ELK4 neutralization sensitizes glioblastoma to apoptosis through downregulation of the anti-apoptotic protein Mcl-1

ELK4 neutralization sensitizes glioblastomato apoptosis through downregulationof the anti-apoptotic protein Mcl-1

Bryan W. Day†, Brett W. Stringer†, Mark D. Spanevello, Sara Charmsaz,Paul R. Jamieson, Kathleen S. Ensbey, Jacinta C. Carter, Joanne M. Cox,Vicky J. Ellis, Christopher L. Brown, David G. Walker, Po L. Inglis,Suzanne Allan, Brent A. Reynolds, Jason D. Lickliter, and Andrew W. Boyd

Brain Cancer & Leukaemia Foundation Research Units, Queensland Institute of Medical Research, Herston,

4029, Queensland, Australia (B.W.D., B.W.S., M.D.S., S.C., P.R.J., K.S.E., J.C.C., J.M.C., S.A., J.D.L., A.W.B.);

Queensland Brain Institute, University of Queensland, St Lucia, 4067, Queensland, Australia (B.W.D., M.D.S.,

A.W.B.); The ESKITIS Institute for Cell and Molecular Therapies, Griffith University, Nathan, 4109,

Queensland, Australia (V.J.E., C.L.B.); BrizBrain & Spine Research Foundation, Auchenflower,

4066, Queensland, Australia (D.G.W.); Cancer Services, Royal Brisbane and Women’s Hospital, Herston,

4029 Queensland, Australia (P.L.I., S.A.); McKnight Brain Institute, University of Florida, Gainesville, Florida,

32611, USA (B.A.R.); Department of Medicine, University of Queensland, Herston, 4029, Queensland,

Australia (A.W.B.)

Glioma is the most common adult primary brain tumor.Its most malignant form, glioblastoma multiforme(GBM), is almost invariably fatal, due in part to theintrinsic resistance of GBM to radiation- and che-motherapy-induced apoptosis. We analyzed B-cell leu-kemia–2 (Bcl-2) anti-apoptotic proteins in GBM andfound myeloid cell leukemia–1 (Mcl-1) to be thehighest expressed in the majority of malignant gliomas.Mcl-1 was functionally important, as neutralization ofMcl-1 induced apoptosis and increased chemotherapy-induced apoptosis. To determine how Mcl-1 was regu-lated in glioma, we analyzed the promoter and identifieda novel functional single nucleotide polymorphism in anuncharacterized E26 transformation-specific (ETS)binding site. We identified the ETS transcription factorELK4 as a critical regulator of Mcl-1 in glioma, sinceELK4 downregulation was shown to reduce Mcl-1 andincrease sensitivity to apoptosis. Importantly the pres-ence of the single nucleotide polymorphism, which

ablated ELK4 binding in gliomas, was associated withlower Mcl-1 levels and a greater dependence on Bcl-xL. Furthermore, in vivo, ELK4 downregulationreduced tumor formation in glioblastoma xenograftmodels. The critical role of ELK4 in Mcl-1 expressionand protection from apoptosis in glioma defines ELK4as a novel potential therapeutic target for GBM.

Keywords: apoptosis, ELK4, glioblastoma, glioma,Mcl-1.

Treatment of glioblastoma multiforme (GBM),the most malignant primary adult braintumor, involves surgical resection followed

by radiation and chemotherapy.1 Therapy is almostnever curative, due in part to the infiltrative natureof these tumors and their intrinsic resistance to radi-ation and chemotherapy. Even with optimal treatment,the median survival is less than 15 months, with only10% of patients surviving 2 years without diseaserecurrence.2 This dismal situation highlights a pressingneed to identify new therapeutic targets to improvepatient outcome.

One important target for molecular therapy is theprogrammed cell death, or apoptotic, machinery ofthe cell. This is regulated by both pro-apoptotic and

Corresponding Author: Dr. Bryan W. Day, Queensland Institute of

Medical Research, P.O. Royal Brisbane Hospital, Queensland, 4029,

Brisbane, Australia ([email protected]).

†B.W.D, B.W.S. contributed equally to this paper.

Received August 4, 2010; accepted July 6, 2011.

Neuro-Oncology 13(11):1202–1212, 2011.doi:10.1093/neuonc/nor119 NEURO-ONCOLOGYAdvance Access publication August 16, 2011

# The Author(s) 2011. Published by Oxford University Press on behalf of the Society for Neuro-Oncology.All rights reserved. For permissions, please e-mail: [email protected].

at University of Q

ueensland on July 19, 2012http://neuro-oncology.oxfordjournals.org/

Dow

nloaded from

anti-apoptotic (Bcl-2, Bcl-xL, Bcl-w, Mcl-1, and Bfl/A1) B-cell leukemia–2 (Bcl-2) proteins. Imbalancesbetween these underlie a number of neoplastic malig-nancies.3 The p53 tumor suppressor gene, which isfrequently mutated in GBM, normally inhibits thepro-apoptotic BH3-only proteins Puma and Noxa,preventing apoptosis.4–6 Agents that mimic BH3 pro-teins therefore might trigger apoptosis in tumors,even those harboring p53 mutations. One suchagent, ABT-737, which induces cell death in tumorcell lines, is a potent inhibitor of Bcl-2, Bcl-xL, andBcl-w (Ki ≤ 1nM) but has far lower affinity formyeloid cell leukemia–1 (Mcl-1) (Ki . 1 mM).7

Overexpression of Mcl-1 has been shown to attenuateABT-737 sensitivity both in a mouse lymphoma modeland in acute myeloid leukemia,8,9 whereas ABT-737has been shown to induce apoptosis in human carci-noma cell lines when Mcl-1 was neutralized.10,11

