Molecular Cell Biology

Identification of Novel RAS Signaling TherapeuticVulnerabilities in Diffuse Intrinsic Pontine GliomasRobert F. Koncar1,2, Brittany R. Dey1,2,3, Ann-Catherine J. Stanton1,2,Nishant Agrawal1, Michelle L.Wassell1,2, Lauren H. McCarl1,2, Abigail. L. Locke1,Lauren Sanders4, Olena Morozova-Vaske4,5, Max I. Myers1, Ronald L. Hamilton6,Angel M. Carcaboso7, Gary Kohanbash1,2, Baoli Hu1,2, Nduka M. Amankulor2,James Felker8, Madhuri Kambhampati9,10, Javad Nazarian9,10,11, Oren J. Becher12,13,C. David James13,14,15, Rintaro Hashizume13,14,15, Alberto Broniscer8, Ian F. Pollack1,2,and Sameer Agnihotri1,2,16

Abstract

© 2019 American Association for Cancer Research

P

P

P

P

P

P

RAS-GDP RAS-GTP

GRB2 SOS

MEK5

MEKK2

MYC

TranscriptionP

RTK

Growth Factor

ERK5

PERK1/2

H3K27M?

intrinsic pontine glioma.

Diffuse intrinsic pontine gliomas (DIPG) are incurablebrain tumors with an aggressive onset. Apart from irradia-tion, there are currently no effective therapies available forpatients with DIPG, who have a median survival time of lessthan one year. Most DIPG cells harbor mutations in genesencoding histone H3 (H3K27M) proteins, resulting in aglobal reduction of H3K27 trimethylation and activation ofoncogenic signaling pathways. Here we show that theH3K27M mutations contribute to RAS pathway signaling,which is augmented by additional RAS activators includingPDGFRA. H3K27M mutation led to increased expression ofreceptor tyrosine kinases (RTK). A RAS pathway functionalscreen identified ERK5, but not ERK1/2, as a RAS pathwayeffector important for DIPG growth. Suppression of ERK5decreased DIPG cell proliferation and induced apoptosisin vitro and in vivo. In addition, depletion or inhibition ofERK5 significantly increased survival of mice intracraniallyengrafted with DIPG cells. Mechanistically, ERK5 directlystabilized the proto-oncogene MYC at the protein level.Collectively, our data demonstrate an underappreciated roleof H3K27M in RAS activation and reveal novel therapeutictargets for treating DIPG tumors.

Significance: These findings identify the H3K27Mmutation as an enhancer of RAS activation in DIPG and ERK5 as a novel,immediately actionable molecular target.

Graphical Abstract: http://cancerres.aacrjournals.org/content/canres/79/16/4026/F1.large.jpg.

1John G. Rangos Sr. Research Center, Children's Hospital of Pittsburgh,Pittsburgh, Pennsylvania. 2Department of Neurological Surgery, Universityof Pittsburgh School of Medicine, Pittsburgh, Pennsylvania. 3SUNY DownstateMedical Center, New York, New York. 4Department of Molecular, Cell, andDevelopmental Biology, University of California Santa Cruz, Santa Cruz,California. 5University of California Santa Cruz Genomics Institute,Santa Cruz, California. 6Department of Pathology, University of PittsburghCancer Institute, Pittsburgh, Pennsylvania. 7Institut de Recerca Sant Joan deDeu, Barcelona, Spain. 8Pediatric Neuro-Oncology Program, UPMC Children'sHospital of Pittsburgh, Pittsburgh, Pennsylvania. 9Children's National HealthSystem, Washington, D.C. 10Department of Genomics and Precision Medicine,George Washington University School of Medicine and Health Sciences,Washington, D.C. 11Department of Oncology, University Children's HospitalZ€urich, Z€urich, Switzerland. 12Division of Hematology, Oncology and Stem CellTransplant, Ann & Robert H. Lurie Children's Hospital of Chicago, Chicago,Illinois. 13Department of Biochemistry and Molecular Genetics, Robert H. LurieCancer Center, Feinberg School of Medicine, Northwestern University, Chicago,

Illinois. 14Department of Neurological Surgery, Robert H. Lurie NCI Comprehen-sive Cancer Center, Northwestern University, Chicago, Illinois. 15Feinberg Schoolof Medicine, Northwestern University, Chicago, Illinois. 16Department ofNeurobiology, University of Pittsburgh, Pittsburgh, Pennsylvania.

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

R.F. Koncar and B.R. Dey contributed equally to this article.

Corresponding Author: Sameer Agnihotri, Children's Hospital of Pittsburgh,University of Pittsburgh School of Medicine, Pittsburgh, PA 15224. Phone: 412-692-5509; Fax: 412-692-5921; E-mail: sameer.agnihotri@pitt.edu

Cancer Res 2019;79:4026–41

doi: 10.1158/0008-5472.CAN-18-3521

�2019 American Association for Cancer Research.

CancerResearch

Cancer Res; 79(16) August 15, 20194026

on December 1, 2020. © 2019 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst June 14, 2019; DOI: 10.1158/0008-5472.CAN-18-3521

IntroductionPrimary brain and CNS tumors are the leading cause of cancer-

related mortality in children ages 0–14. Most of these deaths aredue to pediatric high-grade gliomas (PHGG), which are themost commonly diagnosed type of malignant brain tumor inchildren (1). PHGGs are largely incurable with a poor mediansurvival post diagnosis of 9–15 months (1, 2). A particularlyaggressive type of PHGG that originates primarily in the midlineand pons is known as diffuse intrinsic pontine glioma (DIPG).Presenting nearly exclusively in children, DIPG has a two-yearsurvival rate postdiagnosis of only 10% and a median survivaltime of 9–12 months (3). This poor prognosis is partly due tothe tumor's anatomic location, preventing surgical excision,and narrowing therapeutic options to blood–brain barrier–penetrating agents (4). Furthermore, radiotherapy providesonly temporary benefits while chemotherapy and targetedtherapeutics in clinical trials have proven relatively ineffectivein treating DIPG (5).

As molecular profiling techniques have progressed, our under-standing of cancer biology is shifting from a histopathologicperspective to one informed by the genomics and epigeneticsmodulating disease pathogenesis. This reevaluation of tumorbiology has revealed key differences between PHGGs and theircorresponding adult tumors, as well as the molecular heteroge-neity among PHGGs (6). It has also indicated conceptual short-comings associated with traditional treatment of PHGG, whichhelp explain the ineffectiveness of treatments used in previousclinical trials (6–8).

The discovery of novel DIPG-associated hotspot mutations(K27M) in the genes encoding histone H3.3 and H3.1 haveincreased our understanding that pediatric and adult HGGspossess distinct genetic as well as epigenetic characteristics (9).Among PHGGs, those with K27M or G34R H3mutations exist asdistinct tumor subtypes with unique clinical, biological, andpathologic features (6, 9). H3K27M mutations in DIPG lead toan altered epigenetic state characterized by a global reduction ofhistone trimethylation (H3K27me3), as the mutant histone sup-presses the polycomb repressive complex 2 (PRC2) throughinteraction with the EZH2 subunit of PRC2 (10). DecreasedH3K27me3 results in the transcriptional activation of severaloncogenic proteins. The discovery of H3K27M mutation effectshas fostered an active interest in using epigenetic-based therapiesto counteract the transcriptional dependencies resulting fromthe H3K27M mutation (11–14). Moreover, H3K27M mutationsappear to be not only required for tumor initiation but alsorequired for tumor maintenance (15, 16).

The RAS–MAPK signaling pathway, which stimulates cell pro-liferation and survival, is comprised of clinically actionable tar-gets, with several inhibitors that target specific pathway signalingmediators already in clinical use (17). RAS, although infrequentlymutated, is often hyperactive in DIPG due to a variety of mechan-isms including: recurrent growth factor activation, mutations inreceptor tyrosine kinases such as platelet-derived growth factorreceptor alpha (PDGFRA), and genetic inactivation of RAS-regulatory proteins such as neurofibromin 1 (NF1; refs. 6, 9, 18).However, the relationship between the H3K27M mutations andRASpathwayhas yet to be explored.Herewe present evidence thattheH3K27Mmutations activate RAS anddemonstrate that severaleffectors of the RAS pathway are clinically actionable targets inDIPG. In addition, we show that one of these targets, ERK5, isupregulated in DIPG and is a potent driver of tumor growth.

Materials and MethodsCell culture

Human neural stem cells (H9 hESC-derived) were purchasedfrom Invitrogen and grown in DMEM/F12 containing penicillin/streptomycin/amphotericin B (1% vol/vol) supplemented withN2 (2% vol/vol; Invitrogen), EGF (20 ng/mL; Invitrogen), andFGF-2 (20 ng/mL; Invitrogen). DIPG cell lines (all with theH3K27M mutation) SU-DIPG-IV (DIPG-IV), SU-DIPG-13p(DIPG-13p), and HSJD-DIPG-007 (DIPG-007) were grown andpassaged in DIPG medium as reported previously (11, 13, 19).SF8628 and immortalized NHAs were grown in DMEM (Invitro-gen) supplemented with 10% FBS (Invitrogen). All cell lineswere confirmed by short tandem repeat profiling and testedMycoplasma negative by PCR. All tumor-derived cell lines usedare described in Supplementary Table S1.

Generation of inducible ERK5 knockdown cellsTo produce lentivirus, 400 mL of Opti-MEM I (Gibco) contain-

ing 2.4 mg pCMV-dR8.2 dvpr, 1.8 mg pCMV-VSV-G, 3 mg piSMARTmCMV/TurboGFP shRNA plasmid (Dharmacon), and 20 mLFuGENE HD (Fugent LLC) was added to a 10-cm plate ofHEK293T cells grown in complete mediaþ 10% FBS. Media werecollected from the HEK293T cells 48 hours after transfection,spun at 2,000� g, and the virus-containing supernatantwas eitherused directly for lentiviral transduction or concentrated withLenti-X lentiviral concentrator (Takara). Cells were transducedby overnight incubation with lentivirus in appropriate growthmedia containing 5–8 mg/mL polybrene (EMD Millipore).Beginning two days after transduction, cells were cultured with0.5–1 mg/mL puromycin (Gibco) for 5–7 days.

Plasmids and plasmid constructionWild-type human H3F3A cDNA was ligated with a puromycin

selection marker and a self-cleaving T2A sequence usingGibson Assembly Cloning and cloned into pLV with an EF1apromoter.MYC (NM_002467.4) orMAPK7 (ERK5,NM_139033)were ligated with T2A mCherry and cloned into pLV underthe control of the EF1a promoter. Cloning and plasmid sequencevalidation were performed by VectorBuilder Services (95050).Site-directed mutagenesis was performed to generate H3F3AK27M, H3F3A G34R, MYC S62D, and ERK5-mutant constructsusing the QuikChange II Site-Directed Mutagenesis Kit andmanufacturer's protocol (Agilent, catalog no. 200523). Lenti-viral ERK5 GIPZ shRNA plasmids were purchased fromDharmacon (80026, catalog no. RHS4531-EG5598 glycerol set:V2LHS_202701, V3LHS_366737, V3LHS_366738, V3LHS_366740,V3LHS_640828, V3LHS_640830).

