Author Contributions: B. R. P.: conception and design, financial support, collection of data, data analysis and interpretation, manuscript writing; M. C. H.: conception and design, collection of data, data analysis and interpretation, manuscript writing; A. M.: provision of study material; C. M. W.: collection of data; Z. Z.: provision of study material; M. A. I.: conception and design, financial support, data analysis and interpretation, manuscript writing, final approval of manuscript.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSEXPRESS May 23, 2011.
Neural progenitor cells (NPCs) have the capacity to proliferate and give rise to all major central nervous system cell types and represent a possible cell of origin in gliomagenesis. Deletion of the tumor suppressor gene Tp53 (p53) results in increased proliferation and self-renewal of NPCs and is a common genetic mutation found in glioma. We have identified inhibitor of DNA binding 2 (Id2) as a novel target gene directly repressed by p53 to maintain normal NPC proliferation. p53(−/−) NPCs express elevated levels of Id2 and suppression of Id2 expression is sufficient to inhibit the increased proliferation and self-renewal which results from p53 loss. Elevated expression of Id2 in wild-type NPCs phenocopies the behavior of p53(−/−) NPCs by enhancing NPC proliferation and self-renewal. Interestingly, p53 directly binds to a conserved site within the Id2 promoter to mediate these effects. Finally, we have identified elevated Id2 expression in glioma cell lines with mutated p53 and demonstrated that constitutive expression of Id2 plays a key role in the proliferation of glioma stem-like cells. These findings indicate that Id2 functions as a proproliferative gene that antagonizes p53-mediated cell cycle regulation in NPCs and may contribute to the malignant proliferation of glioma-derived tumor stem cells. STEM CELLS 2011;1090–1101
Neural progenitor cells (NPCs) residing in the subventricular zone (SVZ) of adult mammals have the ability to give rise to all three major central nervous system (CNS) lineages throughout adulthood [1, 2]. The hypothesis that NPCs have the potential to give rise to brain tumors is underscored by the fact that brain tumors, especially gliomas, are often made up of tumor cells with a variety of characteristics representing multiple cell lineages [3, 4]. Importantly, tumor suppressor genes frequently found to be altered in human glioma, including PTEN, RB1, NF1, and Tp53 (p53), function to regulate normal NPC homeostasis in mice [5–9]. The multipotentiality of NPCs and their ability to continually divide throughout the life span of the organism  suggests the possibility that NPCs may be the cell of origin in glioma, and a number of mouse models of glioma provide strong evidence for this possibility [7, 10, 11]. In several mouse models of glioma, activation of the initiating oncogenic mutation results in altered differentiation, hyperplastic growth, and enhanced self-renewal within NPCs in the SVZ [7, 8, 10]. However, very little is known about how genetic lesions common in brain tumors could drive the oncogenic transformation of multipotent adult NPCs.
We have studied the effect of Tp53 (p53) pathway disruption in NPCs. p53 inactivation is thought to be an early event in gliomagenesis  and is estimated to occur in 87% of such tumors . Inherited germline mutations within the p53 gene, as seen in patients with Li-Fraumeni syndrome, result in an increased incidence of glioma with an early onset . Deletion of p53 in mice results in numerous preneoplastic phenotypes in NPCs including enhanced self-renewal, accelerated growth, and an expanded SVZ [8, 9] suggesting that p53 functions not only as a tumor suppressor but also as an essential NPC regulatory gene. Targeted inactivation of the p53 DNA binding domain in cells expressing glial fibrillary acidic protein (GFAP+ cells) is sufficient to initiate glioma arising in SVZ progenitor cells, although many other GFAP+ cell types, including astrocytes, contain the initiating p53 mutation . These studies support the notion that gliomas frequently arise from SVZ NPCs that have lost p53 function. Understanding how p53 functions to regulate the biology of NPCs should identify critical NPC regulatory pathways that when deregulated could promote brain tumorigenesis.
Id genes are known to promote many key aspects of NPC biology including proliferation and self-renewal [15–17] and two cellular processes critical for tumor growth. Four Id genes (Id1–Id4) exist within the mammalian genome and are thought to exert their function most often by heterodimerizing with basic helix-loop-helix (bHLH) transcription factors through the helix-loop-helix domain [18, 19]. Id proteins lack a basic DNA binding domain and, as a result, function as dominant-negative inhibitors of bHLH DNA binding. Most frequently, this results in the inhibition of differentiation [15, 20] and promotion of cell cycle progression [15, 21]. Recently, our laboratory has identified a dramatic decrease in adult neurogenesis and NPC proliferation after targeted deletion of inhibitor of DNA binding 2 (Id2) , which appears to be critical for embryonic NPC growth and self-renewal .