We show that Mcl-1 is the most highly expressedanti-apoptotic Bcl-2 family member in high-gradeglioma (n ¼ 51). In addition, in analyzing the Mcl-1 pro-moter in GBM cells we identified a novel, functionalG.A single nucleotide polymorphism (SNP) in a consen-sus E26 transformation-specific (ETS) transcriptionfactor binding site. The wild-type (WT) G form of theSNP actively binds nuclear proteins from GBM cells,whereas the A form does not. The SNP correlates withdecreased promoter activity and lower Mcl-1 levels.Importantly, GBM cultures harboring the SNP showedincreased sensitivity to ABT-737 and cisplatin treatment.Furthermore, apoptosis was observed in GBM cell linesand neurosphere lines without the ETS SNP whenMcl-1 was neutralized. We show that ELK4 is the domi-nant ETS domain transcription factor in GBM and bindsto the identified Mcl-1 promoter ETS site. In addition,significant correlation (r ¼ 0.76, P , .0001) betweenELK4 and Mcl-1 mRNA in 51 glioma specimens wasfound. Downregulation of ELK4 by small-interfering(si)RNA resulted in loss of Mcl-1 expression andincreased sensitivity to ABT-737 and cisplatin.Conversely, ELK4 overexpression increased Mcl-1levels and increased resistance to higher concentrationsof chemotherapy. Furthermore, in vivo downregulationof ELK4 reduced Mcl-1 levels and reduced tumor for-mation in both subcutaneous and intracranial gliomaxenograft models. These findings demonstrate ELK4 tobe a critical regulator of Mcl-1 and highlight bothELK4 and Mcl-1 as potential therapeutic targets in GBM.

Materials and Methods

Cell Culture

D series GBM cell lines were obtained from the DukeUniversity Medical Center and U series from theAmerican Type Culture Collection. Cell lines were cul-tured in Roswell Park Memorial Institute–1640medium with 10% fetal bovine serum (JRHBiosciences) at 378C in humidified air/5% CO2.

Neurospheres were generated from a primary GBMsample (L1-NS) and cultured as described.12

Clinical Specimens

Primary specimens were obtained from the Royal Brisbaneand Women’s Hospital and the BrizBrain & SpineResearch Foundation after informed consent from adultpatients diagnosed with high-grade glioma (Table S1).

ELK4 cDNA Expression Construct

ELK4 cDNA was cloned into the plasmid elongationfactor internal ribosome entry site (pEF-IRES)–puro6mammalian expression vector.13

Mcl-1 cDNA Expression Construct

Mcl-1 cDNA was cloned into the pIRES2-DsRed-Express mammalian expression vector (JRH Biosciences).

Noxa Peptide Treatment

Mcl-1 was neutralized using a Noxa peptide asdescribed.14

Mouse Tumor Models

Cells were injected into 5-week-old non-obese diabetic/severe combined immunodeficiency (NOD/SCID) mice.For subcutaneous experiments, animals were euthanizedwhen tumors exceeded 1 cm in diameter. For intracra-nial experiments, animals were monitored for signs oftumor formation (rough coat, hunching, weight loss).Animals were culled and hematoxylin-and-eosin sec-tions were prepared to identify tumors.

DNA Extraction

Genomic DNA was extracted as described.15

Relative Quantitation by Real-time PCR

RNA was extracted using TRIzol (Invitrogen). First strandcDNA was synthesized using random hexamers andSuperscript III (Invitrogen). Real-time PCR was carriedout using SYBR (Synergy Brands) Green PCR MasterMix (Applied Biosystems). Cycling conditions were15 min at 958C and 30 cycles of 30 s at 958C, 30 s at558C, and 30 s at 728C. Full primer sequences are listedin the supplementary materials and methods (Table S2).

Protein Analysis

Immunoblots and immunohistochemistry were per-formed as described.16 Complete immunoblots areshown in Figure S6.

Day et al.: ELK4 and Mcl-1: Targets in glioblastoma

NEURO-ONCOLOGY † N O V E M B E R 2 0 1 1 1203

at University of Q

ueensland on July 19, 2012http://neuro-oncology.oxfordjournals.org/

Dow

nloaded from

Luciferase Reporter Assay

The Mcl-1 promoter was cloned into pGL2-Basic(Promega). Reporter assays were conducted using theDual-Luciferase Reporter Assay System (Promega).

Electrophoretic Mobility Shift Assay

Electrophoretic mobility shift assays (EMSAs) were per-formed with end-labeled double stranded oligonucleo-tides using GBM nuclear extracts as described.17

Chromatin Immunoprecipitation Assay

ChIP assays were performed as described (http://genomics.ucdavis.edu/farnham/protocols/chips.html).Primer sequences and antibodies were used as in thesupplementary materials and methods.

siRNA/shRNA Gene Knockdown

Mcl-1 and ELK4 small hairpin (sh)RNA was clonedinto pSuperior.neo+gfp (Oligoengine). siRNA StealthDuoPak (cat # 1293609) (Invitrogen) and shRNAsequences are listed in the supplementary materials andmethods (Table S1).

Annexin V Staining

Apoptosis was determined by Annexin V cell staining asdescribed.16

ETS SNP Assignment

The novel –116G.A Mcl-1 promoter SNP has SingleNucleotide Polymorphism Database accession numberss107739141.

Statistical Analysis

Student’s t-test determined the probability of difference;P , .05 was considered significant. All statistical testswere 2-sided. The correlation coefficient was determinedusing a Spearman rank.

Results

Mcl-1 is the Dominant Anti-apoptotic Bcl-2–relatedProtein in High-grade Glioma

Expression of anti-apoptotic Bcl-2 family members wasassessed in a series of resected gliomas (tumor type andWorld Health Organization grade listed in Table S1).Expression of Mcl-1, Bcl-xL, Bcl-2 (L12), Bcl-w, Bcl-2,and A1 mRNA (as well as the pro-apoptotic Mcl-1short isoform) was determined by quantitative PCR(qPCR) in 51 resected brain tumors (�75% GBM), aswell as in normal brain (Fig. 1A). While the totalexpression was variable from tumor to tumor, the

expression profile was characterized by substantiallevels of Mcl-1 and to a lesser degree Bcl-xL and Bcl-2(L12) in glioma tissues. In this study 74% (38/51) ofsamples expressed high levels of Mcl-1 compared withnormal total brain RNA, which showed only low levels.Bcl-xL was also significantly elevated in a proportion ofsamples but was greater than Mcl-1 in only 2 cases.Mcl-1 and Bcl-xL protein levels were also evaluated in4 common GBM cell lines (Fig. 1B). Notably, 3 of the 4lines tested expressed Mcl-1 (U87, U118, and D645,but not U373). All 4 lines expressed Bcl-xL, with U87and D645 expressing high levels. Mcl-1 and Bcl-xLlevels also were found to be high in a series of tumor-derived primary GBM early passaged lines (Fig. S1).Immunocytochemical analysis of Mcl-1 in U87, U118,D645, and U373 indicated that Mcl-1 protein expressionparalleled the mRNA expression pattern (Fig. 1C). Takentogether, this expression profile suggested a potentiallysignificant role of Mcl-1 and, to a lesser extent, Bcl-xLin high-grade glioma survival.

