identiﬁcation of foxr2 as an oncogene in medulloblastomatumor and stem cell biology...

Tumor and Stem Cell Biology

Identification of FoxR2 as an Oncogene in Medulloblastoma

Hideto Koso1,2, Asano Tsuhako1, Eli Lyons1, Jerrold M. Ward2, Alistair G. Rust3, David J. Adams3,Nancy A. Jenkins2,4, Neal G. Copeland2,4, and Sumiko Watanabe1

AbstractMedulloblastoma is the most common pediatric brain tumor, and in �25% of cases, it is driven by aberrant

activation of the Sonic Hedgehog (SHH) pathway in granule neuron precursor (GNP) cells. In this study, weidentified novel medulloblastoma driver genes through a transposonmutagenesis screen in the developing brainof wild-type and Trp53 mutant mice. Twenty-six candidates were identified along with established driver genessuch as Gli1 and Crebbp. The transcription factor FoxR2, the most frequent gene identified in the screen,is overexpressed in a small subset of human medulloblastoma of the SHH subtype. Tgif2 and Alx4, 2 newputative oncogenes identified in the screen, are strongly expressed in the SHH subtype of human medulloblas-toma. Mutations in these two genes were mutually exclusive with mutations in Gli1 and tended to cooccur,consistent with involvement in the SHH pathway. Notably, Foxr2, Tgif2, and Alx4 activated Gli-binding sites incooperation with Gli1, strengthening evidence that they function in SHH signaling. In support of an oncogenicfunction, Foxr2 overexpression transformed NIH3T3 cells and promoted proliferation of GNPs, the latter ofwhich was also observed for Tgif2 and Alx4. These findings offer forward genetic and functional evidenceassociating Foxr2, Tgif2, and Alx4 with SHH subtype medulloblastoma. Cancer Res; 74(8); 2351–61. �2014 AACR.

IntroductionMedulloblastoma is the most common malignant brain

tumor of childhood and tends to metastasize throughout thecentral nervous system. Despite overall improvements in sur-vival with multimodal treatment, a substantial proportion ofpatients are still incurable. Moreover, survivors often sufferconsiderable treatment-related morbidities, such as neurocog-nitive deficits related to radiation therapy. To develop bettertherapies, insights into the molecular mechanisms leading tothe development of medulloblastomas are essential. Recentmolecular studies have shown that medulloblastomas arecomposed of at least four distinct subtypes: Sonic Hedgehog(SHH), WNT, Group 3, and Group 4 (1). Tumors characterizedby aberrant activation of the SHH pathway (SHH subtype)originate from cerebellar granule neuron precursor (GNP),whereas tumors with activating mutations in theWNT effectorCTNNB1 (WNT subtype) arise outside the cerebellum from cells

of the dorsal brainstem (2). Althoughmutations in the SHH andWNT pathway seemmutually exclusive, these alterations affectonly 30% to 40% of medulloblastomas (3), suggesting that otherpathways are also operative in medulloblastoma formation. Tobetter understand the pathogenesis of medulloblastomas, acomprehensive understanding of genes and pathways thatregulate tumor formation in the cerebellum is essential.

Genome-wide insertional mutagenesis is an unbiased andhigh-throughput method to profile the landscape of driver genesinamousemodel system(4). TheSleepingBeauty (SB) transposonsystemhas expanded the applicability of insertionalmutagenesisfor the studyof various types of solid tumors (5). To identify genesand pathways that can transform embryonic neural stem andprogenitor cells to brain tumor-initiating cells, we applied SBtransposon mutagenesis to the developing mouse brain. Aprevious SB mutagenesis study screened medulloblastoma can-didate genes by analyzing recurrent insertions in medulloblas-tomas that developed on Ptch1 heterozygous or Trp53 (þ/� and�/�) mutant genetic backgrounds (6) and identified genes thatpromote metastatic dissemination of medulloblastomas (7). Incontrast, we used wild-type or Trp53mutant (R172H/þ) geneticbackgrounds that do not develop medulloblastomas spontane-ously. Transposition of the mutagenic SB transposons inducedmedulloblastomas in the cerebellum even on the wild-typebackground. Here, we describe the identification and functionalanalyses of candidate genes that may regulate the initiationand/or progression of medulloblastoma.

Materials and MethodsMouse strains

Mouse strains used in this study are described previously (8).All manipulations were performed with institutional animal

Authors' Affiliations: 1Division of Molecular and Developmental Biology,The Institute of Medical Science, The University of Tokyo, Tokyo, Japan;2Division of Genetics and Genomics, Institute of Molecular and Cell Biol-ogy, Agency for Science, Technology and Research, Singapore, Singa-pore; 3Experimental Cancer Genetics, Wellcome Trust Sanger Institute,Hinxton, Cambridge, United Kingdom; and 4Cancer Research Program,The Methodist Hospital Research Institute, Houston, Texas

Note: Supplementary data for this article are available at Cancer ResearchOnline (http://cancerres.aacrjournals.org/).

Corresponding Authors: SumikoWatanabe, Institute of Medical Science,University of Tokyo, 4-6-1 Shirokanedai, Minato-ku, Tokyo 108-8639,Japan. Phone: 81-3-5449-5663; Fax: 81-3-5449-5474; E-mail:[email protected]; and Neal G. Copeland, [email protected]

doi: 10.1158/0008-5472.CAN-13-1523

�2014 American Association for Cancer Research.

CancerResearch

www.aacrjournals.org 2351

on August 4, 2021. © 2014 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst March 5, 2014; DOI: 10.1158/0008-5472.CAN-13-1523

http://cancerres.aacrjournals.org/

care and use committee approval (Institute of Molecular andCell Biology).

Cell cultureMouse embryonic fibroblasts (MEF) were generated from

ICR mouse embryos at embryonic day 13.5. Primary cerebellarcultures were prepared from the cerebellum of ICR mice atpostnatal day 5 as described (9). Briefly, the cerebellum wastreated with 0.25% trypsin with 0.1% DNaseI (Invitrogen), andthen dissociated by trituration. The dissociated cells werecultured in DMEM/F12 medium supplemented with 10% FBSand penicillin/streptomycin. MEFs, cerebellar cells, andNIH3T3 cells (American Type Culture Collection, ATCC) wereinfected with retrovirus and selected for puromycin resistance.For cotransduction experiments, doubly infected cells wereselected with puromycin and G418. Foci formation assay andcell-proliferation assay in low serummedia were performed asdescribed (10). Cerebellar cells were plated to 6-well plates forcolony formation assay (more than 50 colonies were countedfrom three samples), or to 8-well chamber slide for immunos-taining (more than 100 Math1-positive cells were counted inthree samples).