Cell proliferation assays, viability assays, and drug EC50

analysisDirect cell counts were performed using the Countess II

FL Automated Cell Counter (Thermo Fisher Scientific). Alamarblue viability (Thermo Fisher Scientific) and the 5-bromo-20-deoxyuridine (BrdU) incorporated cell proliferation assay (CellSignaling Technology catalog no. 6813)were performed using themanufacturer's standard protocol. EC50 curve analysis was per-formed using raw fluorescent Alamar blue values that wereinputted into PRISM 7.0. Drug concentrations were log10 trans-formed and values normalized to percent viability with respect tovehicle-treated cells. EC50 values were interpolated from a log

RAS Pathway Vulnerabilities in Histone-Mutant DIPG

www.aacrjournals.org Cancer Res; 79(16) August 15, 2019 4027

on December 1, 2020. © 2019 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst June 14, 2019; DOI: 10.1158/0008-5472.CAN-18-3521

inhibitor versus normalized response with variable slope using aleast squares fit model.

Small-molecule inhibitorsTG02 was obtained from Tragara Pharmaceutical under a Mate-

rials Transfer Agreement. ERK5-IN-1 (catalog no. 5393/10),BIX02189 (catalog no. 4842/10), XMD 8-92 (catalog no. 4132),and MG132 (catalog no. 6033) were purchased from Tocris.

Apoptosis assayApoptosis was quantified using the Pacific Blue Annexin V

Apoptosis Detection Kit with propidium iodide (BioLegend,catalog no. 640926). Analysis was performed at the Flow Cyto-metry Core, Rangos Research Center, UPMC Children's Hospitalof Pittsburgh (Pittsburgh, PA).

RAS activity assayRAS activity was measured using the RAS activation assay (Cell

Signaling Technology, catalog no. 8821). Briefly, active RASDetection GST-RAF1-RBD fusion protein was used to bind theactivated form of GTP-bound RAS, which was then immunopre-cipitated from 500 mg of total cell lysate with glutathione resin.RAS activation levels were then determined by Western blotanalysis using a RAS Mouse mAb. Activated RAS immune precip-itate was normalized to total RAS from whole-cell lysate.

RAS siRNA screenA total of 5,000 neural stem cells (NSC) control, NSC

H3K27M, and DIPG-007 cells were plated in black 96-wellplates. Twelve hours postplating, cells were transfected with acustom Endoribonuclease-prepared siRNA library (esiRNA;Sigma; Supplementary Table S2). Twenty nanograms of esiRNAwere used per target with X-tremeGENE HP DNA TransfectionReagent (Sigma). Cell viability was assessed 96 hours later bythe Alamar Blue Cell Viability Assay (Invitrogen, catalog no.DAL1025). Data were normalized using the standard z-scoreusing the percent viability difference of NSC control versus NSCH3K27M for each siRNA target and control siRNA DIPG-007versus DIPG-007 siRNA target gene. For significance, z-scorecut-off values were �2 (P < 0.05) in three biological replicates.Validation of esiRNA screen was performed using two individ-ual Dicer-substrate siRNA (DsiRNA) per gene-target (IntegratedDNA Technologies).

Western blotting and immunoprecipitationCell pellets were lysed in PLC lysis buffer containing protease

and phosphatase inhibitors (Roche Inc.). Protein lysates werequantified using the bicinchoninic acid assay (Pierce ChemicalCo.). Thirty micrograms of total protein lysate was loaded in10% SDS-PAGE gels and electrophoresed. Proteins were trans-ferred to polyvinylidene difluoride membranes using a semi-dry transfer apparatus (Bio-Rad). Membranes were blocked for1 hour and probed for various proteins overnight in 5% nonfatmilk or 5% BSA in Tris buffered saline solution, 0.05% Tween-20 (TBST) or PBS solution with 0.05% Tween-20 (PBST).Membranes were washed for 5 minutes in TBST (3�) andincubated with horseradish peroxidase–conjugated antibodiesspecific for the primary antibody (Bio-Rad Laboratories, Inc.).Binding was detected using Chemiluminescence ReagentPlus (PerkinElmer Inc.). Antibodies were used at the followingdilutions:

H3K27M (1:1,000, Abcam catalog no. 190631), H3K27me3(1:1,000, Cell Signaling Technology catalog no. 9733), H3K27Ac(1:1,000, Cell Signaling Technology, catalog no. 4353), b-actin(1:10,000, Cell Signaling Technology, catalog no. 3700), p-ERK1/2 (1:1,000, Cell Signaling Technology, catalog no. 8544), ERK1/2(1:1,000, Cell Signaling Technology, catalog no. 4695), HA(1:1,000, Cell Signaling Technology, catalog no. 3724),RAS (1:1,000, Cell Signaling Technology, catalog no. 3965),p-ERK5 (T218/Y220; 1:1,000, Cell Signaling Technology, catalogno. 3371), p-ERK5 (S731/T733; Abnova, catalog no. PAB15919),ERK5 (1:1,000, Thermo Fisher Scientific, catalog no. PA5-17689),p-MEK5 (S311/T315; 1:1,000, Thermo Fisher Scientific, catalogno. 480024), MEK5 (Thermo Fisher Scientific, catalog no. PA5-15083), MEKK2 (Cell Signaling Technology, catalog no. 19607),GAPDH (Cell Signaling Technology, catalog no. 5174), H3.3(1:500, BioLegend, catalog no. 601901), pAKT ser473 (1:2,000,Cell Signaling Technology, catalog no. 4060), AKT (1:1,000, CellSignaling Technology, catalog no. 4691), MYC S62 (1:1,000,Thermo Fisher Scientific, catalog no. PA5-36671), MYC (1:1,000,BioLegend catalog no. 626802). Immunoprecipitations and coim-munoprecipitationswere performedusing 500mg total lysate usingmanufacturer-specific dilutions. Denaturing immunoprecipita-tions were performed as described previously (20–22).

In vitro kinase assaysPurified active recombinant ERK5 protein was purchased from

Thermo Fisher Scientific (catalog no. A32872) and purifiedrecombinant MYC was purchased from Abcam (catalog no.ab84132). Briefly, purified ERK5 and MYC were resuspended in40 mL of 1� kinase buffer supplemented with 200 mmol/L ATP,0.25 mg substrate, and 1.0 mg purified recombinant and activekinase. Samples were incubated in the presence or absence ofERK5 inhibitors for 30 minutes at 30�C. The reaction wasterminated with 20 mL 3� SDS sample buffer. Samples were thenboiled at 100�C for 5 minutes and then evaluated by Westernblotting using a phospho-MYC-ser62 antibody (1:1,000).1� Kinase Buffer recipe: 25 mmol/L Tris (pH 7.5), 5 mmol/Lb-glycerophosphate, 2 mmol/L DTT, 0.1 mmol/L Na3VO4,10 mmol/L MgCl2.

Cancer signaling antibody arrayThe cancer signaling antibody array was purchased from Full

Moon Biosystems and is an ELISA-based Antibody Array platformcomprising of 269 antibodies targeting proteins involved incancer signaling (Supplementary Table S3). The array involvesfour major steps: (i) protein extraction with nondenaturing lysisbuffer; (ii) biotinylation of protein samples; (iii) incubation oflabeled samples with antibody array; and (iv) detection by dye-conjugated streptavidin. Briefly, biological triplicates of ERK5knockdown or control SU-DIPG IV cells were pooled and assayedon two independent arrays. The array was performed as per themanufacturer's protocol with image quantification and analysisperformed by Full Moon Biosystems data service.

IHCParaffin-embedded blocks were cut in 5-mm sections. Slides

were processed as follows: dewaxed in xylene followed by rehy-dration in a standard alcohol series. Antigen retrieval was bypressure cooking for 20 minutes in citrate buffer (pH 6.0),followed by blocking of endogenous peroxidase in 3% H2O2.The antibodies were added and incubated overnight at 4�C.

Koncar et al.

Cancer Res; 79(16) August 15, 2019 Cancer Research4028

on December 1, 2020. © 2019 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst June 14, 2019; DOI: 10.1158/0008-5472.CAN-18-3521

Antibodies were detected using a secondary-HRP–labeled mouseor rabbit antibody detection system (Dako EnVisionþ System-HRP catalog no. k4001, catalog no. k4003), followed by theaddition of 3,30-diaminobenzidine (DAB) chromagen (VectorLabs) for visualization. Sections were counterstained with hema-toxylin (Thermo Fisher Scientific Inc.) and slides dehydrated in70%, 80%, and 100% ethanol and xylene. Slides were cover-slipped and mounted in Permount (Thermo Fisher ScientificInc.). Antibodies and concentrations for IHC are as follows:Ki67(1:100, Dako, catalog no. F7268). All images were capturedon a Leica DM 100 microscope using Leica Application SuiteSoftware (Version 3.8.0).

Gene expression and copy number analysisCopy number analysis of MAPK7, PDGFRA, and NF1 was

performed using the Mackay and colleagues' dataset (6) usingthe pediatric cBioPortal website (https://pedcbioportal.org/).Using the R2 software (http://r2.amc.nl), we analyzed geneexpression levels in normal brain and two independent, non-overlapping patient cohorts (23, 24).

Real-time quantitative PCRTotal RNA isolationwas performed using an RNA extraction kit

according to the manufacturer's instructions (RNeasy Kit,Qiagen). cDNA was synthesized from 100 ng of total RNA usingtheQuantitect RT kit, which includedDNAse treatment (Qiagen).Quantitative real-time PCR was performed on 10 ng of cDNAtemplate in a final volume of 20 mL using the Chromo4 Real TimePCR detector (MJ Research, a Division of Bio-Rad LaboratoriesLtd.) using SYBR green fluorescence. Real-time PCR data wereanalyzed using Opticon Monitor 3.1.3 analysis software. Dataanalysis was done using the DDCt method with HPRT1 as areference/control gene. Specific primers are available in Supple-mentary Table S4.

