Nature Communications Research

ARTICLE

https://doi.org/10.1038/s41467-022-30396-3 OPEN

InhibitionofmitochondrialcomplexIreverses

NOTCH1-drivenmetabolicreprogrammingin

T-cellacutelymphoblasticleukemia

NataliaBaran1,AlessiaLodi2,YogeshDhungana3,ShelleyHerbrich1,MeghanCollins2,ShannonSweeney2 , RenuPandey2,AnnaSkwarska1,ShraddhaPatel1,MathieuTremblay4,VinithaMaryKuruvilla1,AntonioCavazos1 , MecitKaplan5,MarcO.Warmoes6,DiogoTroggianVeiga7,KenFurudate1,8,ShantiRojas-Sutterin4 , AndreHaman4,YvesGareau4,AnneMarinier4,HelenMa1,KarineHarutyunyan1,MayDaher5 , LucianaMeloGarcia5,GheathAl-Atrash5,SujanPiya1,VivianRuvolo1,WentaoYang9 , SriramSaravananShanmugavelandy10,NingpingFeng11,JasonGay11,DiDu6,JunJ.Yang9,FiekeW.Hoff1 , MarcinKaminski12,KatarzynaTomczak13,R.EricDavis14,DanielHerranz15,AdolfoFerrando16, EliasJ.Jabbour1,M.EmiliaDiFrancesco17,DavidT.Teachey18,TerzahM.Horton19,StevenKornblau1, KatayounRezvani5,GuySauvageau4,MihaiGagea20,MichaelAndreeff1,KoichiTakahashi1, JosephR.Marszalek11,PhilipL.Lorenzi6,JiyangYu21,StefanoTiziani2,TrangHoang4,22& MarinaKonopleva1 ✉

T-cellacutelymphoblasticleukemia(T-ALL)iscommonlydrivenbyactivatingmutationsin NOTCH1 thatfacilitateglutamineoxidation.Hereweidentifyoxidativephosphorylation (OxPhos)asacriticalpathwayforleukemiacellsurvivalanddemonstrateadirectrelationshipbetween NOTCH1,elevatedOxPhosgeneexpression,andacquiredchemoresistance inpre-leukemicandleukemicmodels.DisruptingOxPhoswithIACS-010759,aninhibitorof mitochondrialcomplexI,causespotentgrowthinhibitionthroughinductionofmetabolic shut-downandredoximbalancein NOTCH1-mutatedandlesssoin NOTCH1-wtT-ALLcells. Mechanistically,inhibitionofOxPhosinducesametabolicreprogrammingintoglutaminolysis.WeshowthatpharmacologicalblockadeofOxPhoscombinedwithinducibleknockdownofglutaminase,thekeyglutamineenzyme,conferssyntheticlethalityinmiceharboring NOTCH1-mutatedT-ALL.Weleverageonthissyntheticlethalinteractiontodemonstratethat IACS-010759incombinationwithchemotherapycontainingL-asparaginase,anenzymethat uncoverstheglutaminedependencyofleukemiccells,causesreducedglutaminolysisand profoundtumorreductioninpre-clinicalmodelsofhumanT-ALL.Insummary,thismetabolic dependencyofT-ALLonOxPhosprovidesarationaltherapeutictarget.

Afulllistofauthoraffiliationsappearsattheendofthepaper.

T-cellacutelymphoblasticleukemia(T-ALL)isanaggressivetypeofleukemia,requiringintensivechemotherapy regimens1,2.ThebackboneofstandardofcarechemotherapyforT-ALLincludesvincristine,dexamethasone,and Lasparaginase(VXL)3,4.Despiteimprovedcureratesinpediatric T-ALL,relapseddiseaseoccursin40%ofadultprimaryT-ALL patientsandcarriesapoorprognosis1,2.Atthemolecularlevel, T-ALLisassociatedwithabnormalitiesinoncogenictranscriptionfactors5,6,withthemostfrequentactivatingmutationsin NOTCH1 thatleadtoupregulationofthepro-survival NOTCH1signaling7–9 andmutationsin FBXW7 gene,whichis similartomutationsinNOTCH1PESTdomain,resultsin increasedICN1proteinstability,oftenco-occurringwithmutationsin NOTCH110–12.NOTCH1activationsupportsdifferentiationofhematopoieticprogenitorstowardthymocytesand activatesmultipleanabolicpathwaysrequiredforcellgrowth, proliferation,andsurvivalofbothhealthyandleukemic T-cells6,7,13.Thus,targetingNOTCH1anditsdownstream effectorsisanattractivetherapeuticapproachinT-ALL14–16 However,thedevelopmentofNOTCH1inhibitorsremains challenging,duetoacquiredresistanceandpoortolerability,so targetingmetabolicpathwaysdownstreamofNOTCH1mayoffer analternativetherapeuticapproach.

Leukemiacellsdependonoxidativephosphorylation(OxPhos) tomeettheirenergeticandbiosyntheticdemands17–27.Phenformin,amitochondrialcomplexI/OxPhosinhibitor(OxPhos-i), reduceddiseaseburdeninpreclinicalT-ALLmodelsand improvedoverallsurvival(OS)25.IACS-010759,amorepotent andspecificsmall-moleculecomplexIinhibitor,hasbeeninvestigatedinclinicaltrialsofacutemyeloidleukemia(AML)and advancedcancers(NCT03291938,NCT02882321)28.GlutaminolysisisanothermetabolicfeatureofT-ALLandAMLthatis crucialforanaplerosis,biosyntheticprocesses,andcellularredox maintenance29,30.Basedonpromisingpreclinicalevidenceof antileukemicactivity,theglutaminaseinhibitor(GLS-i)CB-839is beinginvestigatedinongoingclinicaltrials31.Also,partofthe efficacyof L-asparaginaseagainstT-ALLreliesonitsGLS-i activity32

Wereporthereouruseofbioinformatics,genetic,andpharmacologicalapproachestoinvestigatethemetabolicfeaturesof T-ALL.WeidentifyOxPhosasatargetabledependencerequired formitochondrialenergyproduction,whosedisruptioneliminatesdiseaseusingpreleukemicandleukemiccelllines,patient samples,andpatient-derivedxenograft(PDX)models.Wealso establishamechanisticrationaleforcombiningOxPhos-iand GLS-i,byshowingsynergisticefficacyofIACS-010759and Lasparaginase-basedchemotherapyagainstchemoresistant NOTCH1-mutatedT-ALL.

Results

ActivityofOxPhosislinkedtoNotch1statusinmurinepreleukemicandhumanmodelsofT-ALL.Bioinformaticanalysis ofvariousdatasetsfromT-ALL12,33,normalthymocytes34,and murineT-ALLcelllines35 indicatedthatNOTCH1-boundtarget genes(NBTs)wereenrichedforMYCtargets,genesofthe OxPhospathway,thetricarboxylicacid(TCA)cycle,andthe mitochondrialelectrontransportchain(ETC)(Fig. 1a)35,36, possiblylinkingNOTCH1signalingtoMYCandOxPhos.During normalthymocytedifferentiation34,expressionofNBTsincreases fromtheETPstageonward,peakingattheDN3/DN4stage,and remainselevateduptotheimmaturesinglepositive(ISP)stage (Fig. 1b).BeginningattheDN3stage,thereisahighexpression ofNBTsthatareinvolvedinOxPhos,whichsharplydropatthe DPstage(Fig. 1c),togetherwiththeglobalshutdownofcanonical NOTCH1targetgenes.Moreover,followingtheintroductionof

theconstitutivelyactiveNOTCH1oncogene(Notch1/IC) intoDP cells37,genesetenrichmentanalysis(GSEA)showedupregulation ofNBTsexclusivelywithintheOxPhospathway,atboththe preleukemic(p <0.0001)andleukemic(p <0.0001)stages(SupplementaryFig.1a,b).

HavingestablishedapossiblelinkbetweenNOTCH1and OxPhosinmurinepreleukemicT-cells,wenextexamined whether NOTCH1 activationaffectsputativeOxPhosactivityin humanT-ALL,usingRNA-seqdatafromtwoT-ALLclinical trials12,33 andadata-drivennetworkinferencealgorithm (NetBID).Ascomparedtosampleswithout NOTCH1 mutation, therewasahigheractivityofOxPhossignaturegenesin NOTCH1-mutantT-ALL(p = 0.012)alongwithseveralpathways associatedwithmitochondrialmachinery(p = 0.01; Fig. 1d–f, SupplementaryFig.1c–e)andaloweractivityofapoptosis pathways(p = 0.0073; Fig. 1d,g;SupplementaryFig.1c,f).

Furthermore,thecomparisonofdependencescoresfrom DepMapunbiasedgenome-scaleCRISPR-Cas9screeningdata38 betweenthreeT-celllines(PF382,SUPT1,HSB2)toother hematologiccelllines(N = 73)orothercancercelllines (N = 686)showedsignificantlyhighernumberofOxPhosgenes inT-ALL.Inaddition,OxPhosgenesweresignificantlyoverrepresentedamongT-ALLdependentgenes,ascomparedto thoseforhematologicmalignanciesorothercancercelllines (p = 0.03 andp = 0.048,respectively;SupplementaryFig.1g) suggestingtheimportanceofOxPhosgenesforT-ALL survival38.Takentogether,bioinformaticanalysisprovidesevidencethatoncogenicactivationof NOTCH1 isassociatedwith OxPhos,andsuggestedthattreatmentwithOxPhos-icould reducetheviabilityofmurineandhumanT-ALL,potentially dependingon NOTCH1 status.

