−/− (that develop glioblastoma-like tumors) and p53+/− mice (that do not develop glioblastoma-like tumors) and compared their behavior. One possibility was that ENU treatment modified the property of self-renewal. We therefore assessed clonal self-renewal (i.e., the ability of each neurosphere to generate clonal aggregates) in SVZ cells in isolated ENU-treated and untreated p53−/− and p53+/− mice. Clonal self-renewal was rapidly exhausted after three serial passages of neurospheres from ENU-treated p53+/− and from untreated p53−/− and p53+/− mice, thus indicating that these cells were likely fast-proliferating multipotential progenitors (Fig. 9A). In contrast, ENU-treated p53−/−-derived neurospheres retained the ability to self-renew even after three passages (Fig. 9A) and were characterized by a very large size (Fig. 9B). Thus, the proliferative advantage conferred by loss of p53 associated with the enhanced self-renewal induced by ENU exposure modified these properties of adult neural stem cells only in those animals with a high incidence to develop glioblastoma-like tumors.
Another possibility accounting for the abundant presence of immature cells in vivo in ENU-treated p53−/− mice was that ENU treatment modified the ability of adult multipotential SVZ progenitors to differentiate along distinct lineages. We therefore assessed differentiation of neurospheres from ENU-treated and untreated p53−/− and p53+/− mice by immunocytochemistry using antibodies against the cellular markers TuJ1, GFAP, O4, and nestin (Fig. 9C–F). Differentiation proceeded normally in cultures from ENU-treated p53+/− and untreated p53−/− or p53+/− mice, in agreement with the absence of spontaneous tumors. In contrast, the cells isolated from the ENU-treated p53−/− mice did not differentiate along the distinct lineages and retained very immature features, including simple bipolar or unipolar morphology and nestin immunoreactivity (Fig. 9E). Some cells were both nestin+ and GFAP+ immunoreactive and showed the tendency to grow as cellular aggregates (Fig. 9F). Other cells were GFAP+ but with long and bulky processes rather than with the typical flat morphology of cells adhering to the substrate (Fig. 9F). These changes of antigenic properties and morphological appearance resembled those detected in tumoral masses in vivo and were also observed in SVZ cells from p53−/− mice overexpressing the active form of Ras. Although the infection with the viral vector alone did not alter the properties of p53−/− cells (Fig. 9G, H), the overexpression of constitutively active Ras induced a transformed phenotype on these cells, characterized by the persistence of nestin immunoreactivity, growth in aggregates, and immature morphology characterized by bulky processes and small soma (Fig. 9H–J).
This paper addresses the important question of the neoplastic potential of adult SVZ cells. More specifically, it defines how loss of the tumor-suppressor gene p53 modifies the cellular properties of distinct populations of cells in the adult SVZ, changes the dynamics between them, and leads to neoplastic transformation.
The role of neural stem cells and multipotential progenitors as cell-of-origin of glioblastoma has been suggested by the detection of positive immunoreactivity for immature markers such as nestin in human tumors (Dahlstrand et al., 1992; Toda et al., 2001; Ignatova et al., 2002; Bouvier et al., 2003) and by studies in transgenic mice overexpressing activated Ras and Akt in nestin+ cells (Holland, 2000). Because loss of p53 is detected early in glial brain tumors (von Deimling et al., 1992; Watanabe et al., 1996; Hayashi et al., 2004), its mutation is observed in sphere-forming cells isolated from human glioblastoma (Ignatova et al., 2002) and its expression in the adult brain is confined to germinal areas (van Lookeren Campagne and Gill, 1998), we postulated that characterization of the changes induced by loss of p53 in cells of the adult SVZ could lead us to a better understanding of the genesis of glial tumors.
Loss of p53 does not induce spontaneous glial tumors (Donehower et al., 1992; Philipp-Staheli et al., 2004). However, we show here that it induces the formation of periventricular areas of cellular hyperplasia in the adult SVZ, characterized by clusters of GFAP+ stem cells, mature glia, and neuroblasts (supplemental Fig. 1, available at www.jneurosci.org as supplemental material). These regions are highly reminiscent of the “microtumors” around the ventricles that were originally described at the early stages of glioma formation in rats prenatally exposed to mutagens (Lantos and Pilkington, 1979). They also resemble the previously described hyperplastic areas observed after intraventricular grafting of astrocytes-like stem cells (Zheng et al., 2002). The increased number of cells in the SVZ of p53−/− mice is not attributable to decreased apoptosis, as predicted by the well known proapoptotic function of p53 (D’Sa-Eipper et al., 2001; Leonard et al., 2001; Katayama et al., 2005). Our results clearly indicate an increase in compensatory apoptosis in p53−/− mice. Apoptotic cells were detected in vivo in those areas exhibiting the most dramatic hyperplastic changes and were likely attributable to upregulation of proapoptotic genes by other p53 family members (S. Gil-Perotin and P. Casaccia-Bonnefil, unpublished observation). We show here that the increased cell number observed in the SVZ of p53−/− mice is consequent to the expansion of the stem cell/progenitor compartment associated with their rapid differentiation toward the neuronal and glial lineages. Thus, loss of p53 confers a proliferative advantage to the slow- and fast-proliferating population of cells in the adult SVZ and favors the progression toward a differentiated phenotype. The balance between generation of extra SVZ cells and their elimination by apoptosis and differentiation accounts for the lack of spontaneous glial tumors in p53−/− mice (Donehower et al., 1992; Philipp-Staheli et al., 2004).
