Research MGMT-Independent Temozolomide Resistance in ... · 29.10.2010 · MGMT-Independent Temozolomide Resistance in Pediatric Glioblastoma Cells Associated with a PI3-Kinase–Mediated

Ther

MGGlioHOX

NathaMartaRui M

Abst

Intro

Glionervousentiamainssurvivhas imof trearadiotuse ofa newsequenin a m

AuthorCentre3PaediSuttonCancer5Life aMinho,

Corresof CanUnitedE-mail:

doi: 10

©2010

www.a

Published OnlineFirst October 8, 2010; DOI: 10.1158/0008-5472.CAN-10-1250
Published OnlineFirst on November 2, 2010 as 10.1158/0008-5472.CAN-10-1250
Canceresearch

apeutics, Targets, and Chemical Biology

MT-Independent Temozolomide Resistance in Pediatricblastoma Cells Associated with a PI3-Kinase–Mediated

R

/Stem Cell Gene Signature

lie Gaspar1,2,3, Lynley Marshall1,2,4, Lara Perryman1, Dorine A. Bax1, Suzanne E. Little1,
Viana-Pereira1,5, Swee Y. Sharp2, Gilles Vassal4, Andrew D.J. Pearson1,3, . Reis5, Darren Hargrave3, Paul Workman2, and Chris Jones1
ractSen

resistaase Mmodelin vitroMGMTexceptofHOXway inhigh-gsultedgenes

tial treatmodest imp

s' Affiliatiofor Canceratric Onco, United K, Institut dend Health SBraga, Port

ponding Aucer ResearKingdom. Pchris.jones@

.1158/0008-

American A

acrjourna

Downloa

sitivity to temozolomide is restricted to a subset of glioblastoma patients, with the major determinant ofnce being a lack of promoter methylation of the gene encoding the repair protein DNA methyltransfer-GMT, although other mechanisms are thought to be active. There are, however, limited preclinical data insystems derived from pediatric glioma patients. We screened a series of cell lines for temozolomide efficacy, and investigated the differential mechanisms of resistance involved. In the majority of cell lines, a lack ofpromoter methylation and subsequent protein overexpression were linked to temozolomide resistance. Anionwas the pediatric glioblastoma lineKNS42. Expression profiling data revealed a coordinated upregulationgene expression in resistant lines, especially KNS42, whichwas reversed by phosphoinositide 3-kinase path-hibition. High levels of HOXA9/HOXA10 gene expression were associated with a shorter survival in pediatricrade glioma patient samples. Combination treatment in vitro of pathway inhibition and temozolomide re-in a highly synergistic interaction in KNS42 cells. The resistance gene signature further included contiguouswithin the 12q13-q14 amplicon, including the Akt enhancer PIKE, significantly overexpressed in the KNS42hese cells were also highly enriched for CD133 and other stem cell markers. We have thus shown an in vitro
line. T
link between phosphoinositide 3-kinase–mediated HOXA9/HOXA10 expression, and a drug-resistant, progenitorcell phenotype in MGMT-independent pediatric glioblastoma. Cancer Res; 70(22); OF1–10. ©2010 AACR.

Tembioavaschedsurvivthe suenzymref. 4)adducthereMGMmethysilenc

duction

blastoma is the most common tumor of the centrals system, affecting patients of all ages, and being es-lly refractory to treatment; the clinical outcome re-dismal regardless of age at diagnosis. The median

al of a patient with glioblastoma is 15 months, and thisproved little in the last four decades (1). The mainstaystment during this time have been surgical resection andherapy, often with nitrosurea-based chemotherapy. Theadjuvant temozolomide has more recently emerged asstandard of care in glioblastoma, with concurrent and

ent in the initial therapy of patients resultingrovement in median survival (2).

Defto resiare elebase ehanceby me(PARPThe

glioblaIn thepredicsurvivnot secontr

ns: 1Paediatric Oncology, and 2Cancer Research UKTherapeutics, The Institute of Cancer Research, andlogy, The Royal Marsden NHS Foundation Trust,ingdom; 4Pharmacology and New Treatments ofCancérologie Gustav Roussy, Villejuif, France; andcience Research Institute (ICVS), Universidade dougal

thor: Chris Jones, Paediatric Oncology, The Institutech, 15 Cotswold Road, Sutton, Surrey, SM2 5NG,hone: 44-0-20-8722-4416; Fax: 44-0-20-8722-4321;icr.ac.uk.

5472.CAN-10-1250

ssociation for Cancer Research.

ls.org

Research. on March 23, 202cancerres.aacrjournals.org ded from

ozolomide is a DNA methylating agent that is orallyilable, crosses the blood:brain barrier, and exhibitsule-dependent antitumor activity (3). The improvedal benefits in glioblastoma are largely restricted tobset of patients lacking expression of the DNA repaire O6-methylguanine-DNA-methyl-transferase (MGMT;. Temozolomide induces cytotoxic O6-guanine methylts that are removed directly by functional MGMT,by producing drug resistance. Downregulation ofT usually occurs in tumors by gene promoter hyper-lation, in which >50% methylation has been shown toe gene expression (5).iciencies in DNAmismatch repair (MMR) are also linkedstance to alkylating agents such as temozolomide (6), asvated levels of Ape1/Ref-1, a major component of thexcision repair (BER; ref. 7) system, with attempts to en-temozolomide-induced cytotoxicity by disrupting BERans of inhibition of poly-(ADP-ribose)-polymerase) proving effective in vitro and in vivo (8).vast majority of the above work has taken place in adultstoma and preclinical models derived from adult patients.pediatric setting, MGMT promoter hypermethylationts for response to alkylating agents (9); however, theal of children treated with adjuvant temozolomide does
em to be improved when compared with historicalols (10–14). The mechanisms of drug resistance in
OF1

1. © 2010 American Association for Cancer

http://cancerres.aacrjournals.org/

pediatdue toWe scrtemoztial mvolvemMGMT

Mate

Cell cAdu

U118MKNS42

turedtion aconsisplemegrowt

GrowTem

benzyPharmterminas pre

Gaspar et al.