OxPhosdownstreamofNOTCH1isessentialforpreleukemic andleukemicstemcellfunction.Ourbioinformaticanalysis suggestedthattreatmentwithOxPhos-icouldreducetheviability ofmurineandhumanT-ALL,potentiallydependingon NOTCH1 status.

Inthemurine SCL-LMO1 transgenicmodel,progressionto acuteleukemiaisdrivenbytheacquisitionof Notch1 gain-offunctionmutations,startingat7–8weeksofage(Fig. 2a; SupplementaryFig.2a),reproducingamajorfeatureofthe humandisease.Indeed,introducingthe Notch1/IC oncogenethat driveselevatedandactivatedNOTCH1inthymocytes(Fig. 2b)is sufficienttotransform SCL-LMO1-inducedpre-LSCsinto hypercompetitiveleukemia-propagatingcellsandtotrigger aggressiveT-ALLwithoutlatency36,39.Todirectlyaddressthe questionwhetherNOTCH1activationaffectstheresponseto OxPhos-i,weharvestedpre-LSCsat1,3,and5weeksofage,i.e. atleastonemonthbeforetheappearanceof Notch1 mutations (SupplementaryFig.2a).ThefunctionalimportanceofNOTCH1 activationwasdirectlyaddressedbycomparingcellsexpressingor nottheconstitutivelyactive Notch1/IC allele.Whenmaintained onstromalcellspresentingtheNOTCH1ligandDL4(MS5-DL4), pre-LSCsfrom fivegeneticmodels: LMO1tg, SCLtgLMO1tg, Notch1tg, LMO1tgNotch1tg,and SCLtgLMO1tgNotch1tg,showed comparablegrowthinhibitionuponOxPhos-iwithIACS-010759, andcomparableIC50 indose–responsecurves(Fig. 2c,d; SupplementaryFig.2b).Ofnote,theinhibitoryeffectofIACS010759onpre-LSCswasobservedwhencellsstartedtoproliferate incultureonday2and3,butnotonday1(Fig. 2e).Inthe absenceofDL4stimulation, LMO1-or SCL-LMO1-inducedpreLSCssurvivedfor2dayswithoutproliferationandwere insensitivetoIACS-010759(Fig. 2d,f).Undertheseconditions, the Notch1 transgenewassufficienttoconferresponsivenessto IACS-010759,eventhoughcellproliferationwasminimal.

HALLMARKMYCTARGETS

HALLMARKOXPHOS

REACTOMETCACYCLE&RESP.ETC

REACTOMERESP.ETC&ATPSYNTH.

KEGGOXPHOS

NOTCH1-boundgenes FDRqval(-log10)

0102030

b ac defg

Fig.1OxidativephosphorylationdownstreamofNOTCH1isessentialforpreleukemicstemcellsfunction.a GenesetenrichmentanalysisofNOTCH1boundgenesinMolecularSignaturesDatabase.ThelistofgenesinwhichNOTCH1bindingwasobservedwithin2kboftranscriptionstartsitesinmurine T-ALLcelllineswasextractedasdescribedinMaterialsandMethods.Falsediscoveryrates(FDRs)areshownforoverlapscomputedwithHallmarkgene setsorcanonicalpathways(ReactomeandKyotoEncyclopediaofGenesandGenomes).TCA:tricarboxylicacid;ETC,electrontransportchain. b Dynamics ofNOTCH1targetgeneexpressionduringthymocytedifferentiation.ThenumbersofNOTCH1targetgenesthatincreased>1.3-foldateachtransition(gray bars,leftaxis).GeneexpressiondataarefromtheImmunologicalGenomeProject.GSEAwasconductedonthegenesetswhoseexpressionincreasedat eachtransitionalstage.FDRvaluesoverlapwiththeOxPhosgeneset(Hallmark)werecomputedasabove(greenbars,rightaxis).ETP,earlyT-lineage precursor;DN3a,(CD4CD8CD25+CD44CD28),DN3b,(CD4CD8CD25+CD44CD28+). c HeatmapofexpressionofNOTCH1-boundgenes withintheOxPhospathway.Ultralowinput(ULI)RNA-seqdatawereobtainedfromIMMGEN.ETP,earlythymocyteprogenitor(LinlowCD25CD44+Kit+), DN2,doublenegative2(LinlowCD25+CD44+Kit+),DN3(CD4CD8CD25+CD44),DN4(CD4CD8CD25CD44CD28+),ISP,immaturesingle positive(CD4CD8+CD24hiTCRlo),DP(CD4+CD8+TCRloCD24hi),T4(CD4+CD8TCRhi),T8(CD4CD8+TCRhi). d GSEAoftheTARGETgeneset (TherapeuticallyApplicableResearchtoGenerateEffectiveTreatments)(N = 265),indicatedenrichmentofgenesrelatedtoglutaminemetabolism, mitochondrialmetabolismaswellastranslationanddownregulationofapoptosis-relatedgenesinpatientswith NOTCH1-mutations(eachverticalbarin xaxisisgenerankinthepathwaylistand y-axisrepresentsrunningenrichmentscore). e TheOxPhosgenesignatureinT-ALLpatientsfromTARGETcohort. f EnrichmentofmitochondrialtranslationgenesinT-ALLpatientsfromTARGETcohort. g Enrichmentofapoptosis-relatedgenesinT-ALLpatientsfrom theTARGETcohort;for(e), (f)and(g): two-sided t-test;Thecenterlinerepresentsthemedianandwhiskersrepresentsmaximum(Q3 + 1.5*IQR)and minimumvalue(Q1+1.5*IQR).

GrowthinhibitionwasnotduetoconcurrentapoptosisinpreLSCsnordecreasedviabilityofthesupportingstromalcells (SupplementaryFig.2c–f),indicatingthatOxPhox-ispecifically actsonpre-LSCs.Althoughcellsexpressingthe Notch1/IC allele wereinhibitedbyIACS-010759,thecurvesdidnotreachthe samelevelsofmaximuminhibitionobservedinDL4-stimulated cultures(Fig. 2d,SupplementaryFig.2b).Wethencompared drugsensitivitiesbycomputingtheareasunderthecurves (AUC).AsshowninFig. 2g,drugsensitivitywashighestinDL4stimulatedculturesandall fivegenotypesclustertogether.Inthe absenceofDL4exposure,pre-LSCsseparatedintotwoclusters: LMO1-and SCL-LMO1-inducedpre-LSCsclusterwithcontrol

Pre-leukemic -ALL

SCLtgLMO1tg

Notch1 status wt mut Wild

culturesexposedtoDMSOwhereasallthree Notch1/IC genotypes clustertogether,awayfromDMSOcontrols.Nonetheless,drug sensitivityremainedlowercomparedtoDL4-stimulatedcells.

TogetherwithNOTCH1-occupancyofgenesoftheOxPhos pathway35 (Fig. 1a),theresponsetoOxPhos-iobservedhere suggeststwolevelsofdrugsensitivity:onedirectlyregulatedvia NOTCH1bindingtoOxPhosgenes,andtheotheroneassociated

withDL4-inducedcellproliferation.Together,thegeneticmodels andtheMS5-DL4co-culturemodelestablishadirectlink betweenNOTCH1activationandOxPhos-isensitivity.

Next,weaskedwhether NOTCH1 statusimpactsmitochondrialrespirationinhumanmodelsofT-ALL.Oxygenconsumptionrate(OCR)measurementsinT-lymphocytesandT-ALLcell linesseparatedby NOTCH1 mutationstatusshowedanincreased

Fig.2NOTCH1activationaffectstheresponsetoOxPhos-inhibitionofprimarypre-LSCs.a Notch1 genemutationin SCLtgLMO1tgpreleukemicand leukemicthymocytes. b OverexpressionofNOTCH1proteinby flowcytometry.Thymocytesfromwt-and NOTCH1tgmicewerestainedwithanti-NOTCH1 antibodyoranisotypecontrol. c dose–responseofprimarypre-LSCstoIACS-010759,co-culturedonMS5-DL4stromalcellsfollowedby48hdrug treatment(mean±SD, n = 3independentexperimentsintriplicates). d Dose–responseofpre-LSCstoIACS-010759treatment,co-culturedon MS5stromalcellsfollowedby48hdrugtreatment(mean±SD, n = 3.independentexperimentsintriplicates). e EffectsofIACS-010759treatment (133nMorDMSO)onviabilityofpre-LSCsculturedonMS5-DL4(mean±SD, n = 3independentexperiments). f EffectsofIACS-010759treatment (133nMorDMSO)onviabilityofpre-LSCsculturedonMS5(mean±SD, n = 3independentexperiments). g Pre-LSCsensitivitytoIACS-010759forthe indicatedgenotypes.TheAreaunderthecurves(AUC)werecomputedfromdose–responsedataillustratedin(c).ShownistheAUCdifferenceobtained betweenDMSO(blackring)anddrug-treatedcells(p <0.00001,multipleunpaired t tests, n = 3micepergenotype,intriplicates);with(green)and withoutDL4(blue),inpre-LSCswithorwithoutNOTCH1oncogeneintheabsenceofDL4. h Basaloxygenconsumptionrate(OCR)inT-lymphocytes (n = 7), NOTCH1-wt(n = 3)and-mutantcelllines(n = 7)byMitoStressTestassay(mean±SD, n = 3independentmeasurementsforeachline,4 replicates);one-wayANOVA;*p <0.05;***p <0.001. i IC50forbasalOCRinhibition(SupplementaryFigs.3cand4)forT-lymphocytes(n = 4)and NOTCH1-wt(n = 3)and-mutatedT-ALLcelllines(n = 7)(mean±SD,n = 3independentexperimentsforeachline,4replicates);one-wayANOVA; *p <0.05;***p <0.001. j IC50forviabilityinhibitionforT-lymphocytes(n = 5), NOTCH1-wt(n = 3)and-mutantcelllines(n = 7)treatedwithIACS-010759 (0–123nM,96h);(mean±SD, n = 3independentexperiments);one-wayANOVA;ns-nosignificance,**p <0.005. k ViablecellnumberinT-lymphocytes (n = 7), NOTCH1-wt(n = 3)and-mutant(n = 7)T-ALLcelllines,treatedwithIACS-010759(1,10,100nM,96h),(mean±SD, n = 3independent experimentsperline,3replicates);two-wayANOVA:ns-nosignificance,*p <0.05;**p <0.005;***p <0.001;and****p <0.0001. l ROS(MFI)treatedwith IACS-010759(1,10,100nM;96h)inT-lymphocytes(n = 6),T-ALLcelllineswithwt-(n = 3)andmutant NOTCH1(n = 7);(mean±SD, n = 3independent experimentsperline,3replicates);two-wayANOVA:ns-nosignificance,*p <0.05;**p <0.005;***p <0.001;and****p <0.0001.