Immunohistochemical analysisImmunohistochemistry was performed as described (8).

Each tumor was obtained from a different mouse. Anti-Math1antibody is a generous gift from Dr. J. Johnson (University ofTexas Southwestern Medical Center, Dallas TX; ref. 11). Thedetection of senescence-associated b-galactosidase activitywas performed as described (12).

Production of retrovirus stockRetroviral expression vectors (pMXs-IRES-Puro and pMXs-

IRES-Neo) were obtained from Cell Biolabs, Inc. A plasmidcoding for full-length Foxr2 was obtained from OriGene. AtruncatedGli1was excised from full-lengthGli1 (OriGene) withKpnI (located in exon 5) andXhoI to remove theN-terminal 125amino acids required for interaction with Sufu (13). A trun-cated form of Alx4 was PCR-amplified using primers in Sup-plementary Table S1. A cDNA insert for Tgif2was excised frompMX-Tgif2-IRES-EGFP (14). Retroviral production was per-formed as described (8).

Transfection assayHEK-293T cells (ATCC) were plated into 24-well plates and

transfection was carried out according to the manufacturer'sprotocol (GeneJuice; Millipore). Cells were harvested 2 daysafter transfection. Luciferase assay were performed accordingto themanufacturer's protocol (Promega). Luciferase activitieswere normalized by total protein amounts.

Identification of common insertion sites and the analysisof fusion transcripts

Common insertion sites (CIS) were identified as described(8). To analyze fusion transcripts, cDNA synthesis wasperformed using RevertraAce (Toyobo) and subsequent PCRwas performed using the primers listed in SupplementaryTable S1. PCR products were TA-cloned into pGEM-Teasy

(Promega) and sequenced. Real-time PCR analysis was car-ried out on Light Cycler 1.5 (Roche) using primers inSupplementary Table S1.

ResultsSleepingBeautymutagenesis promotesmedulloblastomaformation in the cerebellum

To identify novel driver genes for medulloblastoma, wemobilized SB in the brains of wild-type and Trp53R172Hmutantmice. Trp53R172H is an inducible lox-stop-lox allele with bothdominant-negative and gain-of-function properties (15). Brief-ly, Nestin-cre (Nes-Cre) transgenic mice (16) were crossed withTrp53R172H mutant mice (15) to generate compound hetero-zygous Nes-cre/þ; Trp53R172H/þ mice. These mice were thencrossed to mice homozygous for a transposon concatemercarrying up to 350 copies of the mutagenic transposon T2/Onc2 (17) and an inducible LSL-SB transposase (SBase) tar-geted to theRosa26 locus (18) to generate triple (Nes-cre/þ; T2/Onc2/þ; SBase/þ) and quadruple (Trp53R172H/þ; Nes-cre/þ;T2/Onc2/þ; SBase/þ) mice. Trp53R172H was used as a sensi-tizing mutation because mutant Trp53 is one of the key driversof medulloblastoma (19), whereas Nes-cre was used to activateTrp53R172H and SB transposition in neural stem and progenitorcells. Triple transgenic mice showed enlargement of the headat approximately 2 to 4 months of age and a gait disorder,whereas necropsy of these mice revealed tumors within thecerebellum (Fig. 1A–C). In contrast, control animals (T2/Onc2/þ; SBase/þ) remained tumor free (Fig. 1A, black line). Hema-toxylin and eosin (H&E) staining of tumors showed classicalfeatures of medulloblastoma, including small round tumorscomposed of sheets of undifferentiated cells with minimalcytoplasm (Fig. 1D and E). Tumors were negative for gliomamarkers such as GFAP and Nestin (Fig. 1F and G), providingfurther evidence they are medulloblastomas. Trp53R172H/þ;Nes-cre/þ mice also succumbed to tumors but the tumorswere of skin cell origin (15).Trp53R172Hmutantmicewith activeSB transposition also developed medulloblastomas in thecerebellum but at a higher frequency than in triple transgenicmice lacking Trp53R172H (Fig. 1A, blue line; Table 1, 59% vs. 40%;P¼ 0.026, Fisher exact test). PCR analysis of medulloblastomagenomic DNA showed that the wild-type Trp53 allele was stillpresent in tumors (Fig. 1H). This was not unexpected becauseTrp53R172H is a dominant negative Trp53 allele that can inac-tivate wild-type Trp53 (15). Together, these results indicatethat mobilized transposons can promote the formation ofmedulloblastoma in the cerebellum.

Sequence analysis of SB insertion sites inmedulloblastomas

To identify driver genes for medulloblastoma, we PCR-amplified and sequenced the transposon insertion sites from17 wild-type and 27 Trp53R172H mutant tumors (Table 1). Intotal, 7,806 and 13,057 nonduplicated SB insertions wereidentified in wild-type and Trp53R172H mutant tumors, respec-tively (Table 1). Insertions were then analyzed using the GKCmethod (20) to identify common insertion sites (CIS). CISs areregions in cancer genomes that harbor a higher number oftransposon insertions than predicted by random chance and

Koso et al.

Cancer Res; 74(8) April 15, 2014 Cancer Research2352




therefore most likely to harbor candidate cancer genes. TheseCISs were then annotated to the nearest mouse gene (CISgenes; ref. 20). Thirteen CIS genes were identified in wild-typetumors (Fig. 2A) and 18 CIS genes were identified in Trp53R172H

mutant tumors (Fig. 2B). Five CIS genes, including Foxr2, Tgif2,Alx4, Crebbp, and Wac, were identified in both wild-type andTrp53R172H mutant tumors (Fig. 3A), indicating that thesegenes regulate medulloblastoma formation irrespective of thegenetic background.To determine whether the CIS genes are enriched in sig-

naling pathways important in cancer, we made use of the

DAVID Bioinformatics Resources (http://david.abcc.ncifcrf.gov/). A Gene Ontology term enrichment analysis of thecombined CIS gene list (26 genes in Fig. 3A) identified pro-cesses associated with transcriptional regulation as beinghighly significant (Fig. 2C). Ten out of 26 CIS genes (38%) areinvolved in transcription, and include transcription factors(Foxr2, Tgif2, Alx4, Gli1, and Hivep3), transcription cofactors(Crebbp and Trim33), a histone acetyltransferase (Crebbp), ahistone methyltransferase (Whsc1l1), a regulator of histoneH2B ubiquitination (Wac), and a component of the SWI/SNFchromatin remodeling complex (Arid1b).