CanceOF2


ric high-grade glioma are poorly understood, in partthe lack of availability of suitable models of the disease.eened a series of pediatric and adult glioma cell lines forolomide efficacy in vitro, and investigated the differen-echanisms of resistance involved, highlighting the in-ent in pediatric cells of processes outside of the usual/MMR/BER axis.

rials and Methods

ulturelt glioblastoma cell lines A172, LN229, SF268, U87MG,
G, and U138MG, and pediatric glioma cell lines SF188, temoz
r Res; 70(22) November 15, 2010

Research. on March 23, 202cancerres.aacrjournals.org Downloaded from

as previously described (15). For the spheroid forma-ssay, cells were grown in neurosphere medium, whichted of NDiff RHB-A medium (Stem Cell Sciences) sup-nted with epidermal growth factor and fibroblasth factor 2, each at 20 ng/mL.

th inhibition studiesozolomide was obtained from Apin Chemicals, O6-lguanine from Calbiochem, and PI-103 from Pirameda or synthesized in-house. Growth inhibition was de-ed using the sulforhodamine B (16) or MTS (17) assayviously described. To attempt reversion of resistance to
olomide, O6-benzylguanine was added at the highest
of cell growth inhibition,

ureadozoMT2, U68)479s wecytorhoplottforrelatthylaMSressozoepts, wnotistanF18tedloguwinureistanS42detssaozolomide is on a log10 scale.

, UW479, Res259, and Res186 were obtained and cul- nontoxic concentration (10–15%

FigandtemMGA17SF2UWcellandsulfareblotcormeandexptemexccelldoresC, StreaanashonatresKNwasB atem

1. © 2010 American Associa

1. Sensitivity of pediatricult glioma cell lines tolomide and relationship tostatus. A, adult (LN229,118MG, U138MG, U87MG,and pediatric (SF188, KNS42,, Res186, Res259) gliomare treated with temozolomide,otoxicity was assessed by thedamine B assay. IC50 valuesedona log10scale.B,WesternMGMT protein expressioned with extent of promotertion as assessed by MS-PCR-MLPA. In most casesion correlates withlomide resistance, with theion of U87MG and KNS42hich are hypermethylated,express the protein, and aret to temozolomide.8 and KNS42 cells werewith MGMT substratee O6-benzylguanine,g the MGMT-dependentof temozolomide (TMZ)ce in SF188, but notcells. Growth inhibitionerminedby thesulforhodaminey. Concentration of

Cancer Research

tion for Cancer


20 μmwere talonecombiand Tthree

PromCell

Epitections.was pligatioout asturer'smethyof 5-Acontromina I

WesteImm

(15) u(1:500

gies),ence),(1:1,00(Novup110δphospGAPD

mRNACell

otide aaccessformeGenePsoftwatifiedmit.edenrichsubsetenrich24-houIllumiE-TABGenomcross-calculclinicadatasepediatsion n

ImmuCD

cytomVantacence(AC13tively.tems)(DAPIcarrie

Resu

Sensiin viton Mof prWe

our pRes186U138MRes259lues ocells wwith I

Figureadult gand BEUW479expressinvolve

Temozolomide Resistance in Pediatric Glioma

www.a


ol/L). For the assessment of combination effects, cellsreated with increasing concentrations of drugs eitheror concurrently at their equipotent molar ratio andnation indices were calculated by the method of Choualalay (18). All values are given as mean ± SD of at leastindependent experiments.

oter methylation analysisline DNA was treated with sodium bisulphite using thet kit (Qiagen) according to the manufacturer's instruc-Methylation-specific (MS) PCR for theMGMT promotererformed as described previously (19). MS–multiplexn-dependent probe amplification (MLPA) was carriedpreviously reported (15) according to the manufac-instructions (MRC-Holland; ref. 20). HOXA9/HOXA10lation was assessed by comparing expression profilesza-2′-deoxycytidine–treated cells with vehicle-treatedls on Illumina Human-6 v2 Expression BeadChips (Illu-nc.), ArrayExpress accession number E-TABM-858.

rn blot analysisunodetection was performed as previously described

d in both pathways. *, temozolomide-resistant cell lines.

sing antibodies against MGMT (1:500; Zymed), MLH1; Pharmingen), MLH3 (1:500; Santa Cruz Biotechnolo-

Wespecif

acrjournals.org


MSH2 (1:500; Calbiochem), MSH3 (1:250; BD Biosci-MSH6, PMS2 (both 1:500; BD Bioscience), PARP1/20; Cell Signaling), XRCC1 (1:500; Cell Signaling), APE1s Biochemicals), p85, p110α (Cell Signaling), p110β,(Santa Cruz), PIKE-A/PIKE-L (all 1:1,000; Abcam),ho-AktSer473, Akt (both 1:1,000; Cell Signaling), andH (1:2,000; Chemicon).