OCRinasubsetofcellswith NOTCH1 wildtype(wt)(p = 0.03; Fig. 2h)andinallcelllinesharboring NOTCH1 mutation (p = 0.0004; Fig. 2h)comparedwithOCRofhealthy T-lymphocytes.Albeitthe NOTCH-mutantcellsonaverage showedhigherbasalandmaximalOCR(SupplementaryFig.3a), thesedifferencesdidnotreachstatisticalsignificance.AllT-ALL celllinesshowedhighexpressionofmitochondrialcomplexes I–V,withnoclearcorrelationbetweenmitochondrialprotein assemblyleveland NOTCH1 status(SupplementaryFig.3b).

Asexpected,IACS-010759inhibitedOxPhosacrossT-ALLcell lines(n = 11),reducingbasalandmaximalOCRinboth NOTCH1-wtand-mutantcelllines(p = 0.05and p = 0.005, respectively,SupplementaryFigs.3c–d,4),withthelattershowing greatersensitivitytoOCRinhibitionwhencomparedto T-lymphocytes(p = 0.0006, Fig. 2i,SupplementaryFig.3e). Consistentwiththis,IACS-010759producedstrikingdosedependentinhibitionofATPproductioninT-ALLlines (SupplementaryFig.3f),withthelowestIC50 in NOTCH1mutantlinesversusT-lymphocytes(p = 0.0006, Fig. 2j),while onlymoderateATPandviabilityreductionoflessthan20%atthe highestdosewasseeninhealthyT-lymphocytes(Supplementary Fig.3f–h).TheresponsetoIACS-010759observedinmatured T-lymphocyteswasnotincreasedbystimulationwithDLL4. NeitherATPproduction,cellnumber,norapoptosisratewere significantlychangeduponcombinedDLL4stimulationand IACS-010759treatment(SupplementaryFig.3i–k).ReducedATP productionuponOxPhos-iwasparalleledbydecreasedviability inT-ALLcelllines,PDXmodelsandprimaryT-ALLsamples (SupplementaryFig.3g,h).Attheclinicallyachievableconcentrationof10nM,IACS-010759reducedviabilitysubstantiallyin both NOTCH1-wt(p = 0.01)and NOTCH1-mutatedcells (p <0.0001,Fig. 2k),andminimallyaffectedtheviabilityof healthyT-lymphocytes(Fig. 2k).Similareffectswereseenat 100nMdoseofIACS-010759,producingsignificantreductionin both NOTCH1-wt(p = 0.0093) andmutants(p <0.0001,Fig. 2k).

OxPhos-iin NOTCH1-mutatedcellsalsoproducedtimedependentchangesintotallevelsandphosphorylationstatusof multiplemetabolism-relevantproteins(SupplementaryFig.5g), includingactivationofmTORandglycolyticenzymes,activation ofLKB1andAMPKinresponsetoATPdepletion,and upregulationofDNAdamage(pH2AX)inresponsetoenergetic stressandROSaccumulation.GSEAofOxPhos-i-induced transcriptionalchangesrevealeddownregulationofmitochondrialgenesetsinvolvingtranscriptionalprocesses,theETC,and cytochromecrelease(p = 0.046;SupplementaryFig.5h).

Insummary,ourdataprovidedirectevidencethatliganddependentactivationofNOTCH1signalingbyDLL4orbythe NOTCH1 oncogeneupregulatestheOxPhospathwayandconfers responsivenessofprimarypre-LSCsandhumanT-ALLcellsto OxPhos-i.

OxPhos-Iinducesdistinctmetabolicreprogrammingin NOTCH1-mutatedT-ALLtowardreductivemetabolismof glutamine.Toinvestigatefurthertheroleof NOTCH1 inregulatingmetabolisminT-ALLcells,weanalyzedtheimpacton cellularprocessesofsupplementationorwithdrawalofessential nutrientsglutamine(Gln),glucose(Glc),andpyruvate(Pyr). SupplementationwithGlnalone,butnotwithGlc,contributed to50%ofATPgenerationby NOTCH1-mutantcelllines (p = 0.0064;Fig. 3a,SupplementaryFig.6a).TheadditionofPyr withGlnfurtherimprovedATPgenerationto80%(p = 0.001) andincreasedbasalandmaximalOCRtothelevelobservedin completemedium(Gln + Glc+Pyr; p = 0.001;Fig. 3b,SupplementaryFig.6b),buthadnoimpactonextracellularacidification rates(ECAR).

ThepresenceofGlnandPyrwithoutGlcin NOTCH1-mutant cellscontributedtoadistinctaerobicmetabolicphenotypewith predominantOxPhosactivity,asindicatedbythebasalOCR/ ECARratio,whiletheadditionofGlcclusteredthemetabolic profiletowardanenergetic/glycolyticphenotype(Fig. 3c).In contrast, NOTCH1-wtcellsshowedtheirmostefficientATP productionandgrowthunderGlcsupplementation(p = 0.001; SupplementaryFig.6c).Theseobservationsrevealeddivergent

TheseresultscorrelatedwithbasalOCRchanges(Supplementary Fig.5a, R2 = 0.4572, p = 0.0079).Ofnote,IACS-010759increased significantlythelevelofreactiveoxygenspecies(ROS),where NOTCH1-mutationcontributedto8-foldhigherand NOTCH1wt2-foldhigherROSincreasewhencomparedtoT-lymphocytes (p = 0.0012), leadingtohigherROSaccumulationat10nMand 100nMdoses (p = 0.05and0.0036)andindicatingadistinct propensitytoundergoredoximbalanceuponOxPhos-i(Fig. 2l). ROSincreasecorrelatedwithviabilityreduction(R2 = 0.559, p = 0.0021),indicatingthatT-ALLcellshaveineffectivemechanismsofcopingwithoxidativestress(SupplementaryFig.5b). DLL4incubationdidnotproducesignificantchangesin respirationofmaturedT-lymphocytes(SupplementaryFig.5c±f), suggestingthatDLL4stimulationofNOTCH1playsarolemainly attheearlystagesoflymphocytedevelopment.

nutrientutilization,with NOTCH1-wtcellsrelyingstronglyon Glc,while NOTCH1-mutantsweredependentonGlntoactivate OxPhosandgenerateATP.

ToinvestigatewhetherT-ALLcellsswitchtoglycolysistocope withOxPhos-i,wemeasuredtheeffectsofIACS-010759onbasal andmaximalECAR,inthepresenceofGln,beforeandafterGlc supplementation(Fig. 3d;SupplementaryFig.7a,b).WithGln

only,both NOTCH1-wtand-mutantreliedonlyonOxPhos(high OCR,lowECAR)(Fig. 3e,f),andIACS-010759inhibitedGlndrivenOCRinadose-dependentmanner(SupplementaryFig.7).

AcuteGlcadditionproducedanimmediateshifttowardmaximal glycolyticactivity(p = 0.005, Fig. 3e,f),howeverthisswitch towardglycolysiswassignificantlyhigherfor NOTCH1-wtthan inmutant,therebyestablishinganewmetabolicequilibrium