Figure 1. Transposon mutagenesisinduces medulloblastoma. A,Kaplan–Meier survival curvesshow that mice with mobilizedtransposons display shortenedsurvival compared with micewithout mobilized transposons. B,tumors were found in thecerebellum (arrows). C–E, H&Estaining of tumors show classichistologic features ofmedulloblastomas. Arrows,multinucleated giant cells. Bars,100 mm. F and G, tumors werenegative for GFAP (F, left) andNestin (G, left). GFAP and Nestinstaining were observed in corticalastrocytes (F, right) and theventricular zone of embryonicbrain (G, right), respectively. Bars,100 mm. H, PCR-based detectionof the recombined and activatedTrp53R172H mutant allele. TheTrp53R172H mutant allele wasdetected in the cerebellum (Ce)of double transgenic animals(Nes-cre/þ; Trp53R172H/þ) butnot in control (Ctrl) animals(Trp53R172H/þ). Both Trp53 WTand Trp53R172H mutant alleleswere observed in 6 MBs (T1–T6).

Table 1. Analysis of transposon insertions in medulloblastomas

GenotypeMice

necropsya MedulloblastomabTumorsamplesc

Number ofinsertionsd

Number ofCIS genes

T2/Onc2/þ; SBase/þ (n ¼ 5) 5 0 (0%)Nes-cre/þ; T2/Onc2/þ; SBase/þ (n ¼ 43) 19 17 (40%) 17 7,806 13Trp53R172H/þ; Nes-cre/þ (n ¼ 10) 10 0 (0%)Trp53R172H/þ; Nes-cre/þ; T2/Onc2/þ; SBase/þ (n ¼ 58) 37 34 (59%) 27 13,057 18

Total 44 20,863 26

aThe number of mice that were sent to necropsy.bThe number of mice with medulloblastoma at histologic analysis.cThe number of tumor samples used for the analysis of transposon insertions. Each tumor is from a different mouse.dInsertions in the transposon donor chromosomes were not included because of problem caused by local hopping.

Medulloblastoma Cancer Gene Screen

www.aacrjournals.org Cancer Res; 74(8) April 15, 2014 2353




GLI1, a downstream effector of SHH signaling, is often ampli-fied in medulloblastomas of the SHH subtype (21). Gli1 wasidentified as a CIS gene in Trp53 mutant tumors (Fig. 3A). AllGli1 insertions were located upstream of exon 6, in the sametranscriptional orientation as Gli1 (Fig. 3B), suggesting that SBmight be upregulating Gli1 expression from the MSCV pro-moter contained within the transposon, consistent with itsknown role as an oncogene. To confirm this, we PCR-amplifiedSB–Gli1 fusion transcripts with primers that recognize thesplice donor site located downstream of the MSCV promoterand the coding region of Gli1. Sequencing of the amplifiedtranscripts showed that the MSCV promoter was spliced toexon 6 of Gli1 in both tumors examined with Gli1 insertions(Fig. 3H), resulting in the overexpression of N-terminal trun-cated transcripts (Gli1-dN) that lack the suppressor of fused(Sufu)-binding domain (13). Sufu has an inhibitory effect onGli1transcriptional activity by the sequestration of Gli1 to thecytoplasm (22). A similar truncated GLI1 is expressed in humantumor cell lines (23). Overexpression of Gli1-dN promotedfocus formation in NIH3T3 cells (Fig. 5A) as well as growthin low serum (Fig. 5B). Taken together, these data indicatethat derepression of Gli1 from Sufu-mediated suppressionplays an important role in the formation of medulloblastoma.

Themost frequentlymutated gene inmedulloblastomaswasthe Foxr2 transcription factor (Fig. 2A and B). Foxr2 is amember of the forkhead box transcription factor family. MostFoxr2 insertions are sense oriented and located upstream ofthe protein-coding region of Foxr2 (Fig. 3C). Sequencing of thefusion transcripts showed that theMSCVpromoterwas splicedto exon 2 of Foxr2 (Fig. 3I), resulting in the overexpression of afull-length Foxr2 transcript (Fig. 3L). We also examined thepatterns of transposon insertions in the other four genes thatwere identified irrespective of genetic background (Fig. 3A).Tgif2 insertions are all sense-oriented and located upstream ofthe initiating codon (Fig. 3D). Sequencing of the fusion tran-

scripts showed that the MSCV promoter was spliced to exon 2(Fig. 3J), again driving the expression of a full-length Tgif2transcript. Tgif2 is a member of the TALE homeodomainprotein family. Alx4 belongs to the family of aristaless-likehomeobox genes. Alx4 insertions are similarly sense-orientedand located upstream of the gene or within intron 1 (Fig. 3E).Sequencing of the fusion transcripts showed that the MSCVpromoter was spliced to Alx4 exon 2 (Fig. 3K), inducing theexpression of a truncated protein that may initiate from anATG codon located in exon 2.

In contrast, Crebbp and Wac insertions were mostlydistributed throughout the genes, and there was little ori-entation bias (Fig. 3F and G). The average numbers of Crebbpand Wac insertions in tumors are 1.15 and 1.18 respectively,suggesting a haploinsufficient role for Crebbp and Wac intumor suppression. Crebbp functions as a transcriptionalcoactivator, and haploinsufficiency of CREBBP is responsiblefor Rubinstein–Taybi syndrome, an autosomal congenitaldisorder characterized by developmental and pathologicphenotypes, including predisposition to tumors includingmedulloblastoma (24, 25). Somatic heterozygous nonsenseCREBBP mutations have been identified in medulloblasto-mas (26). WAC is a functional partner of histone H2Bubiquitin ligase RNF20/40, and the depletion of RNF20/40or WAC abolishes H2B monoubiquitination (27). Althoughsignificant loss of RNF20/40 is catastrophic to cells, moresubtle changes in RNF20/40 expression because of partialknockdown allow cells to proliferate and manifest dramaticgenomic instability (28). These data support a haploinsuffi-cient role for Crebbp and Wac in medulloblastomas.

A previous transposon screen that analyzed for recurrentinsertions in medulloblastomas, which developed on a Ptch1heterozygous genetic background (6) did not identify Gli1,Foxr2, Tgif2, or Alx4, raising the possibility that these geneshave overlapping function with Ptch1 deficiency. We also

Figure 2. CIS genes identified inmedulloblastomas. A, thirteen CISgenes were identified in wild-typetumors. B, eighteen CIS geneswere identified in Trp53 mutanttumors. P values were calculatedas described (20). Prediction ongene function indicates whetherthe T2/Onc2 insertion would causetranscriptional activation of a gene(drives) or would disrupt genetranscription (disrupts) on the basisof the position and orientation ofthe T2/Onc2 insertion relative togene transcription. C, GO analysisof biological processes of CISgenes in medulloblastomas.

