RESEARCH ARTICLE

Loss of p53 Cooperates with K-rasActivation to Induce Glioma Formation

in a Region-Independent Manner

Diana Marcela Mu~noz,1 Takyee Tung,1 Sameer Agnihotri,1 Sanjay Singh,1 Abhijit Guha,1,2

Gelareh Zadeh,1,2 and Cynthia Hawkins1,3

Gliomas are recognized as a heterogeneous group of neoplasms differing in their location and morphological features. Thesedifferences, between and within varying grades of gliomas, have not been explained solely on the grounds of an oncogenicstimulus. Interactions with the tumor microenvironment as well as inherent characteristics of the cell of origin are likely asource of this heterogeneity. There is an ongoing debate over the cell of origin of gliomas, where some suggest a progenitor,while others argue for a stem cell origin. Thus, it is presumed that neurogenic regions of the brain such as the subventricularzone (SVZ) containing large numbers of neural stem and progenitor populations are more susceptible to transformation. Ourstudies demonstrate that K-rasG12D cooperates with the loss of p53 to induce gliomas from both the SVZ and cortical region,suggesting that cells in the SVZ are not uniquely gliomagenic. Using combinations of doxycycline-inducible K-rasG12D and p53loss, we show that tumors induced by the cooperative actions of these genes remain dependent on active K-ras expression,as deinduction of K-rasG12D leads to complete tumor regression despite absence of p53. These results suggest that the inter-play between specific combinations of genetic alterations and susceptible cell types, rather than the site of origin, are impor-tant determinates of gliomagenesis. Additionally, this model supports the view that, although several genetic events may benecessary to confer traits associated with oncogenic transformation, inactivation of a single oncogenic partner can underminetumor maintenance, leading to regression and disease remission.

GLIA 2013;00:000–000Key words: gliomagenesis, mouse model, K-ras, inducible expression system, region of origin

Introduction

For decades, it was widely assumed that differentiated glia

were the only cells capable of malignant transformation as

the adult brain was thought to be mitotically inactive. How-

ever, the demonstration of functional neurogenesis in the

adult central nervous system (CNS) provided new possibilities

for the candidate cell of origin of CNS neoplasms (Eriksson

et al., 1998; Gage, 2000; Sanai et al., 2004). These progeni-

tor populations reside in multiple regions of the adult brain,

including the subventricular zone (SVZ) and the dentate

gyrus within the hippocampus (Eriksson et al., 1998; Sanai

et al., 2004). Although gliomas can originate throughout the

white matter and are not limited to the SVZ, this region has

long be proposed as a source of gliomas, as many are either

periventricular or contiguous with the SVZ (Bohman et al.,

2010b; Lim et al., 2007a). This notion has gained increasing

popularity with the discovery of cells with stem-like proper-

ties within human gliomas (Singh et al., 2003, 2004; Venere

et al., 2011). Although the concept of cancer stem cells and

the idea that gliomas arise from stem or progenitor cells are

distinct, many studies have focused on the neural stem cell

(NSC) population in the SVZ as the cells of origin for

gliomas.

Several lines of evidence suggest that NSC-rich regions

in the brain are more susceptible to malignant transforma-

tion. For example, GFAP and Nestin-Cre mediated

View this article online at wileyonlinelibrary.com. DOI: 10.1002/glia.22563

Published online Month 0, 2013 in Wiley Online Library (wileyonlinelibrary.com). Received Jan 24, 2013, Accepted for publication July 17, 2013.

Address correspondence to Cynthia Hawkins, Department of Paediatric Laboratory Medicine, The Hospital for Sick Children, 555 University Avenue, Toronto, ON,

Canada M5G 1X8. E-mail: cynthia.hawkins@sickkids.ca

From the 1The Arthur and Sonia Labatt Brain Tumor Research Centre, Hospital for Sick Children Research Institute, University of Toronto, Toronto, ON, Canada;2Division of Neurosurgery, Toronto Western Hospital, University of Toronto, Toronto, ON, Canada; 3Division of Pathology, The Hospital for Sick Children, Toronto,

ON, Canada.

Gelareh Zadeh and Cynthia Hawkins are co-senior authors.

VC 2013 Wiley Periodicals, Inc. 1

inactivation of the tumor suppressors Nf1 and p53 induces

glial progenitor proliferation and ultimately malignant astro-

cytomas; in these models the earliest identifiable areas of

tumor are confined to the SVZ (Alcantara Llaguno et al.,

2009; Zhu et al., 2005). Similar results are observed when

deletion of Nf1, p53, and Pten are targeted to the SVZ

which, as opposed to targeting non-neurogenic regions, leads

to glioma formation with 100% penetrance (Alcantara Lla-

guno et al., 2009). Collectively, this suggests that NSCs/pro-

genitors in the SVZ are uniquely susceptible to

transformation. However, recent studies suggest that this sus-

ceptibility may be shared by cycling progenitors distributed

throughout the cortex (CTX) and white matter (Gensert and

Goldman, 2001; Liu et al., 2011). This idea is supported by

animal models where over-expression of PDGFB induced the

development of tumors when targeted to the forceps minor

of the corpus callosum in rats or to the CTX of an Ink4a/Arf

null transgenic mouse model (Assanah et al., 2006; Hambard-

zumyan et al., 2011).

In light of these conflicting viewpoints and observations,

we developed an inducible mouse model in which the expres-

sion of activated K-ras in combination with loss of p53 is

induced after Cre-mediated excision. Using this model, we

have targeted the SVZ and cortical regions to test the glioma-

genic potential of these two distinct anatomical regions to the

same oncogenic stimulus. We set out to investigate (1)

whether the cell populations residing in these regions are sus-

ceptible to transformation; (2) To what extent the phenotype

and behaviour of the tumor is determined by the region of

origin; and (3) is tumor viability dependent on the continued

expression of Ras?