We have discovered a molecular mechanism that mediates, at least in part, p53-dependent hyperplasia of NPCs [8, 9]. We identified elevated expression of Id2 in p53(−/−) NPCs and demonstrated that Id2 expression alone is sufficient for maintaining the increased growth and self-renewal of p53(−/−) NPCs. We also found evolutionarily conserved p53 binding sites within the Id2 promoter and established that p53 binding to one of these directly represses Id2 transcription. Finally, we have identified elevated Id2 expression in glioma cell lines with mutant p53 and demonstrated that constitutive expression of Id2 plays a key role in the proliferation of glioma derived stem-like cells. These data indicate that p53 functions to regulate both NPC and glioma stem-like cell proliferation and self-renewal by directly modulating Id2, a known regulator of proliferation in NPCs . Our data support a model by which loss of p53 function, common in the majority of gliomas, drives cell cycle transit by increasing Id2 expression thus potentiating the malignant transformation of NPCs.
MATERIALS AND METHODS
Tissue Harvest and Culture
Primary NPCs were isolated from P0 mouse forebrains and cultured in Dulbecco's modification of Eagle's medium (DMEM)/F12 containing 10 ng/ml epidermal growth factor (EGF) and B27 as described previously . To prepare glioma stem-like cells, tumor cell lines were replated in NPC proliferation media without EGF.
Proliferation and Apoptosis Analyses
A total of 2 × 104 NPCs growing in 24-well plates were counted using trypan blue exclusion, except for the experiment shown in Figure 1A, in which 5,000 cells were plated. To determine cell cycle distribution NPCs stained with propidium iodide (PI) were examined by flow cytometry using Modfit. Carboxyfluorescein diacetate succinimidyl ester (CFSE, Invitrogen, Carlsbad, CA, www.invitrogen. com) labeling was performed as described previously . Apoptosis was quantified by live cell flow cytometry using Annexin-V/PI (BD Biosciences, San Jose, CA, www.bdbiosciences.com) and analyzed using FlowJo (Ashland, OR, www.flowjo.com). TUNEL staining was performed after plating on laminin-coated chamber slides.
NPCs were dissociated, plated in proliferation media for at least 24 hours, and treated with the indicated amounts of gamma radiation at 10 Gy/minute in a 137Cs irradiator.
To differentiate NPCs, cells were plated on laminin coated chamber slides and placed in DMEM/F12 containing L-glutamine, 10% fetal bovine serum, and B27(with vitamin A, Invitrogen) for 14 days and immunostained with the indicated antibodies. For cytospins, 5 × 104 dissociated NPCs were centrifuged onto poly-L-lysine slides at 600 rpm and immunostained with mouse anti-Ki-67 antibody (Leica Microsystems. Wetzlar, Germany, www.leica-microsystems.com), detected by fluorescence, and quantified.
DNA corresponding to the human Id2 (hId2) promoter was purchased from Switchgear Genomics (Menlo Park, CA, www.switchgeargenomics.com). Mutagenesis of Id2 luciferase reporters was performed using the Stratagene (Santa Clara, CA, www.genomics.agilent.com) Quik Change kit. Each of the two human Id2 luciferase (hId2-luc) constructs deleting different binding sites also lacked a 10-bp region of high G/C content 57 bp upstream from the human Id2 transcriptional start site. pBMN-Id2 encoded an N-terminal flag-tagged human Id2 cDNA. All constructs were verified by DNA sequencing.
Luciferase reporter constructs, 800 ng, were cotransfected with 20 ng of pRL-TK Renilla and 200 ng of pMAX-green fluorescent protein (GFP) into 293 T cells using LipoD293 (SignaGen, Rockville, MD, www.signagen.com). After 24 hours of incubation, cells were infected with recombinant adenovirus encoding LacZ or p53. Luciferase activity was assayed using Dual-Luciferase Assay (Promega, Madison, WI, www.promega.com).