Koso et al.





looked for co-occurring and mutually exclusive mutations inpairs of CIS genes. A significant co-occurrence of mutations inTgif2 and Alx4 genes was observed (Fig. 3M and N; P ¼ 0.042after multiple testing correction), whereas Tgif2 and Alx4insertions were mutually exclusive to Gli1 insertions (Fig. 3M).

Expression of GLI1, FOXR2, TGIF2, and ALX4 in humanmedulloblastomasBecause the patterns of transposon insertions predict

oncogenic functions for GLI1, FOXR2, TGIF2, and ALX4 inmedulloblastoma, we next determined whether expression ofthese genes was upregulated in human medulloblastomausing a microarray dataset for 285 medulloblastomas (29)that was accessible through the open access database R2(http://r2.amc.nl). Not surprisingly, GLI1 is strongly exp-ressed in medulloblastomas of the SHH subtype (Supplemen-tary Fig. S1A). Interestingly, most tumors that stronglyexpress TGIF2 also belong to the SHH subgroup (Supplemen-tary Fig. S1B). ALX4 and FOXR2 are also strongly expressed in

medulloblastoma, but in more limited subsets of tumors(Supplementary Fig. S1C and S1D), and once again most ofthese tumors are of the SHH subgroup. We also found thatmost of the tumors highly expressing ALX4 also expressedTGIF2 at high levels (Supplementary Fig. S1E), showingsignificant co-occurrence of ALX4 and TGIF2 in medulloblas-tomas (P < 0.001, Fisher exact test). Taken together, thesedata indicate that these 4 putative oncogenes are overex-pressed in a subset of human medulloblastomas that belongto the SHH subgroup.

Foxr2, Tgif2, and Alx4 enhance transcriptional activity ofGli1

Because FOXR2, TGIF2, andALX4 are expressed in a subset ofSHHsubgroupmedulloblastomas,wenext testedwhether thesegenes can activate a SHH responsive element, the Gli-bindingsite (Gli-BS; ref. 30). Transfection of a vector expressing full-length Gli1 increased reporter expression in HEK-293T cells;however, expression of the reporter with mutated Gli-binding

Figure 3. Patterns of transposoninsertions in medulloblastomacandidate genes. A, thirteen and18CIS genes were identified in wild-type and Trp53 mutant tumors,respectively. Five CIS genes wereshared between the wild-type andTrp53 mutant tumors. B–G, SBinsertions are shown for wild-typeand Trp53 mutant tumors for theCIS genes Gli1 (B), Foxr2 (C), Tgif2(D), Alx4 (E), Crebbp (F), and Wac(G). Transposon insertions arelocated in the sense (red arrows) orantisense orientation (whitearrows) relative to transcription.H–K, fusion transcripts were PCRamplified with primers designed torecognize the splice donor sitelocated downstream of the MSCVpromoter in T2/Onc2 and thecoding region of Gli1 (H), Foxr2 (I),Tgif2 (J), or Alx4 (K), andsequenced. L, expression level ofFoxr2 in tumors with Foxr2insertions (n ¼ 4) and tumorswithout (n ¼ 4). Expression levelswere normalized using b-actin asan internal control, and the averagewas set to a value of 1. Data arepresented as box-whisker plotsshowing the minimum, 25thpercentile,median, 75th percentile,and the maximum (P-value:Student t test, one-tailed). M, co-occurrence of insertions in Foxr2,Tgif2, Alx4, and Gli1 in 44 tumors.N, significance (P-value) of co-occurring insertions wasdetermined using the Fisher exacttest for each combination.






site (mGli-BS; ref. 30) was not changed (Fig. 4A). Transfection ofvectors expressing Foxr2, Tgif2, or Alx4 alone did not affectreporter expression; however, cotransfection of these vectors

along with the Gli1 expression vector increased Gli1 reporterexpressionabove that seenwith theGli1 expressionvector alone(Fig. 4A and B). This upregulation of Gli1-dependent reporter

Figure 4. Foxr2, Alx4, and Tgif2cooperate with Gli1 and activateGli-dependent transcription. A,expression of Gli-binding siteluciferase (Gli-BS-Luc) wasupregulated by transfection withGli1 expression vector (the averageluciferase activity from the Gli1expression vector was designatedas 100), whereas no activation wasobserved with mutated Gli-bindingsite luciferase (mGli-BS-Luc;ref. 30). Transcriptional activationof Gli-BS-Luc was upregulated bycotransfection with Foxr2, Alx4, orTgif2 expression vectors. Datarepresent means� SEMwith threesamplesper group. TheamountsofDNA (mg) used for transfection areshown. �, P < 0.05 and ��, P < 0.01(Student t test, one-tailed). B, theexperiments shown in A wererepeated three times, and datarepresent means � SEM with 9samples per group. ��, P < 0.01(Student t test, one-tailed). C,dose-dependent cooperativeeffects were observed for Foxr2,Alx4, and Tgif2. Data representmeans � SEM with 9 samples pergroup. ��, P < 0.01 (Student t test,one-tailed). D, the expression ofPtch1 and Ptch2 in NIH3T3 cellsoverexpressing Foxr2, Alx4, orTgif2. Expression was normalizedusing b-actin as an internal control.Data represent means � SEM withthree samples per group. �,P<0.05and ��,P<0.01 (Student t test, one-tailed). E, cooperation betweenAlx4 and Tgif2 on activation of Gli-dependent transcription. Datarepresent means � SEM with 9samples per group. ��, P < 0.01(Student t test, one-tailed).

Koso et al.





expression was observed dose dependently (Fig. 4C). Asexpected, expression of endogenous SHH target genes inNIH3T3 cells, Ptch1, and Ptch2 (31) were upregulated by Foxr2,Tgif2, or Alx4 (Fig. 4D). Cotransfection of Tgif2 and Alx4induced stronger luciferase reporter expression comparedwith Tgif2 or Alx4 alone (Fig. 4E), providing evidence forfunctional cooperation between these genes. Taken together,these results demonstrate cooperative functions of thesegenes (Foxr2, Tgif2, or Alx4) and Gli1 in Gli-dependent tran-scriptional activation.

Functional analysis of candidate medulloblastomadriver genesTo examine the oncogenic functions of candidate genes, we

first measured their transforming activity in NIH3T3 cells. We

infected NIH3T3 cells with retroviruses expressing full-lengthFoxr2, Tgif2, truncated Alx4, or an empty vector. Foxr2 over-expression induced morphologic changes (Fig. 5C), and pro-duced a significantly greater number of NIH3T3 foci, whereastruncated Alx4 did not promote and Tgif2 suppressed fociformation (Fig. 5D). Analysis of cell growth in low-serummediaalso showed that Foxr2-expressing cells grew more efficiently(Fig. 5E). These data indicate transforming activity of Foxr2in NIH3T3 cells.

