13
Molecular and Cellular Pathobiology Cic Loss Promotes Gliomagenesis via Aberrant Neural Stem Cell Proliferation and Differentiation Rui Yang, Lee H. Chen, Landon J. Hansen, Austin B. Carpenter, Casey J. Moure, Heng Liu, Christopher J. Pirozzi, Bill H. Diplas, Matthew S.Waitkus, Paula K. Greer, Huishan Zhu, Roger E. McLendon, Darell D. Bigner,Yiping He, and Hai Yan Abstract Inactivating mutations in the transcriptional repression factor Capicua (CIC) occur in approximately 50% of human oligodendrogliomas, but mechanistic links to pathogenesis are unclear. To address this question, we generated Cic-de- cient mice and human oligodendroglioma cell models. Genetic deciency in mice resulted in a partially penetrant embryonic or perinatal lethal phenotype, with the production of an aberrant proliferative neural population in surviving animals. In vitro cultured neural stem cells derived from Cic conditional knockout mice bypassed an EGF requirement for proliferation and displayed a defect in their potential for oligodendrocyte differentiation. Cic is known to participate in gene suppression that can be relieved by EGFR signal, but we found that cic also activated expression of a broad range of EGFR-independent genes. In an orthotopic mouse model of glioma, we found that Cic loss potentiated the formation and reduced the latency in tumor development. Collectively, our results dene an important role for Cic in regulating neural cell proliferation and lineage specication, and suggest mech- anistic explanations for how CIC mutations may impact the pathogenesis and therapeutic targeting of oligodendroglioma. Cancer Res; 77(22); 6097108. Ó2017 AACR. Introduction Oligodendrogliomas are a histologic type of diffuse glioma (WHO grade IIIII) that account for between 5% and 20% of gliomas (1) and frequently display characteristic loss of whole chromosome arms 1p and 19q (1p/19q-codeletion; refs. 24). Relative to other diffuse glioma subtypes, oligodendrogliomas are associated with better prognosis due in part to the tumor's increased chemosensitivity (5). However, tumor recurrence remains the most common cause of death for oligodendroglioma patients, as these tumors often recur or progress to higher grades. Previous studies from ours and other groups have revealed recur- rent inactivating mutations of Capicua (CIC, on chromosome arm 19q) and far upstream element binding protein (FUBP1, on chromosome arm 1p), identifying these genes as putative tumor suppressors for oligodendroglioma (69). Somatic mutations of these two genes, together with IDH1R1/2 mutations, 1p/19q loss, and TERT promoter mutations, dene oligodendroglioma as a specic brain tumor subtype (6, 7, 10, 11). Interestingly, whereas TERT promoter and IDH mutation occur across multiple diffuse glioma subtypes, somatic mutations of CIC and FUBP1 are exclusively restricted to oligodendrogliomas, suggesting a possible link of CIC and FUBP1 to oligodendrocyte lineage in neurogenesis and tumorigenesis. CIC is a member of the Sox-related high-mobility group (HMG) subfamily and is highly conserved among species (12). Previous evidence suggests that cic acts primarily as a transcrip- tional repressor consisting of two highly conserved domains: the HMG box, which mediates DNA binding and nuclear localiza- tion, and a C-terminal motif C1, which co-operates with the HMG-box for DNA binding (1216). In Drosophila, cic transduces the signaling of receptor tyrosine kinases (RTK), Torso, and the EGFR, and by doing so regulates cell proliferation and cell lineage specication during development (13, 15, 1719). Besides oligodendrogliomas, CIC is also linked to other cancers and noncancer diseases in which its function as a transcriptional regulator is believed to be involved (14, 20). However, the mechanism underlying the role of CIC inactivation in oligoden- droglioma genesis remains unclear, in part due to the lack of appropriate animal and cell models of oligodendroglioma. In this study, we generated Cic knockout mouse models, mouse neural stem cell (NSC) lines, and isogenic CIC knockout human oligo- dendroglioma cell lines to investigate the effect of Cic inactivation (loss) on neural development and on oligodendroglioma path- ogenesis. We reveal aberrant neural development in vivo and defective NSC differentiation/proliferation in vitro upon Cic knockout, illuminate genes/pathways regulated by Cic via both EGFR-dependent and -independent mechanisms, and present direct evidence to support the tumorigenic role of Cic loss in oligodendroglioma genesis in vivo. Materials and Methods Generation of CIC tm2a (KOMP)Wtsi mice and Cic ox/ox mice KOMP plasmid clone PRPGS00139_B_H08 was transfected into 129/B6 hybrid embryonic stem cells (ESC) by Department of Pathology and the Preston Robert Tisch Brain Tumor Center at Duke, Duke University Medical Center, Durham, North Carolina. Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/). Corresponding Authors: Hai Yan, Duke University Medical Center, The Preston Robert Tisch Brain Tumor Center at Duke, 203 Research Drive, MSRB-1, Room 199B, Durham, NC 27710. Phone: 919-668-7850; Fax: 919-684-8756; E-mail: [email protected]; and Yiping He, Department of Pathology, Duke University Medical Center, DUMC-3156, 199A-MSRB, Research Drive, Durham, NC 27710. Phone: 919-684-4760; Fax: 919-684-8756; E-mail: [email protected] doi: 10.1158/0008-5472.CAN-17-1018 Ó2017 American Association for Cancer Research. Cancer Research www.aacrjournals.org 6097 on November 18, 2020. © 2017 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from Published OnlineFirst September 22, 2017; DOI: 10.1158/0008-5472.CAN-17-1018

Cic Loss Promotes Gliomagenesis via Aberrant Neural Stem ... · In vitro proliferation assay and drug study Cellswereplatedonlaminin-coated96-wellplatesatadensity of 1,000–3,000

  • Upload
    others

  • View
    5

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Cic Loss Promotes Gliomagenesis via Aberrant Neural Stem ... · In vitro proliferation assay and drug study Cellswereplatedonlaminin-coated96-wellplatesatadensity of 1,000–3,000

Molecular and Cellular Pathobiology

Cic Loss Promotes Gliomagenesis via AberrantNeural Stem Cell Proliferation and DifferentiationRui Yang, Lee H. Chen, Landon J. Hansen, Austin B. Carpenter, Casey J. Moure,Heng Liu, Christopher J. Pirozzi, Bill H. Diplas, Matthew S.Waitkus, Paula K. Greer,Huishan Zhu, Roger E. McLendon, Darell D. Bigner, Yiping He, and Hai Yan

Abstract

Inactivating mutations in the transcriptional repressionfactor Capicua (CIC) occur in approximately 50% of humanoligodendrogliomas, but mechanistic links to pathogenesisare unclear. To address this question, we generated Cic-defi-cient mice and human oligodendroglioma cell models.Genetic deficiency in mice resulted in a partially penetrantembryonic or perinatal lethal phenotype, with the productionof an aberrant proliferative neural population in survivinganimals. In vitro cultured neural stem cells derived from Cicconditional knockout mice bypassed an EGF requirement forproliferation and displayed a defect in their potential for

oligodendrocyte differentiation. Cic is known to participate ingene suppression that can be relieved by EGFR signal, but wefound that cic also activated expression of a broad range ofEGFR-independent genes. In an orthotopic mouse model ofglioma, we found that Cic loss potentiated the formation andreduced the latency in tumor development. Collectively, ourresults define an important role for Cic in regulating neuralcell proliferation and lineage specification, and suggest mech-anistic explanations for how CIC mutations may impact thepathogenesis and therapeutic targeting of oligodendroglioma.Cancer Res; 77(22); 6097–108. �2017 AACR.

IntroductionOligodendrogliomas are a histologic type of diffuse glioma

(WHO grade II–III) that account for between 5% and 20% ofgliomas (1) and frequently display characteristic loss of wholechromosome arms 1p and 19q (1p/19q-codeletion; refs. 2–4).Relative to other diffuse glioma subtypes, oligodendrogliomas areassociated with better prognosis due in part to the tumor'sincreased chemosensitivity (5). However, tumor recurrenceremains themost common cause of death for oligodendrogliomapatients, as these tumors often recur or progress to higher grades.Previous studies from ours and other groups have revealed recur-rent inactivatingmutations ofCapicua (CIC, on chromosome arm19q) and far upstream element binding protein (FUBP1, onchromosome arm 1p), identifying these genes as putative tumorsuppressors for oligodendroglioma (6–9). Somatic mutations ofthese two genes, together with IDH1R1/2mutations, 1p/19q loss,and TERT promoter mutations, define oligodendroglioma as aspecific brain tumor subtype (6, 7, 10, 11). Interestingly,whereas TERT promoter and IDHmutation occur across multiplediffuse glioma subtypes, somatic mutations of CIC and FUBP1

are exclusively restricted to oligodendrogliomas, suggesting apossible link of CIC and FUBP1 to oligodendrocyte lineage inneurogenesis and tumorigenesis.

CIC is a member of the Sox-related high-mobility group(HMG) subfamily and is highly conserved among species (12).Previous evidence suggests that cic acts primarily as a transcrip-tional repressor consisting of two highly conserved domains: theHMG box, which mediates DNA binding and nuclear localiza-tion, and a C-terminal motif C1, which co-operates with theHMG-box for DNA binding (12–16). InDrosophila, cic transducesthe signaling of receptor tyrosine kinases (RTK), Torso, and theEGFR, and by doing so regulates cell proliferation and cell lineagespecification during development (13, 15, 17–19).