In this study, we show that expression of K-rasG12D and

loss of p53 in the adult mouse brain are sufficient for full

and rapid malignant transformation regardless of the region

of origin. Additionally, deinduction of mutant K-ras expres-

sion alone is sufficient for regression of the resulting tumors

with consequent prolonged survival. These results indicate

that the interplay between specific combinations of genetic

mutations and susceptible cell types, regardless of the region

of origin, is what defines the tumorigenic potential of a spe-

cific cell population.

Material and Methods

MiceThe following genotypes were used Rosa (Stopflox/flox – rtTA); TRE-

K-rastg/2; p53flox/flox; Z/DsRedtg/2 (indicated in the text as: Tg/Dox)

fed with doxycycline (Dox). Control mice are the pool of phenotypi-

cally indistinguishable mice with genotypes Rosa (Stopflox/1 – rtTA);

TRE-K-rastg/2; p53flox/1; Z/DsRedtg/2 or Rosa (Stopflox/flox – rtTA);

TRE-K-rastg/2; p53flox/flox; Z/Dsredtg/2 without Dox induction (indi-

cated in the text as: Control). All mice were intercrossed from single

transgenic or conditional knockout mice. After generating quadruple

homozygous mouse lines, they were kept homozygous for all alleles

and in-bred within the same colony. Further description of the Rosa

(Stopflox/flox – rtTA) line is found in the report by Belteki et al.

(2005), the TRE-K-ras line is further described by Fisher et al.

(2001), the Z/DsRed line is described by Vintersten et al. (2004) and

the p53f/f is further described by Jonkers et al. (2001).

GenotypingPCR was performed in 20 lL reaction mixtures containing standard

PCR buffer, 10 mM MgCl2, 200 lM dNTPs, 200 nM primers, 5U

Taq polymerase, and 100 ng genomic DNA isolated from mouse ear

biopsies. The primers for detection of the targeted Rosa26 allele con-

taining the STOP cassette were as follows: ROSA5 50-GAGTTCTC

TGCTGCCTCCTG-30; and RTTA3 50-AAGACCGCGAAGAGT

TTGTC-30. The reaction resulted in a 215 bp band. The wild-type

Rosa26 allele was detected by an amplicon of 322 bp using primers

ROSA5 and ROSA3: 50-CGAGGCGGATACAAGCAATA-30. For

detection of the K-RasG12D transgene the primers K-Ras fw 50-GGG

AATAAGTGTGATTTGCCT-30 and mp-1 rv: 50-GCCTGCGAC

GGCGGCATCTGC-30 with an amplicon of 300 bp. The primers

for detecting the p53 flox allele were as follows: fw: 50-

CACAAAAACAGGTTAAACCCAG-30 rv: 50-AGCACATAG GAG

GCAGAGAC-30 and generated an amplicon of 370 bp for the LoxP

allele and 288 bp for the wild type allele. Reactions were amplified

with the following PCR protocol: initial denaturation for 10 min at

94�C followed by 35 cycles of 1 min at 94�C, 45 s at 62�C

(Rosa26-knock-in allele) or 64�C (Rosa26-wild-type allele) 57�C

(TRE-K-Ras) and 60�C (p53f/f ) and 1 min at 72�C and a final

extension of 5 min at 72�C. PCR products were detected on 2%

agarose gel. The Z/DsRed mice were genotyped by X-gal staining.

In brief, ear punches were washed and fixed for 10 min in 0.2% glu-

taraldehyde and stained in X-gal staining solution (X-gal (in

DMSO, 40 mg/mL) 1mg/mL, K3Fe(CN)6 5 mM, K4Fe(CN)6 5

mM and MgCl2 2 mM in 13 PBS).

Dox InductionControl and/or transgenic mice were given Dox in food pellets (Bio-

serv, Frenchtown, NJ) at a concentration of 6 g/kg. Dox administra-

tion was started the day of Adeno-Cre injection. Mice were left on

Dox until a phenotype was observed. For MRI experiments, mice

were left on Dox for 3 weeks and after MRI confirming lesions, a

cohort of mice were placed back on regular food with no Dox.

Preparation and Administration of AdenovirusThe Ad5-Cre-IRES-GFP was constructed and propagated as

described in reference (Luo et al., 2007). In brief, the construct nls-

Cre was removed from PCAGGS (a kind gift from Dr. Andras

Nagy) with an XbaI, EcoRI digest, blunted and ligated to a BglII

site of a pAdenoVactor-CMV5-IRES-GFP (Q-BIOgene, Carlsbad,

CA) an adenoviral transfer plasmid. The pAdenoVector (containing

the gene of interest) and a pAdenoVector DE1/E3 (containing the

viral genome) were cotransfected into electrocompetent BJ5183 bac-

teria. Recombination between these two plasmids produces an ade-

novirus that can be efficiently packaged in 293 cells stably expressing

2 Volume 00, No. 00

the E1 region necessary for viral packaging. The recombined viral

vector is recovered after transfection, plaque purified and verified by

restriction digest. A viral stock was purified be cesium chloride band-

ing twice. The titer was 1 3 109 pfu/mL.