expression profiling analysisline expression profiling by Affymetrix U133 oligonucle-rrays has been previously described (ref. 15; ArrayExpression number E-TABM-579). Supervised analysis was per-d using an absolute signal-to-noise metric of >1.5 inattern software (http://www.broad.mit.edu/cancer/re/genepattern/). Coordinate gene regulation was iden-using gene set enrichment analysis (GSEA; www.broad.u/gsea/), with a nominal P value cutoff of 0.001. "Coreed" genes are defined as belonging to the leading-edgewithin the gene set, and thus contribute themost to thement result. Assessment of HOX gene expression afterr treatment with PI-103 at 5 × IC50 was carried out usingna HT-12 BeadChips (ArrayExpress accession numberM-890). Affymetrix U133 expression data fromTheCancere Atlas (TCGA) glioblastoma study (21) was assessed forcorrelations of probesets corresponding to HOXA9 byating Pearson's correlation coefficients in R. GSEA andl correlations were further carried out on a publishedt (22) of Affymetrix U133 expression array profiling of 78ric high-grade gliomas (Gene Expression Omnibus acces-umber GSE19578; http://www.ncbi.nlm.nih.gov/geo/).

nofluorescence and flow cytometry133 protein expression was measured by both flowetry using a BD fluorescence-activated cell sortingge SEDiVa system (BDBiosciences) and immunofluores-on cytospin preparations, using anti-CD133 antibody3/1, Miltenyi Biotec) at 1:50 and 1:100 dilution, respec-Cells were costained with nestin (196908, R & D Sys-and visualized with 4′, 6-diamidino-2-phenylindole; Vector Laboratories Inc.). Cell cycle analysis was alsod out by flow cytometry.

lts

tivity of glioma cell lines to temozolomidero is largely but not exclusively dependentGMT promoter methylation and lackotein expressionfirst determined the response to temozolomide in vitro ofanel of five pediatric (SF188, KNS42, UW479, Res259,) and six adult (A172, LN229, SF268, U87MG, U118MG,G) glioma cell lines. Four lines (A172, LN229, SF268,) were classed as temozolomide sensitive, with IC50 va-f between 10 and 20 μmol/L (Fig. 1A). The remainingere resistant to treatment with the alkylating agent,C50 values of >500 μmol/L.

2. Assessment of DNA MMR and BER pathways in pediatric andlioma cell lines. Western blot for proteins involved in MMR (A)R (B). Although deficiencies in MSH3 are noted in U138MG,, and Res186, and there is some variability in PARP and XRCC1ion, U87MG and KNS42 have normal expression of proteins

assessedMGMT promoter methylation by methylation-ic PCR and MLPA, and protein expression by Western

Cancer Res; 70(22) November 15, 2010 OF3



Gaspar et al.

CanceOF4


r Res; 70(22) November 15, 2010 Cancer Research

Research. on March 23, 2021. © 2010 American Association for Cancercancerres.aacrjournals.org Downloaded from


blot (FMGMRes259correland tediatrictivity imechawe trewith tO6-benincrea20 μmSF188IC50 indepenzolomKNS4the in

DysreexplaMGMEnz

MSH3(Fig.resist(U138in KNSof prominvestPARP1UW47expreslines, i

Identsignain pedUsin

entialllines (spite tlines.and CDmoduautophimmu

Whdinat(MSigcell linP = 0.the Ha GSEgenescell linenrichwithinHOXCAs

beenreporincludfurtheand ousion pwith tHOXAmerouby thigenomsignatincludnon–huous ggenomTo

markeKNS42expresconsidotherexprestemozized bnestinlevelscytochused tusuallpositivitive cin oth

Figureclusteri(Res25priori. Egenes dalong wgenes pin Mura


www.a


ig. 1B). Extensive methylation resulted in an absence ofT protein expression in LN229, A172, U87MG, SF268,, and KNS42 cells. There was, for the most part, a directation between MGMT methylation/lack of expressionmozolomide sensitivity. An exception to this was the pe-glioblastoma KNS42 cell line, which displayed insensi-n the absence of MGMT protein, implying that alternatenisms of resistance must be operative. To confirm this,ated pediatric glioblastoma SF188 and KNS42 cellsemozolomide in the presence of the substrate analoguezyl guanine (O6BeG), which depletes the enzyme andses cytotoxicity (ref. 23; Fig. 1C). Treatment withol/L O6BeG increased the efficacy of temozolomide incells nearly 40-fold (IC50 without O

6BeG = 194 μmol/L;the presence of O6BeG = 5 μmol/L), thus confirming thedence of these cells on MGMT in conferring temo-ide resistance. By contrast, no such effect was seen in2 cells, showing the MGMT-independent nature ofsensitivity.

gulation of MMR and BER proteins do notin resistance to temozolomide inT-deficient KNS42 cellsymes involved in DNA MMR (MLH1, MLH3, MSH2,, MSH6, PMS2) were evaluated by Western blot2A). Although we identified three temozolomide-ant lines with abrogated expression of MSH3MG, UW479, and Res186), there was no deficiency42 or U87MG cells. These data correlated well with levelsotermethylation assessed byMS-MLPA (15).We further

igated components of the BER pathway, including/2, XRCC1, and APE1. Although temozolomide-resistant9 and Res186 seemed to lack PARP1/2 and XRCC1sion, there were no apparent alterations in other gliomancluding KNS42 or U87MG (Fig. 2B).