We next analyzed the effects of Foxr2 overexpression onMEFs. Normal MEFs undergo senescence as a result of serialpassage in culture (32). As expected, control MEFs becameenlarged and ceased to proliferate after several passages, andthese cells expressed senescence-associated b-galactosidase(Fig. 5F andG). However, Foxr2-expressingMEFs retained their

Figure 5. Functional analysis of CISgenes. A–E, NIH3T3 cells wereinfected with retrovirusesexpressing Gli1-dN, Foxr2, Alx4,Tgif2, or empty vector control.These cells were used for the focusformation assay (A and D) and cellgrowth assay in low serum (B andE).Bars, 250mm(C).Data representmeans � SEM with threesamples per group. �, P < 0.05 and��, P < 0.01 (Student t test, one-tailed). Data are representative oftwo independent experiments.F–I, MEFs were infected withretroviruses expressing Foxr2 orempty vector, and seriallypassaged. Senescence-associated b-gal expression wasexamined at passage 2 (F). Foxr2-expressing MEFs showedprolonged proliferation comparedwith control (G). The proportionof b-gal–expressing cells in controland Foxr2-expressing MEFs (H).The expression of p16Ink4 waslower in Foxr2-expressing MEFsthan in control (I). Expression wasnormalized using b-actin as aninternal control. Data representmeans � SEM with threesamples per group. �, P < 0.05 and��, P < 0.01 (Student t test,one-tailed). Data arerepresentative of two independentexperiments.






small cell shape and continued to proliferate (Fig. 5F and G),and the proportion of b-galactosidase expressing cellsdecreased (Fig. 5H). p16Ink4a, a cyclin-dependent kinase inhib-itor, is also expressed in most senescent cells (32). As expected,the expression level of p16Ink4a increased during passages ofMEFs; however, its expression was significantly lower in MEFsoverexpressing Foxr2 (Fig. 5I). Taken together, these resultsindicate that overexpression of Foxr2 prevents MEFs fromentering senescence.

Cerebellar GNPs are considered to be the cell-of-origin of theSHH subgroup of medulloblastomas (3). We therefore nextexamined the effects of overexpression of Foxr2, truncatedAlx4, and Tgif2 on cultures of primary cerebellar cells. Thesecultures consisted primarily of GNPs and astrocytes (9). At 1week after plating, colonies consisting of GNPs were distin-guishable from surrounding large flat astrocytes because oftheir small cell shape and cytoplasmic granules (Fig. 6A–C).These small granular cells expressed Math1, a specific marker

Figure 6. Overexpression of medulloblastoma candidate genes promotes proliferation of GNPs. A, cerebellar cells infected with retroviruses expressingFoxr2, Alx4, Tgif2, or empty vector. Arrowheads, large colonies consisting of GNPs. Bars, 250 mm. B, magnified views of GNP colonies. Bars, 50 mm. C,GNP colonies contained cytoplasmic granules that were visible under differential interference contrast (DIC) microscopy (left). These granularcells expressed Math1, but were negative for S100b (right). Arrow, the nucleus of an astrocyte. Nuclei were counterstained with DAPI. Bars, 50 mm.D–F, colony size and proliferation of GNPs were compared among control, Foxr2, Alx4, and Tgif2-expressing GNPs. Foxr2-expressing GNPsgenerated larger colonies than control (D; Mann–Whitney test, P < 0.001). The fraction of Ki67-expressing Math1-positive GNPs was increased byFoxr2 overexpression (E and F). Data represent means � SEM with three samples per group. ��, P < 0.01 (Student t test, one-tailed). Data arerepresentative of two independent experiments. G, cerebellar cells infected with retroviruses expressing Alx4 and Tgif2, or Alx4 and empty vector.Most of the control GNPs remained single (arrows), whereas cotransduction of Alx4 and Tgif2 induced colonies consisting of GNPs (arrowheads).Bars, 250 mm (top) and 50 mm (bottom). H and I, GNPs transduced with Alx4 and Tgif2 generated larger colonies than control (Mann–Whitneytest, P < 0.001; H). The fraction of Ki67-expressing Math1-positive GNPs was increased by cotransduction of Alx4 and Tgif2 (I). Data representmeans � SEM with three samples per group. ��, P < 0.01 (Student t test, one-tailed). Data are representative of two independent experiments.

Koso et al.





for GNPs (9, 11), but were negative for an astrocyte marker,S100b (Fig. 6C). The colony size was larger in Foxr2-expressingGNPs than control cells (Fig. 6A, B, and D). The expression ofKi67, a marker associated with cellular proliferation, was alsoincreased by Foxr2 overexpression (Fig. 6E and F). Thesefindings demonstrate a growth promoting function for Foxr2in primary GNPs. Although Alx4 or Tgif2 alone did not promoteproliferation (Fig. 6D and F), cotransduction of Alx4 and Tgif2induced colonies with larger sizes than Alx4 alone (Fig. 6G andH). Ki67 expression was also increased by cotransduction ofAlx4 and Tgif2 (Fig. 6I), providing evidence for cooperationbetween these genes in GNPs.

DiscussionTransposon-based insertional mutagenesis allows for unbi-

ased, whole genome screens for cancer genes, and has beenused successfully to model many different types of cancer inmice (5). In this study, we used transposon mutagenesis tomodel medulloblastoma. Using Nes-cre transgenic mice, wemobilized T2/Onc2 transposons in neural stem and progenitorcells of the developing brain. As a result, mutagenic transpo-sons induced medulloblastomas in the cerebellum. Strikingly,we observed medulloblastomas even on a wild-type geneticbackground. These tumors are invaluable because they havethe potential to identify driver genes that can initiate tumor-igenesis rather than just accelerate tumorigenesis induced bymutation of another gene, such as a mutation of Ptch1.Subsequent sequence analysis of the transposon integrationssites in these tumors identified knownmedulloblastoma-asso-ciated genes Gli1 and Crebbp as well as genes that have notbeen associated with medulloblastoma, including Foxr2, Tgif2,Alx4, and Wac.A previously published transposon screen identified medul-

loblastoma candidate genes by analyzing for recurrent inser-tions in medulloblastomas that developed on heterozygousPtch1 mutant or Trp53 mutant (þ/� and �/�) backgrounds(6). Ptch1 is a negative regulator of the SHH signaling pathway(33) and not surprisingly, Ptch1 heterozygous mice spontane-ously develop medulloblastomas (34). CIS genes identified onthe Ptch1 mutant background were shown to promote meta-static dissemination of medulloblastoma (7). The previousstudy (6) also identified Ptch1 in medulloblastomas inducedon Trp53 mutant backgrounds. The identification of Ptch1 onthe Trp53�/� background is consistent with a previous reportdescribing cooperation between Trp53 loss and Ptch1 hetero-zygosity formedulloblastoma formation (35). In contrast, it hasbeen reported that Ptch1 heterozygosity does not acceleratemedulloblastoma formation on Trp53 heterozygous back-ground (35). This may explain why Ptch1 was not identifiedin our screen on the Trp53R172H/þ heterozygous mutantbackground. We also used a wild-type background, which didnot develop medulloblastoma spontaneously. Therefore, someof the CIS genes identified in our screen are likely to representdriver genes involved in the initiation of medulloblastoma.Comparison of our CIS genes with those identified in the Ptch1and Trp53mutant screen showed that 5 genes were identifiedin both screens, includingArid1b, Crebbp, Trim33,Map2k4, and