The Ad5-CMV-Cre-IRES-GFP adenovirus was injected stereo-

tactically into the SVZ or CTX of adult transgenics. Two microliters

of purified virus (2 3 106 pfu/mL) were injected unilaterally into the

SVZ or CTX using the following coordinates A, L, D: 0.5, 1.3, 1.2 to

target the SVZ and 1, 1.6, 0.6 to target the CTX. Injections were

administered with a 26G Hamilton syringe (Hamilton, Reno, NV)

Western BlotBrains were homogenized in lysis buffer for 20 min on ice prior to

centrifugation at 14,000 rpm for 10 min at 4�C. Equal amounts of

total protein as determined using a BCA protein assay kit (Pierce,

Rockford, IL) were separated on SDS-PAGE gel. After blocking in

5% dry milk in Tris-buffered saline, blots were incubated overnight

at 4�C with primary antibodies Ras (Chemicon 1:10000), phospho-

ERK (Cell signaling 1:2000), total ERK (Cell signaling 1:2000),

phospho-AKT (Cell Signaling 1:2500), total-AKT (Cell signaling

1:2000) prior to peroxidase-conjugated secondary antibody (Bio-

Rad, Hercules, CA) and enhanced chemiluminescent signal was

detected with Western Lightning Plus (PerkinElmer, Waltham, MA).

Tissue PreparationBrains were collected and fixed in 10% formalin prior to paraffin

embedding and sectioning. For frozen sections, brains were obtained

following intracardial perfusion with 0.9% NaCl and 4% PFA.

Brains were fixed overnight in 4% PFA at 4�C and then transferred

into 30% sucrose for 36 h before embedding in OCT medium for

snap freezing.

ImmunostainingFive-micron paraffin sections were deparaffinized and subjected to

heat induced antigen retrieval using sodium citrate pH 6.0. Sections

were then incubated in 5% serum blocking solution prior to a 1 h

incubation with primary antibodies at 37�C: GFAP (Dakocytoma-

tion 1:800), Nestin (Neuromics 1:200), Ki67 (Dakocytomation

1:25), and phospho-ERK (Cell Signaling 1:2000). Immunofluores-

cence analysis was performed on 16 lm cryosections, these were per-

meabilized with 0.1% Triton X-100 in PBS prior to blocking and

application of primary antibodies against GFAP (Dakocytomation

1:800), Nestin (Neuromics 1:200). Fluorescent detection was per-

formed with Alexa-labeled secondary antibodies (Molecular Probes,

Eugene, OR), whereas horseradish peroxidase-conjugated secondary

antibodies (Vector Laboratories, Burlington, ON, Canada) were

detected using a Vecta Stain Elite ABC development (Vector Labora-

tories, Burlington, ON, Canada).

MRI ImagingIndividual mice were subjected to a T2 weighted MRI assessment

for the detection of brain tumors 3 weeks post injection. In brief,

mice were anesthetized with 2% isoflurane and images were obtained

using a Bruker BioSpec 70/30 7 Tesla animal MRI/MRSI system

(Siemens, Mississauga, ON, Canada). After the MRI, mice were

divided into two cohorts where Dox was either continued or

removed. Tumor volume was calculated using MIPAV (NIH,

Bethesda, MD), by manually identifying the volume of interest on

multiple MRI slices. Once the volume of interest was defined statis-

tics of volume were calculated.

Statistical AnalysisKaplan-Meier curves for mouse glioma latency were made using

GraphPad Prism 4.0 software (GraphPad Inc., San Diego, CA) and

analyzed with standard log rank test. Each experiment was per-

formed with samples from at least three animals from independent

litters. Statistical significance (P < 0.05) was determined by

unpaired t-test using GraphPad Prism 4.0 software (GraphPad, San

Diego, CA).

Results

Adeno-cre Targeted Cortical and SVZ RegionsIn this study we used mice bearing a conditional allele (LoxP

sites) of p53, in combination with a conditional and induci-

ble K-rasG12D gene, as deregulation of p53 and RTK/RAS/

PI(3)K pathways have been shown to be an obligatory events

in most, and perhaps all, glioblastoma tumors (TCGA, 2008;

von Deimling et al., 1995). Although Ras mutations are rare

in GBM’s and high levels of Ras are mostly due to the aber-

rant expression and overactivity of membrane tyrosine kinase

receptors such as EGFR and PDGFR or through loss of Nf-

1, recent studies have reported Ras mutations in �1% of

human GBMs (TCGA, 2008), with K-Ras mutations

accounting for 0.4% of tumors presented in the TCGA pro-

visional data (Cerami et al., 2012; Gao et al., 2013). As a

result, our mouse model may represent the group of patients

where high levels of Ras are not due to Nf-1 loss or overex-

pression of receptor tyrosine kinases.

To induce site-specific recombination, adenovirus carry-

ing Cre-recombinase were stereotactically delivered to specific

regions of the adult mouse brain. Recombination resulted in

deletion of p53 and expression of the tetracycline transcrip-

tional activator, which, in the presence of Dox, induces

expression of active K-rasG12D (Fig. 1A).

To directly examine whether K-rasG12D alone or in

cooperation with p53 loss is sufficient to induce astrocytoma

formation we targeted Cre to specific regions in the adult

mouse brain: the CTX and SVZ (Fig. 1B). First, we deter-

mined the distribution of cells that underwent recombination

after targeting both the CTX and SVZ using the Cre reporter

transgene expressed by our mouse model; in this case cells

that undergo recombination are tagged with DsRed (Fig. 1C–

E). We confirmed expression of DsRed in the SVZ one week

post injection (Fig. 1Ci,ii). To confirm that SVZ progenitor

cells were among the cells that underwent recombination, we

looked for DsRed expression in the olfactory bulb (Fig.