Animal xenograft studiesAll animal procedures were carried out ethically according to

animal user protocols approved by Institutional Animal CareCommittee. Four- to 6-week-old NOD-SCIDmale or femalemice(NOD-SCID-Prkdcscid) were injected with 1 � 105 SU-DIPG-13pcells or 1� 105 SF8628DIPG cells. Cells were resuspended in 2 mLof PBS and injected into the pons/midbrain using a stereotacticframe (Stoelting) and automated cell injector (Stoelting) withcells delivered over 4 minutes. Coordinates were as follows fromthe Lambda suture (x¼ 0.8mm, y¼�0.8mm, z¼�5.0mm). Tendays postinjection, mice were randomized into three groups:vehicle, ERK5-IN-1 (50mg/kg), and TG02 (20mg/kg). Mice weretreated for two cycles (5 days on, with 2 days off per cycle). Micewere sacrificed at sign of neurologic duress and brains wereextracted and fixed in 4% PFA.

Statistical analysisStatistical analysis was performed in GraphPad Prism 7.0. All

in vitro experimentswere performed inbiological triplicates unlessotherwise stated. Values reported are the mean and SEM. ANOVAwas conducted formultigroup comparisons followed by a post hocTukey test or post hoc Dunnett test to identify differences withingroups. Z-score analysis was performed to identify significantgenes from the siRNA screen and from the cancer phosphopro-teomic array. Survival analysis was performed using the log-ranksurvival test. For direct pairwise comparisons where appropriate,

an unpaired two-tailed Student t test was used. Significance wasestablished as �, P < 0.05; ��, P < 0.01; ���, P < 0.001.

ResultsH3K27M activates RAS in neural stem cells

H9-derived NSCs were stably transfected with lentivirusexpressing the mutant H3F3A K27M (H3K27M) and a select-able puromycin marker. Introduction of H3K27M promotedincreased proliferation from 96 hours onwards, as comparedwith NSC-expressing wild-type H3F3A or empty vectorcontrols (Fig. 1A). H3K27M-mutant protein expression wasconfirmed by immunoblotting (Fig. 1B). As expected andpublished previously by others, expression of H3K27M inNSCs leads to the reduction of H3 trimethylated lysine 27(H3K27me3, Fig. 1B).

Given the relevance of activated RAS in cell growth, we hypoth-esized that H3K27M may promote proliferation through activa-tion of RAS. A pan-RAS activity pull-down assay confirmed thatH3K27M-mutant NSC lines harbored activated RAS as well asactivated downstream signaling mediator phospho ERK1/2(Fig. 1C). Increased active RAS was confirmed by densitometry(Fig. 1D). Stable expression of H3 WT did not increase activatedRAS (Fig. 1C and D).

We next performed siRNA knockdown of three major RASisoforms (NRAS, KRAS, and HRAS) in NSCs expressing emptyvector control, H3 WT, or H3K27M. Compared with scrambledcontrol siRNA, loss of HRAS, KRAS, and NRAS hadmodest effectson cell proliferation in NSCs expressing empty vector control orwild-type H3 but had a significant effect on proliferation inH3K27M NSCs (Fig. 1E–G; Supplementary Fig. S1A–S1C). Com-pared with control or H3 WT, H3K27M-expressing NSCs treatedwith RAS siRNA had the greatest degree of apoptosis as evaluatedby activated cleaved caspase 3/7 (Supplementary Fig. S1D).Reduced RAS expression was confirmed by quantitative reversetranscriptase real-time PCR (qRT-PCR; Supplementary Fig. S1E–S1G). DIPG cells harbor amplification of PDGFRA and loss ofNF1, both ofwhich result in RAS activation.We hypothesized thatH3K27Mexpressionwouldbe additivewith knowndrivers of RASsignaling. Interestingly, overexpression of PDGFRA in NSCsincreased activated RAS comparably with that of H3K27M, andcombined PDGFRA þ H3K27M expression was additive withrespect to effect on activated RAS levels (Fig. 1H and I; �, P < 0.05).To date, the status of activated RAS has never been evaluated inDIPG at the biochemical level or comparedwith established high-grade glioma cultures. Active RAS pull downs in DIPG linesharboring H3K27M mutations have comparable active RAS tohemispheric high-grade gliomas (Fig. 1J and K). Interestingly,CNMC-XD-760, an H3 wild-type DIPG cell line, had the leastactive RAS (Fig. 1J and K).

The Polycomb repressive complex 2 (PRC2) has histonemethyltransferase activity and trimethylates H3K27 via EZH2subunit activity. H3K27M inhibits EZH2 function with respectto H3K27 trimethylation (H3K27me3). We investigated whetherpharmacologic inhibition of EZH2 would have similar effects onH3K27 methylation as ectopic H3K27M expression. EZH2 inhi-bition by GSK343 (5 mmol/L) resulted in loss of H3K27me3 andreduced cell proliferation (Supplementary Fig. S1H and S1I).In addition, EZH2 inhibition, but not H3K27M, resulted inactivation of the CDKN2A locus as indicated by p16 proteinimmunoblot results (Supplementary Fig. S1H and S1I). Also,

RAS Pathway Vulnerabilities in Histone-Mutant DIPG

www.aacrjournals.org Cancer Res; 79(16) August 15, 2019 4029

on December 1, 2020. © 2019 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst June 14, 2019; DOI: 10.1158/0008-5472.CAN-18-3521

EV

WT

H3.

3

H3.

3 K

27MA B

D E

C

0 48 96 1440

200

400

600

800

Hours

EV

Scramble

HRAS

KRAS

NRAS

*

0 48 96 1440

200

400

600

800

Hours

H3 WT

Scramble

HRAS

KRAS

NRAS

*

F

0 48 96 1440

200

400

600

800

1,000

Hours

Scramble

HRAS

KRAS

NRAS

H3K27M

*

H I

G

β-ACTIN

ActiveRASTotalRAS

J

Con K27M

PDGFRA

Dual

NSC

Con

PDGFRA

K27MDual

0

2

4

6

8

Den

sito

met

ry: f

old

chan

ge

*** K

Ce

ll co

un

t (×

1,0

00

)

Ce

ll co

un

t (×

1,0

00

)

Ce

ll c

ou

nt

(×1

,00

0)

0 24 48 72 96 120 1440

50

100

150

200

250

Time (h)

Ce

ll co

un

t (×

1,0

00

)

NSC con

NSC H3.3WT

NSC H3.3 K27M*

*

0

1

2

3

4

5

Den

sito

met

ry: f

old

chan

ge(A

ctiv

e R

AS

/Tot

al R

AS)

***

EV

WT

H3.

3

H3.

3 K

27M

EV

WT

H3.

3

H3.

3 K

27M

H3K27M

H3K27me3

H3

β-ACTIN

pERK1/2

ERK1/2 WC

L

Active

RAS

Total

RAS

CN

MC

-XD

-76

0

SU

-DIP

G IV

SU

-DIP

G 1

3P

HS

JD-D

IPG

00

7

SF

86

28

HS

JD-G

BM

00

1

SU

-pcG

BM

2

Active

RAS

Total

RASH3 WT MU MU MU MU WT WT

Den

sito

met

ry:

fold

cha

nge

CN

MC

-XD

-760

SF8

628

SU

-DIP

G IV

SU

-DIP

G V

IS

U-D

IPG

13P

HS

JD-G

BM

001

SU

-pcG

BM

20

1

2

3

4

5

*

Figure 1.

H3K27M increases activated RAS signaling. A,Direct Trypan blue automated cell count of control and H3K27M-expressing cells over 120 hours (5 days).B,Western blot analysis of H9 NSCs confirming wild-type and H3K27M-mutant expression, loss of H3K27me3, and total H3. NSC control cells were transducedwith empty vector. C,Active GTP-bound RAS immunoprecipitation in control, wild-type, and H3K27M-expressing cells. WCL, whole-cell lysate. D, Densitometryratio of active GTP-bound RAS immunoprecipitation to total RAS in NSC empty vector (EV) control and NSC H3K27M pooled clones from C. E–G, siRNA targetingH-RAS, K-RAS, N-RAS in control (EV), wild-type H3, and H3K27M NSC cells. H and I,Western blot and densitometry ratio of active GTP-bound RASimmunoprecipitation to total RAS in NSC control (Con), NSC PDGFRA, NSC H3K27M, and NSCs cotransfected with PDGFRA and H3K27M (Dual). J and K,Western blot and densitometry ratio of active GTP-bound RAS immunoprecipitation to total RAS in H3WT DIPG cell line, H3K27M cells, and hemisphericpGBM cells. � , P < 0.05; ��� , P < 0.001.

Koncar et al.

Cancer Res; 79(16) August 15, 2019 Cancer Research4030

on December 1, 2020. © 2019 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst June 14, 2019; DOI: 10.1158/0008-5472.CAN-18-3521

EZH2 inhibition did not result in RAS activation as observed withH3K27M, but rather in reduced activated RAS (SupplementaryFig. S1H).

Deregulation of RTKs in H3K27M neural stem cells and DIPGsWe hypothesized that H3K27M mutations deregulate recep-

tor tyrosine kinases (RTK) involved in gliomagenesis as apotential source of RAS activation. qRT-PCR confirmed upre-gulation of RTK growth factors EGF, PDGFA, but not PDGFB(Fig. 2A; �, P < 0.05). We also observed increased RNA expres-sion of major glioma RTKs PDGFRA and EGFR (Fig. 2B; �, P <0.05). We confirmed that H3K27M NSCs, but not control or H3wild-type–expressing NSCs, had increased total and activatedPDGFRA (Fig. 2C).

siRNA screening identifies novel effectors of RAS signalingimportant to DIPG cells

The RASpathway is critical for tumor cell conversion of externalsignals from mitogens into signal transduction events that pro-mote cell growth and proliferation (Fig. 2D). Given the impor-tance of H3K27M in activating RAS signaling and promotingcell proliferation, we set out to identify effectors of RAS signalingthat are critical for its proliferative effect. To accomplish this,we performed a targeted siRNA screen focused on 294 genesknown to directly activate, inactivate, or cooperate with RASpathway signaling (Fig. 2D; Supplementary Table S2). The via-bility of NSC control, NSC H3K27M, and DIPG-007 cells wasassessed at 96 hours post siRNA transfection. We next performedz-score analysis of three individual experiments to identifygene suppressions that selectively target NSCs-H3K27M cells(Fig. 2E). Twenty-six genes were identified that when silencedled to inhibition of NSC-H3K27M growth (viability change >�40%, z-score <�2; �, P < 0.05), and 15 genes were identified forwhich inhibition accelerated growth (viability change >40%,z-score > 2; �, P < 0.05; Fig. 2E; Supplementary Table S2). Tocomplement our NSC data, we performed a validation siRNAscreen in DIPG-007 cells and identified 27 genes that whensilenced led to inhibition of NSC-H3K27M growth (viabilitychange > �50%, z-score < �2; �, P < 0.05) and 6 genes whereinhibition accelerated growth viability change >30%, z-score >2;�, P < 0.05; Fig. 2F; Supplementary Table S2). Twenty-six geneswere identified in both screens (Fig. 2G). Interestingly, AKT2,AURKA, AURKB, MYC, and PDGFRA, previously identified asimportant oncogenes in DIPG, were identified by our approach.We next rescreened and validated our top 10 candidates basedon greatest viability decline at day 5 treatment, using two inde-pendent siRNAs per gene target to reduce the potential of off-target effects that can be caused by pooled siRNA, with MYC andPDGFRA used as positive controls in DIPG-007 cells (Fig. 2H;Supplementary Fig. S2A–S2F). siRNA knockdown efficiency wasvalidated byqRT-PCR and reduced cell viabilitywas confirmed forall 10 genes using a second set of siRNAs and in NSC and NSCH3K27M cells (Supplementary Fig. S2A–S2F).