ification of a HOX/stem cell gene expressionture associated with temozolomide resistanceiatric glioblastoma cellsg expression microarrays, we identified 135 genes differ-y expressed between sensitive and resistant glioma cellFig. 3A). Included in this list wereMGMT and PARP2, de-hese enzymes not explaining the resistance in all cellAlso included were several kinases, including MAPK9K6,which may prove suitable targets for pharmacologic

lation; PIK3C3 (Vps34), suggesting a possible link to the
agic response (24); and genes encoding elements of thene response such as IL10 and IL16.
pressed by far the greatest levels of other stem cell markerssuch as nestin, SOX2, and musashi-1 (Affymetrix U133, data

3ng9,ne

t and colleagues (26), HOXA9 and HOXA10.

acrjournals.org


en we applied GSEA to our data, we observed coor-ed differential expression of the HOX_GENES setDB C2:curated gene sets) in resistant versus sensitivees (Fig. 3B), with an enrichment score of 0.54 [nominal01; false discovery rate (FDR) q = 0.403]. Furthermore,OX_GENES list was also identified as significant usingA “Preranked” analysis based on differentially expressedbetween KNS42 alone and temozolomide-sensitivees (Fig. 3C). In these analyses, the genes in both core-ed lists, which contribute to the leading-edge subsetthe gene set (25), included HOXA9, HOXA10, HOXB13,4, HOXC10, HOXC11, HOXC13, HOXD1, and GBX2.coordinated expression of HOX genes had recentlynoted in glioblastoma clinical samples (26), and wasted as evidence of a “self-renewal” signature as ited the stem cell marker PROM1 (CD133), we soughtr evidence for this in a published glioblastoma datasetr cell linemodels. When we investigated TCGA expres-rofiles of 163 glioblastomas for genes that correlatedhe top-ranking HOX gene in our KNS42 GSEA list,9, we noted a remarkable parallel expression of nu-s other homeobox genes (Fig. 4A). Of the top 47 geness analysis, 18 were homeobox genes found at 9 distinctic loci, and 7 were included in the self-renewalure of Murat and colleagues (26). These latter genesed PROM1. Intriguingly, the vast majority of theomeobox genes identified by this analysis are contig-enes found commonly amplified in glioblastoma at theic locus 12q13-q14.determine whether an enrichment of the stem cellr CD133 may be playing a role in the resistance ofcells to temozolomide, we assessed the levels of mRNAsion relative to the other cell lines (Fig. 4B), and notederably higher levels of PROM1 in KNS42 cells than in anyline in our panel. Of note, the only other two cell lines tos PROM1 at above background levels were the similarlyolomide-resistant U87MG and SF188. This was visual-y immunofluorescent staining for CD133, colabeled with(Fig. 4C). We had previously reported the relatively highof stem cell markers in SF188 and KNS42 by immuno-emistry (15), and to more accurately quantify this, wehe more sensitive flow cytometry analysis to reveal any high degree of expression in KNS42, with 17.0% of cellse for CD133. There were also high levels of CD133-pos-ells in the SF188 line (4.8%) compared with 0.0% to 0.02%er cell lines. KNS42 cells grown as monolayers also ex-

. Expression profiling reveals a HOX/stem cell signature–associated temozolomide resistance in glioma cell lines. A, heatmap showing hierarchicalof 135 differentially expressed genes between resistant (UW479, Res186, KNS42, SF188, U87MG, U118MG; gray highlight) and sensitive

SF268, A172, LN229; yellow highlight) high-grade glioma cell lines. B, GSEA highlighting coordinated differential expression of gene sets defined ariched in temozolomide-resistant cell lines were the HOX_GENES set (enrichment score = 0.54, nominal P = 0.01; FDR q = 0.403). C, rankedrived from GSEA of KNS42 versus temozolomide-sensitive cell lines, and all temozolomide-resistant versus sensitive lines. Genes are provided

ith their rank in the total gene list and rank enrichment metric, and running enrichment scores are provided for both analyses. Gray highlight,resent within the core enrichment signature; green highlight, genes present in both analyses; bold, genes also present in the HOX gene signature




not shassociassaysspherorenewpackedthan nformedcells sassocia

ExprephospinhibtemozTrea

resulte

and HKNS42lack ostudytransca phosgeneti(27), wactivethe duresultand H(Fig. 5expreswas o

Figureexpresslisted, icolleagPROM1exprestight th

Gaspar et al.

CanceOF6


own). Consistent with the cancer stem/progenitor cell–ated gene expression profile, neurosphere formationshowed that KNS42 cells form tight three-dimensionalids, which may be serially passaged and undergo self-al. In contrast, SF188 cells form smaller, more looselyspheres, whereas U87MG grow as cell aggregates rathereurospheres per se. Neither UW479, Res259, nor Res186spheres under these conditions. Taken together, KNS42

eem to have a significant cancer stem/progenitor cell–ted gene expression signature and biological phenotype.