1Di,ii), as it has been demonstrated that interneurons in the

Month 2013 3

Munoz et al.: Murine Model for the Development of Gliomas

FIGURE 1: Targeting the CTX and SVZ in the adult mouse brain. A: Schematic diagram of mouse crosses, Mice carrying a targeted inser-tion of the conditional rtTA transgene at the ROSA26 locus, were bred with transgenics carrying a Dox-inducible rtTA-dependent TetO-K-RasG12D line also carrying a Cre reporter (Z/Red) transgene. Homozygous and heterozygous double transgenics were then bred tomice carrying the floxed p53 allele. Stereotactic injections of Adeno-Cre leads to expression of the rtTA in addition to a p53 deletion,addition of Dox results in the formation of an active transactivator and expression of K-RasG12D. B: Schematic of SVZ-olfactory bulb neu-rogenesis. Top, coronal section of the adult mouse brain. The orange box indicates the SVZ and the blue box the CTX, regions thatwere targeted with adenovirus. Bottom, coronal sections of the olfactory bulb (OB) indicating the region where newly born neuroblastsinitially arrive from the SVZ. The purple box indicates the granular cell layer (GCL) and the pink box the glomerular layer (GL). CTX: Cor-tex, SVZ: subventricular Zone. C: (i) Immunofluorescence of coronal sections through the adult SVZ, one week post injection (Orangebox) and (ii) High magnification image of the area surrounded by the hatched white box in (Ci). Red cells (Dsred) represent Cre-recombination. Sections are counterstained with DAPI (blue). CC: corpus callosum, SVZ: Subventricular. Scale bars represent 100 lm (Ci)and 38 lm (Cii). D: Immunofluorescence of coronal section through the olfactory bulb 1week post injection. (i) Granule cell layer (purplebox); (ii) glomerular layer (pink box). Red cells (DsRed) represent progeny from Cre-recombined progenitors induced by adenovirus injec-tion. Sections are counterstained with DAPI (blue). GCL: Granular cell layer GL: Glomerular layer. Scale bars represent 100 lm (Di) and38 lm (Dii). E: Immunofluorescence of coronal section through the adult CTX. (i) The expression of DsRed demonstrates cortical target-ing. (ii) High magnification image of the area surrounded by the hatched white box in (Ei). Red cells (DsRed) represent Cre-recombination. Sections are counterstained with DAPI (blue) CC: corpus callosum, CTX: CTX. Scale bars 100 lm (Ei) and 38lm (Eii). F:Ipsilateral SVZ showing no recombination. Sections are counterstained with DAPI (blue). CC: corpus callosum, SVZ: Subventricular zone.Scale bars represent 50 lm (Di, Diii) and 21 lm (Diii).

granule cell layer and glomerular layer in the adult mouse

olfactory bulb derive from progenitor populations in the SVZ

(Doetsch and Alvarez-Buylla, 1996; Doetsch et al., 1999).

Similarly, when the cortical region was targeted, expression of

DsRed was restricted to this area (Fig. 1Ei,ii) and there was

no expression of DsRed in the ipsilateral SVZ (Fig. 1F).

Cooperation Between Hyperactive K-ras and p53Loss is Necessary for the Induction of Brain TumorsIndependent of the Region of OriginTo determine whether K-rasG12D alone or in combination

with loss of p53 could lead to tumor formation in different

region of the adult mouse brain, we induced these genetic

alterations in the SVZ and CTX by stereotactic injection of

Adeno-Cre into mice carrying activated K-ras alone or in

combination with heterozygous or homozygous loss of p53.

Mice expressing K-rasG12D alone or in combination with

heterozygous loss of p53 did not develop tumors. Only mice

expressing K-rasG12D and a homozygous loss of p53 developed

malignant gliomas (Table 1). This occurred regardless of

whether the SVZ or the CTX was targeted. Targeting of either

the SVZ or cortical region in homozygous transgenics led to

tumor formation with equal incidence (Table 1). Tumors aris-

ing from both regions had severe nuclear pleomorphism and

marked mitotic activity (Fig. 2A), with a large number of Ki67

positive cells indicating robust proliferation (Fig. 2B) and were

immunoreactive for glial and progenitor markers, GFAP and

Nestin, markers for glial tumors (Fig. 2B) (Kleihues et al.,

1995). Consistent with the activation of the Ras pathway, some

tumor regions showed robust p-ERK expression (Fig. 2B).

Despite these histopathological similarities, a significant differ-

ence in overall survival was observed. Mice where the SVZ was

targeted had a significantly shorter median survival of 43 days

compared with 83 days for mice whose tumors originated from

the CTX (log rank, P < 0.0001; Fig. 2C).

Taking advantage of the Cre reporter transgene in our

model, we co-stained recombinant cells (DsRed positive) with

antibodies against GFAP and Nestin two weeks post injection

to characterize the cell types being targeted. Interestingly,

these experiments demonstrated that targeted cells in the

CTX represent one population characterized by the expression

of GFAP. Conversely, cells targeted in the SVZ represent three

populations of cells, those that express GFAP or Nestin alone

and those that are double-labeled with both markers (Fig. 3).

These results demonstrate that somatic expression of

K-rasG12D combined with p53 loss in the SVZ or CTX is

sufficient to induce tumors with similar incidence and histo-

logical characteristics despite differences in the marker profiles

of the initial targeted cells.

In Early Stages of Gliomagenesis, Tumors AreDependent on K-ras Expression Despite Loss ofp53We next asked whether the tumors depended on continuous

expression of K-rasG12D for continuous growth or survival.

To examine this dependence, we observed a cohort of mice

for prolonged periods after Dox withdrawal. MRI imaging

was used to confirm tumor formation before Dox removal;

mice were imaged 3 weeks post Adeno-Cre injection and

Dox induction. Mice were then divided into two cohorts,

those where Dox was removed from their diet and those that

continued on Dox as survival controls. Mice were reimaged 5

weeks post injection to determine changes in tumor burden

(Fig. 4A).