Chromatin was prepared from NPCs fixed with 1% formaldehyde, sheared by sonication (average 500 bp), immunoprecipated with anti-p53 antibody (5 μg, FL-393, Santa-Cruz Biotechnology, Santa Cruz, CA, www.scbt.com), and processed using a chromatin immunoprecipitation (ChIP) kit (Millipore, Billerica, MA, www.millipore.com). Polymerase chain reaction (PCR) primer sequences included:
p53 Site 1: 5′TCGCATCACTTTGCCACCTACACT (sense), 5′CGAATTGTAGGAACACTGTGCGGT (antisense).
p53 Site 2: 5′TGCTCCAAGTTGCAAAGCTTCACG (sense), 5′GGCAAATTGAGTACAGTGTGCGCT (antisense).
p53 Site 3: 5′ACCAATGGGAGAATTCGCCTGGTA (sense), 5′ACTGAAGGCTTTCATGCTGCTCGT (antisense).
RNA was reverse-transcribed using iScript (Bio-Rad, Hercules, CA, www.bio-rad.com). Controls were incubated without reverse transcriptase. cDNA was quantified using MyiQ Sybr Green qPCR (Bio-Rad). Fold change was determined after normalization to β-actin (ΔΔ−ct). Primer sequences included:
Lysates (100 μg) of NPCs were analyzed by western blot using anti-Id2 1:500 (Santa Cruz, sc-489) or anti-Id2 1:500 (Cell Signaling Technology, Danvers, MA, www.cellsignal.com, D39E8), anti-p53 1:1,000 (Santa-Cruz, sc-6243), anti-p21 1:500 (Santa-Cruz, sc-6246), anti-total poly ADP ribose polymerase (PARP) 1:2,000 (Cell Signaling Technologies, 9542), or anti-β-actin 1:5,000 (Sigma, St. Louis, MO, www.sigmaaldrich.com, A-1978) antibodies. For densitometry, images were quantified by relative pixel area using Photoshop (Adobe, San Jose, CA, www.adobe.com).
NCI-60 Microarray Meta-Analysis
Microarray gene expression values  were downloaded using the NCI cell miner webpage (http://discover.nci.nih.gov/cellminer), and the accompanying p53 mutational sequence data was obtained. Average gene expression was evaluated using a one-sided t test.
Experiments were performed using cells isolated from Trp53Tm-Tyj (p53(−/−)) C57BL/6 mice . The Id2CAG-FID2 mouse was created within our laboratory (unpublished data).
Elevated and Aberrant Expression of Id2 in p53(−/−) NPCs
Multipotent NPCs can be cultured as neurospheres in vitro and give rise to all three CNS lineages upon differentiation (Supporting Information Fig. 1). Our laboratory and others have observed that targeted deletion of the tumor suppressor p53 is associated with increased proliferation of NPCs (Fig. 1A) [8, 9]. The molecular mechanisms mediating the increased growth of p53(−/−) NPCs are largely unknown. Given the importance of p53 inactivation in brain tumor pathogenesis and the potential for NPCs to give rise to such tumors , we sought to better understand the effect of p53 loss on NPC proliferation. We determined that p53 deletion altered the cell cycle distribution of NPCs in culture. p53(−/−) NPCs exhibited a significant decrease in the fraction of cells in the G0/G1 stage of the cell cycle and an increase in the percentage of cells in S-phase as compared to wild-type (WT) NPCs (Fig. 1B). This supports a role for p53 in restraining the proliferation of NPCs and is consistent with the observations of p53-dependent regulation of adult stem cells from neuronal  and mammary tissues .
Our laboratory has previously identified a dramatic decrease in the proliferation of Id2(−/−) NPCs cultured as neurospheres in vitro . We noticed that the increased growth rate and percentage of cells in S-phase in p53(−/−) NPCs contrasted sharply with the decreased proliferation of Id2(−/−) NPCs (unpublished data and ). This led us to hypothesize that both Id2 and p53 may function to regulate NPC proliferation by modulating the G1–S transition. To investigate if altered Id2 expression occurs in p53(−/−) NPCs, we examined steady state Id2 levels in WT and p53(−/−) NPCs cultured as neurospheres. Western blot analysis revealed a substantial increase in Id2 protein expression in p53(−/−) neurospheres (Fig. 1C) along with a significant increase in Id2 mRNA in p53(−/−) NPCs detected by quantitative PCR (qPCR) (Fig. 1D). The increase in Id2 mRNA we observed under steady state conditions in p53(−/−) NPCs suggested that p53, either directly or indirectly, could modulate Id2 mRNA levels. To further investigate the extent to which p53 might repress Id2, we treated WT and p53(−/−) NPCs with 2 Gy of gamma irradiation to increase p53 protein expression and transcriptional activity (Fig. 1E) (reviewed in ). We observed a decrease in Id2 protein 4–12 hours after treatment with gamma irradiation (Fig. 1F), which corresponds to the kinetics of p53 activation (Fig. 1E). In contrast, p53(−/−) NPCs express elevated levels of Id2 and after irradiation and DNA damage, p53(−/−) NPCs fail to demonstrate decreased Id2. Taken together, these data support a model in which p53 functions to inhibit Id2 expression and p53 loss leads to inappropriately high levels of Id2 expression.