ERK5 is active and expressed in DIPGStrikingly, three of our validated targets MAP3K2 (MEKK2),

MAP2K5 (MEK5), and MAPK7 (ERK5) are interconnected andform the MEKK2–MEK5–ERK5 signaling cascade that signals inparallel with the more commonly studied RAF–MEK1/2–ERK1/2signaling cascade (Fig. 3A).We confirmed that theH3K27MNSCsexpressed elevated phospho ERK5 compared with empty vector

control NSCs (Fig. 3B). Although low in frequency, we observed4% of DIPG tumors have high-level amplification of the MAPK7gene. Interestingly, MAPK7 gene amplification was mutuallyexclusive with PDGFRAmutation or amplification and also exclu-sive with NF1 mutation or deletion, as indicated in a dataset ofH3.3 K27M–mutant PHGGs for which mutation and copy num-ber information are available (Fig. 3C). We observed no ampli-fication or mutations in this dataset for MAPK3 (ERK1), MAPK1(ERK2),MAPK6, (ERK3),MAPK4, (ERK4),MAP3K1 (MEKK1) oractivators of ERK5, namely MAP2K5, MAP3K2, and MAP3K3.Moreover, we found increased total and phospho ERK5 in fiveDIPGpatient samples comparedwith three normal pediatric pons(Fig. 3D; �, P < 0.05). Immunoblotting confirmed increased totalMEKK2, activated MEK5 (phospho Ser311, Thr315), total ERK5,and activated ERK5 (phospho Thr218, Tyr220 and phosphoSer731 and Thr733) in DIPG cell cultures compared with NSCsempty vector and NSC H3 WT overexpressing controls (Fig. 3E).Furthermore, we observed robust nuclear ERK5 protein expres-sion in the cytoplasm and nucleus of DIPG-IV andDIPG-007 cells(Fig. 3F). We observed significant increased ERK5 transcription inDIPG compared with normal brain (Supplementary Fig. S3A andS3B; �,P<0.05). Expanding our analysis into all PHGG subgroupsrevealed two additional amplification events of ERK5 in H3WTPHGG and no amplifications of ERK1 or ERK2 (SupplementaryFig. S3C). Given the importance of ERKs in glioma and cancer, weexpanded our analysis and observed that GBM had one of thehighest ERK5 RNA expression levels of all adult cancers fromTCGA RNA-seq data (Supplementary Fig. S4A).

Loss of ERK5 inhibits growth of DIPGTo assess the long-term effect of ERK5 knockdown, we gener-

ated doxycycline-inducible shRNA-stable DIPG cell lines(SF8628, DIPG-IV, and DIPG-13p). For all three lines, doxycy-cline treatment (2 mg/mL) resulted in complete ERK5 knockdownat the protein level (Fig. 4A) and reduced cell proliferation, asmeasured by Alamar blue viability assay and 5-bromo-20-deox-yuridine (BrdU) incorporation (Fig. 4B–D; �, P < 0.05). A signif-icant increase in caspase 3/7 cleavage, which is indicative ofapoptosis, was observed on day 5 in ERK5 knockdown cellscompared with control cells (Supplementary Fig. S4B; �, P <0.05). Interestingly, loss of ERK5 resulted in reduction of phos-phorylation of ERK1/2 but not phosphorylation of AKT (Ser473and Thr308), both of which are key signaling mediators of majorglioma proliferation and survival pathways, including for DIPG(Fig. 4A). To validate the ERK5 knockdown phenotype in vivo, wegenerated anorthotopicDIPGxenograft by injectionofDIPG-13pcells into the midbrain of NOD-SCID mice. Mice were random-ized into two groups, a control group and one receiving doxycy-cline administered through drinking water (2 mg/mL) to inducesustained ERK5 knockdown. Mice receiving doxycycline hadsignificantly longer overall survival compared with the controlgroup (Fig. 4E; P < 0.05). Immunoblotting confirmed absence ofERK5 in the doxycycline group compared with the control group(Fig. 4F).

Unlike ERK1/2, ERK5 contains both a kinase domain and atransactivation domain. To ascertain which domain or domainsare responsible for ERK5 function, we generated lentiviral con-structs containing HA-tagged complete or partial ERK5 codingsequence: wild-type, kinase dead (D200A), transactivationdomain deleted (delta TAD), and kinase dead with deltaTAD (Fig. 4G). We also generated another inducible ERK5

RAS Pathway Vulnerabilities in Histone-Mutant DIPG

www.aacrjournals.org Cancer Res; 79(16) August 15, 2019 4031

on December 1, 2020. © 2019 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst June 14, 2019; DOI: 10.1158/0008-5472.CAN-18-3521

shRNA–stable cell line in SF8628 cells, inwhich the shRNA targetsthe 30UTR of ERK5 to prevent gene silencing of transduced ERK5constructs (Fig. 4H). Reintroduction of ERK5 in knockdowncells was confirmed postinfection with HA immunoblotting in

SF8628 cells (Fig. 4I). Full-length ERK5 was able to rescue theproliferation defect caused by shRNA-mediated suppression, withthe kinase dead and delta TAD constructs partially rescuinggrowth defects as evaluated by BrdU incorporation (Fig. 4J and

Figure 2.

A RAS-targeted siRNA screen identifies novel RAS effector genes that drive H3K27M growth.A, qRT-PCR, for growth factors in control, wild-type–expressing,and H3K27M NSC cells. B, qRT-PCR for growth factors in control, wild-type–expressing, and H3K27M NSCs. C,Western blot confirmation of EGFR and PDGFRAactivity in H3K27M NSCs. D, Schematic of RAS signaling pathway and targeted siRNA screen. E and F, NSC control, NSC H3K27M, and DIPG-007 cells weretransfected with a pooled library of 294 siRNAs. Cell viability was assessed by the Alamar blue cell viability assay 120 hours after siRNA transfection. Data werenormalized using the standard z-score method. Significance of potential hits were determined using z-score cut-off values� 2 (dotted line), which correspondedto a P value of <0.05 in all three biologic replicates (Rep 1–3). G, Venn diagram of overlapping targets and unique targets between H3K27M NSC cells andDIPG-007 cells. H, Re-screen of top targets using siRNA and direct Trypan blue automated cell count of DIPG-007 cells over 7 days. � , P < 0.05.

Koncar et al.

Cancer Res; 79(16) August 15, 2019 Cancer Research4032

on December 1, 2020. © 2019 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst June 14, 2019; DOI: 10.1158/0008-5472.CAN-18-3521

K; �,P<0.05). Expressionof double dead (kinase dead, delta TAD)ERK5 had no proproliferative effect (Fig. 4J and K).

Pharmacologic inhibition of ERK5 inhibitsDIPG tumor growthand promotes apoptosis

We tested various ERK5 inhibitors that are specific for theireffects on H3K27M cells: ERK5-IN-1 (ERK5-specific inhibitor),BIX02189 (dual MEK5 and ERK5 inhibitor), XMD8-92 (dual

MEK5 and ERK5 inhibitor), and TG02 (CDK9 and ERK5 inhib-itor). We generated EC50 curves for DIPG-007 (Fig. 5A), DIPG-IV(Fig. 5B), SF8628 (Supplementary Fig. S5A), and DIPG-13pcells (Supplementary Fig. S5B). EC50 values were lowest forERK5-IN-1 (EC50s from �0.5–1 mmol/L) and TG02 (EC50sfrom 50–100 nmol/L). We next compared the EC50 values ofERK5-IN-1 and TG02 to a DIPG histone wild-type H3 cell line,CNMC-XD-760 and hemispheric glioblastoma. Interestingly,

A BGrowth factors

A/B/CRAF

MEK1/2

ERK1/2

MEKK2

MEK5

ERK5

*

*

Cell proliferation

Cell migration

*

C

CDK

NS

C E

V

NS

C H

3 W

T

NS

C H

3K27

M

SF8

628

SU

-DIP

G 1

3pE

p-ERK5 (T218/Y220)

F

ERK5

p-MEK5 (S311/T315)

β-ACTIN

MEK5

N C N C

ERK5

DIPG IV DIPG 007

Histone H3

GAPDH

pERK 1/2

ERK 1/2

MEKK2

p-ERK5 (S731/T733)

No alterations Amplification Deep deletion Mutation

H3.3 K27M

ERK5 (4%)

NF1 (9%)

PDGFRA (18%)

ERK2 (MAPK1)

ERK1 (MAPK3)

MEK1 (MAP2K1)

MEK2 (MAP2K2)

MEK5 (MAP2K5)

MEKK2 (MAP3K2)

EV

WT

H3

.3

WT

H3

.3 K

27

M

p-ERK5 S731/T733

p-ERK5 T218/Y220

ERK5

β-ACTIN

D

ERK5

P1

P2

P3

D1

D2

D3

D4

D5

p-ERK5 (S731/T733)

β-ACTIN

Pons DIPG012345678

Fold

cha

nge

ER

K5/

β-A

CTI

N

*

SU

-DIP

G IV

HS

JD-D

IPG

-007

SU

-DIP

G V

I

Figure 3.

ERK5 is upregulated and active in DIPG. A, Schematic of the ERK5 signaling cascade. Red asterisks indicate candidate genes from siRNA screen (Fig. 2G). B,Western blot of phospho and total ERK5 in NSC cell–expressing empty vector (EV), wild-type H3.3, and H3K27M. C,Oncoprint map of genomic alterations in H3.3K27M–mutant pediatric HGG patients. D,Western blot and densitometry of pERK5 and ERK5 in five DIPG patient samples compared with three pediatric ponsfrom nontumor patients. E,Western blot of the MAPK signing network in patient-derived DIPG cultured cells compared with empty vector NSCs and NSCsexpressing wild-type H3 or H3K27M. F,Western blot after a cellular fractionation assay for ERK5. GAPDHwas used to confirm cytoplasmic purity and H3.3 fornuclear purity. � , P < 0.05.