Besides oligodendrogliomas,CIC is also linked to other cancersand noncancer diseases in which its function as a transcriptionalregulator is believed to be involved (14, 20). However, themechanism underlying the role of CIC inactivation in oligoden-droglioma genesis remains unclear, in part due to the lack ofappropriate animal and cellmodels of oligodendroglioma. In thisstudy, we generated Cic knockout mouse models, mouse neuralstem cell (NSC) lines, and isogenic CIC knockout human oligo-dendroglioma cell lines to investigate the effect ofCic inactivation(loss) on neural development and on oligodendroglioma path-ogenesis. We reveal aberrant neural development in vivo anddefective NSC differentiation/proliferation in vitro upon Cicknockout, illuminate genes/pathways regulated by Cic via bothEGFR-dependent and -independent mechanisms, and presentdirect evidence to support the tumorigenic role of Cic loss inoligodendroglioma genesis in vivo.

Materials and MethodsGeneration of CICtm2a(KOMP)Wtsi mice and Cicflox/flox mice

KOMP plasmid clone PRPGS00139_B_H08 was transfectedinto 129/B6 hybrid embryonic stem cells (ESC) by

Department of Pathology and the Preston Robert Tisch Brain Tumor Center atDuke, Duke University Medical Center, Durham, North Carolina.

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

Corresponding Authors: Hai Yan, Duke University Medical Center, The PrestonRobert Tisch Brain Tumor Center at Duke, 203 Research Drive, MSRB-1, Room199B, Durham, NC 27710. Phone: 919-668-7850; Fax: 919-684-8756; E-mail:[email protected]; and Yiping He, Department of Pathology, Duke UniversityMedical Center, DUMC-3156, 199A-MSRB, Research Drive, Durham, NC 27710.Phone: 919-684-4760; Fax: 919-684-8756; E-mail: [email protected]

doi: 10.1158/0008-5472.CAN-17-1018

�2017 American Association for Cancer Research.

CancerResearch

www.aacrjournals.org 6097

on November 18, 2020. © 2017 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst September 22, 2017; DOI: 10.1158/0008-5472.CAN-17-1018

Page 2: Cic Loss Promotes Gliomagenesis via Aberrant Neural Stem ... · In vitro proliferation assay and drug study Cellswereplatedonlaminin-coated96-wellplatesatadensity of 1,000–3,000

electroporation (21). Targeted ESC clones were screened by PCRand chimeric mice were generated by tetraploid blastocyst com-plementation performed by the Duke Transgenic Mouse Facility.F0 chimeric mice were further bred to C57BL/6 wild-type mice togenerate F1 heterozygous Cictm2a(KOMP)Wtsi mice (Cicnull/þ). F1mice were inbred to generate homozygousCicnull/null mice. In thiscase, viable homozygous Cictm2a(KOMP)Wtsi offspring were under-represented due to the partial embryonic lethality. F1 mice werethen bred to Flp-expressing mice (FLPo-10 from The JacksonLaboratory) to remove the Frt-flanked gene trap cassette in theCictm2a(KOMP)Wtsi allele, resulting in a conditional allele (Cicflox/þ;ref. 21). Further inbreeding of these Cicflox/þ mice allowed gen-eration of homozygous Cicflox/flox mice, which allow furtherconditional knockout in the presence of Cre expression.

GenotypingTissue was lysed in lysis buffer (100 mmol/L Tris-HCl pH 8;

5 mmol/L EDTA pH 8; 0.2% SDS; 200 mmol/L NaCl; ProteinaseK, and water) and DNA was ethanol precipitated. Pellet wasresuspended in 100 mL sterile H2O. One to two microliters ofdilutedDNAwere used for PCR (15mL reaction volume). PCRwasperformed as followed: 95�C for 5minutes, 30 cycles of (95�C for15 seconds, 60�C for 15 seconds, 72�C for 15 seconds), 72�C for5 minutes. Genotyping primers can be found in SupplementaryTable S1. FFPE genomicDNAwas extracted by theGeneReadDNAFFPE Kit (Quagen; catalog no. #180134) according to the man-ufacturer's instructions, and further genotyped to exclude chime-ric expression of cic in brain tissue.

Mouse neonatal NSCs and virus transductionPostnatal day 4 (P4) Cicflox/flox or Cicþ/þ neonatal mice were

anesthetized by hypothermia and then decapitated with sharpscissors. The subventricular zone (SVZ) was microdissected frombrains and dissociated into single cells by trituration. These cellswere then cultured as neurospheres in EGF/FGF containingserum-free neural stem cell medium (StemCell Technologies;catalog no. 05702) following the manufacturer's instructions. Toobtain Cic�/� NSC lines, Cicflox/flox NSC lines were transducedwith Cre-expressing adenovirus (Vectorbiolabs; catalog no. 1769)at a multiplicity of infection (MOI) of 10–100. Removal of thestop cassette was verified by PCR. Cicþ/þNSC lines were similarlytransduced with the Cre-expressing adenovirus and served as thecontrol (Cicþ/þ NSC lines). These cells were used for all in vitroassays (proliferation assay, drug resistance assay, andmicroarray).

In vitro differentiation of NSCs and immunofluorescentstaining

Cic�/� or Cicþ/þ NSCs were cultured for 7 days in 8-chambervessel glass slides (BD Falcon; ref. no. 354108) in either completeproliferation medium (StemCell Technologies; catalog no.05702) supplemented with EGF (20 ng/mL) and FGF (10 ng/mL)or differentiation medium (StemCell Technologies; catalog no.05704). Cells were fixed by formalin, blocked with blockingbuffer (1% BSA, 10% goat serum in PBS), and stained with thefollowing primary antibodies: Nestin (StemCell Technologies;catalog no. 60051, 1:50), GFAP (StemCell Technologies; catalogno. 60048, 1:200orCell Signaling Technology; catalogno. 12389,1:100), Olig2 (Millipore; catalog no. AB9610, 1:100), Ki67 (BD;catalog no. 550609, 1:100), b3 tubulin (StemCell Technologies;catalog no. 01409, 1:1,000), or CNPase (Cell Signaling Technol-ogy; catalog no. 5664, 1:100). After blocking with primary

antibodies, cells were washed three times in PBS and stainedwith appropriate Alexa Fluor secondary antibodies (1:500) for30minutes at room temperature. Cells were washedwith PBS andcounterstained with DAPI before being mounted with mountingmedium (Vectashield H-1400). Slides were imaged with a NikonECLIPSE TE2000-E fluorescent microscope.

In vitro proliferation assay and drug studyCells were plated on laminin-coated 96-well plates at a density

of 1,000–3,000 cells/well. At designated time points, CCK8reagent (Dojindo, CK04-20) was added into each well accordingto manufacturer's instructions. After incubation for 3–5 hours,OD450 absorbance of each well was screened by a microplatereader (Tecan, infinite M200PRO). For cell response to U0126,cells were plated out for 24 hours before U0126 was added intothe wells.

Gene expression microarray analysisFour Cicþ/þ and three Cic�/� neonatal NSC lines (established

from different animals independently) were cultured in 6-wellplates in the standard proliferation medium (StemCell Technol-ogies; catalog no. 05702) with or without EGF (20 ng/mL) for24 hours. RNA was then extracted by RNA Miniprep Kit (Zymo;catalog no. 11-328) following the manufacturer's instruction andsubmitted to Duke Sequencing and Genomic TechnologiesShared Resource for microarray hybridization (Affymetrix Gene-Chip MO 2.0 ST; catalog no. 902118). Gene expression micro-array data have been submitted to the Gene Expression Omnibusdatabase under accession GSE95012. The Robust MultichipAverage (RMA) method was applied for data processing andnormalization by R. For the most differentially expressed genesin response to EGF depletion, fold changes of gene expression(EGF� vs. EGFþ) of both Cicþ/þ and Cic�/� cells were calculatedand averaged (designated as fold change 1 and fold change2), anddifference of fold changes 1 and 2 (Dfold change) were calculated.Most differentially expressed genes in Cic�/� versus Cicþ/þ NSCsin response to EGF depletion were selected if Dfold change (log2ratio) was equal or greater than 1. The enrichment score is �log(P value or EASE score). EASE score (or P value) is a modifiedFisher exact test calculated by the same method as pathwayanalysis by DAVID (22). To Stratify genes regulated differentlyby EGFR and/or Cic, genes with expression absolute fold changeby EGF (EGF� vs. EGFþ, in Cicþ/þ cells) equal or greater than 2and absolute fold change by Cic (Cic�/� vs. Cicþ/þ) in EGF-depleted condition equal or greater than 2 were defined asEGF-regulated Cic downstream targets (group I, 44 genes); geneswith expression absolute fold change by EGF (EGF� vs. EGFþ, inCicþ/þ cells) equal or greater than 2 and absolute fold change byCic (Cic�/� vs. Cicþ/þ) in both EGFþ and EGF-depleted condi-tions smaller than 2 were defined as non-Cic–related EGF down-stream targets (group II, 183 genes); genes with expression abso-lute fold change by Cic (Cic�/� vs. Cicþ/þ) in EGF-depletedcondition equal or greater than 2 and absolute fold change byEGF (EGF� vs. EGFþ, in both Cicþ/þ and Cic�/� cells) smallerthan 2 were defined as non-EGF–regulated Cic downstreamtargets (group III, 192 genes). Cell type–specific genes are definedaccording to http://www.stanford.edu/group/barres_lab/brain_rnaseq.html (23). Fold enrichment of astrocytes, neurons, newlyformed oligodendrocytes, myelinating oligodendrocytes, OPC,microglial, and endothelial cells are calculated as FPKMofone celltype divided by average of FPKM of all other cell types. Top 500

Yang et al.