Wac, providing further evidence that these genes are medul-loblastoma disease genes. The patterns of transposon inser-tions predict tumor suppressor functions for these genes. Infact, recurrent mutations in cancers have been reported forTRIM33 (36), ARID1B (37), and MAP2K4 (38). Interestingly,genes such as Gli1, Foxr2, Tgif2, and Alx4 were not identifiedin the Ptch1 and Trp53mutant screen (7), raising the possibilitythat these genes might function in SHH signaling and weretherefore not identified because the SHH pathway was alreadyactivated by either Ptch1 germline or transposon-inducedsomatic mutations.

The most frequently mutated gene identified in our screenwas Foxr2. Fox proteins can both activate and repress geneexpression through the recruitment of cofactors or repressors,primarily histone deacetylases. Deregulation of Fox proteins iscommonly associated with cancer progression (39). FOXR2 hashigh homology to FOXR1, which can functionally replace MYCand drive proliferation of a neural crest cell line (40). A recentmutagenesis screen identified Foxr2 as a proto-oncogene inmalignant peripheral nerve sheath tumors (41). This studyprovides the first forward genetic evidence associating Foxr2with medulloblastoma. Foxr2 insertion pattern predicts anoncogenic role for Foxr2. Consistent with this, Foxr2 over-expression showed transforming activity in NIH3T3 cells,suppressed senescence in primary fibroblasts and promotedthe proliferation of GNPs. Interestingly, Foxr2 was stronglyexpressed in a small subset ofmedulloblastomas that belongedto the SHH subgroup of medulloblastoma.

The analysis of co-occurring and mutually exclusive pairs ofinsertions in CIS genes showed that Gli1 insertions weremutually exclusive to insertions in Tgif2 and Alx4. In addition,we found a significant co-occurrence between insertions inTgif2 and Alx4. Analysis of the expression of these genes inhuman medulloblastoma showed that TGIF2 and ALX4 arestrongly expressed in a subset ofmedulloblastomas that belongto the SHH subgroup. This again raises the possibility that Tgif2and Alx4 function in SHH signaling. Consistent with this,expression of Gli1 is reduced in Tgif2 conditional null mice,and deficiency of Gli3, a potent repressor of SHH signaling,partially rescues the Tgif2 null phenotype (42). Interestingly,the Alx4 null phenotype can also be partially rescued by loss ofGli3, further suggesting that Alx4 functions in SHH signaling(43). Our studies showing that TGIF2 and ALX4 genes arestrongly expressed in the SHH subgroup of medulloblastomaalso supports this notion.

Importantly, we found that Foxr2,Tgif2, andAlx4 cooperatedwith Gli1 and activated Gli-dependent transcription, demon-strating the involvement of these genes in SHH signaling.Tumors with mutations in members of the SHH pathwayconstitute nearly 25% ofmedulloblastomas (26, 44) that belongto the SHH subgroup of medulloblastomas, and the aberrantactivation of the SHH pathway is considered to be causallyrelated to this subgroup. Although Alx4 and Tgif2 promotedGli-dependent transcription, these genes did not promoteproliferation of GNPs by itself. In contrast, cotransduction ofAlx4 and Tgif2 enhanced Gli-dependent transcription andpromoted proliferation. A previous study showed that addi-tional factors define an activation threshold for Gli target genes






(45), raising a possibility that similarmechanismsmay regulatea proliferative response to the Gli transcriptional activity inGNPs. The SHH signaling pathway has been implicated in awide variety of human tumors (46). Analysis of the expressionlevel of FOXR2, TGIF2, and ALX4 in human cancer cell linesusing the CCLE database (47) revealed that nearly 7% of cancercell lines showed higher expression of FOXR2 than 1.5 inter-quartile ranges above the upper quartile (Supplementary Fig.S1F). Interestingly, FOXR2 strongly expressing tumors includecancers associated with SHH signaling such as multiple mye-loma, melanoma, and glioma (46), raising a possibility thatFOXR2 may also be involved in other cancers associated withSHH signaling. ALX4 and TGIF2 were strongly expressed in 2%and 0.1% of cancer cell lines, respectively (Supplementary Fig.S1G and S1H). Although we cannot distinguish the expressionof truncated ALX4 from that of full-length ALX4, nearly half ofthe cell lines that strongly express ALX4 belonged to the lungnon–small cell carcinoma, in which SHH signaling plays anessential role (48).

Interestingly, nearly 40% of genes identified in our screenwere associated with transcription, consistent with a pre-vious screen for medulloblastoma disease genes (49). Tran-scription factors play essential roles in coordinated geneexpression during development. Recently, the aberration ofgene expression has been implicated in mechanisms ofcancer formation in the pediatric setting (26, 50), raisingthe possibility that aberrations of gene expression mayperturb the normal developmental program and causederegulated proliferation of stem or progenitor cells. Thefinding of transcriptional regulators in our screen supportsthe notion that deregulation of the developmental programis a core mechanism of pathogenesis in medulloblastomas(3). Taken together, our findings implicate FOXR2, TGIF2,and ALX4 as potential therapeutic targets for medulloblas-

tomas as well as for cancers that involve SHH signaling.Future studies will be required to determine the exact rolesof these genes in a wide variety of human cancers.

Disclosure of Potential Conflicts of InterestNo potential conflicts of interest were disclosed.