Fig.3MitochondrialcomplexIinhibitioninducesdistinctmetabolicreprogrammingfollowedbyreductivemetabolismofglutaminein NOTCH1mutatedcelllines,uncoveringapotentialmetabolicmechanismofacquiredresistanceunderOxPhosblockade.a ATPlevelsfor NOTCH1-mutated T-ALL(n = 4)treated96hinmediawith:nonutrients,pyruvate(Pyr),glutamine(Gln),glucose(Glc),Gln+Pyr,Glc+Pyr,Glc+Gln,Glc+Pyr+Gln, (mean±SD, n = 4independentexperiments,3replicates);one-wayANOVA:*p <0.05;**p <0.005;***p <0.001;****p <0.0001. b BasalOCRof4 NOTCH1-mutatedT-ALLcelllinestreated24hasin(a),(mean±SD, n = 3independentexperiments,4replicates);one-wayANOVA:*p <0.05; **p <0.005;***p <0.001;****p <0.0001. c Metabolicphenotype(basalOCR/basalECARratio)asin(a);(mean±SD, n = 3independentexperiments,4 replicates). d Extracellularacidificationrate(ECAR)in NOTCH1-mutatedT-ALLcelllinePF-382afterIACS-010759treatment(0–123nM,4h);Glycolysis StressTest(mean±SD, n = 3independentexperiments,4replicates). e BasalECARof NOTCH1-wt(n = 3)and-mutated(n = 6)celllinestreated24hin mediawithGlnonly(blue)andafteracuteinjectionofGlc(black);(GlycolysisStressTest),(mean±SD, n = 3independentexperiments,4replicates); two-wayANOVA:***p = 0.0002. f BasalOCRin NOTCH1-wt(n = 3)and-mutated(n = 6)celllinestreatedasin(e);(mean±SD, n = 3independent experiments,4replicates);two-wayANOVA:*p = 0.0250. g BasalECARof NOTCH1-mutated(n = 5)and-wt(n = 3)celllinestreatedwithIACS-010759 (0–123nM,4h);(mean±SD, n = 3independentexperiments,4replicates);two-tailed t test:****p <0.0001. h Glycolyticcapacityin NOTCH1-mutated (n = 5)and-wt(n = 3)celllinestreatedwithIACS-010759(0–123nM,4h),GlycolysisStressTest;(mean±SD, n = 3independentexperiments,4 replicates);one-wayANOVA:****p <0.0001. i ATPproduction(%)for NOTCH1-mutant(n = 8)and-wt(n = 3)celllinestreatedwithIACS-010759 (0–123nM,4h),ATPRateAssay;(mean±SD, n = 3independentexperiments,4replicates);one-wayANOVA:****p <0.0001. j Levelsofcitrateand αketoglutarateandcitrate/α-ketoglutarateratioovertimein NOTCH1-mutatedcelllinePF382;(mean±SD, n = 1,4replicates);two-wayANOVA: ****p <0.0001. k Stableisotope-resolvedmetabolomicsultra-high-pressureliquidchromatographyMSanalysis(SIRM)of NOTCH1-mutantPF-382cellline culturedwith13C5,15N2-glutamine,treatedwithDMSOorIACS-010759(10nM,12h);(mean±SD, n = 1,4replicates).two-tailed t-test;*p <0.05; **p <0.005;***p <0.001;and****p <0.0001.

(highOCR,highECAR)(Fig. 3e,f).Thiswasaccompaniedby significantdecreaseofOCRobserveduponGlcinjection, indicatingswitchtowardglycolysiswithreducedOxPhos utilizationin NOTCH1-mutantsandthusmoretightlyregulated nutrientutilization.Furthermore,independentlyofOxPhos activity, NOTCH1-wtcelllinesdisplayedhigherglycolysis (p = 0.0001,Fig. 3g)andglycolyticcapacity,showingnearly 3-foldhigherECARthan NOTCH1-mutantsatbothbaselineand uponOxPhos-i(p = 0.0001, Fig. 3h).

Thesimultaneousreal-timemeasurementofATP(Seahorse) generatedbyglycolysisandOxPhos,inthepresenceofbothGln andGlc,showedacomparableinhibitionofOxPhos-derivedATP byIACS-010759inallcelllines.However,glycolysis-derivedATP generationwaslowerin NOTCH1 mutants(p <0.0001; Fig. 3i; SupplementaryFig.8a,b),confirmingthedistinctroleof NOTCH1-statusinthemaintenanceofmetabolicequilibrium uponOxPhos-i.

IACS-010759,likeotherComplexIinhibitors,hasbeenshown toimpairmitochondriabyblockingtheTCAcycle(thereby reducingconcentrationsofitsmetabolites:citrate, α-ketoglutarate [α-KG],succinate,fumarate,andmalate)andbyinhibitingATP productionanddecreasingaspartatelevels(therebydepleting energyandblockingnucleotidebiosynthesis)17,19,21,25,40.To definethemetaboliceffectsofOxPhos-iin NOTCH1-mutated T-ALLcells,weperformedanuntargetedmetabolomicsanalysis thatshowedaccumulationofnucleosidemonophosphate(NMP) (p = 0.04)andredoximbalance,asindicatedbydiminishedratios (reductivetooxidized)ofglutathioneandcoenzymeQ10 (CoQ10)(SupplementaryFig.9a).Also,OxPhos-iat10nM decreasedlevelsofcitrate,succinate, L-glutamine,and L-aspartate, slightlyincreasedfumarateandmalate,andtransientlyincreased α-KG(p <0.0001; SupplementaryFig.9b,c).Sincetheelevated αKG/citrateratioindicatesutilizationofreductiveOxPhos41,42,we nextanalyzedthedynamicsof α-KG/citrateratiochangesupon OxPhos-i.Indeed,aspikeinthe α-KG/citrateratiowasobserved after12h(p <0.0001; Fig. 3j; SupplementaryFig.9c),indicatinga switchtoreductiveOxPhosmetabolismuponOxPhosblockade. Tofurthersupportreductiveoxidativephosphorylationasa mechanismofcounteractingOxPhosblockade,wenextperformedstableisotope-resolvedmetabolomics(SIRM)tofollow thefateofincorporationofisotopically-enriched 13C5,15N2glutamine(13C5,15N2-Gln)or 13C6-glucose(13C6-Glc)in NOTCH1-mutant(Fig. 3k;SupplementaryFig.10)and NOTCH1-wtcelllines(SupplementaryFig.11a,b).Asexpected,

OxPhos-idecreasedlevelsofTCAintermediatesin NOTCH1mutantcells(Fig. 3k).WealsoobservedsimultaneousincorporationofGlcanditsutilizationinglycolysis,indicatedbya moderatelyincreasedrateoflactateenrichment(Supplementary Fig.10,11a).Furthermore,theincreaseof 13C515N2-Gln isotopomerfractions(M+5citrate,M+4malate,andM+4 fumarate)indicatedanenrichmentofGlnincorporation(Fig. 3k), supportingacompensatoryGlninfluxintotheTCAcycleupon OxPhosblockade.

Takentogether,theseresultsdemonstrateanincreaseinGln oxidativephosphorylationin NOTCH1-mutatedcelllinesand selectiveboostingtoutilizeGlninthereductiveoxidative phosphorylationpathwayuponOxPhos-i,offeringapossible mechanismofmetabolicescapein NOTCH1-mutatedT-ALL.

Blockingtheglutaminepathwayislethalto NOTCH1-mutant T-ALLcellswhencombinedwithcomplexIinhibition.Because reductivemetabolismofGlnwasincreasedbyOxPhos-i,we examinedfurthertheimportanceofGlnmetabolisminT-ALL. GSEAofTARGETT-ALLRNA-seqdatasetsshowedthatgenes involvedinGlnmetabolismareexpressedmorehighlyby NOTCH1-mutanttumors(p = 0.0067;Fig. 4a).Theseobservationspromptedthehypothesisthatthedualinhibitionofboth OxPhosandGlnmetabolismwouldfacilitatetheefficacyof OxPhos-iagainstT-ALL.WetestedthishypothesisbydeterminingthesensitivitytoIACS-010759ofcelllinesandprimary T-ALLsamples,undervaryingconditionsofselectivesupplementationofessentialsubstrates(Gln,Glc,andPyr). NOTCH1mutantcelllinesshowedhighersensitivitytoIACS-010759inthe presenceofGln(bluecurve)andglutamine+pyruvate(orange curve),withaverageIC50sof0.677nMand0.552nM(p <0.05 and p <0.0005respectively,Fig. 4b-c;SupplementaryFig.12a–c). TheseresultsindicatethatGlnandPyr,intheabsenceofGlc, supportOxPhosandsensitize NOTCH1-mutatedT-ALLcellsto OxPhos-i.Incontrast,Gln+Pyrsupplementationreducedthe sensitivitytoOxPhos-iof NOTCH1-wtcells(Supplementary Fig.12a,b).In NOTCH1-mutatedT-ALL,thegrowth-inhibitory effectofOxPhos-iinGln/Pyr-supplementedmediumwas reducedbytheadditionofGlc(blackcurve)(Fig. 4b,c; SupplementaryFig.12a–c).ExposuretoGlc,aloneorinthe presenceofGln,switched NOTCH1-mutatedcellsfromOxPhos towardsglycolysis(browncurve), flatteningthedose–response curveuponOxPhos-i,indicatingreducedsensitivitytoIACS010759;however,thisincreasedthesensitivityof NOTCH1-wtcells

• CONTROL

• IACS-010759

• CB839

• IACS-010759/CB839

toOxPhos-i(p <0.0001;Fig. 4b,c;SupplementaryFig.12a–c). TheseresultssuggestthatthemaintenanceofOxPhosfunction dependsonGlnin NOTCH1-mutatedcellsandthatGlniscrucial forthesensitivitytoOxPhos-i(p = 0.0001,Fig. 4c;Supplementary Fig.12a,b).

WenextreassessedtheimpactofOxPhos-ionT-ALLcellsin thecontextofinhibitingGLS,akeyenzymeinGlnutilization.