Cancer Res; 77(22) November 15, 2017 Cancer Research6098

on November 18, 2020. © 2017 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst September 22, 2017; DOI: 10.1158/0008-5472.CAN-17-1018

Page 3: Cic Loss Promotes Gliomagenesis via Aberrant Neural Stem ... · In vitro proliferation assay and drug study Cellswereplatedonlaminin-coated96-wellplatesatadensity of 1,000–3,000

candidate genes of each cell typewere documented. Genes specificto "newly formed oligodendrocytes" and "myelinating oligoden-drocytes" are merged as "oligodendrocytes" for simplicity.

Generation of isogenic CIC knockout HOG cell lines byCRISPR-Cas9

A human oligodendroglioma cell line (HOG) was cultured inIscove's Modified Dulbecco's Medium (Gibco, Invitrogen) sup-plemented with 10% FBS as previously described (24). This cellline is wild type in both IDH1 and IDH2 genes and does not have1p/19q codeletion. The CRISPR/Cas9 system was used for geneediting. pSpCas9(BB)-2A-GFP (PX458) was a gift from FengZhang (Addgene plasmid No. 48138; ref. 25). Double-strandedoligonucleotides were cloned into restriction enzyme BbsI-line-arized PX458 to construct plasmids for CRISPR/Cas9 targeting ofhuman CIC. Three individual sgRNA sequences targeting humanCIC were designed with optimal targeting efficiency andminimaloff-targets (26, 27), and one nontargeting sequence (NTC; ref. 28)was used as a negative control. The sequences of the sgRNAs are:sgRNA-e1: gctactgaccgaacatgccg, sgRNA-e4: ctctaccgcccggaaaacgt,sgRNA-e13: agagctcgtgggcccgtacg, sgRNA-NTC: gtatcctgacctacgcg-ctg. For targeting CIC, three different combinations of sgRNAs(e1þe4, e1þe13, e4þe13) were used for gene knockout via alleledeletions or indels. For transient plasmid transfection, plasmids(two plasmids at 1:1 ratio for achieving the desired gene deletion/mutations) and Transfex (ATCC; catalog no. ACS-4005)weremixedand used for cell transfection according to manufacturer's instruc-tions. One to three days after the transfection, GFPþ cells weresorted via FACS (BD FACSVantage SE cell sorter, Duke CancerInstitute) to obtain a GFPþ population to enrich transfected cells.These cells were further cultured for 1 to 3 days allowing them torecover from the FACS and then serially diluted in 96-well plates forgenerating single-cell clones. Clones were screened for inactivatingalterations by Sanger sequencing. Four subclones bearing differenthomozygous nonsense mutations in CIC were identified andnumbered as HOG_1, HOG_2, HOG_3, and HOG_4, respectively.Ablation of CIC protein level (both long isoformof CIC, CIC-L, andshort isoform of CIC, CIC-S) was validated by CIC Western blotanalysis (Novus Biologicals; catalog no. NB110-59906, 1:1,000).Short tandem repeat (STR) profiling was performed by the CellCulture Facility at Duke to validate that the four isogenic CICknockout lines have consistent STR profiling with the parental line.

Chromatin immunoprecipitation sequencing and data analysisChromatin immunoprecipitation (ChIP) was performed as

described previously (29). Briefly, cells were cross-linked and celllysate was then prepared and subjected to sonication to shear thechromatin. Protein G magnetic beads (New England BioLabs;catalog no. S1430S) conjugated with anti-CIC antibody (NovusBiologicals; catalog no. NB110-59906) were used for immuno-precipitation. ChIP-derived DNA was recovered for library prep-aration. High-throughput sequencing (Illumina HiSeq 4000 sys-tem) was performed by Duke Sequencing and Genomic Technol-ogies Shared Resource, and data analysis was performed by DukeGenomic Analysis and Bioinformatics Shared Resource. ChIP-Seqdata has been submitted to Sequence Read Archive databaseunder the accession number SRP119146. Reads were mapped toGRCh37 version of the human genome using the Bowtie align-ment algorithm (30). For the aligned data, signal tracks weregenerated according to the calculation of the log 10 likelihoodratios between CIC wild-type line ChIP-seq signal and two con-

trols (IgG control andCIC knockout cell line control, respectively)by MACS2 2.1.1. Narrow regions of enrichment were identifiedusingMACS2 2.1.1 with the fragment size of 300 bp based onwetexperiment results and the q-value cutoff as 0.05 (http://github.com/taoliu/MACS/; ref. 31). ChIP-seq signal track was visualizedby the Integrative Genomics Viewer (IGV; refs. 32, 33). Peaklocation data were annotated by PAVIS (34).

Gliomagenesis modeling via orthotopic implantation of NSCsCic�/� and Cicþ/þ NSC lines (one line for each genotype) were

transduced with retrovirus expressing human platelet-derivedgrowth factor subunit B (PDGFB). The retrovirus construct,MigR1-PDGFB, coexpresses GFP and PDGFB from the sametranscript via an internal ribosomal entry site (IRES; Addgene no.27490; ref. 35). MigR1-PDGFB retrovirus particles were producedin 293FT cells, and used to transduce NSCs at an MOI of 10–100.Transduced NSCs were further sorted by FACS (BD FACSVantageSE cell sorter, Duke Cancer Institute) for GFPþ cells. The resultingGFPþ/PDGFB–expressing Cic�/� or Cicþ/þ NSCs were injectedintracranially into the right striatumof 4- to 6-week-oldNSGmiceat a dose of 2.5 � 105 cells per mouse. Mice were sacrificed whenthey showed signs or symptoms of brain abnormalities such asmacrocephaly or lethargy. Brains were harvested, formalin-fixed,and embedded in paraffin for subsequent IHC and immunoflu-orescence staining assays.

The Cancer Genome Atlas data analysisAll The Cancer Genome Atlas (TCGA) gene expression data

were downloaded from the online portal, https://gdc-portal.nci.nih.gov/. CIC copy number and mutation status, as well asIDH1/2 mutation status and histologic subtype were retrievedfrom cBioportal (36, 37), and matched to cases with gene expres-sion data available. The 2016 TCGA Merged Glioma cohort wasused for all analyses (38). All patient cases with complete infor-mation available (i.e., confirmed genotype of IDH, CIC) wereused for each analysis unless otherwise stated, which consistedof 100 oligodendroglioma patients. A total of 748 genes weregenerated for heatmap by using a t test to compare expressionlevels of each gene betweenCIC�/� andCICþ/� cases and filteringfor all genes with a t test P <2.5 � 10�3.

Statistical analysisQuantification of Olig2þ cells in the adult survival Cicþ/þ and

Cic�/� mice were manually performed using two slides from eachanimal (n¼ 2 for each genotype) and a two-tailed unpaired Studentt test was used to assess significant differences between conditions.For all in vitroproliferation assays byCCK8, at least two independentcell lines of each genotype were included and experiments wereconducted at least twice. Two-tailed unpaired Student t test analysiswas performed to compare the Cic�/� lines to the control lines(three technical replicates were performed for each line). qPCR forgene expression analysis was performed with two technical repli-cates and the DDCt method to determine normalized gene expres-sion. For Kaplan–Meier curve, log-rank test P value was calculated.For all microarray data analysis, expression absolute fold changecutoff was set to equal or greater than 2. For all experiments, errorbars represent � SE; �, P < 0.05; ��, P < 0.01; ���, P < 0.001.

Ethics statementAll animal studies were approved by the Duke Institutional

Animal Care and Use Committee, an institution accredited by the

Loss of Cic Promotes Gliomagenesis

www.aacrjournals.org Cancer Res; 77(22) November 15, 2017 6099

on November 18, 2020. © 2017 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst September 22, 2017; DOI: 10.1158/0008-5472.CAN-17-1018

Page 4: Cic Loss Promotes Gliomagenesis via Aberrant Neural Stem ... · In vitro proliferation assay and drug study Cellswereplatedonlaminin-coated96-wellplatesatadensity of 1,000–3,000

Association for Assessment and Accreditation of LaboratoryAnimal Care (AAALAC), International. Protocol number: A072-13-03 and A037-16-02.

ResultsKnockout of Cic results in aberrant hyperproliferative cells inthe adult SVZ

Previous studies have revealed the essential role of CIC indevelopment and the neurodegeneration disorder SCA1 in hu-mans (20). To determine the effects of aCic inactivation in neuraldevelopment, we generated a Cic knockout mouse model using atargeting plasmid, Cictm2a(KOMP)Wtsi (Fig. 1A; SupplementaryFig. S1A; ref. 21). Heterozygous founder mice were crossedto obtain homozygous Cic knockout progeny in which bothalleles are Cictm2a(KOMP)Wtsi (abbreviated as Cicnull/null hereafter)(Supplementary Fig. S1B). In addition, conditional knockoutCicflox/flox mice can be generated by breeding F1 mice to Flp-expressingmice (Fig. 1A; Supplementary Fig. S1C). Among 80 live

pups, only 6 (7.5%) of them, including one runt that died shortlyafter birth, were homozygous knockout, likely due to a partialembryonic or perinatal lethality in Cicnull/null animals.

We first determined the effect of Cic knockout on neuraldevelopment by examining the brain from viable, postnatalCicnull/null animals. As expected, at postnatal day 4 (P4), bothwild-type control (Cicþ/þ) and Cicnull/null mice displayed activepostnatal neurogenesis, as validated by the expression of Ki67around the lateral ventricles (Supplementary Fig. S2). By P28,although Ki67 expression was no longer detected in the SVZ inwild-type animals, as expected (39), active cell proliferation (Ki-67-positive) was readily detectable in Cicnull/null mice (Fig. 1B).Furthermore, the population of GFAP-expressing cells was greatlyelevated in the SVZ of Cicnull/null mice compared with those in thewild-type mice (Fig. 1C). As GFAP-expressing type B radial cellsrepresent the dominant NSC population in the SVZ (40), theseresults suggest Cic loss leads to an aberrant hyperproliferativepopulation in the adult SVZ. Consistent with these hyperproli-ferative phenotypes, DAPI staining of the brain tissue sections at

Figure 1.