Authors' ContributionsConception and design: H. Koso, N.A. Jenkins, N.G. Copeland, S. WatanabeDevelopment of methodology: N.A. Jenkins, N.G. CopelandAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): H. Koso, A. Tsuhako, J.M. Ward, D.J. Adams, N.A.Jenkins, N.G. CopelandAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis):H. Koso, E. Lyons, A.G. Rust, D.J. Adams, N.A. Jenkins,N.G. Copeland, S. WatanabeWriting, review, and/or revision of the manuscript: H. Koso, E. Lyons, J.M.Ward, N.A. Jenkins, N.G. Copeland, S. WatanabeAdministrative, technical, or material support (i.e., reporting or orga-nizing data, constructing databases): E. Lyons, N.A. Jenkins, N.G. CopelandStudy supervision: N.A. Jenkins, N.G. Copeland

AcknowledgmentsThe authors thank K. Rogers and S.M. Rogers for rodent necropsy. The authors

also thank P. Cheok, N. Lim, and D. Chen for their help with tumor monitoring.C.C.K. Yew is acknowledged for the bioinformatics analysis of transposon insertionsites. We are grateful to Dr. J. Johnson for providing the anti-Math1 antibody.

Grant SupportThis work was supported by the Japan Society for the Promotion of Science,

the Biomedical Research Council, the Agency for Science, Technology andResearch, Singapore, and the Cancer Prevention Research Institute of Texas(CPRIT). D.J. Adams and A.G. Rust are supported by Cancer Research-UK and theWellcome Trust. N.G. Copeland and N.A. Jenkins are also both CPRIT Scholars inCancer Research.

The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely to indicate thisfact.

Received June 6, 2013; revised January 31, 2014; accepted February 22, 2014;published OnlineFirst March 5, 2014.

References1. Northcott PA, Korshunov A, Pfister SM, Taylor MD. The clinical

implications of medulloblastoma subgroups. Nat Rev Neurol 2012;8:340–51.

2. Gibson P, Tong Y, Robinson G, Thompson MC, Currle DS, Eden C,et al. Subtypes of medulloblastoma have distinct developmentalorigins. Nature 2010;468:1095–9.

3. Gilbertson RJ, Ellison DW. The origins of medulloblastoma subtypes.Annu Rev Pathol 2008;3:341–65.

4. Kool J, Berns A. High-throughput insertional mutagenesis screens inmice to identify oncogenic networks. Nat Rev Cancer 2009;9:389–99.

5. Copeland NG, Jenkins NA. Harnessing transposons for cancer genediscovery. Nat Rev Cancer 2010;10:696–706.

6. Wu X, Northcott PA, Dubuc A, Dupuy AJ, Shih DJ, Witt H, et al. Clonalselection drives genetic divergence of metastatic medulloblastoma.Nature 2012;482:529–33.

7. Mumert M, Dubuc A, Wu X, Northcott PA, Chin SS, Pedone CA, et al.Functional genomics identifies drivers of medulloblastoma dissemi-nation. Cancer Res 2012;72:4944–53.

8. Koso H, Takeda H, Yew CC, Ward JM, Nariai N, Ueno K, et al.Transposon mutagenesis identifies genes that transform neural stemcells into glioma-initiating cells. Proc Natl Acad Sci U S A 2012;109:E2998–3007.

9. Kenney AM, Rowitch DH. Sonic hedgehog promotes G(1) cyclinexpression and sustained cell cycle progression in mammalian neu-ronal precursors. Mol Cell Biol 2000;20:9055–67.

10. Park HJ, Wang Z, Costa RH, Tyner A, Lau LF, Raychaudhuri P. An N-terminal inhibitory domain modulates activity of FoxM1 during cellcycle. Oncogene 2008;27:1696–704.

11. Helms AW, Johnson JE. Progenitors of dorsal commissural inter-neurons are defined by MATH1 expression. Development 1998;125:919–28.

12. Debacq-Chainiaux F, Erusalimsky JD, Campisi J, Toussaint O. Pro-tocols to detect senescence-associated b-galactosidase (SA-bgal)activity, a biomarker of senescent cells in culture and in vivo. NatProtoc 2009;4:1798–806.

13. DunaevaM,MichelsonP, KogermanP, Toftgard R.Characterization ofthe physical interaction ofGli proteinswith SUFUproteins. JBiol Chem2003;278:5116–22.

14. Satoh S, Watanabe S. TGIF, a homeodomain transcription factor,regulates retinal progenitor cell differentiation. Exp Eye Res 2008;87:571–9.

15. Olive KP, Tuveson DA, Ruhe ZC, Yin B, Willis NA, Bronson RT, et al.Mutant p53 gain of function in two mouse models of Li-Fraumenisyndrome. Cell 2004;119:847–60.

16. Tronche F, Kellendonk C, Kretz O, Gass P, Anlag K, Orban PC, et al.Disruption of the glucocorticoid receptor gene in the nervous systemresults in reduced anxiety. Nat Genet 1999;23:99–103.

17. Dupuy AJ, Akagi K, Largaespada DA, Copeland NG, Jenkins NA.Mammalian mutagenesis using a highly mobile somatic SleepingBeauty transposon system. Nature 2005;436:221–6.

Koso et al.





18. Dupuy AJ, Rogers LM, Kim J, Nannapaneni K, Starr TK, Liu P, et al. Amodified sleeping beauty transposon system that can be used tomodel a wide variety of human cancers in mice. Cancer Res 2009;69:8150–6.

19. Parsons DW, Li M, Zhang X, Jones S, Leary RJ, Lin JC, et al. Thegenetic landscape of the childhood cancer medulloblastoma. Science2011;331:435–9.

20. March HN, Rust AG, Wright NA, ten Hoeve J, de Ridder J, Eldridge M,et al. Insertional mutagenesis identifiesmultiple networks of cooperat-ing genes driving intestinal tumorigenesis. Nat Genet 2011;43:1202–9.

21. Zhu H, Lo HW. The human glioma-associated oncogene homolog 1(GLI1) family of transcription factors in gene regulation and diseases.Curr Genomics 2010;11:238–45.

22. Kogerman P, Grimm T, Kogerman L, Krause D, Unden AB, SandstedtB, et al. Mammalian suppressor-of-fused modulates nuclear-cyto-plasmic shuttling of Gli-1. Nat Cell Biol 1999;1:312–9.

23. Shimokawa T, Tostar U, Lauth M, Palaniswamy R, Kasper M, ToftgardR, et al. Novel human glioma-associated oncogene 1 (GLI1) splicevariants reveal distinct mechanisms in the terminal transduction of thehedgehog signal. J Biol Chem 2008;283:14345–54.

24. Roelfsema JH, Peters DJ. Rubinstein-Taybi syndrome: clinical andmolecular overview. Expert Rev Mol Med 2007;9:1–16.

25. Taylor MD, Mainprize TG, Rutka JT, Becker L, Bayani J, Drake JM.Medulloblastoma in a child with Rubenstein-Taybi syndrome: casereport and review of the literature. Pediatr Neurosurg 2001;35:235–8.