To evaluate the possibility that Id2 expression is diminished because of loss of cells by apoptosis after DNA damage, we quantified the number of apoptotic cells following irradiation (2 Gy) and observed that this dose does not induce apoptosis in WT or p53(−/−) NPCs (Supporting Information Fig. 2). To examine further the effect of p53 on Id2 expression, we introduced exogenous WT p53 into p53(−/−) NPCs using an adenovirus-derived vector. Adenoviral introduction of p53 suppressed Id2 protein to undetectable levels (Supporting Information Fig. 3) and is consistent with a role for p53 in inhibiting Id2 expression in NPCs.
Inhibition of Id2 Expression is Sufficient to Prevent the Hyperplastic Growth and the Enhanced Self-Renewal of p53(−/−) NPCs
We hypothesized that Id2 was a key mediator of the accelerated proliferation of NPCs, which occurs after loss of p53 [8, 9]. To test this hypothesis, we inhibited the expression of Id2 in p53(−/−) NPCs by retroviral infection of short hairpin RNAs (shRNAs) targeting Id2 (Origene, HuSH system) and examined the proliferation of these NPCs. Western blot analysis of neurosphere cell cultures stably expressing Id2 shRNAs (p53(−/−)/sh-Id2) demonstrated decreased expression of Id2 as compared to levels detected in cells infected with a scrambled, nonspecific control shRNA (p53(−/−)/sh-Scramble; Fig. 2A). Early passage p53(−/−)/sh-Id2 NPCs routinely formed significantly smaller neurospheres than p53(−/−)/sh-Scramble NPCs (Fig. 2B), suggesting a decrease in proliferation after inhibition of Id2. To characterize the observed decrease in proliferation, p53(−/−)/sh-Scramble and p53(−/−)/sh-Id2 neurospheres were dissociated, fixed to glass slides, and immunostained using an antibody against the proliferation marker Ki67 . We observed a significant decrease in the frequency of Ki67+ cells in p53(−/−)/sh-Id2 NPCs when compared with control cells (Fig. 2C). To further quantify the alterations in growth kinetics following Id2 inhibition in p53(−/−) NPCs, we examined early passage p53(−/−)/sh-Id2 NPCs in a growth assay. After 1 week of growth, each independent p53(−/−)/sh-Id2 culture grew significantly slower than control p53(−/−)/sh-Scramble NPCs (Fig. 2D). This decreased level of proliferation was indistinguishable from the proliferation rate observed in WT NPCs. The inhibition of Id2 expression did not alter the expression of the cyclin-dependant kinase inhibitor, p21, in either p53(−/−)/sh-Id2 or Id2(−/−) NPCs (Supporting Information Fig. 4). Similarly, we did not observe changes in the percentage of cells positive for nestin expression (Supporting Information Fig. 5). To determine if decreased Id2 expression alters the differentiation of p53(−/−) NPCs, we differentiated p53(−/−)/sh-Scramble and p53(−/−)/sh-Id2 cultures and found no indication that Id2 levels affected the capacity of these cells to form neurons, astrocytes, or oligodendrocytes (Supporting Information Fig. 6).
The ability of NPCs to self-renew, as measured in an assay for secondary spheres, is critical to the expansion of neurospheres in vitro and evidence suggests that loss of p53 increases the rate of NPC self-renewal . Recently, Id proteins were implicated in regulating the self-renewal rate of NPCs . To determine if Id2 can affect the self-renewal of p53(−/−) NPCs, we plated p53(−/−)/sh-Id2 and p53(−/−)/sh-Scramble NPCs in secondary sphere formation assays. NPCs were dissociated and plated at a limiting dilution of approximately one cell per well in 96-well plates to determine the frequency with which individual cells would reform neurospheres. We observed that suppression of Id2 expression reduced the self-renewal rate of p53(−/−) NPCs to levels comparable to WT (Fig. 2E).