ComparedtotreatmentwithIACS-010759alone, GLS knockdowncombinedwithOxPhos-iinhibitedcellgrowth,reduced cellviability(SupplementaryFig.13a),reducedbasaland maximalOCR(Fig. 4d;SupplementaryFig.13b),andresulted ininhibitionofthec-MYC,Akt/mTOR,andAMPKpathways, initiationoftheDNAdamageresponsepathway(γ-H2AX),and activationoftheapoptosispathway(cleavedPARPandcleaved

NATURECOMMUNICATIONS|https://doi.org/10.1038/s41467-022-30396-3ARTICLE

Fig.4BlockadeofglutaminepathwaythroughglutaminaseknockdownandglutaminaseinhibitionincombinationwithcomplexIinhibitioninduces metabolicshutdowninT-ALLcellsinvitro.a EnrichmentofglutaminemetabolismcomponentsinT-ALLTARGETcohort;(GeneOntologyanalysis);twosided t test;Thecenterlinerepresentsthemedianandwhiskersrepresentsmaximum(Q3+1.5*IQR)andminimumvalue(Q1+1.5*IQR). b Viabilityof NOTCH1-mutatedT-ALLcelllines(96h,0–100nMofIACS01759)uponfollowingconditions:nonutrients,pyruvate(Pyr),glutamine(Gln),glutamine +pyruvate(Glut+Pyr),glucose+pyruvate(Glc+Pyr),glucose+glutamine+pyruvate(Glc+Glut+Pyr),normalizedtoDMSO. c AUCuponIACS-010759 treatment(byCTG)calculatedfrom(b),normalizedtoDMSOfromGln+Glc+Pyr,for NOTCH1-wt(blue)(n = 3)and NOTCH1-mutated(black)T-ALLcell lines(n = 4)(mean±SD, n = 3independentexperiments, n = 3replicates/condition).Two-wayANOVA;*p = 0.02;***p = 0.0002;****p <0.0001; d RepresentativeOCRresponsein NOTCH1-mutatedT-ALLcelllinesJURKATsubjectedtodoxycycline-inducedknockdownofGLS(blue)andtreatment withIACS-010759(10nM)for4h(red),orthecombination(green)(mean±SD, n = 3biologicalreplicates, n = 4replicates/condition). e Cellviabilityin NOTCH1-mutatedT-ALLpatientsamples(n = 4)treatedwithCB-839(1 μM)andIACS-01759(10nM)after96h,byFL(mean±SD, n = 4independent patientsamples, n = 3replicates/condition);one-wayANOVA;*p = 0.039;**p = 0.0017;****p <0.0001. f Reactiveoxygenspecies(ROS)-MFIofin NOTCH1-mutantT-ALLcelllines(n = 3)treated96hwithCB-839(1 μM)andIACS-010759(10nM).(mean±SD, n = 3)byFL;one-wayANOVA;*p = 0.0202;***p = 0.0005;****p <0.0001. g BasalOCRin NOTCH1-mutatedT-ALLcelllinesaftertreatmentwithvehicle,IACS-010759(10nM,4h),CB-839 (1 μM,12h),orIACS-010759/CB-839combination(mean±SD, n = 7celllines, n = 3independentexperiments, n = 4replicatespercondition);one-way ANOVA:***p = 0.0003;****p <0.0001. h SIRMultra-HPLCMSof NOTCH1-mutatedT-ALLcelllinePF-382afterenrichmentwith13C5 15N2-glutamineand treatmentwiththecombinationof10nMIACS-010759and1 μMCB-839orvehiclefor12h. i Geneexpressionanalysisin NOTCH1-mutantT-ALLafter treatmentwiththecombinationof10nMIACS-010759and1 μMCB-839comparedtovehicle;(x-generankinthepathwaylist;y-runningenrichment score).

caspase-3)(SupplementaryFig.13c).Furthermore,treatment withtheGLS-iCB-839incombinationwithOxPhos-iledtoa synergisticdropincellviability(SupplementaryFig.14a), reflectedbyanAUCdecrease(p = 0.001; SupplementaryFig.14b) andbydeltaBLISSindexvalues(SupplementaryFig.14c).The efficacyofdualinhibitionwasconfirmedacrosscelllines (SupplementaryFig.14d,e)andprimary NOTCH1-mutated T-ALLblasts(p <0.0001, Fig. 4e),bydecreasedcellnumberand increasedapoptosisratefortheGLS-i/OxPhos-icombination.

SupplementationwithPyrandGlnfailedtoreversethe inhibitoryeffectsofdualinhibition(SupplementaryFig.14f).

GiventhatGlncontributestotheredoxhomeostasis,wetested theimpactofthedualblockadeonROSinductionandshowed thatGLS-iplusOxPhos-ielevatedROSgreaterthanOxPhos-i alone(p <0.0001; Fig. 4f).Finally,theGLS-i/OxPhos-icombinationadditivelyreducedbasalandmaximalOCR,butonlyin NOTCH1- mutantcells(p <0.0001; Fig. 4g;Supplementary Figs.15a,b,16a,b).Ofnote,dualGLS-i/OxPhos-iblockade onlymoderatelyreducedthecellularrespiration(24h)andATP levels(72h)ofhealthybonemarrowcells(Supplementary Fig.16c,d)anddidnotsignificantlyaffectcolonyformation (SupplementaryFig.16e,f).

Theevidenceofsuccessfulblockadeofglutaminolysisby GLS-iuponOxPhos-iwascon fi rmedbySIRMinboth NOTCH1 -mutated(Fig. 4h;SupplementaryFigs.17,19a)and NOTCH1 -wtcelllines(SupplementaryFig.18a,b,19b).GLS-i/ OxPhos-iseverelyreducedthelevelsofTCAcycleintermediates citrate, α -KG,andsuccinate( p <0.0001),anddecreasedthe levelsoffumarate( p = 0.012),malate( p = 0.0003),aspartate ( p <0.0001),andglutamate( p <0.0001),accompaniedby relativeaccumulationofGln(Fig. 4 h,SupplementaryFigs.17, 18a,b,19a,b).Theseresults demonstrateaveryslowTCA turnoverindicativeofcellularexhaustion(Fig. 4 h;SupplementaryFig.17).Similareffectswereobservedfor NOTCH1 -wt cells,withoutsignifi cantaccumulationofGlninthelabeled fraction(SupplementaryFig.18a,b).Thispotentmetabolic shutdownwasre flectedbyGSEA,indicatingthedownregulationofgenesinvolvedintheTCAcycle,ETC,ATPresponse (Fig. 4 i;SupplementaryFig.20a,b),andmitochondrialactivity (SupplementaryFig.20a),accompaniedbyanupregulationof apoptosisactivationinresponsetoDNAdamageandcatabolic processes(Fig. 4 i;SupplementaryFig.20a,b).Giventhatredox balanceiscontrolledbyglutathioneconcentration 43, 44 ,we testedwhetherN-acetylcysteine(NAC)supplementationwould rescuetheeffectsofOxPhos-ialoneorcombinedwithGLS-ion

viability,apoptosis,andROS production.Surprisingly,cotreatmentwithNACat0.5or2.0mM 45 increasedthe cytotoxicityofIACS-010759orIACS-010759/CB-839combinationinT-ALLcells,decreasingcellnumber,inducingapoptosis, andtriggeringastrongeraccumulationofROS.Thesedata suggestthatNACsupplementationisunabletoquenchtheROS accumulationandglutathionedepletionuponOxPhos-i/GLS-i treatment,makingthosechangesirreversible.Onthecontrary, NACcompletelyrescuedthegrowth-inhibitoryeffectsofCB839alone(SupplementaryFig.21a– c).

Together,thesedatademonstratethatdualOxPhosand glutaminepathwayblockadeinducedprofoundmetabolicshutdowninT-ALLcellswith NOTCH1 mutations,inturnresulting inapoptosis.

BlockadeofglutaminemetabolismwithcomplexIinhibition improvesdepthoftumorburdenreductioninvivo.WeevaluateddualOxPhos/GLSinhibitioninvivo,usinga Notch1mutated GLSfl/fl murineT-ALLmodel29 inC57BL/6mice (Fig. 5a).Followingestablishedleukemiaengraftmentinperipheralbloodatday7(SupplementaryFig.22a,b),micereceived vehicleorIACS-010759therapy,aloneorwithconcomitant tamoxifenadministrationtoinduce GLS knockoutselectivelyin leukemiccells.Afteronly5daysofdualpharmacologic/genetic blockade,diseaseburdenwasreducedinbothperipheralblood (PB)(Fig. 5b)andbonemarrow(BM)(p = 0.0001; Fig. 5candf), withsubstantialclearanceofleukemiacellsinthespleen (p = 0.013;Fig. 5d),reductionofsplenomegaly(Fig. 5e)and clearanceofleukemiacellsintheliver(Fig. 5f).Consistentwith invitroobservations,metabolomicanalysisofPBfromleukemic micesubjectedtodualintervention,revealedanaccumulationof NMP,withdecreasedATPandTCAintermediates,indicativeof profoundmetabolicreprogrammingthatsuppressedallcritical aminoacidandnucleotidebiosynthesispathways(Ingenuity PathwayAnalysis,IPA)(Fig. 5g,SupplementaryFig.22a,b).The long-termevaluationfoundthatleukemiaprogressionwassignificantlydelayedinmicetreatedwitheitherIACS-01075or subjectedto GLS knockout(p = 0.0001;Fig. 5h).Thisantileukemiceffectwassignificantlyenhancedunderdualinhibition (p <0.0001;Fig. 5h),withnocirculatingleukemiacellsinperipheralblooduptoday220(Fig. 5i).Takentogether,ourresults demonstratethatblocking GLS incombinationwithcomplexI inhibitionprovidesapowerfulapproachtoabrogateleukemia progressioninvivo.

OxPhos-IaddedtoVxlchemotherapyinducesprominent metabolicchangesinvivoandimprovesoverallsurvivalin T-ALLPDXmodels.Since L-asparaginase,acomponentofthe VXLstandard-of-carechemotherapy,containsGLS-inhibiting activity,wenextfocusedonmetabolicconsequencesinducedby VXL-chemotherapyinT-ALL.First,wedeterminedtheeffect invitroonOxPhosactivityofeachVXLagentusedaloneor

combinedwithOxPhos-i.Dexamethasoneandvincristine increasedOCRsignificantly(p = 0.0001;SupplementaryFig.23a) andledtomoderatereductionsofcellviability(Supplementary Fig.23b,c). L-asparaginasesimilarlyinducedOCR(p = 0.0001; SupplementaryFig.23a)butdecreasedviabilitypotentlyas monotherapy(SupplementaryFig.23d).Whencombinedwith OxPhos-i,alldrugssynergizedtoreduceviability,withthemost

Fig.5ComplexIinhibitionincombinationwithglutaminasedeletionimpedes Notch1-mutantT-ALLleukemiadevelopment,improvestumorburden reduction,andextendsoverallsurvivalinamurinemodel.a SchematicofstudydesignutilizingmurineNotch1-mutatedGLS fl/fl b GFP+mCD45+ leukemiacells(%)inPBfrommicebearingmurine Notch1-mutatedGLS fl/fl T-ALLcellsafter5daysoftreatmentasshownin(a)(mean±SD, n = 3 individualmicepertreatmentarm);one-wayANOVA;**p = 0.0025. c GFP+mCD45+ leukemiacells(%)inBMfrommiceasin(b)(mean±SD, n = 3 individualmicepertreatmentarm);one-wayANOVA,**p = 0.001;and***p = 0.0005. d GFP+mCD45+ leukemiacells(%)inspleensfrommiceasin (b)(mean±SD, n = 3individualmicepertreatmentarm);one-wayANOVA,*p = 0.013. e Spleensharvestedfrommicebearingmurine Notch1-mutated GLS fl/fl T-ALLcellsfollowing5daysoftreatmentwithvehicle;IACS-010759;tamoxifen;orcombinationofIACS-010759andtamoxifen(n = 3). f H&E stainingofBM,spleenandliverfromonerepresentativemousefromeachgroup,selectedbasedoncomplementaryFLresultsfrom(c)and(d).Scalebars represent100 μm;fromthesameexperimentasFLdata. g HeatmapofMSanalysisofmetabolitesinPBofmicebearingmurine Notch1-mutatedGLS fl/fl T-ALLcellsafter5daysoftreatmentwithvehicle;IACS-010759;tamoxifen;orIACS-010759andtamoxifen;(meanlogoffold-changeratiooverthelevel ofmetabolitesmeasuredinmicetreatedwithvehicle);(mean±SD, n = 4individualmice/group). h Kaplan–Meiersurvivalcurvesofmicetransplantedwith murine Notch1-mutatedGLS fl/fl T-ALL.Mice(n = 5pertreatmentarm)weretreatedwithvehicle;tamoxifen(1mg/mouseover5days);IACS-010759 (5daysperweek5mg/kgMo–Fr)orconcomitantlywithIACS-010759andtamoxifen;treatmentwasinitiatedafterdetectionofGFP+ cellsinPBbyFLat day7;log-ranktest,***p <0.005,****p <0.0001. i WeeklytumorburdenmonitoringbydetectionofGFP+mCD45+ leukemiacellsinPBinmiceas indicatedin(h);thetimeframeoftreatmentislabeledbytheyellowwindowsonthegraph;(mean±SD, n = 5individualmicepertreatmentarm).

potenteffectobservedintheVXLcombination(Supplementary Fig.23e).Further,incomparisontoOxPhos-iorVXLalone,the VXL/OxPhos-icombinationpotentlyreducedbothbasaland maximalOCRonlyin NOTCH1 mutants(p <0.0001;SupplementaryFigs.24a,b;25a,b).Incontrast,theimpactofbasaland maximalOCRandECARinhealthybonemarrowcellswas moderateafterpre-treatmentwithOxPhos-i/VXLfor24h (SupplementaryFig.25c),didnotsignificantlyimpactATP productionafter72h(SupplementaryFig.25d)anddidnotaffect thecolony-formingcapacity(SupplementaryFig.25e).

VXL/OxPhos-i,likeGLS-i/OxPhos-i,ledtometabolicshutdownasmeasuredbySIRM(SupplementaryFigs.26a,b,27a,b), withexpectedchangesinTCAmetabolitescitrate, α-KG, succinate(p < 0.0001)andmoderatechangesinmalate(SupplementaryFig.28a,b).The L-AsparaginedrivenglutaminedepletingeffectswereobservedinmediaobtainedafterincubatingcellswithVXLandVXL/OxPhos-i,withareductionof glutamineleveldownto70%ofcontrolfor NOTCH1-mutated cells(p = 0.0004,SupplementaryFig.26c)and NOTCH1-wtcells (p = 0.02,SupplementaryFig.27c),respectively.Further,completedepletionofasparagine(p <0.0001),accumulationof glutamate(p <0.0001)andaspartate(p <0.0001)wereobserved inmediawithVXLtreatment(SupplementaryFigs.26c,27c). ThistranslatedintointracellulardepletionofAsparagine (p <0.0001)anddecreaseinAspartatelevelinVXL/OxPhos-i treatedcells(p = 0.0003;SupplementaryFig.28a,b).

SimilartoresultsobtainedforCB-839,weobservedrescueafter treatmentofJURKATandPF382cellswithNACof2mMonlyin VXLtreatment;inconditionscontainingOxPhos-i,NAC treatmentledtodeeperreductionofviablecellnumber (p <0.0001),increasedapoptosisrate(p <0.0001)andenhanced ROSaccumulation(p <0.0001)(SupplementaryFig.29a–c).

Next,weinvestigatedtheeffectofVXL/OxPhos-ionthe metabolicandtranscriptomicprofilesofT-ALLPDXswith advanceddisease.Thepharmacodynamicprofileofmicebearing 70-80%ofleukemiacellsinPBwasdeterminedafter12hof treatmentwithIACS-010759,VXL,orVXL/IACS-010759. Metabolomeanalysisrevealedprofounddownregulationof pathwaysrelatedtoOxPhosincellsderivedfromPB,spleen, andBMafterOxPhos-i(Fig. 6a;SupplementaryFigs.29d,30,31), facilitatedfurtherbyVXL.LeukemiacellsfromPB,BM,and spleendisplayedastrikingreductionincitricacid,isocitricacid, α-KG,succinate,fumarate,andmalate,aswellaslesserdecreases inaspartateandGlnafterVXL/OxPhos-icombinationtherapy (Fig. 6a;SupplementaryFigs.29d,30,31).Theseresultsindicate effectiveblockadeofglutaminolysisandinductionofredox imbalance,evidencedbylowlevelsofglutathioneandhomocysteine.VXL/OxPhos-ireducednucleotidetriphosphates(NTP),

leadingtoblockadeofDNAandRNAsynthesis(Fig. 6a; SupplementaryFigs.29d,30,31).PathwayanalysisbyIPA confirmedthatVXL/OxPhos-iupregulatedAMPKanddownregulatedthesirtuinpathway,CoQ10biosynthesis,andnucleotidebiosynthesis(SupplementaryFig.29e),asreflectedfurtherby OCRreduction(p <0.0001;SupplementaryFig.32a)anddownregulatedpathwaysshowedbyGSEA(SupplementaryFig.32b).

Theimpactofthesemetabolicchangeswasfurthervalidatedby single-cellCyTOFanalysis,whichconfirmedthereductioninPB leukemiacellsbyonly12hoftreatmentinvivo.Fractionsof T-ALLcellsexpressingCD45,CD7,andCD3werereducedand showedreducedexpressionofKi67andotherintracellular markerssuchasNOTCH1,p-p38,c-MYC,p-ERK,and p-H2AX(Fig. 6b,c).GSEAofglobalgeneexpressionchanges occurringinvivoafterVXL/OxPhos-ishowedaglobalshutdown ofallmitochondria-relatedprocesses,indicatingthatthis treatmentnegativelyimpactedawidespectrumofmetabolic andcellularprocessesinleukemiacells(Fig. 6d;Supplementary Fig.32b).

Finally,wedeterminedtheshort-term(5days)andlong-term effectsofVXL/OxPhos-iondiseaseprogressionandsurvivalin micebearinghumanT-ALL(SupplementaryFig.33a).Inallthree NOTCH1-mutatedT-ALLPDXmodels,OxPhos-i/VXLproduced aprofoundreductionofspleensize(SupplementaryFig.33b), spleenweight(p <0.0001;SupplementaryFig.33c),andtumor burdenreductioninPB(SupplementaryFig.33d),spleen(Fig. 6e; SupplementaryFig.33f),andBM(SupplementaryFig.33e,f). Diseaseprogression,asmeasuredbycirculatingleukemiaburden, was2–3-foldslowerwithVXL/OxPhos-i(Fig. 6f),translatinginto longeroverallsurvivalintheVXL/OxPhos-imicecohort (p <0.0001;Fig. 6g).Takentogether,ourdataindicatethat OxPhos-icombinedwithstandard-of-carechemotherapycontaining L-asparaginase,withitsGLS-inhibitoryactivity,effectively inducesmetabolicandtranscriptomiccellularcatastrophethat impedesleukemiaprogressionandleadstotheextensionof survival.

Discussion

Ourstudycharacterizedthemetabolicfeaturesof NOTCH1mutatedand NOTCH-wtT-ALLandelucidatedthelinkbetween NOTCH1 andmitochondrialactivation.Our findingssuggestthat targetingOxPhosinT-ALLisaneffectivewaytocontroldisease burdeninvivoandthatcompensatorymetabolicmechanisms uponOxPhos-icanbeovercomebyinhibitingreductivemetabolismoftheglutaminepathway.

OxPhosisawell-studiedtargetinmalignancies,withupregulatedOxPhosactivitybeingobservedinAMLandotherhematologicandsolidtumors.IACS-010759,apotentsmallmolecule

withselectiveinhibitoryeffectsonOxPhos,hasbeenstudiedin theseindications20,21,28,46.

WepreviouslyshowedthatT-pre-LSCsaremoreresistantto mainstaychemotherapyagentsthanthebulkofleukemiccells47 Wenowprovidedirectevidencethatpre-LSCsinaT-ALL murinemodelaresensitivetoOxPhos-ibyIACS-010759inthe presenceofinducedorconstitutiveNOTCH1signaling.We,

therefore,proposethatthesynergyofOxPhos-iandVXLchemotherapywilleffectivelytargetbothdrug-resistantpre-LSCs andbulk NOTCH1-dependentleukemiccells.Supportingthelink betweenmutationofthe NOTCH1 oncogeneandOxPhosactivity,geneexpressionanalysisfromtwoindependentpatient cohortsdemonstratedenrichmentofOxPhosrelatedgenesinthe transcriptomeof NOTCH1-mutatedT-ALL.Further,DepMap a

Fig.6VXLandIACS-010759induceprofoundmetabolicdamageinvivoandleadstoimprovementoftumorburdenreductionandoverallsurvivalin PDXmodelsinvivo.a MSHeatmapofmetabolitesinBMofmicetransplantedwithNOTCH1-mutatedPDX80after12hoftreatmentwithvehicle,IACS010759,VXL,orthecombinationofIACS-010759andVXL(n = 2). b CyTOFheatmapofspleensamplesfromPDX80modeltreatedwithvehicle,IACS010759,VXL,orthecombinationofIACS-010759andVXL(n = 2). c CyTOFUMAPforCD45,CD3,CD5,CD7andKi67markersinspleensamplesfrom PDX80modeltreatedwithvehicle,IACS-010759,VXL,orthecombinationofIACS-010759andVXL(n = 2). d Geneexpressionanalysisofspleensamples fromPDX80modelaftertreatmentwiththecombinationofIACS-010759andVXLcomparedtovehicletreatment. e Leukemicengraftment(%ofhuman CD45+ leukemiacells)inspleenatday5oftreatmentwithvehicle,VXL,IACS-010759,orthecombinationofIACS-010759andVXL,asmeasuredbyFL in3PDXmodels(fromlefttoright:PDXD115,PDXCU76,PDX80)(mean±SD, n = 4individualmicepertreatmentarm);one-wayANOVA;*p = 0.01; **p = 0.0016;***p = 0.0002,****p <0.0001. f Tumorburdendevelopmentin3PDXmodels(fromlefttoright:PDXD115,PDXCU76,PDX80)measured asweeklydetectionofcirculatinghumanCD45+ cellsinperipheralblood(PB)andexpressedaspercentofnormalizedhumantosumofhumanand murineCD45+ cellsinmiceundergoingtreatmentwithvehicle,IACS-010759(5mg/kgoncedaily,5dayson/2daysoff),VXL(onceweekly),orthe combinationofIACS-010759andVXL(mean±SD, n = 10micepertreatmentarmforPDXD115andPDXCU76and n = 8micepertreatmentarmforPDX 80). g Kaplan–MeiersurvivalcurvesofmicetransplantedwithPDXmodels(fromlefttoright:PDXD115,PDXCU76,PDX80)andtreatedwithvehicle, IACS-010759,VXL,orthecombinationofIACS-010759andVXL(n = 8).TreatmentwasinitiatedoncethelevelofcirculatingleukemiacellsinPBreached 0.5–1%;thetimeframeoftreatmentislabeledbytheyellowwindowsonthegraphs.log-ranktest:***p <0.0006****p <0.0001. NATURECOMMUNICATIONS|https://doi.org/10.1038/s41467-022-30396-3ARTICLE

dependencyanalysisandmetabolicanalysisbySeahorsetechnologyindicatedstrongerOxPhosdependencyandhigheroxygen consumptionratesin NOTCH1-mutatedcellscomparedtowt counterparts.Thisevidenceisinlinewithdatathatdemonstrated adependencyonETCuponphenforminandmitochondriatargetedantioxidantMVEtreatmentinT-ALLJurkatcells25,40

TargetingmitochondriainT-ALLissupportedbymechanistic studieswithafewagents(avicin,ABT-737,mitoTEMPOand oligomycin),butthesearchforcompoundseligiblefortranslationalstudiesisstillongoing.WeprovideevidencethatOxPhos dependencyin NOTCH1-mutatedT-ALLimpactsATPproductionandimpairsproliferation.Aspreviouslydemonstratedfor AML,OxPhos-iwithIACS-010759blockedmitochondrial activityinT-ALLandexhibitedpotentantileukemicactivityas monotherapyinvitro.

ItshouldbenotedthatIACS-010759alsoimpactedviability andreducedOCRof NOTCH-wtcellscomparedwithhealthy lymphocytes.Althoughnotstudiedhere,alternativemechanisms, inadditiontoNOTCH1genemutation,areknowntoactivate NOTCHsignalinginT-ALL10–12;andwhilemetabolicphenotypesdescribedinthisstudyaremostprominentin NOTCHmutatedcells,OxPhosclearlycontributestometabolic fitnessof NOTCH1 unmutatedcells.InlinewithdatareportedforIACS010759inAMLandforphenforminandMB1-4748 inT-ALL, comprehensivemetabolomicstudiesrevealedthatIACS-010759 causedanenergeticcrisisbydepletionofATPandNAD/NADH ratio,confirmingon-targeteffects,withblockadeofRNAand DNAsynthesisthroughdepletionofotherNTPs.Attheproteomicandtranscriptomiclevels,IACS-010759inducedAMPK, Akt,andDNAdamagepathways,withfurtheractivationofglycolyticenzymesinvitroandactivationofAMPKandDNA damageinvivo.WealsodemonstratedROSinduction,redox imbalanceandDNAdamageuponcomplexIinhibitioninT-ALL cells,yetthesedidnottranslateintoapoptosis,indicatingthe abilityofT-ALLcellstocopewiththemitochondrialblockade. Nevertheless,IACS-010759demonstratedantileukemicactivity invivodelayingaprogressionandsignificantlyextendingsurvival inamurinemodelof Notch1-mutatedT-ALLandhuman NOTCH1-mutatedPDXmodels.

Todissectmechanismsofmetaboliccompensation,we examinedtheglycolyticcapacityofT-ALLsamples.Ofnote,we observedthatwhileallT-ALLcel lsupregulateglycolysis,the levelofupregulationisstronglydependenton NOTCH1 -status, whereby NOTCH1 -mutatedcellsexhibitedalowercapacityto utilizeglycolysisuponOxPhosblockade.Instead,IACS-010759 causedastrongreductionofTCAintermediates α -KG,succinate,oxaloacetate,andcitrate,andamoderatereductionof fumarateandmalate.Inparallel,wefoundaprofoundchange

inthe α -KG/citrateratio,indicatingcompensatoryreprogrammingthroughrecruitmentofglutaminereductivemetabolism. Thiswassupportedbynutrientdepletionassays,togetherwith SeahorseandSIRM,whichindicatedthatleukemiccellsutilize glutaminolysisasasourceofenergytofueltheTCAcycleand recruitreductivemetabolismofglutaminetocopewithmetabolicstressunderOxPhos-i.Wedemonstratedthatactivationof reductiveglutaminemetabolismreducedtheef fi cacyofIACS010759asasingleagent,allowingcellstoescapethecomplexI blockade.These fi ndingsareconsistentwiththepreviously reportedresultsdemonstratingthatglutaminesustainstheTCA cycleinT-ALL29 , 49 ,andresemblesthereportedrelianceonGln metabolismobservedfollowingFLT3inhibitortreatmentin AML 50 – 53.This fi ndingwasfurthersupportedbytranscriptomicanalysesdemonstratingenrichmentofglutaminerelatedsignaturesin NOTCH1 -mutatedT-ALL.Notably, Herranzetal.identi fi edT-ALLasaglutaminolysis-dependent diseaseshowingthattheblockadeofglutaminemetabolismin T-ALLledtoautophagyinduction 29 ,whileinthestudyof Nguyenetal.glutaminedepletiontriggeredapoptosisinT-ALL cells49 .Altogether,ourworkprovidesstrongevidencethat T-ALLcellshaveauniquevulnerabilityrelatedtobothOxPhos andglutaminolysis.

Toexploitthismetabolicreprogrammingtherapeutically,we examineddualinhibitionofOxPhosandglutaminolysis.We demonstratedthatthegeneticsilencingofGLSincombination withcomplexIinhibitioncaneradicate NOTCH1-mutatedTALL,producingasimilarextensionofsurvivalasreporteddirect co-targetingNotchby γ-secretaseinhibitor29.Givenpotential challengesofthecombinatorialuseoftwonovelsmallmolecules inclinicaltrials,weutilizedamino-aciddepletionwith L-asparaginase,knowntohaveGLS-inhibitoryactivity.Inaggressive NOTCH1-mutatedhumanPDX-T-ALLmodels,the L-asparaginase-containingVXLregimen,incombinationwithIACS010759,ledtothemetabolicandtranscriptionalcollapseof T-ALLcellsresultinginasignificantreductionintumorburden andprolongedsurvival.Arecentreportdemonstratedenhanced combinatorialanti-tumoractivityofETCinhibitionwithdietary asparaginerestrictionwithasparaginase,throughdepletionof exogenousasparaginethatcommunicatesactiverespirationto ATF4andmTORC154.Whileitremainstobeseenifsimilar mechanismsareoperationalinleukemia,thesynergyobservedin ourstudycouldbebroadlyapplicabletotumorsbeyond NOTCH1-mutatedALL.

Insummary,our findingsprovideacomprehensiveunderstandingoftheroleofoxidativephosphorylationandmetabolic reprogramminginT-ALLsandofferafeasibletranslational strategyofOxPhosinhibitioninclinicaltrials.

Table1Panelofcelllinesusedinthestudy.