26. Robinson G, Parker M, Kranenburg TA, Lu C, Chen X, Ding L, et al.Novelmutations target distinct subgroups ofmedulloblastoma.Nature2012;488:43–8.

27. Zhang F, Yu X. WAC, a functional partner of RNF20/40, regulateshistone H2B ubiquitination and gene transcription. Mol Cell 2011;41:384–97.

28. Chernikova SB, Razorenova OV, Higgins JP, Sishc BJ, Nicolau M,Dorth JA, et al. Deficiency in mammalian histone H2B ubiquitin ligaseBre1 (Rnf20/Rnf40) leads to replication stress and chromosomalinstability. Cancer Res 2012;72:2111–9.

29. Northcott PA, Shih DJ, Peacock J, Garzia L, Morrissy AS, Zichner T,et al. Subgroup-specific structural variation across 1,000 medullo-blastoma genomes. Nature 2012;488:49–56.

30. Sasaki H, Hui C, Nakafuku M, Kondoh H. A binding site for Gli proteinsis essential for HNF-3b floor plate enhancer activity in transgenics andcan respond to Shh in vitro. Development 1997;124:1313–22.

31. Kimura H, Stephen D, Joyner A, Curran T. Gli1 is important for medul-loblastoma formation in Ptc1þ/� mice. Oncogene 2005;24:4026–36.

32. Rodier F, Campisi J. Four faces of cellular senescence. J Cell Biol2011;192:547–56.

33. Pasca di Magliano M, Hebrok M. Hedgehog signalling in cancerformation and maintenance. Nat Rev Cancer 2003;3:903–11.

34. Goodrich LV, Milenkovic L, Higgins KM, Scott MP. Altered neural cellfates and medulloblastoma in mouse patched mutants. Science1997;277:1109–13.

35. Wetmore C, Eberhart DE, Curran T. Loss of p53 but not ARF accel-erates medulloblastoma in mice heterozygous for patched. CancerRes 2001;61:513–6.

36. Hatakeyama S. TRIM proteins and cancer. Nat Rev Cancer2011;11:792–804.

37. Sausen M, Leary RJ, Jones S, Wu J, Reynolds CP, Liu X, et al.Integrated genomic analyses identify ARID1A and ARID1B altera-tions in the childhood cancer neuroblastoma. Nat Genet 2013;45:12–7.

38. Biankin AV, Waddell N, Kassahn KS, Gingras MC, Muthuswamy LB,Johns AL, et al. Pancreatic cancer genomes reveal aberrations in axonguidance pathway genes. Nature 2012;491:399–405.

39. Myatt SS, Lam EW. The emerging roles of forkhead box (Fox) proteinsin cancer. Nat Rev Cancer 2007;7:847–59.

40. Santo EE, EbusME,Koster J, Schulte JH, LakemanA, vanSluis P, et al.Oncogenic activation of FOXR1 by 11q23 intrachromosomal deletion-fusions in neuroblastoma. Oncogene 2011;31:1571–81.

41. Rahrmann EP, Watson AL, Keng VW, Choi K, Moriarity BS, BeckmannDA, et al. Forwardgenetic screen formalignant peripheral nerve sheathtumor formation identifies new genes and pathways driving tumori-genesis. Nat Genet 2013;45:756–66.

42. Taniguchi K, Anderson AE, Sutherland AE, Wotton D. Loss of Tgiffunction causes holoprosencephaly by disrupting the SHH signalingpathway. PLoS Genet 2012;8:e1002524.

43. Panman L, Drenth T, Tewelscher P, Zuniga A, Zeller R. Geneticinteraction of Gli3 and Alx4 during limb development. Int J Dev Biol2005;49:443–8.

44. Pugh TJ, Weeraratne SD, Archer TC, Pomeranz Krummel DA,Auclair D, Bochicchio J, et al. Medulloblastoma exome sequencinguncovers subtype-specific somatic mutations. Nature 2012;488:106–10.

45. WangH, Lei Q,Oosterveen T, Ericson J,MatiseMP. Tcf/Lef repressorsdifferentially regulate Shh-Gli target gene activation thresholds togenerate progenitor patterning in the developing CNS. Development2011;138:3711–21.

46. Merchant AA, Matsui W. Targeting Hedgehog—a cancer stem cellpathway. Clin Cancer Res 2010;16:3130–40.

47. Barretina J, Caponigro G, Stransky N, Venkatesan K, Margolin AA,Kim S, et al. The Cancer Cell Line Encyclopedia enables predictivemodelling of anticancer drug sensitivity. Nature 2012;483:603–7.

48. BermudezO,HennenE,Koch I, LindnerM, EickelbergO.Gli1mediateslung cancer cell proliferation and Sonic Hedgehog–dependent mes-enchymal cell activation. PLoS ONE 2013;8:e63226.

49. ŁastowskaM,Al-AfghaniH, Al-BaloolHH,ShethH,Mercer E,CoxheadJM, et al. Identification of a neuronal transcription factor networkinvolved in medulloblastoma development. Acta Neuropathol Com-mun 2013;1:35.

50. Schwartzentruber J, Korshunov A, Liu XY, Jones DT, Pfaff E, Jacob K,et al. Driver mutations in histone H3.3 and chromatin remodellinggenes in paediatric glioblastoma. Nature 2012;482:226–31.






2014;74:2351-2361. Published OnlineFirst March 5, 2014.Cancer Res Hideto Koso, Asano Tsuhako, Eli Lyons, et al. Identification of FoxR2 as an Oncogene in Medulloblastoma

Updated version

10.1158/0008-5472.CAN-13-1523doi:

Access the most recent version of this article at:

Material

Supplementary

http://cancerres.aacrjournals.org/content/suppl/2014/03/05/0008-5472.CAN-13-1523.DC1

Access the most recent supplemental material at:

Cited articles

http://cancerres.aacrjournals.org/content/74/8/2351.full#ref-list-1

This article cites 50 articles, 15 of which you can access for free at:

Citing articles

http://cancerres.aacrjournals.org/content/74/8/2351.full#related-urls

This article has been cited by 3 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected]

To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/74/8/2351To request permission to re-use all or part of this article, use this link



http://cancerres.aacrjournals.org/lookup/doi/10.1158/0008-5472.CAN-13-1523

http://cancerres.aacrjournals.org/content/suppl/2014/03/05/0008-5472.CAN-13-1523.DC1

http://cancerres.aacrjournals.org/content/74/8/2351.full#ref-list-1

http://cancerres.aacrjournals.org/content/74/8/2351.full#related-urls

http://cancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://cancerres.aacrjournals.org/content/74/8/2351


identiﬁcation of foxr2 as an oncogene in medulloblastomatumor and stem cell biology...

Documents