We next evaluated the effect of Id2 inhibition on the rate of p53(−/−) apoptosis. Infection of Id2 shRNAs with recombinant retrovirus resulted in decreased Id2 levels but inhibition was not below the levels observed in WT NPCs (Fig. 2F). Decreased expression of Id2 resulted in a significant increase in the frequency of TUNEL positive cells 48 hours after treatment (Fig. 2G). These data suggest that Id2 expression, increased as the result of p53 inactivation, contributes to maintain enhanced growth and increased self-renewal of p53(−/−) NPCs as a result of both increased proliferation and decreased apoptosis.
Elevated Expression of Id2 Is Sufficient to Increase the Proliferation and Self-Renewal of WT NPCs
The identification of Id2 as a critical mediator of p53-dependent NPC proliferation implies that Id2 functions to modulate cell cycle progression. Increased expression of Id2 in cells that possess WT p53 should phenocopy the hyperplastic growth and enhanced self-renewal of p53(−/−) NPCs described by others [8, 9]. To examine this possibility, we used a transgenic mouse from our laboratory in which GFP tagged human Id2 was placed under the conditional regulation of the chicken actin gene (CAG) promoter (unpublished data). Under basal conditions, the CAG promoter transcribes the chloramphenicol acetyltransferase (CAT) gene which is flanked by loxP sites (Fig. 3A). Upon Cre-mediated recombination, the floxed CAT gene is excised from the genome placing the Id2 transgene downstream of the constitutively active CAG promoter (Id2CAG-FID2).
To prepare NPCs that overexpressed Id2, we established neurosphere cultures from the brains of P0 Id2CAG-FID2 mice. Dispersed neurospheres were infected with adenovirus-expressing either β-galactosidase or Cre recombinase (Ad-LacZ or Ad-Cre, respectively). To determine if genomic recombination occurred at the Id2CAG-FID2 locus after Ad-Cre infection, we used primers which specifically amplify the recombined allele by priming from the CAG promoter and the Id2 gene (primers AG2 and Id2R, Fig. 3A) or which amplify the CAT gene (CAT2 and CAT3 primers). Genomic PCR analysis of adenoviral-infected neurospheres indicated that transgene recombination occurred only in NPCs that received Ad-Cre (Fig. 3B). However, the persistence of CAT gene PCR products in Ad-Cre infected cultures indicated that recombination was not 100% efficient and suggests that these cultures also contained cells that have not undergone recombination and do not express the Id2 transgene (Fig. 3B). Next, we performed quantitative PCR for total Id2 transcript in Id2CAG-FID2 infected cells and semiquantitative reverse-transcriptase PCR for Cre expression. We observed a dose-dependent increase in total Id2 mRNA expression dependent on the amount of Ad-Cre virus used for infection (Fig. 3C) and western blot analysis of Id2 from Id2CAG-FID2/Ad-Cre neurospheres demonstrated an enhanced expression of Id2 (Fig. 3D) at the predicted weight, 42 kDa (molecular weight Id2: 15kDa, GFP: 27 kDa).
We noted that Id2CAG-FID2/Ad-Cre NPCs routinely formed larger neurospheres than control Ad-LacZ cells with a diameter similar to p53(−/−) NPCs (Fig. 4A, 4B). Therefore, we examined the cell cycle distribution of Id2CAG-FID2 NPCs and found that proliferating Id2CAG-FID2/Ad-Cre NPCs mimic the cell cycle distribution of p53(−/−) NPCs (Fig. 4C). These data indicate that both Id2 gain of function and p53 loss of function result in NPCs with similar cell cycle distributions. To more rigorously evaluate the growth of Id2CAG-FID2 NPCs, we used CFSE staining. We identified a decrease in CFSE fluorescence intensity in Id2CAG-FID2/Ad-Cre infected cells at 2 and 4 days postlabeling (Fig. 4D), suggesting that enhanced expression of Id2 is sufficient to increase the proliferation of NPCs expressing p53 (median fluorescence at day 4: Id2CAG-FID2/Ad-LacZ = 474 and Id2CAG-FID2/Ad-Cre = 169).