ssion of HOXA9/HOXA10 is a result ofhoinositide 3-kinase–mediated demethylation,ition of which synergistically interacts witholomide in KNS42 cells
tmentwith the demethylating agent 5-aza-2′-deoxycytidine time t
sion of stem cell markers CD133 (green) and nestin (red) in KNS42 cells grown asree-dimensional spheroids formed by KNS42 cells in contrast to the other cell line

r Res; 70(22) November 15, 2010


OXA10 in all pediatric cell lines with the exception of, in which no changes were observed, indicative of af methylation in the untreated cells (Fig. 5A). As a recenthas proposed amechanism for this observationwherebyriptional activation of the HOXA cluster is reversible byphoinositide 3-kinase (PI3K) inhibitor through an epi-c mechanism involving histone H3K27 trimethylatione sought to investigate whether this mechanism wasin our system. Treatment for 1, 8, and 24 hours withal PI3K/mTOR inhibitor PI-103 (28–30) at 5 × IC50

ed in significantly reduced expression of both HOXA9OXA10 in KNS42 cells in a time-dependent mannerB). These effects were not due to fluctuations inHOX genesion with the cell cycle, as diminished HOXA9/HOXA10bserved as early as one hour posttreatment, at which
here was no evidence of G1 arrest (Fig. 5C), despite inhi-
d in highly differential levels of expression of HOXA9 bition of PI3K as seen by reduced phospho-Akt levels (Fig. 5D).

4. Coordinated upregulation of HOX genes in primary glioblastomas and a striking degree of CD133 positivity on KNS42 cells. A, genes coordinatelyed with HOXA9 from the TCGA dataset (21) were determined by calculating Pearson's correlation coefficients. All genes with values >0.3 aren rank order. Yellow highlight, homeobox genes; orange highlight, genes also present in the HOX gene/self-renewal signature of Murat andues (26), including PROM1 (CD133; box); gray highlight, genes found within the 12q13-q14 amplicon. B, Affymetrix expression analysis of(CD133) mRNA levels in the panel of glioma cell lines, plotted as relative log2 expression. C, immunofluorescence assay showing extensive

a monolayer. D, neurosphere formation assay highlighting thes studied.

Cancer Research



Nexcific dthat mMutatdid noand Wof PI3trast,proteicompacodedq14 amthe HOthis geFina

stratediatricmozoresultsuredFig. 5Fwas oindex

HOXAsurvivTo

sion inaminetumorlong-tthat wand shlist wGSEAin thecytobaalbeitWith tthe HOscorewas comembcorrelaand H

Figurelevels o5-Aza-2treatmereductiotreatmeof KNSHOXA1specificof PI-103 and temozolomide (TMZ). Cotreatment with PI-103 and temozolomide resulted in a high degree of synergy in MGMT-independent KNS42 cells(combin F188 c


www.a


t we sought to determine whether there was any spe-ysregulation of the PI3K/PTEN system in KNS42 cellsay be responsible for the HOX gene overexpression.ion screening for PTEN, PIK3CA, PIK3R1, and PIK3R3t identify any sequence variations (data not shown),estern blot analysis confirmed a lack of overexpressionK regulatory and catalytic subunits (Fig. 5E). By con-there were significantly elevated levels of the enhancerns PIKE-A and especially PIKE-L in KNS42 cells inrison with the other lines. Both PIKE proteins are en-by the CENTG1 (AGAP2) gene found within the 12q13-plicon coordinately upregulated in association withX cluster, and likely represent a significant target fornomic event in human glioblastoma.lly, we investigated the efficacy of targeting PI3K as agy for overcoming temozolomide resistance in our pe-glioma cells. Combination treatment in vitro of te-

lomide with the dual PI3K/mTOR inhibitor PI-103ed in a highly synergistic interaction in KNS42 as mea-by median effect analysis (combination index = 0.43;). By contrast, in SF188 cells an antagonistic response

ation index = 0.43) as calculated by the median effect analysis. By contrast, S

bserved with the same combination (combination= 1.40).

expre75th p

acrjournals.org


9/HOXA10 expression is associated with shorteral in pediatric high-grade glioma patientsassess the translational relevance of HOX gene expres-pediatric high-grade glioma patient samples, we ex-

d published data detailing expression profiles of 78s arising in childhood (22). Although the number oferm (>3 years) survivors is small, we identified 49 genesere differentially expressed between patients with longort (<1 year) overall survival (Fig. 6A). Included in thisere HOXA2, HOXA5, HOX7, and HOXA9. By applyingto the dataset, we identified coordinated upregulationshort-term survivors of genes at the chromosome 7p15nd, with an enrichment score of 0.68 (nominal P < 0.001)with a high false discovery rate value (FDR q = 0.930).he HOXA cluster found at this locus, running GSEA onXA gene list itself gave a highly significant enrichmentof 0.90 (nominal P < 0.001, FDR q < 0.001; Fig. 6B). Therensiderable correlation between gene expression of allers of theHOXA family in the pediatric samples (Pearson'stion coefficients 0.03–0.90). Taking the values of HOXA9OXA10, and segregating samples into “high” and “low”

ells showed an antagonistic interaction (combination index = 1.401).