Identification of a Novel SNP in an ETS Site in the Mcl-1Promoter

Targeted sequencing of genomic DNA from GBMtissue samples and normal control DNA identified apreviously unknown SNP (dbSNP, ss107739141) inthe Mcl-1 promoter in high-grade glioma. This G.Atransversion was located in a consensus ETS transcrip-tion factor binding site located 116 bp upstream ofthe major transcription start site (Fig. 2A). The2116G.A SNP was identified in the GBM cell linesU251 and U373 and in 2/75 normal control peripheralblood mononuclear cell samples tested. The SNP wasnot identified in any of 37 GBM/astrocytoma tumorsamples tested, consistent with the low frequency innormal samples.

Notably, the cell lines U251 and U373, which har-bored the ETS SNP, had the lowest Mcl-1 mRNAexpression of 12 GBM cell lines tested (Fig. 2B). Toinvestigate whether the SNP might affect Mcl-1 tran-scription, the 2116G.A substitution was investigatedby reporter gene assay in 5 GBM cell lines (Fig. 2C).The 2116G.A SNP resulted in statistically significantreductions (P , .01) in Mcl-1 promoter activity (U87,36%; U118, 50%; U251, 42%; D645, 53%; andU373, 35%) compared with the WT promoter.

To determine whether the 2116G.A SNP affectedDNA binding by nuclear protein(s), EMSAs were per-formed. These revealed complexes formed between anMcl-1 WT promoter probe (nucleotides 2124 to2105) and nuclear extracts from each of 4 GBM celllines (arrowed in Fig. 2D) that were not detected, oronly weakly detected (U118), with a –116G.A SNPprobe. Cold competition EMSA with U118 nuclearextracts demonstrated that these complexes werespecific (Fig. 2D). Taken together, these data demon-strate that the –116G.A SNP specifically ablatesDNA binding by nuclear protein(s) to the Mcl-1 pro-moter ETS site in vitro.

Day et al.: ELK4 and Mcl-1: Targets in glioblastoma

1204 NEURO-ONCOLOGY † N O V E M B E R 2 0 1 1

at University of Q

ueensland on July 19, 2012http://neuro-oncology.oxfordjournals.org/

Dow

nloaded from

ELK4 is the Dominant ETS Family Member inHigh-grade Glioma

Given the identification of a functional SNP within aconsensus ETS binding site in the Mcl-1 promoter, wesought to determine whether this site bound an ETSfamily transcription factor in GBM. To begin, we sur-veyed ETS family member expression in primary GBMtumor samples and common GBM cell lines.Twenty-seven ETS family members were investigated,of which 12 showed detectable expression (Fig. 3A).Notably, 4 of the 5 GBM tumor samples and 4 of the5 GBM cell lines expressed substantial levels of ELK4compared with other ETS family members.

To determine whether ELK4 bound the Mcl-1 pro-moter ETS site, EMSA supershift and chromatin immu-noprecipitation (ChIP) assays were conducted. In thepresence of an ELK4-specific antibody, 2 supershiftedcomplexes (arrowed in Fig. 3B) were detected withnuclear extracts from 4 GBM cell lines. The appearancesof these bands correlated with a decrease in intensity ofthe primary band in all samples. This prompted us to useChIP assays to determine whether ELK4 bound theendogenous Mcl-1 promoter (Fig. 3C). ELK4 bindingto the Mcl-1 promoter was detected using 2 ELK4

antibodies. Greater immunoprecipitation was observedwith the H-167 antibody, which binds ELK4 farthestfrom its ETS DNA binding domain. The H-167 antibodywas selected for testing by qPCR in the U118 line(Fig. 3D) and a four cycle threshold difference betweenthe antibody and control was found. Together theseresults show that ELK4 binds to the Mcl-1 promoterETS site in GBM cells.

ELK4 is a Key Regulator of Mcl-1 Expression inHigh-grade Glioma

To further investigate the relationship between ELK4and Mcl-1 expression in high-grade glioma, we usedqPCR to determine the correlation between ELK4 andMcl-1 mRNA in 51 high-grade glioma samples(Fig. 4A). This showed a highly significant positive cor-relation (Spearman r ¼ 0.76, P , .0001). Sequenceanalysis determined that none of the glioma samplestested contained the 2116G.A SNP. Interrogation ofthe larger The Cancer Genome Atlas database of 424GBM specimens further confirmed the positive corre-lation between ELK4 and Mcl-1 expression in GBM(Spearman r ¼ 0.14, P , .002).

Fig. 1. Mcl-1 is highly expressed in glioblastoma. (A) Bcl-2 family anti-apoptotic protein mRNA expression was determined using qPCR in 51

high-grade glioma tissue samples. Mean of duplicate samples expressed per 1000 copies of b-actin. (B) Mcl-1 and Bcl-xL protein levels were

determined using western blotting. (C) Immunocytochemistry was performed to detect Mcl-1 expression. All GBM cell lines expressed Mcl-1

except U373.

Day et al.: ELK4 and Mcl-1: Targets in glioblastoma

NEURO-ONCOLOGY † N O V E M B E R 2 0 1 1 1205

at University of Q

ueensland on July 19, 2012http://neuro-oncology.oxfordjournals.org/

Dow

nloaded from

We also compared the level of ELK4 and Mcl-1mRNA in 3 GBM cell lines, 2 of which contained the –116G.A SNP (Fig. 4B). Significant and comparableELK4 expression was detected in all 3 lines. However,U251 and U373, both of which are heterozygousfor the –116G.A (ETS) SNP, expressed significantlyless Mcl-1 than U87. Indeed, Mcl-1 expression was infact significantly lower from the remaining WT Mcl-1allele.

To further determine whether ELK4 was the criticalETS family member regulating Mcl-1 expression inGBM, we used siRNA to downregulate ELK4 in 2GBM cell lines (Fig. 4C). There was substantial

Fig. 2. Identification of a novel, functional Mcl-1 promoter SNP in

high-grade glioma. (A) The Mcl-1 promoter region depicting the

insertion site of the 6-bp or 18-bp repeat, the 2116G.A SNP,

and the region used in reporter assays. (B) Mcl-1 mRNA

expression in glioma cultures. Mean of duplicate samples

expressed relative to 108 transcript copies of 18S ribosomal

mRNA. U373 and U251 are heterozygous for the novel

2116G.A SNP and show lower Mcl-1 mRNA levels. (C)

Luciferase reporter assays. WT Mcl-1 promoter activity compared

with the 2116G.A SNP in U87, U118, U251, D645, and U373.

Columns, mean of triplicate samples, +SD (*¼ P , .01). (D)

Electrophoretic mobility shift assay (EMSA) conducted using

U118, U251, T46, and D645 nuclear extracts showing reduced

binding of the 2116G.A SNP probe to nuclear extracts

(arrowed). EMSA cold competition assay performed using U118

nuclear extracts demonstrating this complex was competed out

only in the presence of excess unlabeled wild-type probe (arrowed).

Fig. 3. ELK4 is the dominant ETS family member in high-grade glioma

and binds the Mcl-1 promoter ETS site. (A) ETS family mRNA

expression was measured in 5 GBM tumor samples and cell lines.

Mean of duplicate samples expressed relative to 108 transcript

copies of 18S ribosomal mRNA. ELK4 is the dominantly expressed

ETS family member. (B) EMSA supershift assay demonstrating ELK4

bound the Mcl-1 promoter ETS probe. Nuclear extracts for U87,

U118, U251, and D645 were investigated. Supershifted bands

formed in the presence of an ELK4 polyclonal antibody (arrowed) for

all 4 samples with corresponding reduction in the intensity of the

arrowed band in Fig. 2D. (C) Chromatin immunoprecipitation assay

(ChIP) using 2 ELK4-specific antibodies confirmed that ELK4 bound

to the Mcl-1 promoter in the U118 GBM cell line. (D) ChIP results

for U118 were quantitated using qPCR showing a DCt value of 4.

Day et al.: ELK4 and Mcl-1: Targets in glioblastoma

1206 NEURO-ONCOLOGY † N O V E M B E R 2 0 1 1

at University of Q

ueensland on July 19, 2012http://neuro-oncology.oxfordjournals.org/

Dow

nloaded from

downregulation of ELK4 following 24 h treatment withsiRNA in both U87 and U118 cells. Significantly, thereduction in ELK4 was associated with loss of Mcl-1expression in both cases. To further confirm this, wedownregulated ELK4 using an alternate siRNAsequence in U87 cells. Results highlighted a commensu-rate reduction in Mcl-1 levels (Fig. S2). To determinewhether increased levels of ELK4 would have the oppo-site effect on Mcl-1 expression, ELK4 cDNA was stablyoverexpressed in U87 and U118 GBM cells (Fig. 4D).Results show an increase in Mcl-1 expression accompa-nying the increase in ELK4. Together, these resultsdemonstrate that ELK4 is a key regulator of Mcl-1expression in glioma cells.

Mcl-1 is a Critical Survival Determinant in GBMCultures

To investigate the potential role of Mcl-1 in GBMresistance to chemotherapy-induced apoptosis, 5GBM cell lines were treated with either cisplatin (cis-diamine dichloroplatinum (II) cisplatinum), ABT-737,or both agents combined (Fig. 5A). U87, U118, andD645 lines were largely resistant to these treatments;however, the low Mcl-1 –116G.A ETS SNP lines,U251 and U373, showed greater sensitivity (P , .05)to both ABT-737 and cisplatin as well as to theseagents combined. To confirm that Mcl-1 was confer-ring a survival advantage, we inhibited Mcl-1 functionusing a Noxa peptide previously shown to neutralizeMcl-114 (notably A1, the other target of Noxa, is notsignificantly expressed in glioma, Fig. 1). U87 cellswere treated with Noxa or control peptide alone andin combination with ABT-737 or cisplatin (Fig. 5B).Results show increased apoptosis (.20%) (P , .05)following Noxa treatment compared with the controlpeptide. Increased apoptosis also occurred with Noxatreatment in the presence of ABT-737 or cisplatin treat-ment, though the additional increases observed weremodest.

To further substantiate the observed apoptosis whenMcl-1 was inhibited, we also used shRNA to downregu-late Mcl-1 in U87 and U373 cells (Fig. 5C). Mcl-1 down-regulation induced apoptosis in U87 (29%), whereas thealready low Mcl-1 expression line, U373, was onlymildly affected (Fig. 5C).

Overexpression of ELK4, resulting in elevated Mcl-1,was anticipated to confer increased survival in GBMcells. To investigate this, we overexpressed ELK4 inU118 cells. As expected, this resulted in increasedMcl-1 expression (shown in Fig. 4C). U118 controlcells and ELK4 overexpressing cells were treatedwith cisplatin. Results show increased protection fromapoptosis when ELK4 was overexpressed (17% apopto-sis) (P , .05) relative to the control cells (25% apopto-sis), highlighting the protective effect of elevated ELK4and hence Mcl-1 against increased concentrations ofchemotherapy.

A primary derived GBM tumorsphere line (L1-NS)was also tested for inhibition of Mcl-1 cultured underneurosphere conditions. Results show markedly increasedapoptosis (.50%) (P , .05) following Noxa peptidetreatment compared with control (Fig. 5B). Once again,only modest increases in apoptosis accompanied theaddition of ABT-737 or cisplatin treatment.

The longer-term effects of ABT-737 and cisplatin ontumor initiating cells were investigated using the neuro-sphere assay.18 L1-NS was serially passaged 7 times, andthe total theoretical cell number calculated (Fig. S3).L1-NS control cultures generated 1 × 109 cells over 7weeks; this was reduced to 1.2 × 108 cells withABT-737 treatment. Cisplatin had a greater effect,with no cells remaining after 7 weeks of treatment.Notably, no cells remained after 2 weeks of treatmentwhen these agents were combined. Importantly, theseagents were able to prevent extensive self-renewal and

Fig. 4. ELK4 regulates Mcl-1 expression in high-grade glioma. (A)

ELK4 and Mcl-1 mRNA expression were measured by qPCR in

51 GBM tissue samples. Highly significant correlation (r ¼ 0.76),

(P ¼ .0001) between ELK4 and Mcl-1 expression levels was

found. Mean of duplicate samples expressed per 1000 copies of

b-actin. (B) ELK4 and Mcl-1 expression was measured by qPCR

in 3 GBM cell lines. U251 and U373 were heterozygous for the

2116G.A ETS SNP, whereas U87 was free of the SNP. U251

and U373, although expressing equivalent levels of ELK4,

expressed significantly less Mcl-1 compared with U87. (C) ELK4

was downregulated using siRNA. U87 and U118 were treated

with siRNA for 24 h and Mcl-1 and ELK4 protein expression was

determined by western blot. A decrease in ELK4 was observed

for both sequences compared with the control. ELK4

downregulation ablated Mcl-1 levels. (D) ELK4 cDNA was stably

overexpressed in U87 and U118 cells. ELK4 and Mcl-1 protein

expression was determined by western blot. An increase in Mcl-1

expression was observed accompanying the increase in ELK4.

Day et al.: ELK4 and Mcl-1: Targets in glioblastoma

NEURO-ONCOLOGY † N O V E M B E R 2 0 1 1 1207

at University of Q

ueensland on July 19, 2012http://neuro-oncology.oxfordjournals.org/

Dow

nloaded from

generation of a large number of progeny in primaryGBM neurospheres.

ELK4 Downregulation Sensitizes GBM Cells toApoptosis and Reduces in vivo Tumor Formation

To determine whether stable reduction of ELK4 wouldincrease GBM sensitivity to ABT-737 or cisplatin treat-ment, the more effective ELK4 siRNA sequence #2was cloned into the pSuperior shRNA vector andstable U87 clones generated. Following stable

knockdown of ELK4, U87 cells were treated withABT-737, cisplatin, or ABT-737 + cisplatin (Fig. 5D).As expected, stable downregulation of ELK4 sensitizedthese cells to both agents. Combined treatment withABT-737 + cisplatin (1 mM) or cisplatin (10 mM)increased apoptosis .20% and .30%, respectively(P , .05). To confirm that increased apoptosis wasindeed mediated by Mcl-1, the stable ELK4 shRNAknockdowns were rescued by overexpression of Mcl-1(Fig. 5D). When ELK4 shRNA Mcl-1 rescued cellswere again challenged with ABT-737 or cisplatin,

Fig. 5. Mcl-1 and ELK4 neutralization induce apoptosis and increase sensitivity to therapies in glioma. (A) U87, U118, D645 (WT Mcl-1

promoter), U373, and U251 (2116G.A ETS SNP) were treated with ABT-737, cisplatin, or ABT-737 and cisplatin combined. Annexin V

staining conducted 72 h posttreatment determined the level of apoptosis. (B) Mcl-1 was inhibited using a Noxa peptide approach in U87

and L1-NS cultures. Annexin V staining was conducted 48 h post-treatment with Noxa peptide (10 mM) alone and in combination with

ABT-737 or cisplatin and compared with control peptide (10mM). (C) U87 and U373 GBM cells were transfected with Mcl-1 shRNA or

control shRNA sequences. After 48 hours, eGFP+ve cells were sorted by fluorescence-activated cell sorting and Mcl-1 protein expression

evaluated by western blotting. b-actin levels were used as a loading control. Annexin V staining was conducted 48 h post-transfection.

U118 stably overexpressing ELK4 was treated with cisplatin (10 mM). Annexin V staining conducted 48 h post-treatment. (D) U87 stable

ELK4 shRNA knockdown and Mcl-1 rescued cells were generated. U87 control shRNA, ELK4 shRNA, and Mcl-1 rescue cells were

treated with ABT-737, cisplatin, or ABT-737 and cisplatin combined. Annexin V staining was conducted 48 h post-treatment. Columns,

mean of triplicate samples, +SD (*¼ P , .05).

Day et al.: ELK4 and Mcl-1: Targets in glioblastoma

1208 NEURO-ONCOLOGY † N O V E M B E R 2 0 1 1

at University of Q

ueensland on July 19, 2012http://neuro-oncology.oxfordjournals.org/

Dow

nloaded from

apoptosis returned to normal levels, as observed for cellsexpressing control shRNA Mcl-1.

We also investigated the consequence of reducedELK4 expression on tumor formation using a U87 sub-cutaneous and U118 intracranial murine xenograftmodel. Two U87 ELK4 shRNA clones were tested forsubcutaneous tumor formation in NOD/SCID mice. Inexperiment #1 (clone #1) (Fig. 6A), a reduction intumor formation was observed, with a median survivalof 27 days for the control shRNA versus 40 days forthe ELK4 shRNA (P ¼ .006). In experiment #2 (clone#2), a smaller number of cells was injected. In this

experiment, the mice injected with control shRNA hada longer median survival of 67 days, in keeping withthe reduced inoculum. Significantly, however, notumor formation was observed in the animals in whichELK4 was downregulated after 5 months (141 days) (P¼ .004). At autopsy, a small lesion was found in 2 of theELK4 shRNA animals at the site of injection, whereasthe remaining 3 animals were free of tumors. This verymarked difference was investigated further.Quantitative PCR was conducted to determine thelevel of ELK4 and Mcl-1 knockdown in vivo (Fig. 6C).This showed that while clone #1 had a relatively

Fig. 6. ELK4 downregulation reduces tumorigenicity in vivo. (A) Kaplan2Meier survival analysis shows significant survival extension of ELK4

shRNA U87 GBM cells vs control shRNA. Two independent clones were tested. For clone #1 (experiment #1, 2 × 106 cells injected

subcutaneously), ELK4 shRNA median survival was 40 days vs control 27 days (P ¼ .006, n ¼ 7). For clone #2 (experiment #2, 1 × 106 cells

injected subcutaneously), ELK4 shRNA median survival was 141 days (experiment terminated) vs control 67 days (P ¼ .004, n ¼ 5). (B)

qPCR expression data showing reduction of both ELK4 and Mcl-1 mRNA levels in both experiment #1 and experiment #2. Knockdown of

ELK4 and Mcl-1 correlated with increased survival in vivo. (C) Aperio images of U87 subcutaneous xenograft tumors showing tumor

morphology (hematoxylin and eosin), apoptosis (cleaved caspase-3), and proliferation (Ki67). Necrotic and apoptotic cells were observed in

ELK4 shRNA treated tumors. (D) Kaplan–Meier survival analysis shows significant survival extension of ELK4 shRNA U118 GBM cells vs

control shRNA using an intracranial xenograft model. ELK4 shRNA median survival was 38 days vs control 28 days (P ¼ .01, n ¼ 6).

Day et al.: ELK4 and Mcl-1: Targets in glioblastoma

NEURO-ONCOLOGY † N O V E M B E R 2 0 1 1 1209

at University of Q

ueensland on July 19, 2012http://neuro-oncology.oxfordjournals.org/

Dow

nloaded from

modest reduction in ELK4 mRNA (34%), whichreduced Mcl-1 to equivalent levels (30%), a more signifi-cant ELK4 knockdown was achieved in experiment #2(48%), which had a more dramatic effect on Mcl-1mRNA levels (80% reduction). Thus, tumor formationappeared to correlate with both the level of ELK4 andMcl-1 knockdown. In addition, we preparedhematoxylin-and-eosin tumor sections and measuredlevels of apoptosis and proliferation (Fig. 6C). In bothcontrol animal cohorts, tumor cells were viable withprominent angiogenesis. ELK4-downregulated tumors,however, lacked blood vessel formation, with prominentareas of necrosis and apoptosis, as detected usingcleaved caspase-3. Only mild changes in proliferationbetween the 2 groups was observed using Ki67 staining.Given the unique microenvironment of the brain, wealso assessed knockdown mediated by ELK4 shRNAusing the U118 GBM cell line with an intracranial xeno-graft model. Similar to subcutaneous results, a reductionin tumor formation was observed, with a median survi-val of 28 days for control shRNA compared with 38days for ELK4 shRNA–transfected cells (P ¼ .01).

Discussion

GBM is one of the most lethal of all human cancers, anddespite recent advances in therapy, the outlook hasimproved very little. Identifying and targeting key geneproducts offers a rational approach to therapy, andrecent work suggests that glioblastoma patients in par-ticular may benefit from molecularly targeted thera-pies.19 Here we identify two potential GBMtherapeutic targets: the anti-apoptotic protein Mcl-1and, as our results show, a critical regulator of itsexpression, the ETS family transcription factor ELK4.Importantly, ETS family members are implicated in theregulation of Bcl-2 family anti-apoptotic proteins.20

FLI1 blocks apoptosis in primary erythroblasts by regu-lation of Bcl-2, whereas Rel ETS family members havebeen shown to directly regulate Bcl-xL, promoting survi-val.21,22 Moreover, it has been shown that serum respon-sive factor and ELK1 act in concert to mediate theexpression of Mcl-1 in hematopoietic cells.23

Elevated expression of Bcl-2 family anti-apoptoticproteins in tumors can confer a survival advantage oncancers and contribute to the resistance of malignantcells to conventional cytotoxic therapy. Here we showMcl-1 to be highly expressed in a significant majorityof both human primary gliomas and GBM cultures.This observation, based on 51 samples, is supported byinterrogation of the National Cancer InstituteRembrandt database, albeit using an alternative analyti-cal method. Of 405 glioma specimens, 205 show upre-gulation (.2-fold) and 92 show intermediateexpression of Mcl-1. Furthermore, a significant differ-ence in survival (P ¼ .0417) was shown between theupregulated and intermediate populations.

Importantly, inhibition of Mcl-1 sensitized GBM celllines to apoptosis and increased sensitivity to ABT-737or cisplatin. The dominant role of Mcl-1 in glioma is

further supported by the ineffectiveness of ABT-737alone, a small molecular weight inhibitor of Bcl-2,Bcl-xL, and Bcl-w, but not of Mcl-1.

Given the unique dependence of glioma on Mcl-1and its very high expression in these tumors, we inves-tigated the molecular basis for elevated Mcl-1expression in GBM. Genomic analysis of the Mcl-1 pro-moter revealed that guanine-cytosine–rich insertions of6 bp and 18 bp, as reported previously in chronic lym-phocytic leukemia (CLL), were common to both GBMtissue samples and in normal control samples. Althougha previous study24 suggested that these insertionsresulted in increased promoter activity in leukemiccells, we found no significant increases in promoteractivity in GBM cells (Fig. S4). Furthermore, no corre-lation was found between Mcl-1 mRNA expression andthe presence of insertions in GBM, consistent with morerecent findings in CLL.25,26 Taken together, these datasuggest that these insertions are irrelevant to gliomabiology.

More significantly, though, in the course of thisanalysis, we identified a novel SNP (–116G.A) in theMcl-1 promoter in GBM, the presence of which resultedin decreased promoter activity. Two GBM cell lines,identified as being heterozygous for the –116G.ASNP, also exhibited reduced Mcl-1 levels. Significantly,this SNP occurs in a potential ETS transcription factorbinding site that is bound by ELK4 in GBM cells.These results provide strong evidence that ELK4 is akey regulator of both Mcl-1 promoter activity andMcl-1 expression in GBM and is therefore a potentialtarget for therapies aimed at downregulating Mcl-1and enhancing sensitivity to both ABT-737 and cisplatintreatment.

To assess whether ELK4 regulates expression ofMcl-1 in tumors other than glioma, mRNA expressionanalysis is now under way. Interestingly, preliminaryfindings show a highly significant positive correlationbetween ELK4 and Mcl-1 mRNA in kidney, lung,thyroid, and colon tumors (Fig. S5). These findings,while requiring further follow-up, are extremely positiveand suggest that ELK4 could be a potential target toinhibit Mcl-1 in a variety of human tumors.

Both ELK4 and Mcl-1 are identified in this study aspotential targets for therapeutic intervention in high-grade gliomas. Indeed, Mcl-1 has been targeted byothers using a variety of conventional and experimentalagents to enhance the efficacy of ABT-737 in diversehuman tumor cells, including leukemia (U937 andOCI-AML3), cervical (HeLa), breast (MCF-7), andsmall cell lung carcinomas (NCIH196 andNCI-H146).7,8,27,28 What then might be the advantagein targeting ELK4 in GBM? Consideration of therespective phenotypes of Mcl-1 and ELK4 knockoutmice suggest it could be associated with fewer sideeffects than targeting Mcl-1 directly. Mcl-1 knockoutmice, for example, die at early embryonic stages,29

with Mcl-1 critical for lymphocyte and hematopoieticstem cell survival.30,31 Conditional deletion of Mcl-1in mice also shows it to be essential for neutrophilsurvival,32 whereas inhibition of Mcl-1 expression by

Day et al.: ELK4 and Mcl-1: Targets in glioblastoma

1210 NEURO-ONCOLOGY † N O V E M B E R 2 0 1 1

at University of Q

ueensland on July 19, 2012http://neuro-oncology.oxfordjournals.org/

Dow

nloaded from

anti-sense oligonucleotides results in human macro-phage apoptosis.33 In contrast, ELK4-deficient miceare viable with no gross physical abnormalities,although some immune suppression was noted with areduction in single-positive thymocytes and peripheralT cell numbers.34 Given this milder phenotype, targetingELK4 may provide a more attractive target for downre-gulating Mcl-1 and sensitizing GBM to apoptosis. The invitro and in vivo experiments provide proof-of-principlethat suppression of ELK4 can indeed significantly com-promise glioma cell survival and tumorigenicity.

In summary, we provide evidence that elevated Mcl-1expression is a key factor providing protection fromapoptosis in GBM. Elevated expression of Mcl-1 inGBM in turn depends critically on expression of ELK4,suggesting that ELK4-mediated Mcl-1 overexpressionis a key step in the oncogenic pathway leading toGBM. Both ELK4 and Mcl-1 accordingly are identifiedas potential targets for therapeutic intervention inGBM. The novel, functional Mcl-1 promoter ETSSNP, although identified at low frequencies, clearlyreduces Mcl-1 expression and may prove a useful prog-nostic marker for GBM survival outcome or a markerthat determines treatment strategies.

Supplementary Material

Supplementary material is available online at Neuro-Oncology (http://neuro-oncology.oxfordjournals.org/).

Acknowledgments

We thank Dr. TeongChuah,RoyalBrisbane and Women’sHospital, for primary clinical specimens; Dr. JamesDoecke, QIMR, for statistical analysis; AbbottLaboratories for ABT-737; A/Prof David Ashley and Dr.Andrea Muscat, Murdoch Childrens Research Institute,Melbourne for cell lines.

Conflict of interest statement: None declared.

Funding

National Health and Medical Research Council projectgrant I.D 552489. Cancer Council Queensland grantI.D 552483.

References

1. Behin A, Hoang-Xuan K, Carpentier AF, Delattre JY. Primary brain

tumours in adults. Lancet. 2003;361(9354):323–331.

2. Stupp R, Mason WP, van den Bent MJ, et al. Radiotherapy plus conco-

mitant and adjuvant temozolomide for glioblastoma. N Engl J Med.

2005;352(10):987–996.

3. Cory S, Huang DC, Adams JM. The Bcl-2 family: roles in cell survival

and oncogenesis. Oncogene. 2003;22(53):8590–8607.

4. Jeffers JR, Parganas E, Lee Y, et al. Puma is an essential mediator of

p53-dependent and -independent apoptotic pathways. Cancer Cell.

2003;4(4):321–328.

5. Shibue T, Takeda K, Oda E, et al. Integral role of Noxa in p53-mediated

apoptotic response. Genes Dev. 2003;17(18):2233–2238.

6. Watanabe K, Tachibana O, Sata K, Yonekawa Y, Kleihues P, Ohgaki H.

Overexpression of the EGF receptor and p53 mutations are mutually

exclusive in the evolution of primary and secondary glioblastomas.

Brain Pathol. 1996;6(3):217–22; discussion 223–214.

7. Oltersdorf T, Elmore SW, Shoemaker AR, et al. An inhibitor of Bcl-2

family proteins induces regression of solid tumours. Nature.

2005;435(7042):677–681.

8. Konopleva M, Contractor R, Tsao T, et al. Mechanisms of apoptosis

sensitivity and resistance to the BH3 mimetic ABT-737 in acute

myeloid leukemia. Cancer Cell. 2006;10(5):375–388.

9. van Delft MF, Huang DC. How the Bcl-2 family of proteins interact to

regulate apoptosis. Cell Res. 2006;16(2):203–213.

10. Chen S, Dai Y, Harada H, Dent P, Grant S. Mcl-1 down-regulation

potentiates ABT-737 lethality by cooperatively inducing Bak activation

and Bax translocation. Cancer Res. 2007;67(2):782–791.

11. Lin X, Morgan-Lappe S, Huang X, et al. “Seed” analysis of off-target

siRNAs reveals an essential role of Mcl-1 in resistance to the small-molecule

Bcl-2/Bcl-XL inhibitor ABT-737. Oncogene. 2007;26(27):3972–3979.

12. Rietze RL, Reynolds BA. Neural stem cell isolation and characterization.

Methods Enzymol. 2006;419:3–23.

13. Hobbs S, Jitrapakdee S, Wallace JC. Development of a bicistronic vector

driven by the human polypeptide chain elongation factor 1alpha pro-

moter for creation of stable mammalian cell lines that express very

high levels of recombinant proteins. Biochem Biophys Res Commun.

1998;252(2):368–372.

14. Chen L, Willis SN, Wei A, et al. Differential targeting of prosurvival Bcl-2

proteins by their BH3-only ligands allows complementary apoptotic

function. Mol Cell. 2005;17(3):393–403.

15. Miller SA, Dykes DD, Polesky HF. A simple salting out procedure for

extracting DNA from human nucleated cells. Nucleic Acids Res.

1988;16(3):1215.

16. Lickliter JD, Cox J, McCarron J, et al. Small-molecule Bcl-2 inhibitors

sensitise tumour cells to immune-mediated destruction. Br J Cancer.

2007;96(4):600–608.

17. Dignam JD, Lebovitz RM, Roeder RG. Accurate transcription initiation

by RNA polymerase II in a soluble extract from isolated mammalian

nuclei. Nucleic Acids Res. 1983;11(5):1475–1489.

18. Reynolds BA, Rietze RL. Neural stem cells and neurospheres–

re-evaluating the relationship. Nat Methods. 2005;2(5):333–336.

19. Mellinghoff IK, Wang MY, Vivanco I, et al. Molecular determinants of

the response of glioblastomas to EGFR kinase inhibitors. N Engl J Med.

2005;353(19):2012–2024.

20. Vickers ER, Kasza A, Kurnaz IA, et al. Ternary complex factor-serum

response factor complex-regulated gene activity is required for cellular

proliferation and inhibition of apoptotic cell death. Mol Cell Biol.

2004;24(23):10340–10351.

21. Chen C, Edelstein LC, Gelinas C. The Rel/NF-kappaB family directly

activates expression of the apoptosis inhibitor Bcl-x(L). Mol Cell Biol.

2000;20(8):2687–2695.

22. Lesault I, Quang CT, Frampton J, Ghysdael J. Direct regulation of BCL-2

by FLI-1 is involved in the survival of FLI-1-transformed erythroblasts.

Embo J. 2002;21(4):694–703.

Day et al.: ELK4 and Mcl-1: Targets in glioblastoma

NEURO-ONCOLOGY † N O V E M B E R 2 0 1 1 1211

at University of Q

ueensland on July 19, 2012http://neuro-oncology.oxfordjournals.org/

Dow

nloaded from

23. Townsend KJ, Zhou P, Qian L, et al. Regulation of MCL1 through a serum

response factor/Elk-1-mediated mechanism links expression of a viability-

promoting member of the BCL2 family to the induction of hematopoietic

cell differentiation. J Biol Chem. 1999;274(3):1801–1813.

24. Moshynska O, Sankaran K, Pahwa P, Saxena A. Prognostic significance of a

short sequence insertion in the MCL-1 promoter in chronic lymphocytic

leukemia. Journal of the National Cancer Institute. 2004;96(9):673–682.

25. Coenen S, Pickering B, Potter KN, Johnson PW, Stevenson FK, Packham

G. The relevance of sequence insertions in the Mcl-1 promoter in

chronic lymphocytic leukemia and in normal cells. Haematologica.

2005;90(9):1285–1286.

26. Tobin G, Skogsberg A, Thunberg U, et al. Mcl-1 gene promoter insertions

do not correlate with disease outcome, stage or VH gene mutation status in

chronic lymphocytic leukaemia. Leukemia. 2005;19(5):871–873.

27. Dai Y, Grant S. Targeting multiple arms of the apoptotic regulatory

machinery. Cancer Res. 2007;67(7):2908–2911.

28. van Delft MF, Wei AH, Mason KD, et al. The BH3 mimetic ABT-737

targets selective Bcl-2 proteins and efficiently induces apoptosis via

Bak/Bax if Mcl-1 is neutralized. Cancer Cell. 2006;10(5):389–399.

29. Rinkenberger JL, Horning S, Klocke B, Roth K, Korsmeyer SJ. Mcl-1

deficiency results in peri-implantation embryonic lethality. Genes Dev.

2000;14(1):23–27.

30. Opferman JT, Iwasaki H, Ong CC, et al. Obligate role of anti-apoptotic

MCL-1 in the survival of hematopoietic stem cells. Science.

2005;307(5712):1101–1104.

31. Opferman JT, Letai A, Beard C, Sorcinelli MD, Ong CC, Korsmeyer SJ.

Development and maintenance of B and T lymphocytes requires antia-

poptotic MCL-1. Nature. 2003;426(6967):671–676.

32. Dzhagalov I, St John A, He YW. The antiapoptotic protein Mcl-1 is

essential for the survival of neutrophils but not macrophages. Blood.

2007;109(4):1620–1626.

33. Liu H, Perlman H, Pagliari LJ, Pope RM. Constitutively activated Akt-1 is

vital for the survival of human monocyte-differentiated macrophages.

Role of Mcl-1, independent of nuclear factor (NF)-kappaB, Bad, or

caspase activation. J Exp Med. 2001;194(2):113–126.

34. Costello PS, Nicolas RH, Watanabe Y, Rosewell I, Treisman R. Ternary

complex factor SAP-1 is required for Erk-mediated thymocyte positive

selection. Nat Immunol. 2004;5(3):289–298.

Day et al.: ELK4 and Mcl-1: Targets in glioblastoma

1212 NEURO-ONCOLOGY † N O V E M B E R 2 0 1 1

at University of Q

ueensland on July 19, 2012http://neuro-oncology.oxfordjournals.org/

Dow

nloaded from