To further examine alterations in cellular proliferation, we assayed the proliferation of Id2CAG-FID2/Ad-Cre NPCs and observed that Id2CAG-FID2/Ad-Cre NPCs proliferated more rapidly than Id2CAG-FID2/Ad-LacZ cells (Fig. 4E). However, increased Id2 expression did not alter the expression of p21 (Supporting Information Fig. 7). To examine the effect of Id2 on NPC self-renewal, we infected WT NPCs with retrovirus encoding flag-tagged human Id2. Following infection of NPCs with recombinant retrovirus, we used antibiotic selection to prepare polyclonal cultures expressing elevated levels of Id2 (Supporting Information Fig. 8). After 2 weeks of growth as single cells suspended in soft agar [29, 30], we observed a significantly increased number of secondary neurospheres formed from NPCs expressing Id2 (Fig. 4F). We interpret these results to indicate that Id2 can function to increase NPC self-renewal.
We next investigated the effect of Id2 expression on apoptosis 24 hours after radiation (2–10 Gy)-induced DNA damage by AnnexinV/PI flow cytometry. Under these conditions, Id2 expression resulted in decreased PI-positive and AnnexinV/PI double positive cells (Fig. 4G). These data suggest that Id2 expression may function to inhibit programmed cell death and are consistent with the observation that inhibition of Id2 expression results in increased apoptosis in p53(−/−) NPCs (Fig. 2G).
Binding of p53 to the Id2 Promoter Is Required for p53 Repression
Recently, multiple mechanisms of p53-mediated transcriptional repression have been characterized . To determine if p53 could physically associate with the Id2 promoter, we first identified regions of the Id2 promoter evolutionarily conserved between mouse and human by pair-wise sequence analysis 1 kilobase upstream from the transcriptional start site using VISTA  (http://genome.lbl.gov/vista/index.shtml; Fig. 5A, top). We identified three potential p53 binding sites that occurred in evolutionarily conserved regions shared by the human and mouse promoter regions (Fig. 5A, center). These binding sites contain deviations of no more than 3 bp from the 20 nucleotide consensus p53 binding motif (RRRCWWGYYY(0–13 bp spacer)RRRCWWGYYY)  and these deviations are highly conserved between species (Fig. 5A, bottom). The sequences of each of the p53 binding sites contain nucleotide spacers between the decameric half sites (Fig. 5A, bottom). The presence of nucleotide spacer regions occur almost exclusively within binding sites of p53 target genes that are transcriptionally repressed by p53 [34–36], supporting our observations that p53 functions as a transcriptional repressor of Id2.
To determine if p53 can physically associate with these three p53 binding sites, we performed p53 ChIP. Using primers flanking each of the three p53 binding sites to probe for the presence of the corresponding DNA fragments, we found that Id2 promoter DNA corresponding to p53 site 3 immunoprecipitated with p53 but not to sites 1 or 2 (Fig. 5B). We verified the specificity of this immunoprecipitation by examining a known p53 binding site in the promoter of the p53 target gene p21  as well as immunoprecipitating from p53(−/−) extracts. These data indicated that p53 can physically interact with the “Id2 promoter under normal proliferative conditions in vitro.”
To determine if the p53 binding sites located in the Id2 promoter are required for p53-mediated repression, we used a commercially available hId2-luc construct (SwitchGear Genomics) that contained p53 binding sites 2 and 3 (Fig. 5C). To assess the effect of p53 on the hId2-luc reporter, we transfected 293 T cells with the hId2-luc reporter and titrated in increasing amounts of Ad-p53. To avoid the potentially confounding effects of apoptosis, we assayed 293 T cells for luciferase activity 16 hours postinfection before any evidence of cell death was discernable visually and PARP cleavage was minimal  (unpublished data). We observed that increasing amounts of p53 expression resulted in a dose-dependent decrease of hId2-luc reporter expression (Fig. 5D). This result indicates that p53 can act as a transcriptional repressor of Id2 and that this repression occurs within the cloned promoter region containing p53 sites 2 and 3. We then created additional luciferase reporter constructs in which individual sites were ablated using site directed mutagenesis. We began these experiments by deleting either site 2 (hId2-lucΔ site 2) or site 3 (hId2-lucΔ site 3). Addition of Ad-p53 virus led to the repression of hId2-lucΔ site 2 (Fig. 5D) indicating that p53 site 2 does not function to repress Id2 expression. In contrast, hId2-lucΔ site 3 did not demonstrate p53-mediated repression (Fig. 5D), thereby identifying a putative conserved site within the Id2 promoter that functions to repress transcription in a p53-dependent manner. The identification of endogenous p53 physically associated with binding site 3 by ChIP, coupled with the observation that deletion of p53 site 3 abolished Id2 repression (Fig. 5D), provides strong evidence that p53 binding site 3 is a critical cis-acting element mediating p53 repression of Id2 expression.