RAS Pathway Vulnerabilities in Histone-Mutant DIPG

www.aacrjournals.org Cancer Res; 79(16) August 15, 2019 4033

on December 1, 2020. © 2019 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst June 14, 2019; DOI: 10.1158/0008-5472.CAN-18-3521

A B

No Dox

+Dox

ERK5

β-ACTIN

No Dox

+Dox

pERK1/2

ERK1/2

pAKT ser473

1 140 407 578 690 816

NES1 PB1 Kinase Domain PR-1 NLS PR-2 NES2 TAD

N-Terminus C-TerminusERK5 (MAPK7)

NES1 PB1 D200A PR-1 NLS PR-2 NES2 TAD

WT

Kinase

Dead

TAD

Dead

Kinase +

TAD Dead

NES1 PB1 Kinase Domain PR-1 NLS PR-2 NES2

NES1 PB1 D200A PR-1 NLS PR-2 NES2

ED

H J

AKT

No Dox

+Dox (Day 4

)

+Dox (Day 5

)

ERK5

SU-DIPG IV SF8628 SU-DIPG13p

No Dox

+Dox

I

ConERK5

ERK5-KD

ERK5-ΔTA

D

ERK5-KD Δ

TAD

HA

β-ACTIN

β-ACTIN

C

Control

ERK5 knockdown

ERK5

ERK5-KD

ERK5- TAD

ERK5-KD TAD

0 1 2 3 4 50

1

2

3

4

5

Day

Abs

(450

nm

)

0 1 2 3 4 50

1

2

3

4

Day

Abs

(450

nm

)

Control

Dox

0 1 2 3 4 50

1

2

3

4

Day

Abs

(450

nm

)

** *

*

*

*

*

*

ns

K

* * *

0 1 2 3 4 5012345

Day

Abs

(450

nm

) Control

Dox

**

G

F

na

Vehicle

ERK5 KD

0 10 20 30 400

25

50

75

100

Survival (days)

Per

cent

sur

viva

l

Con

Con

Con

+ D

ox+

Dox

+ D

ox

ERK5

β-ACTIN

DIPG-13p DIPG-IV

SF8628

Compared to knockdown

Figure 4.

ERK5 loss inhibits growth in DIPG cells. A, Western blot of control and doxycycline (Dox)-treated cells (2 mg/mL) confirming ERK5 knockdown andanalysis of activated ERK1/2 and AKT. Cells were treated with doxycycline for 5 days. B–D, 5-bromo-20-deoxyuridine (BrdU) cell proliferation assay ofDIPG-13p cells (B), DIPG-IV cells (C), and SF8628 cells (D). Cells were treated with no doxycycline or doxycycline starting 5 days prior to the assay.E, Kaplan–Meier survival curves of DIPG-13p–injected orthotopic xenograft models (n ¼ 5 per group; P < 0.05). ERK5-inducible shRNA for knockdown(ERK5 KD) was accomplished by administration of doxycycline (2 mg/mL) with 5% sucrose in drinking water. F, Western blot in control mice versusdoxycycline mice from E confirming ERK5 KD. G, Schematic of ERK5 rescue constructs with alterations made to the kinase domain (KD, kinase dead),transcriptional activation domain (DTAD, TAD deleted), or both (KD DTAD). H, Western blot in SF8628 cells confirming shRNA targeting ERK5 30UTRinhibits ERK5 at the protein level. I, HA Western blot confirming ERK5 constructs expression in SF8628 cells 72 hours after lentiviral transduction.J, BrdU cell proliferation assay of SF8628 cells transduced with various ERK5-HA–tagged constructs from G. K, Statistical analysis of J compared withdoxycycline-treated cells. � , P < 0.05; na, not applicable; ns, nonsignificant.

Koncar et al.

Cancer Res; 79(16) August 15, 2019 Cancer Research4034

on December 1, 2020. © 2019 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst June 14, 2019; DOI: 10.1158/0008-5472.CAN-18-3521

CNMC-XD-760 had the highest EC50 values for ERK5-IN-1compared to H3 mutant DIPG cells and hemispheric pGBM cellsSU-pcGBM2 andHSJD-GBM-001 (Fig. 5C; �, P < 0.05). NSCs and

NSCs overexpressing wild-type H3 also had higher ERK5-IN-1EC50 values compared with H3K27M DIPG and hemisphericpGBM lines but EC50 values were significantly lower in NSCs

D G

H

1 3 5 7 90

1

2

3

4

Day Day Day

Cel

l cou

nt (×

106 )

Cel

l cou

nt (×

106 )

Cel

l cou

nt (×

106 )

* * ** * ***

ControlERK5-IN-1TG02

1 3 5 7 90

1

2

3

4 ControlERK5-IN-1TG02

* * * ** * * *

E F

β-ACTIN

ERK5

p-ERK5

S731/T733

pMEK5

Con100 n

mol/L

500 nm

ol/L

50 nm

ol/L

100 nm

ol/L

ERK5-IN-1 TG02

p-ERK5

T218/Y220

MEK5

ControlERK5-IN-1TG02

1 3 5 7 90

1

2

3

4

* * * ** * **

I J

Con100 n

mol/L

500 μmol/L

50 nm

ol/l

100 nm

ol/L

ERK5-IN-1 TG02

Ann

exin

V

DIPG-007 DIPG IV

104

103

102

0

105

0 102 103 104 105PI

0 102 103 104 105PI

104

103

102

0

105

Ann

exin

V

0 102 103 104 105PI

104

103

102

0

105

Ann

exin

V

0 102 103 104 105PI

104

103

102

0

105

Ann

exin

V

Ann

exin

V

Ann

exin

V

104

103

102

0

105

0 102 103 104 105PI

0 102 103 104 105PI

104

103

102

0

105

Con

ERK5-IN-1

TG02 Con

ERK5-IN-1

TG020

50

100

Per

cent

(%)

ViableLate apoptosis

Early apoptosisNecrosis

DIPG-007 DIPG IV * * * *

Control

ERK5-IN-1

TG02

-6 -4 -2 0 2 40

25

50

75

100

Log drug (μmol/L) Log drug (μmol/L)

% V

iabi

lity

BIX02189 1.10 μmol/L

XMD8-92 2.25 μmol/L

ERK5-IN-1 600 nmol/L

TG02 93 nmol/L

A B

-6 -4 -2 0 2 40

25

50

75

100

% V

iabi

lity

BIX02189 1.60 μmol/L

XMD8-92 2.35 μmol/L

ERK5-IN-1 500 nmol/L

TG02 60 nmol/L CN

MC

-XD

-760

SU

-pcG

BM

2G

BM

001

SU

-DIP

G13

SU

-DIP

G IV

DIP

G 0

07N

HA

NS

CN

SC

H3

WT

NS

C H

3 K

27M

0

1,000

2,000

3,000

4,000

5,000

C

*

*

H3 WT

H3 K27M

HSJD-DIPG-007 SU-DIPG-IV

EC

50

Val

ues

(ER

K5-

IN-1

nm

ol/L

)

Figure 5.

Pharmacologic inhibition of ERK5 triggers cell death and apoptosis in DIPG cells. A, EC50 dose–response curves at 96 hours in DIPG-007 cells using varyinginhibitors of ERK5: BIX02189, XMD8-92, ERK5-IN-1, and TG02. B, EC50 dose–response curves at 96 hours in DIPG-IV cells using varying inhibitors of ERK5:BIX02189, XMD8-92, ERK5-IN-1, and TG02. C, EC50 values for ERK5-IN-1 at 96 hours in H3K27M or H3WT cell lines. D–F, Direct Trypan blue automated cell countof cells treated with DMSO vehicle (control), ERK5-IN-1 (500 nmol/L), and TG02 (100 nmol/L) for DIPG-007 cells (D), DIPG-IV cells (E), and SF8628 cells (F).G, Annexin-propidium iodide (PI) quantification of DIPG cells treated with ERK5 inhibitors at 48 hours from H. Compared with DMSO vehicle control cells, bothERK5-IN-1 and TG02 had decreased viable cells (Annexin negative, PI negative) and increased populations in early apoptosis (Annexin positive, PI negative), lateapoptosis (Annexin and PI positive) and necrosis (PI positive, Annexin negative). H, Representative Annexin-PI contour plots with quadrant gates showing fourpopulations in DIPG-007 and DIPG-IV cells treated with vehicle (DMSO), ERK5-IN-1, and TG02. I and J,Western blot confirming ERK5-IN-1 and TG02 inhibit theERK5 autophosphorylation site at various doses in DIPG-007 (I) and DIPG-IV (J) cells. � , P < 0.05.

RAS Pathway Vulnerabilities in Histone-Mutant DIPG

www.aacrjournals.org Cancer Res; 79(16) August 15, 2019 4035

on December 1, 2020. © 2019 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst June 14, 2019; DOI: 10.1158/0008-5472.CAN-18-3521

expressing H3K27M (Fig. 5C). Similar results were observed withrespect to TG02 (Supplementary Fig. S5C). Cell counting follow-ing 9 days of treatment for DIPG-007, DIPG-IV, and SF8628 cellsshowed significant decreases in cell number with 500 nmol/L ofERK5-IN-1 and 50nmol/L of TG02, as comparedwith cells treatedwith vehicle (Fig. 5D–F; �, P < 0.05). Annexin-PI flow cytometryconfirmedDIPG-007 andDIPG-IV cells treatedwith ERK5-IN-1orTG02were undergoing significantly higher rates of apoptosis linescompared with controls (Fig. 5G and H; �, P < 0.05). Both ERK5-IN-1 and TG02 were confirmed to inhibit ERK5 autophos-phorylation and ERK5 activity sites (Ser731 and Thr 733) at100 nmol/L and 50 nmol/L, respectively, but did not inhibitERK5 Thr218/Tyr220, which is regulated by MEK5 (Fig. 5I and J).We also observed diminished cell growth and induction ofapoptosis in DIPG-13p and SF8628 cells treated with ERK5-IN-1or TG02 (Supplementary Fig. S5D–S5G). Loss of ERK5 by siRNAor pharmacologic inhibitors in a DIPG histone wild-type H3 cellline, CNMC-XD-760, also had a significant reduction in viabilityalthough the effect was not as pronounced in mutant histoneDIPG cells (Supplementary Fig. S6A and S6B). Loss of ERK5 in ahemispheric pediatric GBM (SU-pcGBM2 EGFR amplified), dis-played similar growth inhibition as our mutant H3K27M cells(Supplementary Fig. S6C). Expression of wild-type HA-taggedERK5 and constitutively active ERK5 (ERK5 T733E) promotedresistance to both ERK5-IN-1 and TG02 as indicated by increasedEC50 values (Supplementary Fig. S6D–S6F) and reduced cleavedCaspase 3/7 levels, with the greatest resistance occurring incells expressing the ERK5 T733E constitutively active mutant(Supplementary Fig. S6G).