Cic knockout results in aberrant hyperproliferative cells in adult SVZ. A, Cic knockout strategy. A gene trap cassette flanked by FRT sites was insertedbetween Cic exon 8 and 9, and exon 9, 10, and 11 were flanked by loxP sites. When treated with Flp recombinase, the gene trap cassette can be removed,resulting in a conditional allele (Cic function is restored). This conditional allele can be further converted into a knockout allele by deleting exon 9, 10, and11 via Cre recombinase. Gray arrows, genotyping primers. B–E, Representative immunofluorescent stained FFPE slides from similar planes of postnatal day 28(P28) Cicþ/þ (n ¼ 2) and Cicnull/null (n ¼ 2) mouse brains for the following markers: Ki67 (�10; B), GFAP (�10; C), Dapi (�20; D), and Olig2 (�10; E).F, Quantifications of Olig2þ cells as shown in E. En2 SA, mouse En2 splicing acceptor; IRES, internal ribosome entry site; LacZ, LacZ reporter; pA,SV40 polyadenylation sequences; hbactP, human b-actin promoter; CC, corpus callosum; LV, lateral ventricle; LSD, lateral septal nucleus.

Yang et al.

Cancer Res; 77(22) November 15, 2017 Cancer Research6100

on November 18, 2020. © 2017 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst September 22, 2017; DOI: 10.1158/0008-5472.CAN-17-1018

Page 5: Cic Loss Promotes Gliomagenesis via Aberrant Neural Stem ... · In vitro proliferation assay and drug study Cellswereplatedonlaminin-coated96-wellplatesatadensity of 1,000–3,000

similar planes revealed thicker lateral ventricular walls inCicnull/null mice compared with those in Cicþ/þ mice (Fig. 1D).Finally, although no differential expression of Nestin or Sox2were observed in the SVZ of Cicþ/þ and Cicnull/null mice, there wasa significantly increase in the number of cells expressing Olig2(Olig2þ) in the dorsal part of lateral septal nucleus (LSD) ofCicnull/null mice (Fig. 1E and F), suggesting an aberrant expansion/migration of the oligodendrocyte progenitor cells (OPC) andproviding a potential link between Cic inactivation and oligo-dendrocyte differentiation. Collectively these results suggestthat Cic inactivation results in aberrant development/expansionof cell populations in the adult SVZ, reminiscent of a preneoplas-tic phenotype.

Cic�/� NSCs bypass the requirement of EGF for proliferationunder hypoxia condition

The observed preneoplastic phenotype raises the possibilitythat Cic knockout leads to aberrant proliferation of NSCs and/orother neural lineage progenitor cells. To overcome the highmortality seen in generating Cicnull/null animals, we crossed theF1 mice (Cicnull/þ) with FLPase-transgenic mice (FLPo-10 fromthe Jackson Laboratory) to achieve Flp-mediated deletion of thegene trap cassette. These resultant mice contained functional Cicalleles (Cicflox), which allowed for subsequent, Cre recombinase-induced Cic knockout (Fig. 1A). NSC cell lines were establishedfrom the SVZ of P4 Cicþ/þ or Cicflox/floxmice followed by adeno-virus-mediated transient Cre expression, resulting in the genera-tion of Cic�/� NSC lines (Supplementary Fig. S1C).

We first analyzed the proliferation of NSCs and found thatunder the standard NSC proliferation medium condition (EGF-supplemented, normoxic culture), both Cicþ/þ and Cic�/� NSCspropagated robustly. As expected, the growth of the NSCs wasdependent on the EGFR signaling, as withdrawal of EGF from the

proliferation medium greatly attenuated the expansion of theNSCs, regardless of their Cic status (Supplementary Fig. S3A andS3B). Although hypoxia in most cases is considered to play animportant role in pathologic conditions such as tumorigenesis, itis alsowell established that a physiologic gradient of oxygen existsin the central nervous system (CNS) and that mild hypoxia is aphysiologic condition in normal brains (41). We therefore testedthe proliferation of NSCs under a mild hypoxic culture environ-ment (3% oxygen). We found that under the standard, EGF-supplemented proliferation media, both Cicþ/þ and Cic�/�NSCsagain propagated robustly (Fig. 2A and B). However, when EGFwas withdrawn from the media, a striking difference in responsewas observed: although the growth ofCicþ/þNSCs was complete-ly attenuated, the growth of Cic�/� NSCs was barely affected bythe EGF withdrawal (Fig. 2A and B; Supplementary Fig. S3C andS3D). In agreement with this resistance to EGFwithdrawal,Cic�/�

NSCs were more resistant to U0126, an inhibitor of the down-stream kinase of EGFR signaling, MEK (Fig. 2C). These results areconsistent with previous findings showing that cic acts as adownstream mediator of the EGFR signaling pathway (13, 17,19). In addition, they provide one potential mechanism under-lying the aberrant proliferation of NSCs in vitro and the preneo-plasia we observed in vivo. The EGF-independent growth ofCic�/�

NSCs under the mild hypoxic condition is particularly interestingas hypoxia/oxygen concentration is a prominent factor in bothphysiologic and tumorigenic conditions in the CNS (41, 42).

Cic�/� NSCs have a compromised oligodendrocytedifferentiation potential and are arrested in an OPC-like state

The aberrant NSC proliferation observed in the above models,together with the knowledge that CIC's inactivating mutationspredominantly occur in human oligodendrogliomas (6–9),prompted us to investigate the effect of Cic knockout on neural

Figure 2.

Cic knockout confers EGFR signaling–independent propagation of NSCs. A, Propagation of two Cicþ/þ NSC lines under mild hypoxic culture condition (3%oxygen) and with or without EGF. B, Propagation of two Cic�/� NSC lines under the same experimental condition as in A. C, Propagation of Cicþ/þ and Cic�/� NSClines in the presence of U0126 for 5 days. Cell numbers of each cell line were normalized to their DMSO-treated controls.

Loss of Cic Promotes Gliomagenesis

www.aacrjournals.org Cancer Res; 77(22) November 15, 2017 6101

on November 18, 2020. © 2017 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst September 22, 2017; DOI: 10.1158/0008-5472.CAN-17-1018

Page 6: Cic Loss Promotes Gliomagenesis via Aberrant Neural Stem ... · In vitro proliferation assay and drug study Cellswereplatedonlaminin-coated96-wellplatesatadensity of 1,000–3,000

lineage specification. To evaluate the differentiation potential ofCicþ/þ and Cic�/� NSCs, we cultured these NSCs under theaforementioned mild hypoxic conditions in the standard NSCproliferation media, or in the differentiation medium. Asexpected, under the proliferation conditions, immunostainingof protein markers revealed that both Cicþ/þ and Cic�/� NSCspositively expressed Nestin and Olig2 (Supplementary Fig. S4A).Furthermore, both NSCs had no detectable expression of GFAP,Tuj1, and CNPase, which are differentiation markers for matureglial cells, neurons, and oligodendrocytes, respectively, confirm-ing their neural stemness/undifferentiated state. After beingcultured in differentiation medium for 7 days, both Cicþ/þ andCic�/� cells lost Nestin expression and displayed strongGFAP expression in the majority of the cells. This suggests thatboth Cicþ/þ and Cic�/� NSCs were readily able to differentiate toglial-like cells (Supplementary Fig. S4B). Interestingly, in responseto the differentiation stimulus, although there were only about4% of Cicþ/þ cells expressing Olig2 (an OPC marker), a signifi-

cantly higher percentage (around 30%) of Cic�/� cells retainedOlig2 expression (Fig. 3A and B; P < 0.001). Furthermore,although cells positively expressing CNPase, a mature oligoden-drocyte marker, were readily detected in the Cicþ/þ cultures, theywere barely detectable in the Cic�/� cultures, even after 7 days ofdifferentiation (Fig. 3C; Supplementary Fig. S5), suggesting thatupon differentiation induction, Cic�/� NSCs are unable to dif-ferentiate intomature oligodendrocytes and insteadblocked in anOPC-like stemness state.

We next performed gene expression profiling analysis to exam-ine how these NSCs may respond differently to the presence/absence of the EGFR signal. Four independent Cicþ/þ and threeindependent Cic�/� NSC lines were cultured in proliferationmediumwith or without EGF for 24 hours before being harvestedfor gene expression profiling analysis. Knocking out of Cic tran-script was further validated by qPCR (Supplementary Fig. S6A).We found that in the control,Cicþ/þNSCs, the withdrawal of EGF(i.e., the absence of EGFR signaling) from the proliferation

Figure 3.

Cic�/� NSCs fail to differentiate into mature oligodendrocyte and are arrested in an OPC-like state. Cicþ/þ and Cic�/� NSCs were maintained in thestandard proliferation media or in differentiation medium for seven days. Immunofluorescent staining was performed. A, Staining (�10) for Olig2 and Dapi.B, Quantification of Olig2þ cells upon differentiation in A. Three randomly chosen fields were quantified for each cell line (two-tailed t test analysis).C, Postdifferentiation staining (�10) for GFAP and CNPase and counterstained with Dapi. D, Most differentially expressed genes in Cic�/� versus Cicþ/þ NSCsin response to EGF depletion. Each column represents fold change (EGF� vs. EGFþ) of gene expression for a sample. Positive value (red) represents higherexpression in EGF-depleted condition compared with EGF-supplemented condition. Genes defined as different neural lineage markers are color-coded on theleft. Enrichment scores [�log(P-value or EASE score)] were calculated for each lineage marker (see Materials and Methods for details). E, A proposedschematic for Cic loss-induced compromised oligodendrocyte differentiation.