TABLE 1: K-rasG12D and p53 Loss Cooperate to Cause Brain Tumors Regardless of the Region of Origin

Genotype Location (numberof mice injected)

Number of mice withtumors (incidence)

rtTAtg/2; TRE-K-rasG12D 1Dox Svz (10) 0 (0%)

CTX (10) 0 (0%)

rtTAtg/tg; TRE-K-rasG12D 1Dox SVZ (10) 0 (0%)

CTX (10) 0 (0%)

rtTAtg/2; TRE-K-rasG12D; p53f/wt 1Dox SVZ (10) 0 (0%)

CTX (10) 0 (0%)

rtTAtg/tg; TRE-K-rasG12D; p53f/f 1Dox SVZ (27) 24 (89%)

CTX 14) 12 (86%)

rtTAtg/tg; TRE-K-rasG12D; p53f/f –Dox SVZ(8) 0 (0%)

CTX (6) 0 (0%)

Munoz et al.: Murine Model for the Development of Gliomas

Month 2013 5

After Dox withdrawal tumors completely regressed

within 2 weeks. This was true for tumors generated either in

the SVZ or CTX. Control littermates left on Dox all suc-

cumbed to their tumors within 6–12 weeks (Fig. 4A,B).

Deinduction of K-rasG12D during early stages of gliomagene-

sis leads to sustained remission, as mice taken off Dox

FIGURE 2: Expression of K-RasG12D in cooperation with p53 loss induces High-Grade astrocytoma formation. A: Representative Hema-toxylin and Eosin (H&E) stained brain sections from transgenic and control littermates. Brain tumors develop in the CTX and SVZ of adultmice expressing K-rasG12D and loss of p53. Tumors arising from both regions had severe nuclear pleomorphism and marked mitoticactivity (arrow). Scale bar 1000 lm (Top), 100 lm (middle), and 25 lm (bottom). B: Tumors expressed traditional markers of astrocyto-mas including GFAP, p-ERK, and Nestin, as well as Ki67 as a measure of proliferation. Scale bar 100 lm. C: Kaplan Meier survival curveof mice injected with Cre recombinase in the SVZ and CTX. Mice with activated K-ras and loss of p53 in the SVZ have a shorter mediansurvival of 43 days compared to those arising in the CTX, with a median survival of 83 days (*log rank, P < 0.0001). Dox induction wasstarted at the time of injection and mice without evidence of tumor were sacrificed 125 days postinjection.

6 Volume 00, No. 00

remained asymptomatic 6 months after Adeno-cre injection

(Fig. 4C).

To confirm that Dox withdrawal led to a decrease in

transgene expression, a group of mice selected from the

cohort where Dox was removed were sacrificed 2 weeks after

Dox withdrawal. A pull-down-based Ras activation assay and

examination of downstream effectors was carried out on fore-

brain lysates from these mice. Dox withdrawal led to a

decline in RasGTP levels within 2 weeks after de-induction,

accompanied by decreases in p-ERK and p-AKT (Fig. 5A).

Additionally, histological examination of dox-withdrawn mice

showed restoration of nearly normal architecture, with a

reduction in GFAP, Nestin, pERK, and Ki67 stains, com-

pared with mice that were left on Dox and sacrificed 5 weeks

post injection (Fig. 5B). Collectively these results indicate

that expression of K-rasG12D is essential for tumor viability

even in the absence of p53. These results have valuable impli-

cations for early treatment of Ras driven gliomas such as low-

grade gliomas, as early treatment to block Ras activity could

lead to tumor remission.

Discussion

Several lines of evidence suggest that neurogenic regions in

the adult brain are more susceptible to malignant transforma-

tion due to their resident NSC population (Sanai et al.,

2005). This has been validated by mouse models where

GFAP and Nestin-Cre mediated inactivation of the tumor

suppressors Nf1 and p53 in the adult brain induced glial pro-

genitor proliferation and ultimately malignant astrocytomas

(Alcantara Llaguno et al., 2009; Zhu et al., 2005). However,

conflicting evidence suggests that SVZ-like cells are not

uniquely gliomagenic, but this capacity may be shared by

cycling progenitors distributed throughout the CTX and

white matter (Gensert and Goldman, 2001; Liu et al., 2011).

This has been corroborated by MRI studies demonstrating

that a majority of GBMs have contrast-enhancing portions

that have no contact with the SVZ (Bohman et al., 2010b;

Lim et al., 2007b).

In light of these conflicting viewpoints we developed a

mouse model, which integrates Cre-Lox mediated and Tet-

regulated expression of K-rasG12D. The Cre-Lox component

of our mouse model allowed us to target specific regions in

the adult mouse brain, to determine whether the cell popula-

tions residing in the CTX and SVZ differ in their tumori-

genic potential and to what extent the phenotype and

behaviour of the tumor is determined by the region of origin.

The Tet regulated component was then used to study whether

tumor viability was dependent on the continued expression of

K-rasG12D despite the loss of p53.

Similar to previous mouse models driven by activated

Ras, it by itself was not sufficient to induce transformation

and cooperating mutations such Akt or the loss on Ink4a/Arf

FIGURE 3: Cells targeted in the SVZ and cortical regions have a different marker profile. Double immunofluorescence analysis ofrecombinant cells (DsRed) in the SVZ and CTX 2 weeks post injection, show that cells targeted in the CTX express GFAP but not Nestin(arrow head). However, cells targeted in the SVZ have three distinct marker profiles, expressing GFAP (arrow head), Nestin (Arrow), orboth (asterisk). Scale bar represents 20 lm.

Munoz et al.: Murine Model for the Development of Gliomas

Month 2013 7

were necessary for the development of highly infiltrative brain

tumors (Holland et al., 2000; Uhrbom et al., 2002). Despite

previous reports demonstrating that the SVZ was the site of

origin for brain tumors induced by loss of Nf1, p53 and Pten

(Alcantara Llaguno et al., 2009) we show that targeting

K-rasG12D expression and p53 loss to either the CTX or SVZ

leads to glioma formation with a similar penetrance. The dis-

crepancies between these two models could be attributed to

the fact that Nf1 deficiency is not functionally equivalent to

oncogenic Ras, in addition to the fact that astrocytes in the

CNS exhibit molecular heterogeneity (Hegedus et al., 2007;

Yeh et al., 2009). It is possible that these two alterations are

not redundant but they rather lead to the activation of dis-

tinct effector genes, leading to a distinct phenotype as has

been shown for some haematological malignancies (Cutts

et al., 2009).