5. HOXA9/HOXA10 expression in KNS42 cells is driven by a lack of promoter methylation in a PI3K-dependent manner. A, relative mRNA expressionf HOXA9 and HOXA10 before and after treatment with 5-Aza-2′-deoxycytidine in pediatric glioma cell lines. An absence of expression changes after′-deoxycytidine treatment in KNS42 cells is indicative of an absence of constitutive promoter methylation. B, HOXA9/HOXA10 expression is reduced bynt with PI-103 in KNS42 cells in a time-dependent manner. Treatment with the dual PI3K/mTOR inhibitor PI-103 at 5 × IC50 for 1, 8, and 24 hours led to an in expression of both HOXA9 and HOXA10 in KNS42 cells as early as 1 hour posttreatment. C, Western blot analysis of phospho- and total Akt afternt of KNS42 cells with PI-103 at 5 × IC50 for 1, 8, and 24 hours. Inhibition of PI3K signaling is observed at the earliest time point. D, cell cycle analysis42 cells after treatment with PI-103 at 5 × IC50 for 1, 8, and 24 hours. There was no G1 arrest evident at the early time points at which reduced HOXA9/0 expression was observed. E, Western blot analysis of PI3K regulatory and catalytic subunits and enhancers in pediatric glioma cells. Expression of theAkt enhancer proteins PIKE-A and PIKE-Lwere considerably elevated in KNS42 cells comparedwith other pediatric glioma cells. F, synergistic interaction

ssers (combined expression greater or less than theercentile of all values, respectively), we showed a




signifigliomarank tof the

Discu

Procepteding aghas mWe idbe resMGMTdoublenism otors thremain

Gengliom

Figuredifferen(green)(enrichmupregulated in short-term survivors. C, Kaplan-Meier plot showing a significantly shorter survival of patients with high levels of HOXA9/HOXA10 geneexpress

Gaspar et al.

CanceOF8


cantly reduced overall survival of pediatric high-gradepatients with high HOXA9/HOXA10 expression (log-

ion (P = 0.0453, log-rank test).

est, P = 0.0453; Fig. 6C) independent of the WHO grade merouKNS42port opatienapprogenetance.to thafectoridentiindepblastoprognHOX

bryon

tumor (P = 0.635, Fisher's exact test).

ssion

moter methylation of the MGMT gene is generally ac-as the major determinant of sensitivity to the alkylat-ent temozolomide in glioblastoma cells, and as suchajor significance in the treatment of these patients.entified the pediatric glioblastoma cell line KNS42 toistant to temozolomide in vitro despite an absence ofexpression, a competent MMR system, and an intact-strand break repair pathway. Clues as to the mecha-f resistance in these cells may help in identifying fac-
at contribute to childhood glioblastoma patients whorefractory to temozolomide treatment.
includappar

r Res; 70(22) November 15, 2010


e expression profiling of a panel of pediatric and adulta cell lines highlighted coordinated expression of nu-s HOX genes in the resistant cell lines, most especially, and provided in vitro model system evidence in sup-f data from temozolomide-treated adult glioblastomats (26). Using a similar expression profiling and GSEAach, Murat and coworkers identified a HOX-dominatedcluster as an independent predictive factor of resis-Integrating the core gene lists of the present studyt dataset highlights HOXA9 and HOXA10 as the key ef-s in both systems. This converges with a recent studyfying the HOXA cluster, and HOXA9 in particular, to beendent negative prognostic markers in adult glio-ma (27). Herein we provide evidence for an additionalostic role in pediatric high-grade glioma.genes are essential in axis determination during em-

ic development and are known to be involved in cancer,

6. High levels of HOXA gene expression are associated with shorter survival in pediatric high-grade glioma patients. A, heatmap representingtially expressed genes between short (<1 year; yellow highlight) and long-term (>3 years; gray highlight) survivors. WHO grade IV (black) and IIItumors are indicated. Light blue, HOXA genes. B, GSEA analysis showing significant enrichment of genes at the chromosome 7p15 locusent score = 0.68, nominal P < 0.001, FDR q = 0.930), and of HOXA genes in particular (enrichment score = 0.90, nominal P < 0.001, FDR q < 0.001),

ing glioblastomas (31, 32); it is not immediatelyent, however, what role they may play in resistance to

Cancer Research



alkylatexertslationindepetion, ateractin vitrcombipatienassociMur

in glioof HOshoweself-re(17.0%monocell mSuch aregula(21). Tcellenwith MWe

cell sigin thesamplwith pin thea pos12q13hancevels inchangvia PI

It isrepairsistanthosetreatmsistantumorthe pr

Discl

N. GS.Y. ShaThe InsdevelopP. Workwith YaintellectP. Workber of, a

Ackn

We tand DrpediatrHealth,

Grant

CancFoundaacknowP. Work

Theof pageaccorda

Refe1. Mr

mid5:4

2. Stucom200

3. Nezol199

4. Heefit997

5. EsDNalk

6. Midlatemo

7. SilbnuresCli


www.a


ing agents. Costa and colleagues propose that HOXA9antiapoptotic and proproliferative effects after upregu-via an epigenetic mechanism controlled by PI3K andndent of mTOR (27). Our findings confirm this observa-nd add to previous evidence showing the synergistic in-ions of temozolomide and PI3K pathway inhibitors ino and in vivo models of adult glioblastoma (33); thisnation may therefore also be beneficial in glioblastomats with a MGMT-independent HOX gene signature-ated mechanism of resistance to temozolomide.at and colleagues suggested strong HOXA10 expressionblastoma-derived neurospheres to be in line with a roleX genes in the glioma stem-like cell compartment, andd that the resistance signature as a whole is evocative ofnewal (26). Of note is the presence of a high proportion) of CD133-positive cells present in our KNS42 celllayer cultures, with additional expression of other stemarkers in coordination with the HOX gene signature.HOX/stem cell signature was also found to be tightly

ted in an analysis of TCGA glioblastoma expression datahe pediatric glioblastoma cell line KNS42 may be an ex-t experimental model for investigating such interactionsGMT-independent treatment failure.also found a remarkable link between the HOX/stemnature and coordinated overexpression of genes with-CDK4 amplicon at 12q13-q14 in glioblastoma patientes. This association was also present, and correlatedoor response to temozolomide chemoradiotherapy,Murat and colleagues dataset (26). Here we providesible mechanism for this coexpression of HOX and-q14 genes in the form of overexpression of the Akt en-r PIKE (CENTG1, AGAP2), also present at very high le-the unamplified KNS42 cells, which may drive the

es in H3K27 methylation of the HOXA cluster mediated3K pathway signaling (27).

ReceOnlineF

clease activity of Ape1/Ref-1 contributes to human glioma cellistance to alkylating agents and is elevated by oxidative stress.n Cancer Res 2002;8:3008–18.

8. Ch1 irep4:1

9. PomechJ C

10. Broradlow

11. NictemneGr

12. Ruin89

13. Hame

14. Laglio

acrjournals.org


apparent that a variety of processes, not all involvingor tolerance of alkyl lesions, may promote alkylator re-ce (23). Along with providing therapeutic guidance forpatients whose tumors are intrinsically resistant toent, characterization of additional determinants of re-ce is necessary to develop new targets for therapy ins that acquire resistance to temozolomide in vivo inesence of hypermethylated MGMT.

osure of Potential Conflicts of Interest

aspar, L. Marshall, L. Perryman, D.A. Bax, S.E. Little, M. Viana-Pereira,rp, A.D.J. Pearson, P. Workman, and C. Jones are or were employees oftitute of Cancer Research, which has a commercial interest in thement of PI3K inhibitors and operates a rewards-to-inventors scheme.man and his team have been involved in a commercial collaborationmanouchi (now Astellas Pharma) and with Piramed Pharma, andual property arising from the program has been licensed to Genentech.man was a founder of, consultant to, Scientific Advisory Board mem-nd stockholder in Piramed Pharma, which was acquired by Roche.

owledgments

hank Dr. Daphne Haas-Kogan (University of California, San Francisco). Michael Bobola (University of Washington) for provision of theic glioma cell lines, and Dr. Michael Hubank (Institute of ChildUniversity College London) for assistance with the expression profiling.

Support

er Research UK (C1178/A10294, C309/A2187, C309/A8274), the Oaktion (L. Marshall), and La Fondation de France (N. Gaspar). Weledge NHS funding to the NIHR Biomedical Research Centre.man is a Cancer Research UK Life Fellow.costs of publication of this article were defrayed in part by the paymentcharges. This article must therefore be hereby marked advertisement innce with 18 U.S.C. Section 1734 solely to indicate this fact.

ived 04/09/2010; revised 09/14/2010; accepted 09/17/2010; publishedirst 10/08/2010.

rencesugala MM, Chamberlain MC. Mechanisms of disease: temozolo-e and glioblastoma-look to the future. Nat Clin Pract Oncol 2008;76–86.pp R, Mason WP, van den Bent MJ, et al. Radiotherapy plus con-itant and adjuvant temozolomide for glioblastoma. N Engl J Med5;352:987–96.wlands ES, Blackledge GR, Slack JA, et al. Phase I trial of temo-omide (CCRG 81045: M&B 39831: NSC 362856). Br J Cancer2;65:287–91.gi ME, Diserens AC, Gorlia T, et al. MGMT gene silencing and ben-from temozolomide in glioblastoma. N Engl J Med 2005;352:–1003.teller M, Garcia-Foncillas J, Andion E, et al. Inactivation of theA-repair gene MGMT and the clinical response of gliomas toylating agents. N Engl J Med 2000;343:1350–4.dlemas DS, Stewart CF, Kirstein MN, et al. Biochemical corre-s of temozolomide sensitivity in pediatric solid tumor xenograftdels. Clin Cancer Res 2000;6:998–1007.er JR, Bobola MS, Blank A, et al. The apurinic/apyrimidinic endo-

eng CL, Johnson SP, Keir ST, et al. Poly(ADP-ribose) polymerase-nhibition reverses temozolomide resistance in a DNA mismatchair-deficient malignant glioma xenograft. Mol Cancer Ther 2005;364–8.llack IF, Hamilton RL, Sobol RW, et al. O6-methylguanine-DNAthyltransferase expression strongly correlates with outcome inildhood malignant gliomas: results from the CCG-945 Cohort.lin Oncol 2006;24:3431–7.niscer A, Chintagumpala M, Fouladi M, et al. Temozolomide afteriotherapy for newly diagnosed high-grade glioma and unfavorable-grade glioma in children. J Neurooncol 2006;76:313–9.holson HS, Kretschmar CS, Krailo M, et al. Phase 2 study ofozolomide in children and adolescents with recurrent central

rvous system tumors: a report from the Children's Oncologyoup. Cancer 2007;110:1542–50.ggiero A, Cefalo G, Garre ML, et al. Phase II trial of temozolomidechildren with recurrent high-grade glioma. J Neurooncol 2006;77:–94.rgrave DR, Zacharoulis S. Pediatric CNS tumors: current treat-
nt and future directions. Expert Rev Neurother 2007;7:1029–42.shford LS, Thiesse P, Jouvet A, et al. Temozolomide in malignantmas of childhood: a United Kingdom Children's Cancer Study



GroJ C

15. Baactdru

16. Skass

17. CotetCo

18. ChmuCh

19. Estvatferhum

20. NyMLco20

21. Mcchawa

22. Papedis

23. BoO6resline

24. WoPI3

25. Suanwid15

26. Musigasglio

27. Coacno45

28. Haevno20

29. RatioCa

30. Womo

31. Abca

32. AbHuan

Gaspar et al.