MYC is a direct target of ERK5We performed a high-throughput ELISA-based antibody array

for quantitative protein phosphorylation profiling using 269phospho antibodies (94 known cancer proteins in total) toidentify ERK5 signaling mediators (Supplementary Table S3).The initial screen was performed using DIPG-IV cells with orwithout ERK5 knockdown. We identified 24 differentially phos-phorylated peptides associated with 15 proteins in ERK5 knock-down DIPG-IV cells compared to controls (Fig. 6A and B).Strikingly, we observed significant reduction of phosphorylationat MYC serine residue 62 (S62), a critical phosphorylation site forMYC stability. We confirmed downregulation of total MYC andphospho MYC (S62) in ERK5 knockdown cells compared tocontrol (Fig. 6C). Moreover, ERK5 inhibition by ERK5-IN-1 orTG02 lead to reduced phospho and total MYC (Fig. 6C) in DIPG-IV and SF8628 cells. MG132, a proteasome inhibitor, restoredprotein expression of total MYC in DIPG-IV cells, indicating thatloss of ERK5 results in proteasomal degradation ofMYC (Fig. 6D).We observed no transcriptional changes of MYC at the RNA level(Supplementary Fig. S7A).

Several kinases including CDK1, ERK1, and ERK2 have beenreported to stabilize MYC at (S62). Our results to this pointsuggest ERK5 may play a role in MYC S62 phosphorylation.Immunoprecipitations of ERK5 from DIPG-007, DIPG-IV, andSF8628 cells coprecipitated MYC, and reverse IPs of MYC co-precipitated ERK5 (Fig. 6E). We performed an in vitro kinase assaywith purified MYC and ERK5 and observed direct phosphoryla-tion on MYC(S62) in the presence of ERK5 and ATP (Fig. 6F).Phosphorylation of MYC(S62) was inhibited in the presence ofERK5-IN-1 and TG02 (Fig. 6F). Introduction of nondegradableMYC (S62D) in DIPG cells significantly rescued proliferation

defects of ERK5 knockdown (Fig. 6G and H; �, P < 0.01). Lastly,MYC S62D promoted resistance to both TG02 and ERK5-IN-1(Supplementary Fig. S7B–S7D; �,P<0.05).DIPG-13p cells harborMYCN amplification and, interestingly, we observed that ERK5loss or inhibition by ERK5-IN-1 or TG02 resulted in MYCNprotein and transcriptional downregulation (SupplementaryFig. S7E and S7F; �, P < 0.05).

ERK5 inhibitors increase survival of mice bearing DIPGxenografts and for mice bearing syngeneic DIPG

We next examined whether ERK5-IN-1 and TG02 could extendsurvival in an orthotopic DIPG xenograft model. Mice bearingSF8628 intracranial xenografts were treated with vehicle, ERK5-IN-1 (50mg/kg), or TG02 (20 mg/kg) for 2 weeks. Mice receivingERK5-IN-1 had a median survival of 40 days compared withvehicle-treated mice whose median survival was 33 days (Fig. 7A;�, P < 0.05). Mice treated with TG02 showed a substantial increaseinmedian survival: 59days (Fig. 7A; �,P<0.05).Hematoxylin andeosin staining confirmed high-grade tumor histology and nogross histology differences between treatment arms, but micetreated with ERK5-IN-1 or TG02 showed reduced Ki-67 staining,a marker of cell proliferation (Fig. 7B–E; �, P < 0.05). Similarresults were obtained in a second xenograft model usingDIPG-13p cells, with ERK5-IN-1–treated mice showing a mediansurvival of 30 days and TG02-treatedmice with amedian survivalof 40 days. Both median survivals were significantly greater thanvehicle control mice whose median survival was 22.5 days(Fig. 7F–J; �, P < 0.05). We confirmed in vivo inhibition of ERK5autophosphorylation and induction of apoptosis by cleavedPARP on mice sacrificed after 5 days of treatment with ERK5-IN-1 or TG02 (Supplementary Fig. S7G). Finally, we used twoestablished cell lines derived from transgenic mouse models ofDIPG. Both lines are p53 null and express PDGFB, differingonly in their H3.3mutation status. Strikingly, H3K27M cells weremore sensitive than H3 wild-type cells to ERK5 knockdown(Supplementary Fig. S7H and S7I).

DiscussionThere are currently no effective treatment options for

patients with DIPG (25). Recent large-scale genomic analysesof DIPG patient samples have redirected the course of treat-ment development toward targeted therapies. Importantly,genomic analyses revealed that an overwhelming majority ofDIPG cases have the H3K27M mutation (6, 18, 26). The highfrequency and predictability of H3K27M mutation in DIPGrender it and its consequences attractive therapeutic targets.Further investigation of the effects of H3K27M in DIPGpathobiology, as we have done in the current investigation,will continue to expose potential therapeutic vulnerabilitiesfor these tumors.

H3K27M is known to result in inhibition of the PRC2 complex,specifically the catalytic subunit of the complex, EZH2. EZH2inhibition results in global reduction of H3K27 trimethylationand attendant alterations in cell transcriptomes (10). Severalstudies have demonstrated therapeutic possibilities associatedwith targeting histone modifiers and transcriptional regulatorsin DIPGmodels. Notably, DIPG growth has proven vulnerable tothe: (i) inhibition of histone demethylases (27); (ii) inhibition ofresidual EZH2 function (28); (iii) inhibition of histone deacety-lases (HDAC) by panobinostat, a drug that is currently being

Koncar et al.

Cancer Res; 79(16) August 15, 2019 Cancer Research4036

on December 1, 2020. © 2019 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst June 14, 2019; DOI: 10.1158/0008-5472.CAN-18-3521

evaluated in a DIPG clinical trial (NCT02717455; ref. 11); and(iv) disruption of transcription by bromodomain inhibitors andCDK7 inhibitors (13). The efficacy indicated by the results of

preclinical success for these targeted therapies suggests that dis-rupting epigenetic and transcriptional effects of H3K27Mmay bean effective strategy for DIPG treatment. Moreover, a recent study

A B

C D

Co

ntr

ol

DO

XC

on

tro

l

DO

XC

on

tro

l

DO

XC

on

tro

lD

OX

MYC

E

SU-DIPG IV SF8628

MYC S62D - - + + - - + +

F

MG132 - - - - + + + +

Co

ntr

ol

DO

X

Co

ntr

ol

DO

X

Co

ntr

ol

DO

X

Co

ntr

ol

DO

X

TG02

ERK5-IN-1

MYC

ERK5

ATP

- - + - -

- + - - -

+ + + + +

+ + + + +

- + + + + Phospho

MYC S62

G

IgG

IP M

YC

IP E

RK

5

IgG

IP M

YC

IP E

RK

5

MYC

ERK5

DIPG 007 SU-DIPG IV SF8628

H

IgG

IP M

YC

IP E

RK

5

ERK5

MYC

β-ACTIN

ERK5

SU-DIPG IV cells

No Dox

(5 days)

Dox

(5 days)

0.0

0.5

1.0

1.5

2.0

2.5

ER

K5

KD

Ph

osp

ho

/To

tal:

Co

ntr

ol P

ho

sph

o/T

ota

l

CDC25A, MYC,MEK1, p44/42 MAPK

ERK8,elF4E,MEK1, PDGFRB

HSP27,HSF1,SRC

JAK2,JUNB,STAT3,STAT1,STAT5A

STAT5A,MYC,STAT3,JAK2,STAT4

Co

n

ER

K5

KD

Co

n

ER

K5

KD

0

20

40

60

Ce

ll d

ou

blin

g t

ime

(h

)

MYC S62D - - + + - - + +

**

*C

on

ER

K5

KD

Co

n

ER

K5

KD

0

20

40

60

80

Ce

ll d

ou

blin

g t

ime

(h)

**

*

Quantify

Phospho:Total(Knockdown)/

Phospho:Total (Control)

Co

ntr

ol

DO

X

ER

K5

-IN

-1

TG0

2

Co

ntr

ol

DO

X

ER

K5

-IN

-1

TG0

2

SF8628 DIPGIV

ERK5 pS731 pT733

MYC pS62

MYC

β-ACTIN

β-ACTIN

ERK5

Figure 6.

ERK5 inhibition reduces MYC protein stability. A, Schematic of identifying differentially expressed phosphoproteins in DIPG-IV cells expressing ERK5 and ERK5knockdown. B, Ratio of phospho:total protein changes in ERK5 knockdown versus ERK5 control DIPG-IV cells. Ratios in blue were downregulated (z-score < 2),while ratios in red were upregulated (z-score > 2). C,Western blot of phospho MYC S62 in DIPG cells treated with doxycycline (ERK5 KD), ERK5-IN-1(500 nmol/L), or TG02 (100 nmol/L). D,Western blot of MYC in DIPG-IV cells treated with doxycycline and MG132 (10 mmol/L). E, Immunoprecipitation of MYCand ERK5 followed byWestern blot of ERK5 and MYC. F, In vitro kinase assay of purified recombinant ERK5 and MYC in presence or absence of ERK5 inhibitors.G,Western blot of DIPG cells retransfected with constitutively active MYC S62D in presence of ERK5 loss. H, Cell doubling times of control and ERK5 knockdownDIPG cells and the effect of MYC S62D rescue. SF8628 cells (left) and DIPG-IV cells (right). � , P < 0.05.

RAS Pathway Vulnerabilities in Histone-Mutant DIPG

www.aacrjournals.org Cancer Res; 79(16) August 15, 2019 4037

on December 1, 2020. © 2019 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst June 14, 2019; DOI: 10.1158/0008-5472.CAN-18-3521

characterizing the transcriptional dependencies of DIPG identi-fied several transcripts and super enhancers of gene targets of theMAPK pathway (13). Most notable were MAP3K2, an upstreamactivator of ERK5, and MEF2A/MEF2C transcription factors,which are both direct targets of ERK5 (29, 30).

Because H3K27M causes global epigenetic and transcriptionalchanges in DIPG cells, we hypothesized that it is likely activatingthe transcription of known oncogenic pathways to sustain tumorgrowth and survival. To address this hypothesis, we investigatedthe impact of H3K27Mmutations on RAS-MAPK/ERK5 signaling.

Figure 7.

ERK5 inhibition increases in vivo survival of DIPG patient-derived xenograft models.A, Kaplan–Meier survival curves of SF8628 orthotopic xenograft models (n¼7 per treatment group). Arrows indicate period of daily treatment (2 cycles of 5 days on 2 days off). B, Sagittal section and hematoxylin and eosin stainingconfirming tumor growth in the midbrain. C, High-power magnification hematoxylin and eosin confirming high-grade glioma histology of tumors. D, Ki67proliferative marker staining of xenograft models treated with vehicle, ERK5-IN-1, or TG02. E, Ki67 quantification from D. Three mice per group with average ofKi67-positive cells taken from 10 high-power fields of view. F, Sagittal section and hematoxylin and eosin staining confirming tumor growth in the midbrain.G, High-power magnification hematoxylin and eosin confirming high-grade glioma histology of tumors. H, Ki67 proliferative marker staining of xenograft modelstreated with vehicle, ERK5-IN-1, or TG02. I, Kaplan–Meier survival curves of DIPG-13p orthotopic xenograft models (n¼ 7 per treatment group). Arrows indicateperiod of daily treatment (2 cycles of 5 days on 2 days off). J, Ki67 quantification from H. Three mice per group with average of Ki67-positive cells taken from10 high-power fields of view. Scale bars, 50 mm. � , P < 0.05.

Koncar et al.

Cancer Res; 79(16) August 15, 2019 Cancer Research4038

on December 1, 2020. © 2019 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst June 14, 2019; DOI: 10.1158/0008-5472.CAN-18-3521

It has been shown by others that RAS-MAPK/ERK activation inDIPG can result from: (i) PDGFRA amplification, (ii) the lossof RAS-GAP NF1, and rarely, (iii) point mutations in K-RAS andN-RAS (6, 31). However, transcriptional effects that would con-nect H3K27M with RAS activation have not been previouslyexplored.

While the primary oncogenic function of H3K27M is thoughtto be PRC2 inhibition and loss of H3K27 trimethylation, weobserved that while either H3K27M expression or treatment withan EZH2 inhibitor reduced H3K27 trimethylation, EZH2 inhibi-tion resulted in decreased active RAS (Supplementary Fig. S1H).Consistent with previous reports, our observation supports thatEZH2 inhibition and H3K27M mutation are not synonymous.EZH2 inhibitors can ablate focal H3K27 trimethylation thatpersists in H3K27MDIPG (28, 32). In addition, EZH2 inhibitioncauses loss of H3K27 dimethylation and trimethylation whileH3K27M disproportionately affects trimethylation (32). Suchdifferences between H3 mutation and EZH2 inhibition appearto be important, given their opposing effects on RAS activity, andworthy of further investigation.

RAS protein activity is known to be important in multiplecancers, yet despite sustained efforts RAS has remained largelyundruggable. Targeting key effectors of RAS signaling may provemore feasible than, andperhaps as effective as, targetingRAS itself.The RAS–MAPK signaling pathway is comprised of a highlydiverse family of proteins with interactions that are extensivelynetworked. To identify potential therapeutic vulnerabilities with-in this network, we used a targeted siRNA screen that revealed lessstudied signaling arms of the RAS pathway as being important toDIPG growth. Among these, ERK5 was of particular interest as atreatment target because, unlike ERK1/2, it has two distinctfunctional domains. Interestingly, we uncovered an unreportedfeedback between ERK5 and ERK1/2, whereby loss of ERK5inhibits activation of ERK1/2.

While the relationship we describe between histone muta-tion and RAS/ERK5 activation is novel, H3K27M is not the soledriver of RAS activation and active ERK5 is not exclusive toDIPG. Other tumors may also be sensitive to ERK5 inhibitionas ERK5 has been shown to have oncogenic activity in othercancers and is known to promote proliferation as well asangiogenesis (33–35). Notably, the H3K27M DIPG and H3WT pediatric hemispheric GBM cell lines we tested had similaractive RAS levels, which were higher than that observed in theH3 WT DIPG cell line. While we were only able to test one H3WT DIPG cell line, the lower relative active RAS supportsH3K27M having a function in RAS activation. The H3 WTDIPG cell line was also the least sensitive to ERK5 inhibition,while the H3 WT pediatric GBM and H3K27M DIPG cell lineswere similarly responsive. While H3K27M mutation results insensitivity to ERK5 loss or inhibition, targeting ERK5 in pedi-atric GBM also warrants consideration. Additional investiga-tion into the role of ERK5 in different H3 WT contexts will be afocus of future work.

ERK5 suppression through inducible shRNA or pharmacologicinhibition attenuated DIPG growth through destabilization ofMYC protein. Our working model posits that MYC and activationof ERK1/2, in associationwith RAS activation, are critical to tumorgrowth. Although ERK5 has been shown to promote phosphor-ylation ofMYC as a compensatory response to ERK1/2 inhibition,our group is the first to identify that direct targeting of ERK5inhibits MYC stability and activity of ERK1/2 (36, 37). Moreover,

we show that ERK5-mediated MYC regulation is important inDIPG pathobiology. However, a nondegradable version of MYCwasnot fully able to rescue antitumor effects of ERK5 suppression,thereby suggesting that additional ERK5 targets may contribute toits proproliferative activity.

BothRASandMYCare considered undruggable,whereas small-molecule inhibitors of ERK5 are available and could be rationallycombined with therapies in DIPG, such as radiation. ERK5inhibition by gene silencing or small-molecule inhibition alsoled to reduction of MYCN RNA and protein in DIPG-13p cells,suggesting that ERK5has importance inMYCN transcription. Thisrelationship between ERK5 and MYCN has been observed inneuroblastoma (33).

Our ERK5 inhibitor experiments included the use of in vivomodels that yielded results showing that treatment with inhi-bitors ERK5-IN-1 or TG02 increased survival of mice bearingpatient-derived xenograft (PDX) tumors. In association withthe inhibitor experiments, we determined that constitutivelyactive ERK5 or MYC confers some, but not complete, resistanceto TG02, indicating that this drug works in part through itseffect on ERK5 and MYC. Importantly, TG02 is blood–brainbarrier penetrant and in clinical trials for several adult cancers(NCT03224104, NCT02942264, and NCT01204164). Interest-ingly, we observed reduced Ki-67, a marker of proliferation, intumors collected from mice weeks after treatment of eitherERK5-IN-1 or TG02 had ceased. This may reflect a potentialtreatment escape known as cancer cell dormancy, where tumorcells recovering from minimal residual disease (MRD) havebeen shown to have a less proliferative state. Moreover, thisphenomenon is known to occur when targeting the RAS–MAPKor PI3K–AKT signaling pathways (38–40). However, we treatedmice 5 days each week for two weeks and this potentialresistance mechanism may possibly be negated by either treat-ment regimen optimization or combination therapies. Target-ing CDKs in DIPG has been considered as a therapeutic strat-egy, and our data support dual CDK and ERK5 inhibition as astrategy that merits consideration.

Our phosphoproteomic array revealed activation of the JAK/STAT pathway in response to ERK5 knockdown. In addition toERK5, TG02 also inhibits JAKs, which may contribute to theapoptotic phenotype in response to TG02 treatment. Interesting-ly, STAT3 is a transcriptional regulator of MEK5, the activator ofERK5 (41). Activation of the JAK/STAT pathway may be a poten-tial compensation mechanism to elicit additional ERK5 activa-tion. Elucidating the relative importance of JAK/STAT activationin the context of ERK5 inhibition requires further investigation, asit may have translational relevance.

Although the impact of the H3K27M mutation is wellcharacterized for its ability to activate and alter both transcrip-tion and epigenetics, its role in cell signaling has not been wellcharacterized. Here, we have investigated the role of H3K27Min cell signaling and determined that this mutation engagesmultiple unreported effectors of RAS signaling that are essen-tial for promoting DIPG tumor growth. Some of these effectorsare therapeutically accessible, and their discovery as key con-tributors to the tumor biology of DIPG could motivate clinicaltrials to assess potential patient benefit from their targetedinhibition.

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

RAS Pathway Vulnerabilities in Histone-Mutant DIPG

www.aacrjournals.org Cancer Res; 79(16) August 15, 2019 4039

on December 1, 2020. © 2019 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst June 14, 2019; DOI: 10.1158/0008-5472.CAN-18-3521

Authors' ContributionsConception and design: R.F. Koncar, R.L. Hamilton, A. Broniscer, I.F. Pollack,S. AgnihotriDevelopment of methodology: R.F. Koncar, M.L. Wassell, A.L. Locke,R.L. Hamilton, G. Kohanbash, J. Nazarian, S. AgnihotriAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): R.F. Koncar, B.R. Dey, A.-C.J. Stanton, N. Agrawal,M.L.Wassell, A.L. Locke, R.L. Hamilton, G. Kohanbash, J. Nazarian, S. AgnihotriAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): R.F. Koncar, B.R. Dey, N. Agrawal, M.L. Wassell,A.L. Locke, L. Sanders, O. Morozova-Vaske, M.I. Myers, R.L. Hamilton,N.M. Amankulor, S. AgnihotriWriting, review, and/or revision of the manuscript: R.F. Koncar, B.R. Dey, A.-C.J. Stanton, N. Agrawal, L.H. McCarl, L. Sanders, M.I. Myers, R.L. Hamilton,B. Hu, N.M. Amankulor, J. Felker, J. Nazarian, C.D. James, A. Broniscer,I.F. Pollack, S. AgnihotriAdministrative, technical, or material support (i.e., reporting or organizingdata, constructing databases): B.R. Dey, A.-C.J. Stanton, N. Agrawal,M.L. Wassell, A.L. Locke, M.I. Myers, B. Hu, M. Kambhampati, O.J. Becher,R. Hashizume, S. AgnihotriStudy supervision: I.F. Pollack, S. Agnihotri

Other (supervision of L. Sanders): O. Morozova-VaskeOther (provided HSJD-DIPG cell models): A.M. Carcaboso

AcknowledgmentsWe would like to thank Dr. Michelle Monje from Stanford University

(Stanford, CA) for providing the DIPG-IV, DIPG-VI, and DIPG-13p cells.We would like to thank Thomas M. Estok and Dr. Kistan Meetze fromTragara Pharmaceuticals for kindly providing TG02. S. Agnihotri was fundedby Children's Trust of Pittsburgh Young Investigator Award, V-Foundation(funded by WWE in Honor of Connor's Cure), and the UPP Foundation ofPittsburgh. I.F. Pollack was funded by the Connor's Cure Fund and theTranslational Brain Tumor Fund of the Children's Hospital of PittsburghFoundation.

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

Received November 7, 2018; revised February 5, 2019; accepted June 11,2019; published first June 14, 2019.

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

CBTRUS statistical report: primary brain and central nervous systemtumors diagnosed in the United States in 2008–2012. Neuro Oncol2015;17 Suppl 4:iv1–iv62. doi: 10.1093/neuonc/nov189.