We then compared tumors that arose from the CTX or

SVZ in terms of their phenotypic characteristics and effect on

survival. We saw no difference in tumor incidence, and

despite necrosis being observed in some of the mice where

the SVZ was targeted, tumors arising from both regions were

otherwise pathologically similar. The only difference between

cortical and SVZ tumors was in overall survival, with mice

where the SVZ was targeted having a significantly shorter

median survival of 43 days compared with 83 days for mice

whose tumors originated from the CTX. Although not tested,

the differences in survival might be attributed to a disparity

in the number of proliferative cells found in the CTX and

SVZ. The SVZ is the largest germinal zone in the adult brain

whereas very few proliferating cells are present in the CTX.

Therefore, the number of cells with tumorigenic potentials

would be greatly different.

FIGURE 4: Dox withdrawal induces tumor regression and sustained remission. A: Representative MRI images of mouse cohorts main-tained or withdrawn from Dox. Images were taken 3 and 5 weeks after intracranial injections to monitor tumor burden. B: Tumor growthcurves show a decrease in tumor volume both in the SVZ and CTX upon Dox removal (*log rank, P < 0.001). C: aplan Meier survivalcurve of mice used in Dox withdrawal experiments. Complete tumor remission was observed in mice taken off Dox 5 weeks after tar-geted transgene expression. Red arrow shows the start of Dox induction at the day of injection. Green arrow shows the time of Doxremoval 3 weeks post injection (log rank, P < 0.001).

8 Volume 00, No. 00

To determine if there were differences in the cell popu-

lations targeted in the CTX and SVZ, we costained recombi-

nant cells expressing DsRed with nestin and GFAP. As

expected, cells targeted in cortical regions had a more differ-

entiated profile, expressing GFAP but not Nestin, whereas

cells targeted in the SVZ had a more progenitor like marker

expression. This is similar to recently published data high-

lighting the dedifferentiation of astrocytes in the CTX by spe-

cific oncogenes (Friedmann-Morvinski et al., 2012). Although

we can not rule out the involvement of other progenitor pop-

ulations like Ng21 cells in tumor formation, our results

highlight the fact that GBMs might arise from regions other

that the SVZ. This is of relevance to human patients as MRI

studies have also shown that a majority of GBMs have no

contact with the SVZ and are widely distributed through the

CTX and white matter (Bohman et al., 2010a; Lim et al.,

2007a).

We then went on to ask whether once established,

tumor cell growth and viability continued to depend on the

presence of both genetic alterations as this has important

therapeutic implications. To answer this, we coupled our

inducible model with MRI imaging to look at tumor growth

after Dox withdrawal to mimic Ras inhibition. In these

experiments we showed that K-ras de-induction, despite con-

tinued loss of p53, led to tumor regression and sustained

remission through down-regulation of downstream signaling

pathways. Our results suggest that tumor regression might

still be achieved in response to inactivation of an initiating

oncogene during early stages of tumor development when

only a few oncogenic mutations are present. One of the

FIGURE 5: Dox withdrawal induces down regulation of RasGTP levels as well as downstream effectors. A: Immunoblotting from totalforebrain lysates of transgenic mice after the SVZ (left) or CTX (right) were targeted. The first lane shows the expression profile of micemaintained on Dox and the second lane, mice 2 weeks after Dox withdrawal. B: Histologic characterization of brains from transgenicmice continued on Dox and sacrificed 5 weeks post induction (top panels) and mice where Dox was removed 3 weeks after induction(bottom panels) and sacrificed after 2 months. No tumors were found on microscopic examination of mice where Dox was removed.Scale bar represents 100 lm.

Munoz et al.: Murine Model for the Development of Gliomas

Month 2013 9

reasons tumors respond to the inactivation of a single onco-

gene might be that certain oncogenic events are uniquely

important for tumor viability during the early process of

tumorigenesis, as has been shown in melanoma and lung can-

cer (Chin et al., 1999; Fisher et al., 2001). Alternatively, the

direct participation of one cooperating oncogene, such as

K-ras, as a proliferative and survival signal could explain why

its loss alone induces cell death and remission. In summary,

our studies support the idea that although several genetic

alterations may be necessary to induce oncogenic transforma-

tion (Hanahan and Weinberg, 2011), inactivation of a single

oncogene may undermine tumor maintenance, leading to

tumor regression and disease remission.

Despite our results showing that de-induction of K-ras

in our model leads to tumor regression, treatments targeted

to Ras activation such as Farnesyl transferase inhibitors have

had disappointing results in a clinical scenario (Feldkamp

et al., 1999; Kohl et al., 1995). One reason for the lack of

efficacy might be related to the patients selected for these

studies as they all had recurrent gliomas (Lobell et al., 2001).

It is possible that tumors dependent on Ras signaling have a

therapeutic window in which they will be responsive to treat-

ment. Due to the genetic instability associated with tumor

progression (Cahill et al., 1998; Lengauer et al., 1998), it is

likely that late stage tumors are no longer dependent on Ras

signaling alone. This suggests that single agent chemothera-

peutic intervention may be more effective at earlier stages of

tumorigenesis, for example for grade 2 astrocytomas.