Yang et al.

Cancer Res; 77(22) November 15, 2017 Cancer Research6102

on November 18, 2020. © 2017 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst September 22, 2017; DOI: 10.1158/0008-5472.CAN-17-1018

Page 7: Cic Loss Promotes Gliomagenesis via Aberrant Neural Stem ... · In vitro proliferation assay and drug study Cellswereplatedonlaminin-coated96-wellplatesatadensity of 1,000–3,000

medium led to the upregulation of lineage differentiation mar-kers, including those of oligodendrocytes, astrocytes, and neu-rons. This is consistent with previous evidence that withdrawal ofmitogen triggers lineage differentiation of NSCs. Interestingly,when Cic�/� NSCs were subjected to the same EGF withdrawalcondition, a different gene expression pattern was observed: thetop genes that displayed themost distinct expression pattern fromthat of Cicþ/þ NSCs were enriched for oligodendrocyte differen-tiation genes, and the expression of these genes were no longerupregulated in Cic�/� cells upon EGF depletion (Fig. 3D; Sup-plementary Fig. S6B). In particular, among the 73 genes showingthemost distinct response to EGF depletion inCic�/�NSCs versusin Cicþ/þ NSCs, 37 of them are well-established neural lineagemarkers as defined by previous studies (23, 43). Among them, 10are OPC markers and 18 are oligodendrocyte markers, includingMyrf, a master regulator of oligodendrocyte differentiation (44),whereas only 2 are astrocyte markers and 7 are neuron markers.The 18 oligodendrocyte marker genes (such as Enpp6, Plp1, Mag,andMyrf) were all upregulated inCicþ/þNSCs in response to EGFdepletion, but displayed no or barely any differential expressioninCic�/�NSCs under the same experimental condition. A distincttrend was also observed for the 10 OPCmarker genes: 6 of the 10OPCmarker genes were upregulated in Cic�/� cells, including thewell-studied OPC marker gene Pdgfra, but showed no such trendof expression upregulation in Cicþ/þ cells (Supplementary Fig.S6C). These findings support a compromised oligodendrocytedifferentiation program inCic�/�NSCs (Fig. 3E), and suggest thatbesides providing an aberrantmitotic signaling,Cic loss leads to acompromised capability of NSCs to differentiate into matureoligodendrocytes; instead, these Cic�/� cells more likely possessOPC-like stemness, a finding that is consistent with the aberrantOPC-like population observed in vivo, and supports the exclusivelink between CIC mutation and human oligodendrogliomas.

Cic regulates expression of novel downstream target genes thatare independent of EGFR signaling

The aberrant differentiation/propagation phenotypes ofCic�/� NSCs led us to further examine the aforementionedgene expression profiles. We identified three groups of genesthat were distinctly regulated in response to EGF depletion andby Cic status (Fig. 4A; Supplementary Table S2). The genes inthe first group (group I, 44 genes) are EGF-regulated Cicdownstream targets. These genes are regulated by EGF inCicþ/þ NSCs but become no longer or less responsive to EGFin Cic�/� NSCs. This is consistent with the well-establishedfunction of Cic as a downstream mediator of EGFR signaling(13, 17, 19). It is worth noting that 28 of these 44 genes areneural lineage differentiation markers defined by previousstudies (23), and among these 28 genes, 23 are OPC oroligodendrocyte markers. This finding suggests that EGF-dependent Cic-regulated differentiation genes are predomi-nantly related to oligodendrocyte lineage differentiation. Thegenes in the second group (group II, 183 genes) are EGFR-regulated but not Cic-related targets, which display differentialexpression change in response to EGF depletion (EGF� vs.EGFþ) but not affected by Cic status (Cic�/� vs. Cicþ/þ). Thisfinding supports the notion that Cic is only one of the down-stream mediators of the EGFR signaling pathway. Most inter-estingly, the genes in the third and largest group of genes(group III, 192 genes) display a very different differentialexpression pattern from the above two groups—they are notaffected by EGF in either Cicþ/þ or Cic�/� NSCs, but displaydifferential expression in Cic�/� versus Cicþ/þ NSCs, suggestingthey are non-EGF–regulated Cic targets. Therefore, this grouprepresents a previously unrecognized, non-EGFR–controlledrole of Cic. Collectively, these results support that Cic doesact as one downstream mediator of the EGFR signal as

Figure 4.

Cic and EGFR signaling regulateoverlapping yet distinct groups ofgenes. A, Three groups of genes thatare distinctly regulated by EGF and byCic. B, Venn diagram showsoverlapping genes (group I, 44 genes)between EGF downstream targets(227 genes) and Cic downstreamtargets (236 genes). C, Gene setenrichment analysis of mRNA profilesrevealed downregulated oxidativephosphorylation as the top pathwayaltered in Cic�/� NSCs.

Loss of Cic Promotes Gliomagenesis

www.aacrjournals.org Cancer Res; 77(22) November 15, 2017 6103

on November 18, 2020. © 2017 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst September 22, 2017; DOI: 10.1158/0008-5472.CAN-17-1018

Page 8: Cic Loss Promotes Gliomagenesis via Aberrant Neural Stem ... · In vitro proliferation assay and drug study Cellswereplatedonlaminin-coated96-wellplatesatadensity of 1,000–3,000

previously reported; however, there is a major, novel part of Cicfunction that is not regulated by EGFR signaling and has notbeen thoroughly investigated (Fig. 4B).

Genes whose expression is affected by Cic loss are involved ina variety of cellular processes (Supplementary Table S3),highlighting a broad and essential role of Cic in NSCs. Inparticular, in the presence of EGF (as in an actual tumorigeniccondition), gene set enrichment analysis (45, 46) revealed thatin the context of Cic loss, the most affected gene set was that ofoxidative phosphorylation (Fig. 4C). This suggests a likelyreduced use of the tricarboxylic acid cycle in the Cic�/� NSCs,thus providing an explanation for the mild hypoxia-specificpropagation advantage of these cells. In further supporting theoncogenic role of Cic knockout, Cic�/� NSCs also display down-regulated expression of multiple genes with tumor suppressorfunction (e.g., Erdr1, F2rl1, Sparch1, and Ajap1), and increasedexpression of oncogenes (e.g., Hoxa1, Wnt5a, and Sostdc1; Sup-plementary Table S2). Collectively, these results suggest that lossof Cic does not simply resemble a constitutive activation of theEGFR signaling pathway, and support the existence of an EGFR-independent oncogenic effect of Cic loss.

Cic loss leads to downregulated expression of a set of genesand dictates a unique gene expression signature in humanoligodendrogliomas

In analyzing differential gene expression, we noted that regard-less of the absence or presence of EGF, there were more genes thatwere downregulated than genes that were upregulated in Cic�/�

NSCs (104 upregulated vs. 88 downregulated; SupplementaryTable S2). Although this can be due to an indirect effect, it isnevertheless unexpected, as Cic is believed to function as atranscriptional suppressor (13–15, 17, 19, 20). To test whetherthis result is specific only toNSCs,we investigated the effect ofCICknockout in a humanoligodendroglioma cell line,HOG(24).Wedesigned sgRNAs targeting human CIC and utilized the CRISPR/Cas9 gene editing to obtain four isogenicCIC-knockout HOG celllines (Supplementary Fig. S7A). Knocking out of the CIC wasvalidated both by genomic Sanger sequencing and Western Blot(Supplementary Fig. S7B and S7C) Global gene expression anal-ysis revealed that again in the HOGmodel, CIC deletion induceddifferential expression of 154 genes (identified by the cutoff of�2or �0.5 fold). Most remarkably, 90% (139 of 154) of thesegenes were downregulated upon CIC knockout (SupplementaryTable S4).

Anti-CIC ChIP-seq analysis of HOG cell lines identified alarge number of loci that were directly bound by CIC (Fig. 5A;Supplementary Table S5). Combined with the HOGmicroarraydata, it appears that among differentially expressed genesbound by CIC, 83% (68 of 82) of them displayed down-regulated expression upon CIC knockout (examples in Fig.5B; Supplementary Fig. S8; and Supplementary Table S5).Collectively, and together with findings from the NSC models,these results support the notion that in addition to the estab-lished transcriptional suppression activity that is regulated byEGFR signaling, CIC likely also functions in the positive reg-ulation of gene transcription.

Finally, to fully evaluate the effect of CIC's absence on thetranscriptional landscape in oligodendrogliomas, we expandedthis gene expression correlation analysis onto a larger set of genesthat demonstrated significant differential expression between theCIC-mutant versus CIC wild-type group of oligodendrogliomas

reported by the TCGA studies. We selected oligodendrogliomaswith IDH1 mutations to ensure a more meaningful comparisonand segregated oligodendrogliomas into those with CIC-intact(CICþ/þ), with heterozygous CIC loss (1p/19q deletion with CICwild-type, designated as CICþ/�), and with homozygous CICmutations (1p/19q deletion with CIC mutations, designated asCIC�/�).We again found an expression signature that is unique tothe CIC�/� group, whereas the other two groups of oligoden-drogliomas (CICþ/þ and CICþ/�) displayed nondistinguishablegene expression signatures (Fig. 5C). It is worth noting that again60% (452 of 748) of those differentially expressed genes aredownregulated when Cic is lost. Although oligodendrogliomaswith CIC mutations exclusively fall into the group of oligoden-drogliomas that carry deletion of 1p/19q, the chromosomal lossdoes not explain the unique gene expression signature displayedin the CIC�/� group, as those differentially expressed genes arelocated across the genome (Supplementary Fig. S9). Together,these results suggest that although the effects of loss of one copyof CIC (via deletion or chromosomal 19q loss) are more difficultto assess, the complete absence of functionalCIC defines a uniquegene expression signature in oligodendrogliomas.