Identification of Id2 As a Bona Fide Target of p53 Repression in Human Cancer Cell Lines
The loss of p53 function occurs commonly in a multitude of tumor types . This fact prompted us to determine if WT p53 repressed Id2 in human cancer cell lines. We sought to determine if high levels of Id2 expression correlated with the loss of p53 function and reduced Id2 expression with WT p53. Therefore, we conducted a meta-analysis of previously compiled microarray data from the NCI-60 panel of cell lines evaluating gene expression changes after DNA damage, a condition that is known to activate p53 activity (Fig. 1E) [24, 27]. We stratified the NCI-60 panel into two groups based on p53 mutational status. This assignment was predictive of elevated expression of the established p53 response gene, p21, after irradiation (Fig. 6A). Next, we evaluated changes in Id2 expression and in the only other Id family member, Id3, which is reported in the dataset. We discovered a significant decrease in Id2 gene expression after irradiation only in tumors that retained WT p53 and could not identify p53-dependent repression of Id3 expression after DNA damage (Fig. 6A). Based on these data, we conclude that p53 repression of Id2 is likely to occur in human cancer cell lines bearing a WT p53 gene and this repression was not seen in all members of the Id gene family.
Based on the similarities between NPCs and glioma cells , we chose to evaluate further the relationship between Id2 and p53 within human glioma cell lines. We selected four well-characterized glioma lines. One of these, U87, has WT p53 and three are p53 mutant lines: U251, SW1088, and A172 . U251 was among the six glioma cell lines in the NCI-60 panel. To verify WT p53 functionality, we treated each cell line with 2 Gy gamma irradiation and screened for p21 induction. As expected, the p53 WT U87 glioma line expresses p21 after DNA damage, while p21 expression was not induced in cell lines with mutated p53 (Fig. 6B). Consistent with our previous findings, we were unable to detect Id2 in the p53 WT U87 cell line, but we did find high levels of Id2 in each of the three p53 mutant glioma lines (Fig. 6B). This suggests that WT p53 can repress Id2 in glioma-derived tumor cell lines, but mutant p53 cannot.
Increased Expression of Id2 Enhances the Proliferation of Glioma Stem Cells
Our current experiments demonstrate that p53 can function to repress Id2 in some human tumor cell lines (Fig. 6) and glioma cell lines with mutated p53 expressing higher levels of Id2 than the p53 WT U87 glioma line (Fig. 6B). Therefore, we tested whether Id2 overexpression could promote the growth of glioma cells in a manner similar to that observed in p53(−/−) NPCs (Figs. 2, 4). Glioma cell lines cultured in defined media known to support the growth of NPCs become enriched for cells with characteristics of tumor stem cells  and mirror phenotypes of primary tumors [39, 42]. Therefore, we used this culture method to study the effects of Id2 expression in glioma stem-like cells. We prepared U251 glioma cells that expressed elevated levels of Id2 using recombinant retrovirus encoding Id2 (U251 +Id2; Fig. 7A). We observed larger sphere size and increased growth of U251 +Id2 glioma stem-like cells when compared with U251 vector control cells (Figs. 7B, 7C). Similar to our observations in NPCs, increased expression of Id2 in glioma stem-like cells results in an increase in proliferation, suggesting that Id2 is a key driver of cell cycle progression in both normal and transformed cells of the CNS.
Identification of the molecular basis of neural stem cell and NPC homeostasis is of critical importance for understanding normal CNS development and the potential of these cells to initiate tumorigenesis [7, 10, 11]. p53 is known to control the size of the neural stem cell compartment in vivo under normal physiologic conditions . We have discovered that p53 directly represses Id2 and modulates NPC proliferation. This previously unrecognized regulatory activity implicates Id2 as a key functional element mediating p53 function in NPCs. Although Id2 has been well-established to promote cell cycle progression, our data suggests that Id2 can also function to inhibit apoptosis in NPCs.
We have evaluated in NPCs previously established mechanisms of Id2-mediated growth promotion through antagonism of the p21 or Rb pathways [43, 44]. Although displaying pronounced alterations in proliferation, neither p53(−/−)/shId2 nor Id2CAG-Cre/Ad-Cre NPCs exhibit changes in p21 expression (Supporting Information Figs. 4, 7). Similarly, we have not observed decreased expression of E2F1 target genes including Ccnd1, Cdc6, and Cdc25a in Id2(−/−) NPCs as would be predicted if increased Rb activity was responsible for the Id2(−/−) NPC growth defect (unpublished data). This could imply that a novel and previously unrecognized mechanism exists for Id2 in promoting the growth of NPCs.