In conclusion, we have demonstrated that the interplay

between specific combinations of genetic mutations and sus-

ceptible cell types, regardless of the region of origin, is an

important determinant of gliomagenesis. Additionally, our

model supports the view that although several genetic events

may be necessary to confer the traits associated with onco-

genic transformation, inactivation of a single oncogenic part-

ner may undermine tumor maintenance, leading to tumor

regression and disease remission if treated during the early

stages of tumor development.

Acknowledgment

This work was funded by a grant from The Canadian Insti-

tutes of Health Research (MOP 102513).

The authors would like to dedicate this work in memory

of Dr. Abhijit Guha who passed away during the final prepa-

ration of the manuscript. He was the leader and primary pro-

genitor of the work described. They thank Dr. Andras Nagy

for providing the ROSA26-rtTA-IRES-EGFP mice, Dr. Tak

Mak for providing p53 flox mice, and Dr. Harold E Varmus

for providing the TRE-K-ras mice. They thank Dr Aaron

Gajadhar for critical comments on the manuscript.

References

Alcantara Llaguno S, Chen J, Kwon C, Jackson E, Li Y, Burns D, AlvarezBuylla A, Parada L. 2009. Malignant Astrocytomas Originate from NeuralStem/Progenitor Cells in a Somatic Tumor Suppressor Mouse Model. CancerCell 15:45–56.

Assanah M, Lochhead R, Ogden A, Bruce J, Goldman J, Canoll P. 2006. Glialprogenitors in adult white matter are driven to form malignant gliomas byplatelet-derived growth factor-expressing retroviruses. J Neurosci 26:6781–6790.

Belteki G, Haigh J, Kabacs N, Haigh K, Sison K, Costantini F, Whitsett J,Quaggin SE, Nagy A. 2005. Conditional and inducible transgene expressionin mice through the combinatorial use of Cre-mediated recombination andtetracycline induction. Nucleic Acids Res 33:e51.

Bohman L-E, Swanson KR, Moore JL, Rockne R, Mandigo C, Hankinson T,Assanah M, Canoll P, Bruce JN. 2010a. Magnetic resonance imaging charac-teristics of glioblastoma multiforme: Implications for understanding gliomaontogeny. Neurosurgery 67:1319–1327; discussion 1327-1328.

Cahill DP, Lengauer C, Yu J, Riggins GJ, Willson JK, Markowitz SD, KinzlerKW, Vogelstein B. 1998. Mutations of mitotic checkpoint genes in humancancers. Nature 392:300–303.

Cerami E, Gao J, Dogrusoz U, Gross BE, Sumer SO, Aksoy BA, Jacobsen A,Byrne CJ, Heuer ML, Larsson E, Antipin Y, Reva B, Goldberg AP, Sander C,Schultz N. 2012. The cBio cancer genomics portal: An open platform forexploring multidimensional cancer genomics data. Cancer Discov 2:401–404.

Chin L, Tam A, Pomerantz J, Wong M, Holash J, Bardeesy N, Shen Q,O’Hagan R, Pantginis J, Zhou H, Horner, JW, 2nd Cordon-Cardo, C. Yanco-poulos, G. 1999. Essential role for oncogenic Ras in tumour maintenance.Nature 400:468–472.

Cutts BA, Sjogren AK, Andersson KM, Wahlstrom AM, Karlsson C, Swolin B,Bergo MO. 2009. Nf1 deficiency cooperates with oncogenic K-RAS to induceacute myeloid leukemia in mice. Blood 114:3629–3632.

Doetsch F, Alvarez-Buylla A. 1996. Network of tangential pathways for neuro-nal migration in adult mammalian brain. Proc Natl Acad Sci USA 93:14895–14900.

Doetsch F, Caill�e I, Lim DA, Garc�ıa-Verdugo JM, Alvarez-Buylla A. 1999. Sub-ventricular zone astrocytes are neural stem cells in the adult mammalianbrain. Cell 97:703–716.

Eriksson PS, Perfilieva E, Bjork-Eriksson T, Alborn AM, Nordborg C, PetersonDA, Gage FH. 1998. Neurogenesis in the adult human hippocampus. NatMed 4:1313–1317.

Feldkamp MM, Lau N, Guha A. 1999. The farnesyltransferase inhibitor L-744,832 inhibits the growth of astrocytomas through a combination of anti-proliferative, antiangiogenic, and proapoptotic activities. Ann N Y Acad Sci886:257–260.

Fisher GH, Wellen SL, Klimstra D, Lenczowski JM, Tichelaar JW, Lizak MJ,Whitsett JA, Koretsky A, Varmus HE. 2001. Induction and apoptotic regres-sion of lung adenocarcinomas by regulation of a K-Ras transgene in the pres-ence and absence of tumor suppressor genes. Genes Dev 15:3249–3262.

Friedmann-Morvinski D, Bushong EA, Ke E, Soda Y, Marumoto T, Singer O,Ellisman MH, Verma IM. Dedifferentiation of neurons and astrocytes by onco-genes can induce gliomas in mice. Science 338 (6110):1080–1084.

Gage FH. 2000. Mammalian neural stem cells. Science 287:1433–1438.

Gao J, Aksoy BA, Dogrusoz U, Dresdner G, Gross B, Sumer SO, Sun Y,Jacobsen A, Sinha R, Larsson E, et al. 2013. Integrative analysis of complexcancer genomics and clinical profiles using the cBioPortal. Sci Signal 6:pl1.

Gensert JM, Goldman JE. 2001. Heterogeneity of cycling glial progenitors inthe adult mammalian cortex and white matter. J Neurobiol 48:75–86.

10 Volume 00, No. 00

Hambardzumyan D, Cheng YK, Haeno H, Holland EC, Michor F. 2011. Theprobable cell of origin of NF1- and PDGF-driven glioblastomas. PLoS ONE 6:e24454.

Hanahan D, Weinberg Robert A. 2011. Hallmarks of Cancer: The Next Gener-ation. Cell 144:646–674.