CIC loss potentiates tumorigenesis in a PDGFB-drivenorthotopic glioma model

The difficulty in obtaining homozygous knockout animalspresents a problem for monitoring the long-term effects ontumorigenesis in vivo. Bearing in mind the established two-hithypothesis in driving tumorigenesis (i.e., a loss of a tumorsuppressor and the gain-of-function of an oncogene), we alter-natively sought to determine the effect ofCic loss in the context ofaberrant oncogenic signaling. Here, we utilized ex vivo culture ofNSCs followed by subsequent orthotopic implantation. Weadapted awell-characterizedmousemodel of oligodendrogliomathat is driven by PDGFB, one of the most common and well-established oncogenes in glioma and one that was previouslyshown to induce glioma-resembling oligodendrogliomas(47, 48). P4 Cicflox/flox or control Cicþ/þ NSCs were harvestedand transduced with Cre-expressing adenovirus to remove theloxP-flanked exons to generateCic�/� or controlCicþ/þNSCs. Theresultant NSCs were transduced with PDGFB-expressing retrovi-rus and implanted into the right striatum of immunocompro-mised NOD/SCID gamma (NSG) mice (Fig. 6A). Mice weresacrificed when they showed signs or symptoms of brain abnor-malities such as macrocephaly or lethargy. We observed that theCicþ/þ,PDGFB-expressingNSCs gave rise to tumorswith amediansurvival of 167 days. Remarkably, in the animals implanted withCic�/�, PDGFB-expressing NSCs, the tumors displayed a shorterlatency with a significantly shorter median survival of 129 days(log-rank P < 0.01; Fig. 6B). Consistent with the previous study(47), both Cicþ/þ and Cic�/�PDGFB-expressing NSCs gave rise tolow-grade, infiltrating tumors. Interestingly, compared with thetumors derived fromCicþ/þNSCs, theCic�/�NSC-derived tumorsdisplayed less dense cellularity, lower mitotic activity, and lessvascular proliferation. In addition, the Cic�/� NSC-derivedtumors had a decreased propensity to infiltrate the subarachnoidspace and form plaque-like tumors, and were more likely toexhibit features found in oligodendrogliomas, including perivas-cular and perineuronal satellitosis, and perinuclear cytoplasmichalos (Fig. 6C). Collectively, these results support that Cic losspotentiates gliomagenesis and facilitates the development oftumors with oligodendroglioma characteristics.

Yang et al.

Cancer Res; 77(22) November 15, 2017 Cancer Research6104

on November 18, 2020. © 2017 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst September 22, 2017; DOI: 10.1158/0008-5472.CAN-17-1018

Page 9: Cic Loss Promotes Gliomagenesis via Aberrant Neural Stem ... · In vitro proliferation assay and drug study Cellswereplatedonlaminin-coated96-wellplatesatadensity of 1,000–3,000

DiscussionIn this study, we provided evidence showing that Cic ablation

causes aberrant development of NSCs and neural lineage cells inthe SVZ, compromises NSC differentiation (in particularly to theoligodendrocyte lineage), confers NSC EGF-independent prolif-eration, and promotes tumorigenesis in an orthotopic gliomamodel.

In our conventional Cic knockout (Cicnull/null) mouse models,although the difficulty in obtaining homozygous knockout ani-mals presents a challenge for studying early neurogenesis indepth, the abnormalities in the SVZ of those viable Cicnull/null

pups were obvious. The SVZ is a heterogeneous reservoir of NSCsand neural progenitor cells, and is at its maximum size at birthwhen the immature brain is at a dynamic state of developmentand undergoing active neurogenesis (49). We observed prolifer-ative cells in the SVZ of both Cicnull/null and Cicþ/þ neonatal mice.

This is indicative of an active and comparable neurogenesis withor without the presence of Cic, likely suggesting that the tran-scriptional suppression role of Cic was blocked at this stage.However, at P28, when the mouse brain is at a mature state andactive neurogenesis is not expected, the continuous presence ofKi67-positive cells was still observed in the SVZ of Cicnull/nullmice,but not in Cicþ/þ mice, suggesting the presence of aberrantproliferative signaling in the adult mouse SVZ when Cic is absent.One prominently proliferative population in the developingbrain is theGFAP-expressing type B radial cells, which are believedto be multipotent NSCs (40). This enlarged GFAP-positive radialcell population in the SVZ, and the thickening of the lateral wallsof the SVZ in the Cicnull/null adult mice compared with that of thecontrol wild-type adult mice, likely indicate an aberrant hyper-proliferative activity in the adult SVZ. Although the exact identityof cell lineages that are most susceptible to Cic loss–drivenoligodendroglioma remains to be determined, these results

Figure 5.

Genes bound by CIC displayed both up- and downregulated expression in CIC-null cells. A, Genome-wide CIC binding loci identified by comparing anti-CICChIP-Seq with control IgG ChIP-Seq signals in the parental HOG cells (top), or by comparing anti-CIC ChIP-seq in the parental HOG cell line with anti-CICChIP-seq in an isogenic CIC�/� HOC derivative line as the negative control (bottom). B, Signal tracks show CIC binding sites for AJAP1, EPDR1, ETV5, PPAP2Bgenes. Statistically significant peaks were identified in regions of transcriptional start site (for EPDR1, PPAP2B, AJAP1), promoter (for ETV5), intron (forETV5), and exon (for AJAP1). Numbers in parentheses following the gene symbols indicate fold change (log2 ratio) of gene expression (KO vs. parental). C, RNA-seqdata analysis of TCGA oligodendroglioma patients. CIC�/� tumors displayed a unique expression signature that is distinct from either CICþ/þ or CICþ/�

tumors (748 genes, 100 patients).

Loss of Cic Promotes Gliomagenesis

www.aacrjournals.org Cancer Res; 77(22) November 15, 2017 6105

on November 18, 2020. © 2017 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst September 22, 2017; DOI: 10.1158/0008-5472.CAN-17-1018

Page 10: Cic Loss Promotes Gliomagenesis via Aberrant Neural Stem ... · In vitro proliferation assay and drug study Cellswereplatedonlaminin-coated96-wellplatesatadensity of 1,000–3,000

provide direct evidence to support a role of Cic ablation ininducing aberrant cell proliferation and elucidate the tumorsuppressive role of Cic.

Our mechanistic studies lead to two intriguing insights regard-ing the biochemical roles of Cic and its tumor suppressor role.First, previous studies using drosophila models suggest Cic func-tions as downstream mediator of EGFR signaling. Generally,certain downstream genes can only be transcribed when EGFRsignaling is activated; the loss of Cic leads to the absence of thissuppression and results in these genes being actively transcribedindependent of EGFR-mediated activation, thus equating the lossof Cic to an activation of EGFR signaling. However, our resultssuggest that Cic loss is distinguishable from EGF-dependentactivation of EGFR signaling pathway, and more likely leads toa cascade of events that are not entirely regulated by the EGFRpathway. Interestingly, we found that in both ourmouseNSC andthe human cell linemodel, a large number of genes bound byCICactually displayed reduced expression when CICwas absent, thusraising an intriguing possibility that CICmay indeed function as atranscriptional activator under some circumstances. Second, intheNSCmodel, the compensationofCicknockout for the absenceof EGF, as assayed by NSC propagation, was only obvious whencells were cultured under mild hypoxic conditions. While we didnot directly assess the situation in an in vivo condition, it suggestsa link between Cic knockout and hypoxia, as mild hypoxia is anintrinsic component to both physiologic neurogenesis and path-ogenic processes such as tumorigenesis (41, 42). Interestingly, aprevious study in Drosophila showed that cic is one of the threegenes that when knocked down, lead to significantly improvedhypoxia tolerance (50). Our finding also raises a possibility thatCic status is a factor for considerationwhendesigning therapeuticsbased on intervention with hypoxic pathways. Further studies inNSCs and in additional tumor models will be necessary to

illustrate the interplay between Cic function and themild hypoxicmicroenvironments, and the therapeutic implication.

The tumor suppressor function of Cic is further underscored bythe broad scope of genes that display differential expression uponCic knockout, including the upregulation of potential oncogenesand the downregulation of those with tumor suppressive roles.This is consistent with recent studies indicating that CIC muta-tions are associated with a more aggressive subtype in oligoden-drogliomas (51, 52). Among these Cic-regulated genes, the onesthat are essential in mediating the pathogenic role of Cic deletionare still unclear. One intriguing candidate is Ajap1, which isinvolved in numerous cellular processes associated with tumor-igenesis and is frequently lost or epigenetically silenced in gliomas(53, 54). Further studies are needed to determine key downstreammediator genes, among numerous other downstream genes iden-tified as differentially expressed in the Cic�/� cells, which areresponsible for the tumorigenic role of Cic loss.

Our in vivo gliomagenesis modeling study raises several impor-tant questions. First, it remains to be determined which specificpopulations/lineages of cells aremost susceptible to Cic deletion-driven transformation. Second, it is noted that although we usedPDGFB as anoncogenic driver for the current study to illustrate thetumorigenic role of the Cic deletion, it is almost certain that CICdeletion can cooperatewithmultiple genetic alterations in drivingoligodendrogliomas, including thosemutations cooccurringwithCIC mutations, such as the IDH1/2 mutations (6, 7). In thisregard, it is also noted that the HOG cell line that was used inthis study, while initially derived from an oligodendrogliomapatient, lacks the IDH1/2 mutations and 1p/19q codeletion(24, 55).Our results should therefore be interpreted in the contextof this limitation, and future studies using more geneticallyrepresentative human oligodendroglioma cell models may benecessary to further define the role of CIC in oligodendroglioma.