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

3. Jones C, Karajannis MA, Jones DTW, Kieran MW, Monje M, Baker SJ, et al.Pediatric high-grade glioma: biologically and clinically in need of newthinking. Neuro Oncol 2017;19:153–61.

4. Hoffman LM, Veldhuijzen van Zanten SEM, Colditz N, Baugh J, Chaney B,Hoffmann M, et al. Clinical, radiologic, pathologic, and molecular char-acteristics of long-term survivors of diffuse intrinsic pontine glioma(DIPG): a collaborative report from the International and EuropeanSociety for Pediatric Oncology DIPG Registries. J Clin Oncol 2018;36:1963–72.

5. Warren KE. Diffuse intrinsic pontine glioma: poised for progress.Front Oncol 2012;2:205.

6. Mackay A, Burford A, Carvalho D, Izquierdo E, Fazal-Salom J, Taylor KR,et al. Integrated molecular meta-analysis of 1,000 pediatric high-grade anddiffuse intrinsic pontine glioma. Cancer Cell 2017;32:520–37.

7. Filbin MG, Tirosh I, Hovestadt V, Shaw ML, Escalante LE, Mathewson ND,et al. Developmental and oncogenic programs in H3K27M gliomas dis-sected by single-cell RNA-seq. Science 2018;360:331–5.

8. Jones C, Baker SJ. Unique genetic and epigenetic mechanisms drivingpaediatric diffuse high-grade glioma. Nat Rev Cancer 2014;14. doi:10.1038/nrc3811.

9. Sturm D, Witt H, Hovestadt V, Khuong-Quang DA, Jones DT, KonermannC, et al. Hotspot mutations in H3F3A and IDH1 define distinct epigeneticand biological subgroups of glioblastoma. Cancer Cell 2012;22:425–37.

10. Lewis PW,MullerMM, KoletskyMS, Cordero F, Lin S, Banaszynski LA, et al.Inhibition of PRC2 activity by a gain-of-function H3 mutation found inpediatric glioblastoma. Science 2013;340:857–61.

11. Grasso CS, Tang Y, Truffaux N, Berlow NE, Liu L, Debily MA, et al.Functionally defined therapeutic targets in diffuse intrinsic pontine glioma.Nat Med 2015;21:555–9.

12. Piunti A, Hashizume R, Morgan MA, Bartom ET, Horbinski CM, Mar-shall SA, et al. Therapeutic targeting of polycomb and BET bromodo-main proteins in diffuse intrinsic pontine gliomas. Nat Med 2017;23:493–500.

13. Nagaraja S, Vitanza NA, Woo PJ, Taylor KR, Liu F, Zhang L, et al. Tran-scriptional dependencies in diffuse intrinsic pontine glioma. Cancer Cell2017;31:635–52.

14. Bitler BG, Aird KM, Garipov A, Li H, Amatangelo M, Kossenkov AV, et al.Synthetic lethality by targeting EZH2methyltransferase activity inARID1A-mutated cancers. Nat Med 2015;21:231–8.

15. Silveira AB, Kasper LH, Fan Y, Jin H, Wu G, Shaw TI, et al. H3.3 K27Mdepletion increases differentiation and extends latency of diffuse intrinsicpontine glioma growth in vivo. Acta Neuropathol 2019;137:637–55.

16. Larson JD, Kasper LH, PaughBS, JinH,WuG,KwonCH, et al.HistoneH3.3K27M accelerates spontaneous brainstem glioma and drives restrictedchanges in bivalent gene expression. Cancer Cell 2019;35:140–55.

17. Bhullar KS, LagaronNO,McGowan EM, Parmar I, Jha A, Hubbard BP, et al.Kinase-targeted cancer therapies: progress, challenges and future direc-tions. Mol Cancer 2018;17:48.

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

19. Vinci M, Burford A, Molinari V, Kessler K, Popov S, Clarke M, et al.Functional diversity and cooperativity between subclonal populations ofpediatric glioblastoma and diffuse intrinsic pontine glioma cells. Nat Med2018;24:1204–15.

20. Smith CJ, Berry DM, McGlade CJ. The E3 ubiquitin ligases RNF126 andRabring7 regulate endosomal sorting of the epidermal growth factorreceptor. J Cell Sci 2013;126:1366–80.

21. Agnihotri S, Wolf A, Munoz DM, Smith CJ, Gajadhar A, Restrepo A, et al.A GATA4-regulated tumor suppressor network represses formation ofmalignant human astrocytomas. J Exp Med 2011;208:689–702.

22. AmanchyR, PeriaswamyB,Mathivanan S, ReddyR, Tattikota SG, PandeyA.A curated compendium of phosphorylation motifs. Nat Biotechnol 2007;25:285–6.

23. Paugh BS, Qu C, Jones C, Liu Z, Adamowicz-Brice M, Zhang J, et al.Integrated molecular genetic profiling of pediatric high-grade gliomasreveals key differences with the adult disease. J Clin Oncol 2010;28:3061–8.

24. Berchtold NC, Cribbs DH, Coleman PD, Rogers J, Head E, Kim R, et al.Gene expression changes in the course of normal brain aging are sexuallydimorphic. Proc Natl Acad Sci U S A 2008;105:15605–10.

25. Fangusaro J. Pediatric high-grade gliomas and diffuse intrinsic pontinegliomas. J Child Neurol 2009;24:1409–17.

26. Khuong-Quang DA, Buczkowicz P, Rakopoulos P, Liu XY, Fontebasso AM,Bouffet E, et al. K27M mutation in histone H3.3 defines clinically andbiologically distinct subgroups of pediatric diffuse intrinsic pontine glio-mas. Acta Neuropathol 2012;124:439–47.

27. Hashizume R, Andor N, Ihara Y, Lerner R, Gan H, Chen X, et al. Pharma-cologic inhibition of histone demethylation as a therapy for pediatricbrainstem glioma. Nat Med 2014;20:1394–6.

28. Mohammad F,Weissmann S, Leblanc B, Pandey DP, Hojfeldt JW, Comet I,et al. EZH2 is a potential therapeutic target for H3K27M-mutant pediatricgliomas. Nat Med 2017;23:483–92.

29. Hayashi M, Kim SW, Imanaka-Yoshida K, Yoshida T, Abel ED, Eliceiri B,et al. Targeted deletion of BMK1/ERK5 in adult mice perturbs vascular

Koncar et al.

Cancer Res; 79(16) August 15, 2019 Cancer Research4040

on December 1, 2020. © 2019 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst June 14, 2019; DOI: 10.1158/0008-5472.CAN-18-3521

integrity and leads to endothelial failure. J Clin Invest 2004;113:1138–48.

30. Wang RM, Zhang QG, Li J, Yang LC, Yang F, Brann DW. The ERK5-MEF2Ctranscription factor pathway contributes to anti-apoptotic effect of cerebralischemia preconditioning in the hippocampalCA1 regionof rats. Brain Res2009;1255:32–41.

31. International Cancer Genome Consortium PedBrain Tumor Project.Recurrent MET fusion genes represent a drug target in pediatric glioblas-toma. Nat Med 2016;22:1314–20.

32. Harutyunyan AS, Krug B, Chen H, Papillon-Cavanagh S, ZeiniehM, De JayN, et al. H3K27M induces defective chromatin spread of PRC2-mediatedrepressive H3K27me2/me3 and is essential for glioma tumorigenesis.Nat Commun 2019;10:1262. doi: 10.1038/s41467-019-09140-x.

33. UmapathyG, ElWakil A,Witek B, Chesler L, Danielson L, Deng X, et al. Thekinase ALK stimulates the kinase ERK5 to promote the expression of theoncogene MYCN in neuroblastoma. Sci Signal 2014;7:ra102. doi:10.1126/scisignal.2005470.

34. Tusa I, Gagliardi S, Tubita A, Pandolfi S,UrsoC, Borgognoni L, et al. ERK5 isactivated by oncogenic BRAF and promotes melanoma growth. Oncogene2018;37:2601–14.

35. Pavan S, Meyer-Schaller N, Diepenbruck M, Kalathur RKR, Saxena M,Christofori G. A kinome-wide high-content siRNA screen identifiesMEK5-ERK5 signaling as critical for breast cancer cell EMT and metastasis.Oncogene 2018;37:4197–213.

36. English JM, Pearson G, Baer R, Cobb MH. Identification of substrates andregulators of the mitogen-activated protein kinase ERK5 using chimericprotein kinases. J Biol Chem 1998;273:3854–60.

37. Vaseva AV, Blake DR, Gilbert TSK, Ng S, Hostetter G, Azam SH,et al. KRAS suppression-induced degradation of MYC is antagonizedby a MEK5-ERK5 compensatory mechanism. Cancer Cell 2018;34:807–22.

38. Townson JL, Chambers AF. Dormancy of solitary metastatic cells.Cell Cycle 2006;5:1744–50.

39. Yeh AC, Ramaswamy S. Mechanisms of Cancer Cell Dormancy–AnotherHallmark of Cancer? Cancer Res 2015;75:5014–22.

40. Gupta PB, Fillmore CM, Jiang G, Shapira SD, Tao K, Kuperwasser C, et al.Stochastic state transitions give rise to phenotypic equilibrium in popula-tions of cancer cells. Cell 2011;146:633–44.

41. Song H, Jin X, Lin J. Stat3 upregulates MEK5 expression in human breastcancer cells. Oncogene 2004;23:8301–9.

www.aacrjournals.org Cancer Res; 79(16) August 15, 2019 4041

RAS Pathway Vulnerabilities in Histone-Mutant DIPG

on December 1, 2020. © 2019 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst June 14, 2019; DOI: 10.1158/0008-5472.CAN-18-3521

2019;79:4026-4041. Published OnlineFirst June 14, 2019.Cancer Res   Robert F. Koncar, Brittany R. Dey, Ann-Catherine J. Stanton, et al.   Diffuse Intrinsic Pontine GliomasIdentification of Novel RAS Signaling Therapeutic Vulnerabilities in

  Updated version

  10.1158/0008-5472.CAN-18-3521doi:

Access the most recent version of this article at:

   

  Overview

Visual 

http://cancerres.aacrjournals.org/content/79/16/4026/F1.large.jpgA diagrammatic summary of the major findings and biological implications:

   

  Cited articles

  http://cancerres.aacrjournals.org/content/79/16/4026.full#ref-list-1

This article cites 40 articles, 9 of which you can access for free at:

   

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

  Subscriptions

Reprints and

  .pubs@aacr.org

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

  Permissions

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

.http://cancerres.aacrjournals.org/content/79/16/4026To request permission to re-use all or part of this article, use this link

on December 1, 2020. © 2019 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst June 14, 2019; DOI: 10.1158/0008-5472.CAN-18-3521