Our findings demonstrate that p53 functions by repressing Id2 through binding to the conserved p53 binding site 3 within the Id2 promoter (Fig. 5). The context in which p53 inhibits target gene expression can be determined, in part, by the composition of the p53 response element itself [31, 45]. Although the exact mechanism of p53 repression of Id2 has not been determined, the structure of the p53 response elements within the Id2 promoter suggests the recruitment of chromatin remodeling complexes, rather than steric hindrance through binding site overlap, as a putative mechanism . Two key characteristics found in the majority of p53 repressed target genes that distinguish these from sites of activation are: (a) The presence of a spacer region between the two decameric half sites [34, 36, 46] and (b) substitutions within the core “CWWG” motif . Both of these attributes occur in the p53 binding site 3 present within the Id2 promoter and are highly conserved between mouse and human. Substitutions in the core motif may affect DNA flexibility and tetrameric p53 binding resulting in conformational changes that may contribute to differential cofactor recruitment [45, 47]. Further elucidation of the mechanisms determining p53-mediated repression, including defining the genetic elements required for this repression, will lead to an enhanced understanding of the coordinated gene expression required for p53 pathway function.
Our observation of increased Id2 expression in p53 deficient NPCs (Fig. 1) may also be of significance for cancer, because inactivation of the p53 pathway has been reported to occur in 87% of glioblastomas , tumors which may arise in NPCs . Because many biological characteristics and molecular pathways are shared by normal NPCs and glioma stem-like cells, the identification of key effector genes downstream of p53 in both these cell types is of great interest. We found that Id2 was highly expressed in glioma-derived cell lines that lacked a functional p53 (Fig. 6), and Id2 can drive the proliferation of glioma-derived tumor stem cells (Fig. 7). These findings complement our observation of the importance of Id2 expression in NPCs and suggest that Id2 is a critical effector gene driving the proliferation of p53 deficient brain tumor stem-like cells and thereby contributing to oncogenesis (Fig. 7).
WT p53 has not previously been recognized as a regulator of Id2 and has not been examined in normal tissue. Fontemaggi et al.  have reported that mutant p53 can activate Id2 expression in the colon cancer cell line, SW480, and in comparable experiments using this same cell line, Yan et al.  have reported repression of Id2 by mutant p53, but not the WT allele. Importantly, these authors did not evaluate Id2 expression in reporter assays which contained our newly identified p53 binding site 3. The finding of Id2 activation by mutant p53 is consistent with our observation that WT p53 is associated with low levels of Id2 expression in the diverse NCI-60 panel of cancer cell lines (Fig. 6A), including cell lines derived from colon tumors. We specifically examined two glioma lines containing hotspot mutations in the same amino acid, R273, as was examined by Fontemaggi et al. , and found that these exhibited high levels of Id2 expression (Fig. 6B). Importantly, these mutations are known to be gain-of-function mutations  suggesting that some mutations within the p53 gene may elevate Id2 expression by converting p53 from a transcriptional repressor to an activator.
Finally, a recent molecular subclassification of gliomas from The Cancer Genome Atlas Project has associated p53 mutations most closely with gene expression signatures similar to oligodendrocyte progenitor cells . Many cellular processes altered in p53 inactivated tumors strikingly coincide with Id2-mediated processes including the promotion of NPC proliferation (this publication, [15, 17, 52]) and inhibition of oligodendrocyte maturation [53, 54]. These studies and our data strongly suggest a role for Id2 in regulating the malignant characteristics of p53 deficient glioma stem-like cells. The identification of key molecular changes that drive NPC and tumor stem cell growth, such as enhanced Id2 expression, have the potential to inform future targeted therapies.
p53 directly represses Id2 transcription to inhibit the proliferation of NPCs and glioma stem-like cells.
We thank Sarah Gilman for expert handing of all animals used in this study and the help of Dr. Jiang Gui with statistical analysis. This study was supported by an NIH fellowship F31NS064634 (to B.R.P.) as well as the Theodora B. Betz Foundation and the Jordon and Kyra Memorial Foundation (to M.A.I.). A.M. is currently affiliated with the Division of Psychiatry, Yoshida General Hospital, Hiroshima, Japan.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors have no potential conflicts of interest.