Prenatal exposure of p53−/− mice to ENU provides a DNA-damaging stimulus that leads to the formation of glioblastoma-like tumor in the adult SVZ (Leonard et al., 2001; Katayama et al., 2005). We show here that the transformed phenotype in ENU-treated p53−/− mice is the consequence of enhanced self-renewal and faster cell division of the relatively quiescent population, associated with expansion and impaired differentiation of multipotential progenitors. Similar effects are observed in vitro in p53−/− cells overexpressing a constitutively active form of Ras.
Therefore, transformation of adult SVZ cells is preceded by the recruitment of the quiescent self-renewing population to the fast-proliferating compartment and their inability to differentiate along distinct lineages. There are several potential explanations for the induced neoplastic changes in the SVZ of p53−/− mice prenatally exposed to ENU. One possibility is that ENU, a potent alkylating agent, induces random mutations in the genome of SVZ cells, and this may affect the function or levels of molecules modulating their behavior. In support of this model is the concept that the susceptibility to develop cancer is dependent on the age of exposure and is directly correlated with the mutagenic effect of ENU (Slikker et al., 2004). Prenatal exposure has been shown to result in a sixfold higher frequency of mutation than adult exposure to the same agent (Slikker et al., 2004). An alternative possibility is that the maturation state of the SVZ cell may result in a differential susceptibility to apoptotic stimuli (Lee et al., 2001) and therefore affect the apoptotic response to mutagenic stimuli (Leonard et al., 2001). Finally, although we hypothesize that neoplastic transformation is a cell-autonomous event triggered by p53 loss of function, we cannot exclude the possibility that it may also result from autocrine or paracrine mechanisms. It is clear, however, from our data and from the literature that neoplastic transformation requires the synergism between p53 loss of function and additional stimuli.
The concept that glioblastoma formation involves the synergism between distinct pathways is supported by several other studies demonstrating tumor formation in p53−/− mice after increased PDGF signaling (Hesselager et al., 2003), prenatal exposure to ENU (Oda et al., 1997; Leonard et al., 2001), or Ras activation (Reilly et al., 2000).
It is important to mention, however, that this cellular mechanism leading to glial tumor formation is specific for SVZ cells and does not apply to other cell types. For example, loss of p53 does not synergize with the constitutive activation of EGF receptor (EGFR) in astrocytes, although these cells have the ability to generate gliomas when expression of constitutively active EGFR is associated with loss of other cell cycle genes such as Ink4a and Arf (Bachoo et al., 2002). It is also worth mentioning that Ink4a/Arf mutations have been also very recently detected in cells isolated from ENU-injected animals (Savarese et al., 2005). Together, these results indicate the existence of at least two mechanisms leading to the genesis of glioblastomas: one that is Ink4a/Arf dependent and p53 independent and another one that is p53 dependent. This explanation is in agreement with several studies in human gliomas reporting that p53 and Inka/Arf mutations or EGFR amplification are mutually exclusive (Sung et al., 2000; Di Sapio et al., 2002). It is also supported by the evidence that EGFR amplification has been frequently observed in association with mutations or deletions of Ink4a/Arf (Costello et al., 1996) but not in association with p53 mutations (von Deimling et al., 1995).
In conclusion, our data indicate that loss of p53 differentially affects the properties of distinct populations of adult SVZ cells by providing a proliferative advantage to slow- and fast-cycling cells. Transformation occurs when multiple converging pathways modulating the cell cycle kinetics cooperatively interact with other pathways affecting the differentiative potential of SVZ cells.
This work was supported by National Institutes of Health Grant NS42925, New Jersey Cancer Commission Grant 05-2414-CCR-EO, and National Multiple Sclerosis Society Grant RG 3553-A5 (P.C.-B.), National Cancer Institute Grants CA096832 and CA71907 and Children’s Brain Tumor Foundation (M.F.R.), Instituto de Salud Carlos III (ISCIII), Madrid Grant G03-056 (J.-M.G.-V.), and Fellowship ISCIII, Reference 01/9513 (S.G.-P.). We thank Dr. A. Berns for the p53flox/flox mice, Dr. C. Eberhart for histological evaluation of the tumors, Dr. J. Downing for retroviral constructs, Dr. C. Sherr, Dr. S. Baker, and Dr. F. Doetsch for critical comments, Dr. A. Alvarez-Buylla for initial encouragement and support, S. Wilkerson, and P. Abano for animal care and genotyping.
↵*M.F.R. and J.-M.G.-V. contributed equally to this work.
- Correspondence should be addressed to Patrizia Casaccia-Bonnefil, Department Neuroscience, R-304, Robert Wood Johnson Medical School, 675 Hoes Lane, Piscataway, NJ 08854. Email: