cancer research global downstream pathway analysis reveals a … · molecular and cellular...

Mole

Gloof OmTO

TatsuMasa

Abst

Intro

Durposedgens,Theygenetition, otion owouldprogretastasin canby boment)

AuthorDivisioInform4DeparSchool

CorresNationaTokyo,81-3-32

doi: 10

©2010

www.a

Published OnlineFirst November 9, 2010; DOI: 10.1158/0008-5472.CAN-10-0384

Canceresearch

cular and Cellular Pathobiology

bal Downstream Pathway Analysis Reveals a Dependencencogenic NF-E2–Related Factor 2 Mutation on the

R

R Growth Signaling Pathway

hiro Shibata1,2, Shigeru Saito1,3, Akiko Kokubu1, Takafumi Suzuki4,
yuki Yamamoto4, and Setsuo Hirohashi1,2
ractIn m

2 (NRactivathe NNRF2cell prgenemammtivatormutatboth iwas aresults

th extrinsand intri

s' Affiliatin, Nationalatics Depatment of Mof Medicine

ponding Al Cancer C104-0045,48-2463; E

.1158/0008-

American A

acrjourna

Downloa

ulticellular organisms, adaptive responses to oxidative stress are regulated by NF-E2–related factorF2), a master transcription factor of antioxidant genes and phase II detoxifying enzymes. Aberranttion of NRF2 by either loss-of-function mutations in the Keap1 gene or gain-of-function mutations inrf2 gene occurs in a wide range of human cancers, but details of the biological consequences ofactivation in the cancer cells remain unclear. Here, we report that mutant NRF2 induces epithelialoliferation, anchorage-independent growth, and tumorigenicity and metastasis in vivo. Genome-wideexpression profiling revealed that mutant NRF2 affects diverse molecular pathways including thealian target of rapamycin (mTOR) pathway. Mutant NRF2 upregulates RagD, a small G-protein ac-of the mTOR pathway, which was also overexpressed in primary lung cancer. Consistently, Nrf2-

ed lung cancer cells were sensitive to mTOR pathway inhibitors (rapamycin and NVP-BEZ235) inn vitro and an in vivo xenograft model. The gene expression signature associated with mutant NRF2marker of poor prognosis in patients with carcinoma of the head and neck region and lung. Theseshow that oncogenic Nrf2 mutation induces dependence on the mTOR pathway during carcino-

s. Our findings offer a rationale to target NRF2 as an anticancer strategy, and they suggest NRF2
genesiactivation as a novel biomarker for personalized molecular therapies or prognostic assessment. Cancer Res;
70(22); 9095–105. ©2010 AACR.

tors (4stressalteragressioptosismolecstressoxidat(NRF2genesNRF

cine zthe sp

duction

ing the multistep carcinogenesis, cancer cells are ex-to multiple environmental stresses such as carcino-hypoxia, host inflammation, or anticancer drugs (1).also have to deal with intrinsic stresses induced byc instability, oncogene-induced abnormal prolifera-r aberrant protein folding (1, 2). Therefore, acquisi-f resistance to these stresses by some cancer cellsbe beneficial for their survival, leading to malignantssion and lethal characteristics such as invasion, me-is, and chemotherapy resistance (3). Oxidative stresscer is one of the important metabolic stresses caused

ic (reperfusion of a hypoxic microenviron-nsic (uncoupled mitochondrial activity) fac-

quencand ptive gubiquNRF2rapidlexposare rachangNRF2and inPre

eficial

ons: 1Cancer Genomics Project and 2PathologyCancer Center Research Institute; 3Chem and Biortment, Infocom Corporation, Tokyo, Japan andedical Biochemistry, Tohoku University Graduate, Sendai, Japan

uthor: Tatsuhiro Shibata, Cancer Genomics Project,enter Research Institute, 5-1-1, Tsukiji, Chuo-ku,Japan. Phone: 81-3-3547-5201 ext. 3123; Fax:-mail: [email protected].

5472.CAN-10-0384

ssociation for Cancer Research.

ls.org

Research. on March 21, 202cancerres.aacrjournals.org ded from

, 5). From the viewpoint of carcinogenesis, oxidativeis a double-edged sword, because it promotes genetictions through DNA damage to drive malignant pro-n while also causing cell damage and inducing apo-(6, 7). Under physiologic conditions, cells possess

ular mechanisms that enable them to adapt to thesees. In multicellular organisms, adaptive responses toive stresses are regulated by NF-E2–related factor 2), a master transcription factor of many antioxidantand phase II detoxifying enzymes (8).2 belongs to the Cap'nCollar subfamily of basic leu-ipper (bZIP)–type transcriptional factors. It recognizesecific antioxidant recognition element (ARE) se-e by forming a heterodimer with the bZIP familyromotes the expression of a wide range of cytoprotec-enes (9). The activity of NRF2 is finely controlled byitin-mediated protein degradation. In unstressed cells,is recognized by an E3 ubiquitin ligase, KEAP1, andy degraded in the proteasome (10, 11). When cells areed to oxidative stress, the cysteine residues of KEAP1pidly modified by electrophiles, resulting in structurale and loss of association with NRF2 (12). The freereleased from KEAP1 then moves into the nucleusduces its specific targets.
vious studies have reported that NRF2 activation is ben-for prevention of cancer, because Nrf2-deficient mice
9095

0. © 2010 American Association for Cancer

http://cancerres.aacrjournals.org/

are sucarcincarcinenzymoxygencells fhave ction oKeap1occur(∼40%and bseemsthe decancer

Mate

DetSupple

Cell bCell

(15, 16ing. Rawere ccDNAThermLipofewere isensitition (cments

In vivFor

cin (mthe nuNMP/weredailywas 1amete

MolecTot

cell camal lutranscevaluaLight-and im(15). ASupple

MicroTen

Molon

Cy3-lapolymProbeoligonHumalogies)the M

StatisAm

corresstatisgrounogy (Ggo.priset enC2 pabroadgene emousthe sicluste1,000to depwas ptest w

Resu

OncomutaWe

epithecDNAtectedtantscDNAof thecumulthe enKeap1Fig. 1AWe

that cthan tciallysiRNAthe mthe mS1A agrowtNRF2whereNRF2activitlung c

Shibata et al.

Cance9096


sceptible to cancer development in chemical-inducedogenesis models and NRF2 promotes detoxification ofogen and decreases DNA damage by inducing phase IIes and reducing oxidative stressors such as reactivespecies (ROS), both mechanisms protecting normal

rom carcinogenesis (13, 14). However, recent studieshallenged this concept and shown that aberrant activa-f NRF2 through either loss-of-function mutations in thegene or gain-of-function mutations in the Nrf2 genes in a wide range of human cancers such as lung), head and neck (∼20%), gallbladder (∼30%), liver,reast cancers (15–20). Thus the KEAP1-NRF2 pathwayto have a dual function in carcinogenesis, althoughtailed biological consequences of NRF2 activation incells remain unknown.

rials and Methods

ails of the experimental procedures are described inmentary Materials and Methods.

iological analyseslines used in this study have been described previously) and were validated by short tandem repeat genotyp-pamycin (Nakarai) and NVP-BEZ235 (LC Laboratories)hemically synthesized and purchased. Mutant NRF2and small interfering RNAs (siRNA; purchased fromo Fisher Scientific) were transfected into cells by usingctamine and RNAiMAX (Invitrogen), and stable clonessolated after G418 selection. Cell proliferation and drugvity were measured by bromodeoxyuridine incorpora-ell proliferation ELIZA, Roche Diagnostic). All experi-were done in triplicate.

o experimentsin vivo assessment of mammalian target of rapamy-TOR) inhibitor, we transplanted 5 × 106 LK2 cells inde mice. NVP-BEZ235 (free base) was formulated inpolyethylene glycol 300 (10:90, v/v). Solutions (5 mg/mL)prepared fresh each day of dosing and were given(20 mg/kg) by oral gavage. The application volume0 mL/kg. Tumor volume was calculated by (tumor di-r)2 × (tumor length/2).

ular analysesal RNA was extracted from cell lines, primary squamousrcinoma (SCC) of the lung, and the corresponding nor-ng tissues by RNAeasy (Qiagen). Quantitative reverseription-PCR (RT-PCR) was performed in triplicate andted using universal probes for each amplicon and theCycler system (Roche Diagnostic). Protein extractionmunoblotting were performed as described previouslyntibodies and siRNA used in this study were listed inmentary Table S1.

array analysis of global gene expression
micrograms of total RNA were reverse-transcribed byey murine leukemia virus reverse transcriptase, and a
Nrf2 mLK2 c

r Res; 70(22) November 15, 2010

Research. on March 21, 202cancerres.aacrjournals.org Downloaded from

beled cRNA probe was synthesized using T7 RNAerase in accordance with the manufacturer's protocol.s were hybridized with a microarray containing 41,000ucleotides covering the whole human genome (Wholen Genome Oligo Microarray, G4112F, Agilent Techno-. All analyzed microarray data are in accordance withIAME guidelines.

tical analysisong the total of 41,000 probes, 12,091 stable probesponding to the MAQC common genes were used fortical analysis. To characterize the molecular back-ds of the gene list, enrichment analysis for Gene Ontol-O) was performed using the GO Term Finder (http://nceton.edu/cgi-bin/GOTermFinder). Preranked generichment analysis was applied for the gene list withthway gene sets and 2,000 permutations (http://www.institute.org/gsea/index.jsp). Two independent sets ofxpression data from head and neck cancer and squa-lung cancer were classified into two groups based ongnatures of the gene list using K-means clustering (Rr package). Then, stable clusters were obtained inrandom trials to avoid artifacts of classification dueendency of initial random seeds. Survival analysis

erformed using the Kaplan-Meier method, and log-rankas used to evaluate disease-free survival.

lts

genic phenotypes of HEK293 cells expressingnt NRF2established stable HEK293 (an immortalized humanlial cell line) clones expressing two mutant NRF2s (T80R and L30F mutants), both of which were de-in human cancer and behave as gain-of-function mu-(15). Clones expressing mutant and wild-type NRF2showed accumulation of NRF2 in the nuclear fractionse clones (Fig. 1A, top). The degree of nuclear NRF2 ac-ation in these clones (samples 3 and 4) was less thandogenous NRF2 accumulation in Nrf2- (sample 6) or- (sample 5) mutated cancer cell lines (refs. 15, 16;, bottom).then tested the cell proliferation activity and foundlones expressing mutant NRF2 grew more rapidlyhe mock or wild-type NRF2-transfected clones (espe-T80R mutants; Fig. 1B). Downregulation of NRF2 byinhibited the proliferation of clones stably expressingutant and wild-type NRF2 gene, whereas growth ofock control was little affected (Supplementary Fig.nd B). We also tested the anchorage-independent cellh of these clones. As shown in Fig. 1C, both mutant-expressing clones produced colonies in soft agar,as the control clone formed no colonies. In addition,mutant clones showed an increase of cell migrationy in vitro (Fig. 1D). We had previously reported twoancer cell lines (EBC1 and LK2) that harbored the
utation (EBC1 cell has the D77V mutation, and
ell has the E79K mutation; ref. 15). Downregulation

Cancer Research



of NRcer cewild-tand Dcells dprevioWe

mutanmunoprodutrol cl5/8, LFTumoferent(>70%accomtics, su

tissueto theand mincreaWe alsels (aservedoverex

GenemutaBec

factorcontriscribe

FigureNRF2.with encytplasprolifermeasurRepresof migr

NRF2 Mutation and mTOR Pathway

www.a


F2 also reduced the proliferation of Nrf2 mutant can-lls but exerted no effect on SQ5 cells that contain lowype NRF2 protein (Fig. 1A; Supplementary Fig. S1C), supporting the contention that Nrf2-mutated cancerepend on the Nrf2 activity for their proliferation asusly reported in Keap1-mutated cancer cells (17).further examined the in vivo tumorigenic activity oft NRF2-expressing cells by s.c. transplantation in im-deficient mice. Both clones expressing mutant NRF2ced tumors at markedly high frequency, whereas con-ones did not form tumors (TR-1: 8/8, TR-2: 8/8, LF1:2: 7/8, control: 0/10; Fig. 2A; Supplementary Table S2).rs expressing mutant NRF2 exhibited poorly dif-iated carcinoma with increased mitotic activitycells being immunopositive for Ki67; Fig. 2B and C),
panied in most cases by highly aggressive characteris- chang
entative pictures of agar wells are shown (left). D, clones expressing mutant NRFated cells (n = 6 in each clone). Representative micrographs of transwells are sh

acrjournals.org


s (Fig. 2D). Moreover, there was occasional metastasisliver, suggesting that mutant NRF2 confers invasiveetastatic activity (Fig. 2E and F), consistent with thesed cell migration activity observed in vitro (Fig. 1D).so noted that these tumors contained many microves-s revealed by CD34 staining; Fig. 2B and G). As ob-under in vitro culture conditions, the tumor cellspressed a NRF2 target, NQO1 in vivo (Fig. 2H).

expression profiling of HEK293 cells expressingnt NRF2ause NRF2 encodes a sequence-specific transcriptional, direct or indirect downstream targets of NRF2 shouldbute to at least a proportion of the phenotypes de-d above. We attempted to overview the genome-wide
es in gene expression induced by mutant NRF2 and
ch as invasion into the surrounding adipose and muscle conducted a microarray analysis (Fig. 3A). Among individual

1. Mutant NRF2 confers oncogenic phenotypes in vitro. A, top, accumulation of nuclear NRF2 in clones expressing mutant (left) and wild (right)The nuclear extract of each clone was blotted with anti-NRF2 antibody. Bottom, NRF2 protein expression in established clones was compareddogenous NRF2 accumulation in Nrf2-mutated (LK2), Keap1-mutated (A549), and both wild (SQ5 and QG56) cancer cell lines. Nuclear (n) andmic (c) fractions were separately examined. Lamin B1 was used as a loading control. Molecular marker is indicated on the right (kDa). B, increasedation of mutant and wild-type NRF2-expressing clones compared with the mock control. Relative number of the cells in each clone wased and plotted (n = 3). C, clones expressing mutant NRF2 showed anchorage independence growth in the soft agar assay (n = 4 in each clone).

2 exhibited elevated migratory activity. The average numberown at the bottom. Data represent the mean ± SD.

Cancer Res; 70(22) November 15, 2010 9097



downsoxidanglutatbe NRsion oplemetargetwhichtivityin vivorect NchromGO

revealmultiptent wto oxid

belongsponsrelate“defenprodu“chem

Prognin huBec

anchotivitymarksinduceical ou

Figureand moinductioto the mSD, n =of NQO or cell

5 M.Y. a

Shibata et al.

Cance9098


tream genes, we observed upregulation of several anti-t genes (Nqo1, Hmox1, Abcc2, and enzymes relating tohione synthesis) that have been previously reported toF2 target genes, and we validated the increased expres-f these authentic targets by quantitative RT-PCR (Sup-ntary Fig. S2A and B). Among the other downstreams, we also observed increased expression of osteopontin,has been reported to be associated with metastatic ac-and angiogenesis in cancer (21, 22), both in vitro and(Supplementary Fig. S1C; Fig. 2I). We also detected di-RF2 binding in the promoter of the onsteopotin gene byatin immunoprecipitation (ChIP) sequencing.5

analyses of control and mutant NRF2-expressing clonesed that persistent expression of mutant NRF2 affectedle signal pathways (Supplementary Table S3). Consis-

16 in TR clones and n = 12 in LF clones). G, detection of tumor-associate1, a NRF2 target, in tumors. I, detection of osteopontin expression in tum

ith its previously known roles in the adaptative responseative stress, mutant NRF2 significantly induced genes

tantSuppleclinicafirst acer satype ond Yoichiro Mitsuishi, unpublished observation.

r Res; 70(22) November 15, 2010


ing to categories such as “response to stress” and “re-e to chemical stimuli.” In addition, expression of genesd to inflammatory processes (“response to wounding,”se response,” “inflammatory response,” and “cytokinection”) and cell motility (“locomotor behavior” andotaxis”) was also altered in NRF2 mutant clones.

ostic significance of the mutant NRF2 signatureman cancerause mutant NRF2 promoted cell proliferation andrage-independent growth in vitro and metastasis ac-and increased angiogenesis in vivo, which are all hall-of cancer malignancy, the molecular signatured by mutant NRF2 may be associated with poor clin-tcome. To test this hypothesis, we extracted the mu-NRF2 signature (complete gene list is shown inmentary Table S4) and examined its association withl data using archives of previous microarray data. Wenalyzed a large cohort (n = 60) of head and neck can-

ll vessels by CD34 staining. H, prominent cytoplasmic expressions. Bar, 100 μm.

2. Mutant NRF2 confers oncogenic phenotypes in vivo. A, representative view of mice bearing transplants (top, TR1 and mock; bottom, LF1ck). B-D, histologic appearance (H&E staining) of tumors formed by clones expressing mutant NRF2. B, the tumors showed solid growth withn of small vessels (arrows). C, high mitotic activity of tumor cells detected by Ki67 staining. D, local invasion of tumors. Tumor cells directly invadeduscle tissues around the rib. E, liver metastasis of tumors. F, the number of metastatic nodules in the liver of mice harboring tumors (mean ±

d sma

mples (23) because NRF2 is frequently mutated in thisf cancer, and thus far little information has been

Cancer Research



availabbasisples inFig. 3shownture ssuppofrom Hture oanalyzthe d0.0803(n = 1tant Nwithou

IdentmutanTo

mutananalysseveraall sign

Consiicant eby oximent (UVB_molecways aPeng_also obstresseture, awas rediversmTORto var

ActivaBec

attracfocusepathw

FiguremutantexpressCasesregulate


www.a


le about its prognostic molecular signature. On theof the mutant NRF2 signature, we classified the sam-to two clusters (Cluster1, n = 45; Cluster 2, n = 15;B) and compared the prognosis between them. Asin Fig. 4B, cases containing the mutant NRF2 signa-

howed significantly poorer prognosis (P = 0.000118),rting the contention that the gene profile extractedEK293 cells is comparable with the malignant signa-

f clinical SCC samples. To validate this result, we thened another cohort of lung SCC cases (24). Althoughifference did not reach statistical significance (P =), probably because of the small size of the cohort6), we observed a trend for tumors harboring the mu-RF2 signature to have a poorer prognosis than thoset the signature (Supplementary Fig. S3).

ification of molecular pathways regulated byt NRF2further elucidate the molecular pathways affected byt NRF2, we then conducted gene set enrichmentis (GSEA). This revealed significant enrichment of
l pathways in clones expressing mutant NRF2 (Table 1; way in
expressing that signature showed significantly worse prognosis. C, representatived by mutant NRF2.

acrjournals.org


stent with the above GO analysis, we detected signif-nrichment of pathways associated with genes induceddative stresses such as ROS (Houstis_ROS) or UV treat-UVB_SCC_UP, UVB_NHEK3_C0, UVB_NHEK3_C3, andNHEK3_ALL). Another noteworthy finding was that theular signature induced by mutant NRF2 included path-ssociated with mTOR signaling (Peng_Rapamycin_DN,Leucine_DN, Peng_Glutamine_DN; ref. 25; Fig. 3C). Weserved other molecular pathways associated with others (hypoxia or viral infection), the progenitor cell signa-nd lipid metabolism (Supplementary Fig. S4). Thus, itvealed that mutant NRF2 induced a set of genes moree than had previously been thought. In particular, thepathway emerged as a new target of NRF2, in addition

ious stress-regulated genes.

tion of the mTOR pathway by mutant NRF2ause the mTOR pathway has been reported to be antive therapeutic target in many solid tumors, we thend on the relationship between the NRF2 and mTORays. We first examined activation of the mTOR path-
HEK293 clones expressing mutant NRF2. Immunoblot
ificant pathways are listed in Supplementary Table S5). analysis revealed increased phosphorylation of ribosomal S6

3. Gene expression profiling of cells expressing mutant NRF2. A, two-dimensional hierarchal clustering of parent, mock, and clones expressingNRF2 by genome-wide gene expression. B, heat map of primary head and neck cancers based on K-means clustering using the mutant NRF2 geneion signature (left). Kaplan-Meier analysis of a head and neck cancer cohort segregated by the mutant NRF2 gene expression signature (right).

panels showing results of gene set enrichment analysis of genes




proteistratesphospto theactiva(Fig. 4cancer(Fig. 4To

ways,NRF2directstreammodupressiray anfamilyof the

the Hknockplemeexpression oprimaof tum(Fig. 4NRF2cells w(Fig. 4tant Nportan

mTORmuta

Figureof S6 inNRF2 sNRF2-eRight, qNRF2-ewere us

Shibata et al.

Cance9100


n, which encodes one of the major downstream sub-of the mTOR pathway, and a mild increase of AKT

horylation in mutant NRF2-expressing cells relativecontrol, whereas no strong activation of mitogen-

ted protein kinase was observed in any of the clonesA). Downregulation of NRF2 expression in Nrf2-mutatedcells was associated with reduced S6 kinase activationB).understand the molecular link between the two path-we returned to the gene expression profiling becauseis a transcriptional factor and does not seem to interactly with mTOR signaling. We searched for the down-targets of mutant NRF2, which are responsible for

lating the mTOR pathway. Among the genes whose ex-on was found to be significantly altered in the microar-alysis, we found that a member of the small G protein

xpressing clones and Nrf2-mutated cancer cells treated with control and Red as loading controls. Molecular marker is indicated on the right (kDa).

, RagD, which encodes a recently discovered activatormTOR pathway (26), was significantly upregulated in

Theof the

r Res; 70(22) November 15, 2010


EK293 cells expressing mutant NRF2 (Fig. 4C). RagDdown reduced the activation of mTOR signaling (Sup-ntary Fig. S5) and NRF2 downregulation reduced RagDsion in Nrf2-mutated cancer cells (Fig. 4C). The expres-f RagD was also quantified in the clinical samples ofry lung SCC and increased in >40% (10 of 24, 41.7%)ors relative to corresponding normal lung tissueC). We found that cell proliferation of both mutantexpression HEK293 clones and Nrf2-mutated canceras reduced by downregulation of RagD expressionD; Supplementary Fig. S6). Collectively, therefore, mu-RF2 upregulates RagD expression and RagD plays an im-t role in the proliferation of Nrf2-mutated cancer cells.

pathway dependence of cell survival in NRF2nt cancer

NAs. Data represent the mean ± SD (n = 3). β-Actin and lamin B1

4. Activation of the mTOR pathway and increase of RagD expression by mutant NRF2. A, immunoblot analysis revealed increased phosphorylationmutant NRF2-expressing clones compared with the mock control. B, Nrf2-mutated cancer cell lines were treated with control siRNA (C) andiRNA (N), and S6 kinase phosphorylation (left) and NRF2 expression (right) were examined. C, left, upregulation of RagD expression in mutantxpressing clones (top). Reduced expression of RagD in NRF2-downregulated cancer cells compared with the control siRNA treatment (bottom).uantitative measurement of RagD mRNA expression in primary lung SCC and corresponding normal lung tissue. D, proliferation of mutant

agD siR

se results prompted us to examine whether inhibitionmTOR pathway has any effect on Nrf2 mutant cancer.

Cancer Research



We firpressimTORHEK29of mTshownducedmocktrol >1nmol/nmol/(SQ5and hcell linmorecer ceresistaof thetem i

previodose (duce tand EcanceBEZ23rentlyinhibiducedcells (0.099μmol/Nrf2-mnot shmutatpressiFig. 5B

Tab

Gen

PenHouParkPGCUVBKREHalmUVBZucPenAddPenHSATCAChoIDX_RuteHypFlecSIGUVBP53TartUVBTSAUbiqWieHSCCMUVB

Abb


www.a


st examined whether the proliferation of clones ex-ng mutant NRF2 was sensitive to inhibition of thepathway. We treated mutant NRF2-expressing

3 clones with rapamycin, a small-molecule inhibitorOR, and measured the resulting cell proliferation. Asin Fig. 5A, rapamycin treatment more effectively re-cell proliferation in Nrf2 mutant clones than in theand wild-type NRF2-expressing ones (IC50 values: con-0 nmol/L, TR1 0.29 nmol/L, TR2 0.3 nmol/L, LF1 0.34L, LF2 0.53 nmol/L, WT1 9.94 nmol/L, WT2 >10L). We then tested low wild-type NRF2 expressingand QG56), high wild-type NRF2-expressing (A549),igh mutant NRF2-expressing (LK2 and EBC1) canceres (Fig. 1A). Nrf2 mutated cancer cell lines exhibitedsensitivity to rapamycin than did Nrf2 wild-type can-ll lines (Fig. 5A). However, Nrf2 mutant cells showednce to rapamycin at a higher dose, probably because
more transformed phenotypes such as feedback sys-nvolving phosphoinositide 3-kinase (PI3K) reported
with Nmice

le h mutant N

e Correc

g_ 0sti 0_R 0

0_S 0BS 0o 0_N 0ch 0g_ 0ya 0g_ 000 0

0les 0TS 0lla 0ox 0hn 0_C 0_N 0Hy 0e_ 0_N 0_H 0ui 0lan 0

reviation: FDR, false discovery rate.

acrjournals.org


usly in other cancer cells (27, 28). Consistently high-1 nmol/L) rapamycin treatment did not robustly re-he phosphorylation of mTOR pathway targets in LK2BC1 cells (Fig. 5B). To overcome this resistance inr cells, we then tested a dual kinase inhibitor (NVP-5), which targets both mTOR and PI3K and is cur-being evaluated in clinical trials. NVP-BEZ235 robustlyted mTOR pathway activation and significantly re-the cell proliferation of Nrf2-mutated lung cancerFig. 5A and B; IC50 values: LK2 0.042 μmol/L, EBC1μmol/L, SQ5 >5 μmol/L, QG56 >5 μmol/L, A549 >5L). Similar drug sensitivity was also observed in anutated head and neck cancer cell line (HO-1-U1; dataown). NVP-BEZ235 also induced apoptosis in Nrf2-ed cancer cells as revealed by an increase in the ex-on of cleaved poly(ADP-ribose) polymerase (PARP;). Finally, we challenged an in vivo treatment model
VP-BEZ235. NVP-BEZ235 was administered p.o. to
bearing s.c. xenografts of LK2 or SQ5 cells, and we

ssion
1. Gene signatures associated wit RF2 expre
FDR (%

000.0020.0120.0110.0130.0160.0160.0180.0160.0160.0160.020.020.0220.0220.0210.0290.0360.0360.0340.0370.0390.0380.0370.0370.037

0. © 2010 Am

Assoc

mTOxi

LipidOxiMetLipidOxi

mT

mTLipidMetLipidLipid

TNFOxi

ProOxi

MetTNF

Cancer Res; 70(22) No

erican Association fo

signature
ted P ) iated pathway
Rapamycin_DN
OR pathway s_ROS dative stress ARALPHA_MOD
metabolism
CC_UP dative stress _TCA_CYCLE abolic signal s_CEBP_UP metabolism HEK3_C0 dative stress i_Epithelial_UP .001 Cancer Leucine_DN OR pathway _K562_HEMIN_Treatment Glutamine_DN OR pathway 100_Biosynthesis_OF_Steroids metabolism
.001
abolic signal terol_Biosynthesis metabolism A_UP_Cluster5 metabolism _HEPATGFSNDCS_UP ia_RCC_NOVHL_UP Hypoxia er_Kidney_Transplant_Well_PBL_UP D40PathwayMap /NF-κB/virus HEK3_C3 dative stress poxiapathway .002 Hypoxia Plasma_Blastic genitor cell HEK2_DN dative stress epatoma_Cancer_DN .004 Cancer tin_Mediated_Proteolysis abolic signal d_Hepatitis_B_Induced /NF-κB/virus aterProgenitors_Adult genitor cell _L 0 0.036 Pro
V_HCMV_Time Course_24HRS_DN 0.002 0.038 TNF/NF-κB/virus_NHEK3_ALL 0 0.041 Oxidative stress

vember 15, 2010 9101

r Cancer


foundLK2 cImmuof mTand incaspasrelativFig. S7

Discu

In tproveand in

markscell limaligand mbe exof themortain vivotant NadditiThe

rectly

Figureor moccancerviable cmTORand 1 nof apopmice, wwhereain phospho-S6 protein and increase in cleaved PARP in NVP-BEZ235 treated tumor (right) compared with the control (left). Bar, 100 μm. D, schematicreprese esent t

Shibata et al.

Cance9102


that this drug significantly inhibited the growth ofells in comparison with placebo treatment (Fig. 5C).nohistologic analysis confirmed significant inhibitionOR activation (indicated by decreased phospho-S6)duction of apoptosis (increase of cleaved PARP ande-3 activation) in NVP-BEZ235–treated LK2 tumorse to the placebo-treated ones (Fig. 5C; Supplementary).

ssion

his study, we found that mutant NRF2 confers im-

ntation of mTOR pathway dependence in Nrf2-mutated cancer. Data repr

d cell proliferation, anchorage-independent growth,vivo tumorigenicity, which are the fundamental hall-

cinogeand id

r Res; 70(22) November 15, 2010


of an oncogene, in an immortalized human epithelialne. Moreover, cells expressing mutant NRF2 showednant phenotypes such as enhanced local invasionetastasis to distant organ. These phenotypes couldplained by acquisition of resistance to or reductionoxidative stress induced by the intrinsic cellular im-lization process and extrinsic environmental factors, but our gene expression analysis revealed that mu-RF2 affects a broader range of biological processes inon to the antioxidant pathway.aforementioned results imply that mutant NRF2 di-or indirectly regulates a set of genes related to the car-

he mean ± SD (n = 3, except in D).

5. mTOR pathway dependence of cell survival in NRF2 mutant cancer. A, left, clones expressing mutant (TR and LF) and wild-type (WT) NRF2k were treated with different concentrations of rapamycin, and the number of viable cells was measured after 3 d and plotted. Middle, Nrf2-mutatedcells (LK2 and EBC1) and Nrf2 wild-type cancer cells (A549, SQ5, and QG56) were treated with rapamycin in a similar way, and the number ofells was plotted. Right, Nrf2-mutated and Nrf2 wild-type cancer cells were treated with different concentration of NVP-BEZ235, a dual inhibitor ofand PI3K, and viable cell number was measured and plotted. B, cancer cells were treated with DMSO (Ctrl), 0.5 μmol/L NVP-BEZ235 (BEZ),mol/L rapamycin (Rap) for 48 h, and phosphorylation of S6 and 4E-BP1 was examined. Cleaved PARP (C-PARP) was examined as an indicatortosis. Total 4E-BP1, S6, and β-actin were used as loading controls. Molecular marker is indicated on the right (kDa). C, top, immunodeficienthich contained s.c. LK2 tumors, were randomly segregated into two groups. NVP-BEZ235 was daily given to mice in the treated group (n = 8),s placebo was given to the control group (n = 8) for 3 wk. The relative tumor size in each group was plotted. *, P < 0.001. Bottom, decrease

nesis process. To clarify this transcriptional networkentify therapeutic targets in Nrf2-mutated cancer, we

Cancer Research



perforway agroupin addNRF2whichstressoverlapreviohepatoto claKeap1pathw(31) aned witlular sThe

pathwstrengclonesbitorssensitenvirocellulanamicautoppossibwith tNRF2expreswhichand wsmalldirectendosanothcogen36). Alry lungene fportedRagDated wplemeof thematinrevealshownmediaCon

mTORdent,of sucported

(37) astratetransccaugh39). Pmechaa duathe meffecticancealso suficial iclear hnegatiular iwouldoccursmutatpromilung cMu

genesinificanstudietracelldomaicanceand enpontinand thgene.batterheparwhichNRF2-to besion. Bencemolecassocithis hture ineck rdepenway inpathwclonesprognmutanIn c

matepathw

6 M.Y. a


www.a


med gene expression profiling and conducted “path-nalysis” to connect the expression patterns of genes with specific phenotypes. This analysis revealed that,ition to the previously known NRF2 targets, mutantaltered the expression of a battery of genes, many ofhad not been previously assigned to the oxidativepathway, in human epithelial cells. This result partlyps (e.g., genes related to lipid metabolism) with theus gene expression analyses of Keap1-deleted mousecytes (29, 30) and our findings of a genome-wide studyrify the in vivo sites of NRF2 binding to chromatin inmutated cancer cells.6 To delineate the molecularays affected by mutant NRF2, we performed GSEAd discovered significant enrichment of genes associat-h the mTOR pathway, lipid metabolism, and other cel-tresses such as hypoxia and viral infection.finding that mutant NRF2 is implicated in the mTORay had not been anticipated. This relationship is alsothened by the fact that NRF2 mutant–expressingand cancer cells are sensitive to mTOR pathway inhi-(rapamycin and NVP-BEZ235). The mTOR pathwayizes cells to changes in multiple extracellular signals/nments (depletion of growth factors and nutrients orr stresses such as hypoxia and DNA damage) and dy-ally regulates translation, ribosome biogenesis, macro-hagy, and other processes (32, 33); thus, it would bele that NRF2-mediated stress adaptation may convergehis environmental sensing system. To determine howmodulates the mTOR pathway, we focused on the genesion data and have revealed one candidate, RagD,is a positive regulator of the mTOR pathway (33, 34)as induced by mutant NRF2. RagD is a member of theG-protein family and enhances mTOR activity throughbinding and by recruiting the mTOR complex to theomal fraction (26), where mTOR is activated by Rheb,er small G protein, which has been shown to be an on-e in a mouse model as well as in chick fibroblasts (35,though the expression of RagD was increased in prima-g cancers, the transcriptional regulation of the Ragamily has been largely unexplored; it has only been re-that amino acid addition triggers its activation (26).

expression in cancer samples was significantly associ-ith Hmox1, another Nrf2 target gene (P = 0.025; Sup-ntary Fig. S8). However, the putative promoter regionRagD gene contains no ARE sequence and our chro-immunoprecipitates sequence (ChIP-seq) analysis hased that RagD is not a direct target of NRF2 (data not). These findings suggest that an additional regulatorytor links the two molecules (Fig. 5D).tinuous NRF2 activation probably stimulates thepathway on which cancer cells have become depen-

making it a potential therapeutic target. Identification
h indirect molecular addiction has been recently re-as NF-κB activation in Kras/Tp53mutated cancer cells
NRF2leculaviewpcancemTORgatednd colleagues in preparation.

acrjournals.org


nd could expand a window of anticancer therapeuticgy especially when targeting oncoproteins encodingriptional factors. Recently, mTOR inhibitors havet much attention as promising anticancer drugs (38,revious studies elucidated the existence of feedbacknism against mTOR inhibitors in cancer (27, 28) andl kinase inhibitor (NVP-BEZ235), which targets bothTOR and PI3Ks, has been recently shown to be moreve in many solid tumors, including Kras mutated lungr or Pik3ca mutated cancers (40–42). The present studypports that the dual kinase inhibition would be bene-n treating Nrf2mutated cancer. However, it remains un-ow mutant NRF2 activates AKT in the presence of theve feedback and further studies to uncover the molec-nteraction of NRF2 and PI3K/AKT/mTOR signalingbe required. Notably, we reported that Nrf2 mutationmore frequently in lung cancers without Egfr and Krasions (15). Therefore, the mTOR pathway may also be asing therapeutic target in the Nrf2-mutated subtype ofancer.tant NRF2 promoted cell motility, local invasion, angio-s, and liver metastasis. GO profiling also revealed sig-t enrichment of genes related to cell motility. Previouss have shown that upregulation of osteopontin, an ex-ular matrix-associated protein with multiple functionalns, is associated with metastatic activity of a range ofrs including head and neck and lung cancers (21, 43)hancement of angiogenesis (22). We found that osteo-was highly expressed in cells expressing mutant NRF2at NRF2 seems to directly activate the osteopontinIn addition to osteopontin, mutant NRF2 induced ay of genes such as matrix metalloproteinase 12 andin-binding epidermal growth factor-like growth factor,are associated with tumor metastasis (44, 45). Thusmediated adaptation to environmental stress seemstightly associated with tumor metastasis and progres-ecause our experimental model showed that the pres-of mutant NRF2 confers metastatic potential, theular signature resulting from NRF2 activation may beated with poor outcome in clinical cases. We testedypothesis and confirmed that the mutant NRF2 signa-s indeed a prognostic factor in SCC of the head andegion and probably that of the lung. Interestingly, an in-dent study focusing on the prognostic molecular path-lung adenocarcinoma has identified the Peng_Rap_DNay, a top-ranking gene set enriched in mutant NRF2, as a signature that is significantly associated with poorosis (46). This also supports our contention that thet NRF2 signature is a marker of poor prognosis.onclusion, our results show the existence of an inti-association between oxidative stress and the mTORay in cancer and further highlight the potential ofactivation as a novel biomarker for personalized mo-r therapies or prognostic assessment. From a clinicaloint, it is important to note that NRF2 activation inr is significantly associated with response to anti-
pathway treatments, which are being widely investi-in clinical trials for many cancers.



Disclo

No p

Grant

A gra

Ministry

19-1, P

and Pro

the NatThe

paymenadvertisthis fac

Refe1. Ba

cenche

2. WhRe

3. Luogen

4. Jolcha92:

5. Sinagi

6. Meind

7. Goge

8. Mopo54

9. ItoerogeRe

10. Itotivato

11. Mcsomnegpre

12. YaofMo

13. RagencerSc

14. Aoandgp

15. ShimpPro

16. Shcoin

17. OhNrfRe

18. Sininte

19. Sjöqu31

Shibata et al.

Cance9104


sure of Potential Conflicts of Interest

otential conflicts of interest were disclosed.

Support

nt-in-aid for the Comprehensive 10-Year Strategy for Cancer Control,

of Health, Labor and Welfare, Japan grant-in-aid for cancer research

rincess Takamatsu Cancer Research Fund research grant 08-24007,

Rece

OnlineF

blom T, Jones S, Wood LD, et al. The consensus coding se-ences of human breast and colorectal cancers. Science 2006;4:268–74.

20. Lelat

21. Huateca

22. DathrOn

23. Chheex

24. Toindpa

25. Pemide

26. Sator32

27. CamTde11

28. Susurap

29. Okletag33

30. Yaacprocin

31. Mogereg

32. Wume

33. Malat

34. KimGT

35. NamTon

36. Jiatra

37. Mesig20

38. SaNa

r Res; 70(22) November 15, 2010


gram for Promotion of Fundamental Studies in Health Sciences of

ional Institute of Biomedical Innovation, Japan.costs of publication of this article were defrayed in part by thet of page charges. This article must therefore be hereby markedement in accordance with 18 U.S.C. Section 1734 solely to indicatet.

ived 01/30/2010; revised 09/15/2010; accepted 09/21/2010; published

irst 11/09/2010.

rencesrtkova J, Rezaei N, Liontos M, et al. Oncogene-induced senes-ce is part of the tumorigenesis barrier imposed by DNA damageckpoints. Nature 2006;444:633–7.itesell L, Lindquist SL. HSP90 and the chaperoning of cancer. Natv Cancer 2005;5:761–72.J, Solimini NL, Elledge SJ. Principles of cancer therapy: onco-e and non-oncogene addiction. Cell 2009;136:823–37.ly C, Morimoto RI. Role of the heat shock response and molecularperones in oncogenesis and cell death. J Natl Cancer Inst 2000;1564–72.gh KK. Mitochondrial dysfunction is a common phenotype inng and cancer. Ann N Y Acad Sci 2004;1019:260–4.na S, Ortega A, Estrela JM. Oxidative stress in environmental-uced carcinogenesis. Mutat Res 2009;674:36–44.ttlieb E, Tomlinson PM. Mitochondrial tumour suppressors: anetic and biochemical update. Nat Rev Cancer 2005;5:857–66.tohashi H, Yamamoto M. Nrf2-1 defines a physiologically im-rtant stress response mechanism. Trends Mol Med 2004;10:9–57.h K, Chiba T, Takahashi S, Ishii T, et al. An Nrf2/small Maf het-dimer mediates the induction of phase II detoxifying enzymenes through antioxidant response elements. Biochem Biophyss Commun 1997;236:313–22.h K, Wakabayashi N, Katoh Y, et al. Keap1 represses nuclear ac-tion of antioxidant responsive elements by Nrf2 through bindingthe amino-terminal Neh2 domain. Genes Dev 1999;13:76–86.Mahon M, Itoh K, Yamamoto M, et al. Keap1-dependent protea-al degradation of transcription factor Nrf2 contributes to theative regulation of antioxidant response element-driven gene ex-ssion. J Biol Chem 2003;278:21592–600.mamoto T, Suzuki T, Kobayashi A, et al. Physiological significancereactive cysteine residues of Keap1 in determining Nrf2 activity.l Cell Biol 2008;28:2758–70.mos-Gomez M, Kwak MK, Dolan PM, et al. Sensitivity to carcino-esis is increased and chemoprotective efficacy of enzyme indu-s is lost in nrf2 transcription factor-deficient mice. Proc Natl Acadi U S A 2001;98:3410–5.ki Y, Hashimoto AH, Amanuma K, et al. Enhanced spontaneousbenzo(a)pyrene-induced mutations in the lung of Nrf2-deficient

t δ mice. Cancer Res 2007;67:5643–8.ibata T, Ohta T, Tong KI, et al. Cancer related mutations in NRF2air its recognition by Keap1-3 E3 ligase and promote malignancy.c Natl Acad Sci U S A 2008;105:13568–73.ibata T, Kokubu A, Gotoh M, et al. Genetic alteration of Keap1nfers constitutive Nrf2 activation and resistance to chemotherapygall bladder cancer. Gastroenterology 2008;135:1358–68.ta T, Iijima K, Miyamoto M, et al. Loss of Keap1 function activates2 and provides advantages for lung cancer cell growth. Cancers 2008;68:1303–9.gh A, Misra V, Thimmulappa RK, et al. Dysfunctional KEAP1-2raction in non-small-cell lung cancer. PLoS Med 2006;3:e420.

e DF, Kuo HP, Liu M, et al. KEAP1 E3 ligase-mediated downregu-ion of NF-κB signaling by targeting IKKβ. Mol Cell 2009;36:131–40.Z, Lin D, Yuan J, et al. Overexpression of osteopontin is associ-d with more aggressive phenotypes in human non-small cell lungncer. Clin Cancer Res 2005;11:4646–52.i J, Peng L, Fan K, et al. Osteopontin induces angiogenesisough activation of PI3K/AKT and ERK1/2 in endothelial cells.cogene 2009;28:3412–22.ung CH, Parker JS, Karaca G, et al. Molecular classification ofad and neck squamous cell carcinomas using patterns of genepression. Cancer Cell 2004;5:489–500.mida S, Koshikawa K, Yatabe Y, et al. Gene expression-based,ividualized outcome prediction for surgically treated lung cancertients. Oncogene 2004;23:5360–70.ng T, Golub TR, Sabatini DM. The immunosuppressant rapamycinmics a starvation-like signal distinct from amino acid and glucoseprivation. Mol Cell Biol 2002;22:5575–84.ncak Y, Peterson TR, Shaul YD, et al. The Rag GTPases bind rap-and mediate amino acid signaling to mTORC1. Science 2008;

0:1496–501.rracedo A, Ma L, Teruya-Feldstein J, et al. Inhibition ofORC1 leads to MAPK pathway activation through a PI3K-pendent feedback loop in human cancer. J Clin Invest 2008;8:3065–74.n SY, Rosenberg LM, Wang X, et al. Activation of Akt and eIF4Ervival pathways by rapamycin-mediated mammalian target ofamycin inhibition. Cancer Res 2005;65:7052–8.awa H, Motohashi H, Kobayashi A, et al. Hepatocyte-specific de-ion of the keap1 gene activates Nrf2 and confers potent resistanceainst acute drug toxicity. Biochem Biophys Res Commun 2006;9:79–88.tes MS, Tran QT, Dolan PM, et al. Genetic versus chemoprotectivetivation of Nrf2 signaling: overlapping yet distinct gene expressionfiles between Keap1 knockout and triterpenoid-treated mice. Car-ogenesis 2009;30:1024–31.otha VK, Lindgren CM, Eriksson KF, et al. PGC-1α-responsivenes involved in oxidative phosphorylation are coordinately down-ulated in human diabetes. Nat Genet 2003;34:267–3.llschleger S, Loewith R, Hall MN. TOR signaling in growth andtabolism. Cell 2006;124:471–84.XM, Blenis J. Molecular mechanisms of mTOR-mediated trans-

ional control. Nat Rev Mol Cell Biol 2009;10:307–18.E, Goraksha-Hicks P, Li L, et al. Regulation of TORC1 by Rag

Pases in nutrient response. Nat Cell Biol 2008;10:935–45.rdella C, Chen Z, Salmena L, et al. Aberrant Rheb-mediatedORC1 activation and Pten haploinsufficiency are cooperativecogenic events. Genes Dev 2008;22:2172–7.ng H, Vogt PK. Constitutively active Rheb induces oncogenicnsformation. Oncogene 2008;27:5729–40.ylan E, Dooley AL, Feldser DM, et al. Requirement for NF-κBnalling in a mouse model of lung adenocarcinoma. Nature
09;462:104–7.batini DM. mTOR and cancer: insights into a complex relationship.t Rev Cancer 2006;6:729–34.
Cancer Research



39. Ench

40. Ko3-kradcan

41. Liupheliccer

42. Seinhcel

43. Co

protum

44. Ho12ea11

45. Keprecobe

46. Eb


www.a


gelman JA. Targeting PI3K signalling in cancer: opportunities,allenges and limitations. Nat Rev Cancer 2009;9:550–62.nstantinidou G, Bey EA, Rabellino A, et al. Dual phosphoinositideinase/mammalian target of rapamycin blockade is an effectiveiosensitizing strategy for the treatment of non-small cell lungcer harboring K-RAS mutations. Cancer Res 2009;69:7644–52.TJ, Koul D, LaFortune T, et al. NVP-BEZ235, a novel dual phos-atidylinositol 3-kinase/mammalian target of rapamycin inhibitor,its multifaceted antitumor activities in human gliomas. Mol Can-Ther 2009;8:2204–10.

rra V,MarkmanB, Scaltriti M, et al. NVP-BEZ235, a dual PI3K/mTOR
ibitor, prevents PI3K signaling and inhibits the growth of cancerls with activating PI3K mutations. Cancer Res 2008;68:8022–30.ppola D, Szabo M, Boulware D, et al. Correlation of osteopontin
nalunCa

acrjournals.org


tein expression and pathological stage across a wide variety ofor histologies. Clin Cancer Res 2004;10:184–90.fmann HS, Hansen G, Richter G, et al. Matrix metalloproteinase-expression correlates with local recurrence and metastatic dis-se in non-small cell lung cancer patients. Clin Cancer Res 2005;:1086–92.rkelä E, Ala-aho R, Klemi P, et al. Metalloelastase (MMP-12) ex-ssion by tumour cells in squamous cell carcinoma of the vulvarrelates with invasiveness, while that by macrophages predictstter outcome. J Pathol 2002;198:258–69.i H, Tomida S, Takeuchi T, et al. Relationship of deregulated sig-ling converging onto mTOR with prognosis and classification of
g adenocarcinoma shown by two independent in silico analyses.ncer Res 2009;69:4027–35.



2010;70:9095-9105. Published OnlineFirst November 9, 2010.Cancer Res Tatsuhiro Shibata, Shigeru Saito, Akiko Kokubu, et al. Growth Signaling Pathway

Related Factor 2 Mutation on the mTOR−of Oncogenic NF-E2 Global Downstream Pathway Analysis Reveals a Dependence

Updated version

10.1158/0008-5472.CAN-10-0384doi:

Access the most recent version of this article at:

Material

Supplementary

http://cancerres.aacrjournals.org/content/suppl/2012/02/22/0008-5472.CAN-10-0384.DC1

Access the most recent supplemental material at:

Cited articles

http://cancerres.aacrjournals.org/content/70/22/9095.full#ref-list-1

This article cites 46 articles, 19 of which you can access for free at:

Citing articles

http://cancerres.aacrjournals.org/content/70/22/9095.full#related-urls

This article has been cited by 7 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] 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/70/22/9095To request permission to re-use all or part of this article, use this link

Research. on March 21, 2020. © 2010 American Association for Cancercancerres.aacrjournals.org Downloaded from


http://cancerres.aacrjournals.org/lookup/doi/10.1158/0008-5472.CAN-10-0384

http://cancerres.aacrjournals.org/content/suppl/2012/02/22/0008-5472.CAN-10-0384.DC1

http://cancerres.aacrjournals.org/content/70/22/9095.full#ref-list-1

http://cancerres.aacrjournals.org/content/70/22/9095.full#related-urls

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

mailto:[email protected]

http://cancerres.aacrjournals.org/content/70/22/9095


cancer research global downstream pathway analysis reveals a … · molecular and cellular...

Documents