Figure 6.

Cic deletion potentiates PDGFB-driven gliomagenesis. A, PDGFB-expressing NSCs with genotype of either Cicþ/þ or Cic�/� were implanted into the rightcortex of 4- to 6-week-old NSG mice (n ¼ 7 for each cell line). B, Kaplan–Meier analysis revealed that Cic deletion event in PDGFB-expressing NSCswas associated with a shorter median survival of implanted animals when compared with the wild-type control group (129 days vs. 167 days, log-ranktest P ¼ 0.0013). C, Hematoxylin and eosin staining of xenograft sections: vascular proliferation in Cicþ/þ tumors (left; arrows; �40); perivascularsatellitosis (middle; arrowheads; �20) and perineuronal satellitosis (arrows) in Cic�/� tumors; perinuclear halos and bland vessels that lack evidenceof proliferation (right; arrows; �40) in Cic�/� tumors.

Yang et al.

Cancer Res; 77(22) November 15, 2017 Cancer Research6106

on November 18, 2020. © 2017 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst September 22, 2017; DOI: 10.1158/0008-5472.CAN-17-1018

Page 11: Cic Loss Promotes Gliomagenesis via Aberrant Neural Stem ... · In vitro proliferation assay and drug study Cellswereplatedonlaminin-coated96-wellplatesatadensity of 1,000–3,000

Finally, it is notable thatCicdeletion canpotentiate gliomagenesisthat is driven by PDGFR signaling, which is itself a potentoncogenic driver for gliomas (47, 48). As PDGF receptor kinaseinhibitors, for example, imatinib, have been under active devel-opment as a strategy for glioma treatment, future studies will berequired to determine how the status of CIC affects tumor cellresponse to such inhibitors. TheCic knockoutmouse and cell linemodels (as described in the current study) provide valuable toolsfor addressing these questions.

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

Authors' ContributionsConception and design: R. Yang, A.B. Carpenter, D.D. Bigner, Y. He, H. YanDevelopment of methodology: R. Yang, L.H. Chen, A.B. Carpenter, C.J. Moure,C.J. Pirozzi, M.S. Waitkus, P.K. Greer, Y. HeAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): R. Yang, L.J. Hansen, A.B. Carpenter, C.J. Moure,C.J. Pirozzi, P.K. Greer, H. Zhu, R.E. McLendon, D.D. Bigner, Y. HeAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): R. Yang, L.H. Chen, L.J. Hansen, A.B. Carpenter,M.S. Waitkus, R.E. McLendon, Y. HeWriting, review, and/or revision of the manuscript: R. Yang, L.J. Hansen,A.B. Carpenter, C.J. Pirozzi, B.H. Diplas, M.S. Waitkus, R.E. McLendon,D.D. Bigner, Y. He, H. Yan

Administrative, technical, or material support (i.e., reporting or organizingdata, constructing databases): R. Yang, H. Liu, P.K. Greer, D.D. Bigner, Y. HeStudy supervision: R. Yang, Y. He, H. Yan

AcknowledgmentsWe thank Cheryl Bock of Duke Transgenic and Knockout Mouse Shared

Resource for generating the knockout mouse models, Duke Sequencing andGenomic Technologies Shared Resource for performing the microarrayexperiments and sequencing for ChIP samples, and David Corcoran ofDuke Genomic Analysis and Bioinformatics Shared Resource for ChIP-seqdata analysis. We thank Jenna Lewis for editorial revisions to themanuscript.

Grant SupportThis work was supported by the NCI grant R01CA140316 to H. Yan. Y. He

was supported by an American Association for Cancer Research (AACR)-AflacCareer Development Award (grant number 13-20-10-HE), a National Com-prehensive Cancer Network Young Investigator Award, and the Circle of ServiceFoundation.

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 indicatethis fact.

Received April 11, 2017; revised July 21, 2017; accepted September 8, 2017;published OnlineFirst September 22, 2017.

References1. Ostrom QT, Gittleman H, Fulop J, Liu M, Blanda R, Kromer C, et al.

CBTRUS statistical report: primary brain and central nervous systemtumors diagnosed in the United States in 2008–2012. Neuro-oncology2015;17 Suppl 4:iv1–iv62.

2. Bromberg JE, van den Bent MJ. Oligodendrogliomas: molecular biologyand treatment. Oncologist 2009;14:155–63.

3. Maintz D, Fiedler K, Koopmann J, Rollbrocker B, Nechev S, Lenartz D, et al.Molecular genetic evidence for subtypes of oligoastrocytomas. J Neuro-pathol Exp Neurol 1997;56:1098–104.

4. Smith JS, Perry A, Borell TJ, Lee HK, O'Fallon J, Hosek SM, et al. Alterationsof chromosome arms 1p and 19q as predictors of survival in oligoden-drogliomas, astrocytomas, and mixed oligoastrocytomas. J Clin Oncol2000;18:636–45.

5. SuzukiH, Aoki K, Chiba K, Sato Y, ShiozawaY, Shiraishi Y, et al.Mutationallandscape and clonal architecture in grade II and III gliomas. Nat Genet2015;47:458–68.

6. Jiao Y, Killela PJ, Reitman ZJ, Rasheed AB, Heaphy CM, de Wilde RF, et al.Frequent ATRX, CIC, FUBP1 and IDH1 mutations refine the classificationof malignant gliomas. Oncotarget 2012;3:709–22.

7. Bettegowda C, Agrawal N, Jiao Y, Sausen M, Wood LD, Hruban RH, et al.Mutations in CIC and FUBP1 contribute to human oligodendroglioma.Science 2011;333:1453–5.

8. Sahm F, Koelsche C, Meyer J, Pusch S, Lindenberg K, Mueller W, et al. CICand FUBP1 mutations in oligodendrogliomas, oligoastrocytomas andastrocytomas. Acta Neuropathol 2012;123:853–60.

9. Yip S, Butterfield YS, Morozova O, Chittaranjan S, Blough MD, An J, et al.Concurrent CIC mutations, IDH mutations, and 1p/19q loss distinguisholigodendrogliomas from other cancers. J Pathol 2012;226:7–16.

10. Cancer Genome Atlas Research Network, Brat DJ, Verhaak RG, Aldape KD,YungWK, Salama SR, et al. Comprehensive, integrative genomic analysis ofdiffuse lower-grade gliomas. N Engl J Med 2015;372:2481–98.

11. Killela PJ, Reitman ZJ, Jiao Y, Bettegowda C, Agrawal N, Diaz LA Jr, et al.TERT promoter mutations occur frequently in gliomas and a subset oftumors derived from cells with low rates of self-renewal. Proc Natl Acad SciU S A 2013;110:6021–6.

12. Lee CJ, ChanWI, Cheung M, Cheng YC, Appleby VJ, Orme AT, et al. CIC, amember of a novel subfamily of the HMG-box superfamily, is transientlyexpressed in developing granule neurons. Brain Res Mol Brain Res2002;106:151–6.

13. Jimenez G, Guichet A, Ephrussi A, Casanova J. Relief of gene repression bytorso RTK signaling: role of capicua in Drosophila terminal and dorso-ventral patterning. Genes Dev 2000;14:224–31.

14. Kawamura-SaitoM, Yamazaki Y, Kaneko K, Kawaguchi N, KandaH,MukaiH, et al. Fusion between CIC and DUX4 up-regulates PEA3 family genes inEwing-like sarcomas with t(4;19)(q35;q13) translocation. HumMol Gen-et 2006;15:2125–37.

15. Astigarraga S, Grossman R, Diaz-Delfin J, Caelles C, Paroush Z, Jimenez G.AMAPKdocking site is critical for downregulation ofCapicua by Torso andEGFR RTK signaling. EMBO J 2007;26:668–77.

16. Fores M, Simon-Carrasco L, Ajuria L, Samper N, Gonzalez-Crespo S,Drosten M, et al. A new mode of DNA binding distinguishes Capicuafrom other HMG-box factors and explains its mutation patterns in cancer.PLoS Genet 2017;13:e1006622.

17. Roch F, Jimenez G, Casanova J. EGFR signalling inhibits Capicua-depen-dent repression during specification of Drosophila wing veins. Develop-ment 2002;129:993–1002.

18. LohrU,ChungHR, BellerM, JackleH. Antagonistic actionof Bicoid and therepressor Capicua determines the spatial limits of Drosophila head geneexpression domains. Proc Natl Acad Sci U S A 2009;106:21695–700.

19. Ajuria L, Nieva C, Winkler C, Kuo D, Samper N, Andreu MJ, et al. CapicuaDNA-binding sites are general response elements for RTK signaling inDrosophila. Development 2011;138:915–24.

20. Lam YC, Bowman AB, Jafar-Nejad P, Lim J, Richman R, Fryer JD, et al.ATAXIN-1 interacts with the repressor Capicua in its native complex tocause SCA1 neuropathology. Cell 2006;127:1335–47.

21. Skarnes WC, Rosen B, West AP, Koutsourakis M, Bushell W, Iyer V, et al. Aconditional knockout resource for the genome-wide study of mouse genefunction. Nature 2011;474:337–42.

22. HuangdaW, ShermanBT, Lempicki RA. Systematic and integrative analysisof large gene lists using DAVID bioinformatics resources. Nat Protoc2009;4:44–57.