CanceOF10


up and French Society for Pediatric Oncology Intergroup Study.lin Oncol 2002;20:4684–91.x DA, Little SE, Gaspar N, et al. Molecular and phenotypic char-erisation of paediatric glioma cell lines as models for preclinicalg development. PLoS ONE 2009;4:e5209.ehan P, Storeng R, Scudiero D, et al. New colorimetric cytotoxicityay for anticancer-drugscreening. JNatlCancer Inst 1990;82:1107–12.ry AH, Owen TC, Barltrop JA, Cory JG. Use of an aqueous solublerazolium/formazan assay for cell growth assays in culture. Cancermmun 1991;3:207–12.ou TC, Talaly P. A simple generalized equation for the analysis ofltiple inhibitions of Michaelis-Menten kinetic systems. J Biolem 1977;252:6438–42.eller M, Hamilton SR, Burger PC, Baylin SB, Herman JG. Inacti-ion of the DNA repair gene O6-methylguanine-DNA methyltrans-ase by promoter hypermethylation is a common event in primaryan neoplasia. Cancer Res 1999;59:793–7.

gren AO, Ameziane N, Duarte HM, et al. Methylation-specificPA (MS-MLPA): simultaneous detection of CpG methylation andpy number changes of up to 40 sequences. Nucleic Acids Res05;33:e128.Lendon R, Friedman A, Bigner D, et al. Comprehensive genomicracterization defines human glioblastoma genes and core path-ys. Nature 2008;455:1061–8.ugh BS, Qu C, Jones C, et al. Integrated molecular profiling ofdiatric high grade gliomas reveals key differences with the adultease. J Clin Oncol 2010;28:3061–8.bola MS, Silber JR, Ellenbogen RG, Geyer JR, Blank A, Goff RD.
-methylguanine-DNA methyltransferase, O6-benzylguanine, andistance to clinical alkylators in pediatric primary brain tumor cells. Clin Cancer Res 2005;11:2747–55.
33. Guph20

r Res; 70(22) November 15, 2010


rkman P, van Montfort RL. Unveiling the secrets of the ancestralkinase Vps34. Cancer Cell 2010;17:421–3.

bramanian A, Tamayo P, Mootha VK, et al. Gene set enrichmentalysis: a knowledge-based approach for interpreting genome-e expression profiles. Proc Natl Acad Sci U S A 2005;102:

545–50.rat A, Migliavacca E, Gorlia T, et al. Stem cell-related "self-renewal"nature and high epidermal growth factor receptor expressionsociated with resistance to concomitant chemoradiotherapy inblastoma. J Clin Oncol 2008;26:3015–24.sta BM, Smith JS, Chen Y, et al. Reversing HOXA9 oncogenetivation by PI3K inhibition: epigenetic mechanism and prog-stic significance in human glioblastoma. Cancer Res 2010;70:3–62.yakawa M, Kaizawa H, Moritomo H, et al. Synthesis and biologicalaluation of pyrido[3′,2′:4,5]furo[3,2-d]pyrimidine derivatives asvel PI3 kinase p110alpha inhibitors. Bioorg Med Chem Lett07;17:2438–42.ynaud FI, Eccles S, Clarke PA, et al. Pharmacologic characteriza-n of a potent inhibitor of class I phosphatidylinositide 3-kinases.ncer Res 2007;67:5840–50.rkman P, Collins I. Probing the probes: fitness factors for smalllecule tools. Chem Biol 2010;17:561–77.ate-Shen C. Deregulated homeobox gene expression in cancer:use or consequence?. Nat Rev Cancer 2002;2:777–85.del-Fattah R, Xiao A, Bomgardner D, Pease CS, Lopes MB,ssaini IM. Differential expression of HOX genes in neoplasticd non-neoplastic human astrocytes. J Pathol 2006;209:15–24.illard S, Clarke PA, Te Poele R, et al. Molecular pharmacology of
osphatidylinositol 3-kinase inhibition in human glioma. Cell Cycle09;8:443–53.
Cancer Research



Published OnlineFirst October 8, 2010.Cancer Res Nathalie Gaspar, Lynley Marshall, Lara Perryman, et al.

/Stem Cell Gene SignatureHOXMediated −Glioblastoma Cells Associated with a PI3-Kinase

MGMT-Independent Temozolomide Resistance in Pediatric

Updated version

10.1158/0008-5472.CAN-10-1250doi:

Access the most recent version of this article at:

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

Subscriptions

Reprints and

[email protected] at

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

Permissions

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

.http://cancerres.aacrjournals.org/content/early/2010/10/29/0008-5472.CAN-10-1250To request permission to re-use all or part of this article, use this link

Research. on March 23, 2021. © 2010 American Association for Cancercancerres.aacrjournals.org Downloaded from


http://cancerres.aacrjournals.org/lookup/doi/10.1158/0008-5472.CAN-10-1250

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

mailto:[email protected]

http://cancerres.aacrjournals.org/content/early/2010/10/29/0008-5472.CAN-10-1250


Documents

Research MGMT-Independent Temozolomide Resistance in ... · 29.10.2010 · MGMT-Independent Temozolomide Resistance in Pediatric Glioblastoma Cells Associated with a PI3-Kinase–Mediated