Hegedus B, Dasgupta B, Shin JE, Emnett RJ, Hart-Mahon EK, Elghazi L,Bernal-Mizrachi E, Gutmann DH. 2007. Neurofibromatosis-1 regulates neuro-nal and glial cell differentiation from neuroglial progenitors in vivo by bothcAMP- and Ras-dependent mechanisms. Cell Stem Cell 1:443–457.

Holland EC, Celestino J, Dai C, Schaefer L, Sawaya RE, Fuller GN. 2000.Combined activation of Ras and Akt in neural progenitors induces glioblas-toma formation in mice. Nat Genet 25:55–57.

Jonkers J, Meuwissen R, van der Gulden H, Peterse H, van der Valk M, BernsA. 2001. Synergistic tumor suppressor activity of BRCA2 and p53 in a condi-tional mouse model for breast cancer. Nat Genet 29:418–425.

Kleihues P, Soylemezoglu F, Schauble B, Scheithauer BW, Burger PC. 1995.Histopathology, classification, and grading of gliomas. Glia 15:211–221.

Kohl NE, Wilson FR, Thomas TJ, Bock RL, Mosser SD, Oliff A, Gibbs JB.1995. Inhibition of Ras function in vitro and in vivo using inhibitors offarnesyl-protein transferase. Methods Enzymol 255:378–386.

Lengauer C, Kinzler KW, Vogelstein B. 1998. Genetic instabilities in humancancers. Nature 396:643–649.

Lim DA, Cha S, Mayo MC, Chen M-H, Keles E, VandenBerg S, Berger MS.2007a. Relationship of glioblastoma multiforme to neural stem cell regionspredicts invasive and multifocal tumor phenotype. Neuro Oncol 9:424–429.

Lim LP, Tay FB, Sum CF, Thai AC. 2007b. Relationship between markers ofmetabolic control and inflammation on severity of periodontal disease inpatients with diabetes mellitus. J Clin Periodontol 34:118–123.

Liu C, Sage Jonathan C, Miller Michael R, Verhaak Roel GW, Hippenmeyer S,Vogel H, Foreman O, Bronson Roderick T, Nishiyama A, Luo L, Zong, Hui.2011. Mosaic analysis with double markers reveals tumor cell of origin in gli-oma. Cell 146:209–221.

Lobell RB, Omer CA, Abrams MT, Bhimnathwala HG, Brucker MJ, Buser CA,Davide JP, deSolms SJ, Dinsmore CJ, Ellis-Hutchings MS, Kral AM, Liu D,Lumma WC, Machotka SV, Rands E, Williams TM, Graham SL, Hartman GD,Oliff AI, Heimbrook DC, Kohl NE. 2001. Evaluation of farnesyl:protein trans-ferase and geranylgeranyl: Protein transferase inhibitor combinations in pre-clinical models. Cancer Res 61:8758–8768.

Luo J, Deng ZL, Luo X, Tang N, Song WX, Chen J, Sharff KA, Luu HH,Haydon RC, Kinzler KW, Vogelstein B, He TC. 2007. A protocol for rapidgeneration of recombinant adenoviruses using the AdEasy system. NatureProtoc 2:1236–1247.

Sanai N, Alvarez-Buylla A, Berger MS. 2005. Neural stem cells and the originof gliomas. N Engl J Med 353:811–822.

Sanai N, Tramontin AD, Quinones-Hinojosa A, Barbaro NM, Gupta N,Kunwar S, Lawton MT, McDermott MW, Parsa AT, Manuel-Garcia Verdugo J,et al. 2004. Unique astrocyte ribbon in adult human brain contains neuralstem cells but lacks chain migration. Nature 427:740–744.

Singh SK, Clarke ID, Terasaki M, Bonn VE, Hawkins C, Squire J, Dirks PB.2003. Identification of a cancer stem cell in human brain tumors. Cancer Res63:5821–5828.

Singh SK, Hawkins C, Clarke ID, Squire JA, Bayani J, Hide T, Henkelman RM,Cusimano MD, Dirks PB. 2004. Identification of human brain tumour initiatingcells. Nature 432:396–401.

TCGA. 2008. Comprehensive genomic characterization defines human glio-blastoma genes and core pathways. Nature 455:1061–1068.

Uhrbom L, Dai C, Celestino JC, Rosenblum MK, Fuller GN, Holland EC.2002. Ink4a-Arf loss cooperates with KRas activation in astrocytes and neuralprogenitors to generate glioblastomas of various morphologies dependingon activated Akt. Cancer Research 62:5551–5558.

Venere M, Fine HA, Dirks PB, Rich JN. 2011. Cancer stem cells in gliomas:Identifying and understanding the apex cell in cancer’s hierarchy. Glia 59:1148–1154.

Vintersten K, Monetti C, Gertsenstein M, Zhang P, Laszlo L, Biechele S, NagyA. 2004. Mouse in red: red fluorescent protein expression in mouse ES cells,embryos, and adult animals. Genesis 40:241–246.

von Deimling A, Louis DN, Wiestler OD. 1995. Molecular pathways in the for-mation of gliomas. Glia 15:328–338.

Yeh TH, Lee da Y, Gianino SM, Gutmann DH. 2009. Microarray analysesreveal regional astrocyte heterogeneity with implications for neurofibromato-sis type 1 (NF1)-regulated glial proliferation. Glia 57:1239–1249.

Zhu Y, Guignard F, Zhao D, Liu L, Burns DK, Mason RP, Messing A, ParadaLF. 2005. Early inactivation of p53 tumor suppressor gene cooperating withNF1 loss induces malignant astrocytoma. Cancer Cell 8:119–130.

Munoz et al.: Murine Model for the Development of Gliomas

Month 2013 11