23. Zhang Y, Chen K, Sloan SA, Bennett ML, Scholze AR, O'Keeffe S, et al. AnRNA-sequencing transcriptome and splicing database of glia, neurons, andvascular cells of the cerebral cortex. J Neurosci 2014;34:11929–47.

24. Post GR, Dawson G. Characterization of a cell line derived from a humanoligodendroglioma. Mol Chem Neuropathol 1992;16:303–17.

25. Ran FA, Hsu PD, Wright J, Agarwala V, Scott DA, Zhang F. Genomeengineering using the CRISPR-Cas9 system. Nat Protoc 2013;8:2281–308.

Loss of Cic Promotes Gliomagenesis

www.aacrjournals.org Cancer Res; 77(22) November 15, 2017 6107

on November 18, 2020. © 2017 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst September 22, 2017; DOI: 10.1158/0008-5472.CAN-17-1018

Page 12: Cic Loss Promotes Gliomagenesis via Aberrant Neural Stem ... · In vitro proliferation assay and drug study Cellswereplatedonlaminin-coated96-wellplatesatadensity of 1,000–3,000

26. Doench JG, Fusi N, Sullender M, Hegde M, Vaimberg EW, Donovan KF,et al. Optimized sgRNA design to maximize activity and minimize off-target effects of CRISPR-Cas9. Nat Biotechnol 2016;34:184–91.

27. Hsu PD, Scott DA, Weinstein JA, Ran FA, Konermann S, Agarwala V, et al.DNA targeting specificity of RNA-guided Cas9 nucleases. Nat Biotechnol2013;31:827–32.

28. Mavrakis KJ, McDonald ER III, Schlabach MR, Billy E, Hoffman GR,deWeck A, et al. Disordered methionine metabolism in MTAP/CDKN2A-deleted cancers leads to dependence on PRMT5. Science 2016;351:1208–13.

29. GuoC,ChangCC,WorthamM,Chen LH, KernagisDN,QinX, et al. Globalidentification ofMLL2-targeted loci revealsMLL2's role in diverse signalingpathways. Proc Natl Acad Sci U S A 2012;109:17603–8.

30. Langmead B, Trapnell C, Pop M, Salzberg SL. Ultrafast and memory-efficient alignment of short DNA sequences to the human genome.Genome Biol 2009;10:R25.

31. Zhang Y, Liu T, Meyer CA, Eeckhoute J, Johnson DS, Bernstein BE, et al.Model-based analysis of ChIP-Seq (MACS). Genome Biol 2008;9:R137.

32. Robinson JT, Thorvaldsdottir H, Winckler W, GuttmanM, Lander ES, GetzG, et al. Integrative genomics viewer. Nat Biotechnol 2011;29:24–6.

33. Thorvaldsdottir H, Robinson JT, Mesirov JP. Integrative Genomics Viewer(IGV): high-performance genomics data visualization and exploration.Brief Bioinform 2013;14:178–92.

34. Huang W, Loganantharaj R, Schroeder B, Fargo D, Li L. PAVIS: a toolfor Peak Annotation and Visualization. Bioinformatics 2013;29:3097–9.

35. Pear WS, Miller JP, Xu L, Pui JC, Soffer B, Quackenbush RC, et al. Efficientand rapid induction of a chronic myelogenous leukemia-like myelopro-liferative disease in mice receiving P210 bcr/abl-transduced bone marrow.Blood 1998;92:3780–92.

36. Gao J, Aksoy BA, Dogrusoz U, Dresdner G, Gross B, Sumer SO, et al.Integrative analysis of complex cancer genomics and clinical profiles usingthe cBioPortal. Sci Signal 2013;6:pl1.

37. Cerami E,Gao J,DogrusozU,Gross BE, Sumer SO, Aksoy BA, et al. The cBiocancer genomics portal: an open platform for exploring multidimensionalcancer genomics data. Cancer Discov 2012;2:401–4.

38. CeccarelliM, Barthel FP,Malta TM, Sabedot TS, Salama SR,MurrayBA, et al.Molecular profiling reveals biologically discrete subsets and pathways ofprogression in diffuse glioma. Cell 2016;164:550–63.

39. Tanaka A, Watanabe Y, Kato H, Araki T. Immunohistochemical changesrelated to ageing in the mouse hippocampus and subventricular zone.Mech Ageing Dev 2007;128:303–10.

40. Doetsch F, Caille I, Lim DA, Garcia-Verdugo JM, Alvarez-Buylla A. Sub-ventricular zone astrocytes are neural stem cells in the adult mammalianbrain. Cell 1999;97:703–16.

41. Erecinska M, Silver IA. Tissue oxygen tension and brain sensitivity tohypoxia. Respir Physiol 2001;128:263–76.

42. Ljungkvist AS, Bussink J, Kaanders JH, van der Kogel AJ. Dynamics of tumorhypoxia measured with bioreductive hypoxic cell markers. Radiat Res2007;167:127–45.

43. Cahoy JD, Emery B, Kaushal A, Foo LC, Zamanian JL, Christopherson KS,et al. A transcriptome database for astrocytes, neurons, and oligodendro-cytes: a new resource for understanding brain development and function.J Neurosci 2008;28:264–78.

44. Emery B, Agalliu D, Cahoy JD, Watkins TA, Dugas JC, Mulinyawe SB, et al.Myelin gene regulatory factor is a critical transcriptional regulator requiredfor CNS myelination. Cell 2009;138:172–85.

45. SubramanianA, TamayoP,Mootha VK,Mukherjee S, Ebert BL,GilletteMA,et al. Gene set enrichment analysis: a knowledge-based approach forinterpreting genome-wide expression profiles. Proc Natl Acad Sci U S A2005;102:15545–50.

46. Mootha VK, Lindgren CM, Eriksson KF, Subramanian A, Sihag S, Lehar J,et al. PGC-1alpha-responsive genes involved in oxidative phosphorylationare coordinately downregulated in human diabetes. Nat Genet2003;34:267–73.

47. Dai C, Celestino JC, Okada Y, Louis DN, Fuller GN, Holland EC. PDGFautocrine stimulation dedifferentiates cultured astrocytes and inducesoligodendrogliomas and oligoastrocytomas from neural progenitors andastrocytes in vivo. Genes Dev 2001;15:1913–25.

48. Uhrbom L, Hesselager G, Nister M, Westermark B. Induction of braintumors inmice using a recombinant platelet-derived growth factor B-chainretrovirus. Cancer Res 1998;58:5275–9.

49. Buono KD, Goodus MT, Guardia Clausi M, Jiang Y, Loporchio D, LevisonSW. Mechanisms of mouse neural precursor expansion after neonatalhypoxia-ischemia. J Neurosci 2015;35:8855–65.

50. Udpa N, Ronen R, Zhou D, Liang J, Stobdan T, Appenzeller O, et al. Wholegenome sequencing of Ethiopian highlanders reveals conserved hypoxiatolerance genes. Genome Biol 2014;15:R36.

51. Chan AK, Pang JC, Chung NY, Li KK, PoonWS, Chan DT, et al. Loss of CICand FUBP1 expressions are potential markers of shorter time to recurrencein oligodendroglial tumors. Mod Pathol 2014;27:332–42.

52. Gleize V, Alentorn A, Connen de Kerillis L, Labussiere M, Nadaradjane AA,Mundwiller E, et al. CIC inactivating mutations identify aggressive subsetof 1p19q codeleted gliomas. Ann Neurol 2015;78:355–74.

53. McDonald JM, Dunlap S, Cogdell D, Dunmire V,Wei Q, Starzinski-PowitzA, et al. The SHREW1 gene, frequently deleted in oligodendrogliomas,functions to inhibit cell adhesion and migration. Cancer Biol Ther 2006;5:300–4.

54. Lin N, Di C, Bortoff K, Fu J, Truszkowski P, Killela P, et al. Deletion orepigenetic silencing of AJAP1 on 1p36 in glioblastoma. Mol Cancer Res2012;10:208–17.

55. Reitman ZJ, Jin G, Karoly ED, Spasojevic I, Yang J, Kinzler KW, et al.Profiling the effects of isocitrate dehydrogenase 1 and 2 mutations on thecellular metabolome. Proc Natl Acad Sci U S A 2011;108:3270–5.

Cancer Res; 77(22) November 15, 2017 Cancer Research6108

Yang et al.

on November 18, 2020. © 2017 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst September 22, 2017; DOI: 10.1158/0008-5472.CAN-17-1018

Page 13: Cic Loss Promotes Gliomagenesis via Aberrant Neural Stem ... · In vitro proliferation assay and drug study Cellswereplatedonlaminin-coated96-wellplatesatadensity of 1,000–3,000

2017;77:6097-6108. Published OnlineFirst September 22, 2017.Cancer Res   Rui Yang, Lee H. Chen, Landon J. Hansen, et al.   Proliferation and Differentiation

Loss Promotes Gliomagenesis via Aberrant Neural Stem CellCic

  Updated version

  10.1158/0008-5472.CAN-17-1018doi:

Access the most recent version of this article at:

  Material

Supplementary

  http://cancerres.aacrjournals.org/content/suppl/2017/09/22/0008-5472.CAN-17-1018.DC1

Access the most recent supplemental material at:

   

   

  Cited articles

  http://cancerres.aacrjournals.org/content/77/22/6097.full#ref-list-1

This article cites 55 articles, 21 of which you can access for free at:

  Citing articles

  http://cancerres.aacrjournals.org/content/77/22/6097.full#related-urls

This article has been cited by 9 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/77/22/6097To request permission to re-use all or part of this article, use this link

on November 18, 2020. © 2017 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst September 22, 2017; DOI: 10.1158/0008-5472.CAN-17-1018