Cellline NOTCH1 statusGlucocorticoid sensitivity Gamma secretase inhibitor sensitivity

JURKATMutationResistantResistant

PF-382MutationResistantResistant 1301MutationSensitiveResistant TALL-1WildtypeResistantSensitive

LOUCYWildtypeResistantResistant P12ICHIKAWA MutationSensitiveResistant

MOLT-3MutationResistantResistant MOLT-4MutationResistantResistant CCRF-CEMMutationSensitiveResistant

KOPT-K1MutationResistantSensitive

SUP-T1WildtypeResistantSensitive

MOLT-16WildtypeSensitiveResistant

DND-41MutationSensitiveSensitive

ALL-SILMutationSensitiveSensitive

HPB-ALLMutationResistantSensitive

Methods

Celllines,primaryT-ALLsamples,andhealthyT-lymphocytes.TheT-ALLcell lines:JURKAT,PF-382,1301,TALL-1,LOUCY,P12-ICHIKAWA,MOLT-3, MOLT-4,CCRF-CEM,SUPT1andKOPT-K1wereobtainedinitiallyfromATCC weremaintainedataninternalcorefacilityofInstituteofScienceinCancer MDACCandauthenticatedbyshorttandemrepeatDNA fingerprintinginFebruary2016.MOLT-16(ACC29),DND-41(ACC525),ALL-SIL(ACC511),and HPB-ALL(ACC483),werepurchasedfromDSMZinMarch2019.MS5andMS5DL4stromacelllines:MousestromalMS5cellswerekindlyprovidedbyK.Mori (NagataUniversity,Japan);VSV-G–producingcellstransfectedwiththehuman DLL4cDNA(MHS4426-98361330,GEHealthcare)wereusedforgenetransductioninMS5cells(48h).AllusedleukemiacelllinesaresummarizedinTable 1 CelllinesweremaintainedifnototherwisespecifiedinRPMI-1640mediumwith 2mMGlutamine(SIGMA)containing10–15%fetalcalfserumand1%penicillinstreptomycin(LifeTechnologiesLaboratories).

BuffycoatswereobtainedfromhealthyblooddonorsprovidedbyMD AndersonBloodBank.Healthybonemarrow(BM)samples,BMandperipheral blood(PB)samplesfromT-ALLpatientswerecollectedduringroutinediagnostic procedures.Sampleswerecollectedafterwritteninformedconsentwasobtained fromallpatientsanddonorsandwasperformedundertheresearchprotocols PA13-1025(Targetingmetabolicpathwaysinleukemia,forALL),LAB02-295and LAB04-0249forhealthybonemarrowandbuffycoatsfromhealthyblooddonors, inaccordancewiththeUniversityofTexasMDAndersonCancerCenter InstitutionalReviewBoardregulationsundertheDeclarationofHelsinki principles.TheresearchprotocolswereapprovedbytherespectiveInstitutional EthicsCommittees.

MononuclearcellswereseparatedbyFicoll-Hypaque(Sigma-Aldrich)density gradientcentrifugation.CD3+ T-lymphocyteswerederivedfrombuffycoatsby CD3+ magneticbeadsisolation(StemCellTechnology).Primarysampleswere maintainedinQBSFmediumcontaining20%fetalcalfserum(GeminiBioProducts)and1%penicillin-streptomycin(LifeTechnologiesLaboratories).Patient characteristicsareshowninTable 2.Allcellsweregrownat37°Cinahumidified atmospherecontaining5%carbondioxide.D115T-ALLPDXcellswerekindly providedbyDr.PatrickZweidlerMcKey.PDXCU76,PDX12780,PDXCU178, PDX2230.andmurineT-ALLmodel(MDX)werekindlyprovidedbyDrAdolfo Ferrando.PDX4911and1921werekindlyprovidedbyDr.DanielLacorazza. PDX80wasgeneratedinourlabfromcellsderivedfromapatientsamplewiththe patient’swrittenconsent.ThecharacteristicsofthePDXsareshowninTable 3

Cellviabilityassay.ALL-SIL,CCRF-CEM,DND-41,HPB-ALL,JURKAT, LOUCY,MOLT-3,MOLT-4,SUP-T1,PF-382,andTALL-1cells,CD3+ T-lymphocytesderivedfromhealthydonorsorbonemarrowcellsderivedfrom healthydonorswereplatedatadensityof10,000cells/wellincompleteRPMI-1640 (SIGMA)supplementedwith10%fetalcalfserumandsubjectedtotreatmentwith IACS-010759(IACS,MDAndersonCancerCenter)atdosesrangingfrom0to 123nMfor96handCB-839atdosesrangingfrom0to30 μM,creatinga10 × 10 matrixoftreatment.CD3+ T-lymphocyteswereadditionallytestedonDLL4 coatedplatesdescribedindetailinDLL4assaysectionandtreatedwithIACS010759.Eachexperimentwasperformedindependently3–5times.Viablecell numbersweremeasuredbyquantifyingATPusingaCellTiter-GloLuminescent CellViabilityAssay(Promega).Dose-responsecurveswereanalyzedusingacurve fittingroutinebasedonnonlinearregressiontocomputetheIC50 value.Thesame

Table2Patientsamplesusedinthestudy.

UPINWBCs%Blasts NOTCH1 statusOthermutations

66617384.650MutationSUZ12,FBXWT, KRAS,WT1

666619029.595MutationCEBPA,NRAS

650667024.269MutationPTEN,SUZ12,TET2, U2AF1,WT1

675991410253MutationJAK3,FBXW7,PHF6

62418322.360MutationEZH2,WT1,NF1,IL7R

681881476.887MutationIL7R

681875231.573Wildtype

60704721.740MutationNRAS

60339662388WildtypeKRAS

40810303193MutationTP53

41864828139MutationTP53

612693210092.7WildtypeEZH,KRAS

UPIN uniquephysicianidentificationnumber, WBC whitebloodcell.

Table3PDXsamplesusedintheinvitroandinvivostudies.

rangeof10drugconcentrationswastestedforeachcelllinetocomputeacurvefreeareaunderthedose–responsecurve(AUC)basedonlinearinterpolationofthe 10datapointsusingabaselineof0(=100%inhibition)andamaximumof10055 TheAUCsgeneratedforeachcelllinewerefurthersummarizedbasedon NOTCH1 status.

Toevaluatetheimpactofglutamine,glucoseandpyruvateonATPproduction, thecellsweregrowinginRPMI1640mediawithoutglutamineorglucose purchasedfromMPBiomedicals(nonutrients),thatweresupplementedwith either2mMGln,Glcat5mMorPyronlyat1mM;orwithcombinationofthese nutrients.Datawerenormalizedtovaluesobtainedfromtriplesupplemented media(Glc+Gln+Pyr).Theresultsof fiveindependentexperimentswereanalyzed andpresentedasaverage±SD.

ToevaluatethesynergisticeffectsofIACS-010759withCB-839,vincristine, dexamethasoneand L-asparaginase,cellswereincubatedinRPMI1640(SIGMA) mediasupplementedwith2mMglutamine.TheMatrixof10×10wascreatedfor eachcombinationandevaluatedbyCTGafter96hofincubation.Forsynergy evaluationincombinationwithCB-839,IACS-010759wasusedintherangeof 0–123nM,vincristine(MPBiomedicals)wasusedintherangeof0–50ng/mL, dexamethasone(Sigma)intherangeof0–370nM,and L-asparaginasefromE.Coli (Sigma;Cat.NrA3809)wasusedintherangeof0–5IU/ml.Forsynergyevaluation incombinationwithVXL,IACS-010759wasusedintherangeof0–50nM, vincristinewasusedintherangeof0–5ng/mL,dexamethasoneintherangeof 0–50nM,and L-asparaginaseintherangeof0–0.5IU/ml.Theresultsof five independentexperimentswerecollected,processed,andvisualizedforsynergy evaluationusingCOMBENEFITsoftware56

Evaluationofcellnumbers,apoptosis,andROS.TheT-AllcelllinesALL-SIL, CCRF-CEM,DND-41,HPB-ALL,JURKAT,LOUCY,MOLT-3,MOLT-4,SUP-T1, PF-382,andTALL-1andCD3+ T-lymphocytesderivedfromhealthydonors,or bonemarrowcellsderivedfromhealthydonorswereplatedatadensityof0.2 × 106 cells/mLincompleteRPMI-1640supplementedwith10%fetalcalfserumand thentreatedwithDMSO(control);IACS-010759atconcentrationsof1nM,5nM, 10nM,30nM,or100nMfor96h;orCB-839at1 μMorVXLfor96h.Theeffect ofIACS-010759treatmentoncellnumberandapoptosiswasevaluatedbyAnnexin V(APC)-DAPIassay,after96hofexposureofthecellstothecompound.The effectofIACS-010759onROSlevelsinviablecellswasevaluatedbystainingcells withH2DCFDA.Briefly,after96hofdrugtreatment,cellswerecollectedand washedtwiceinAnnexinVbindingbufferandresuspendedin100 μLofAnnexin VbindingbufferwithAnnexinV-APCantibody(BDBioscience)ata

concentrationof1 μL/100 μLandH2DCFDAata1:1000dilutionandincubatedin thedarkatroomtemperature(RT)for30min.Cellswerethenwashedwith AnnexinVbindingbufferandresuspendedafterspinningin300 μLAnnexinV bindingbuffercontainingDAPIandcountingbeads(Invitrogen)at5 μL/300 μL. AllconditionswererunintriplicateandevaluatedwithKaluza flowcytometry(FL) softwareorFlowJosoftware.ThepercentageofAnnexinV-positiveeventswas normalizedtothatinDMSO-treatedcontrolsandpresentedasthepercentageof specificapoptosis,calculatedas:(%AnnexinV-positivecellsinthesample % AnnexinV-positivecellsinthecontrol)/(100 %AnnexinV-positivecellsinthe control) × 100.Thenumberoflivingcells(DAPI-negativecells)wasnormalizedto thatinDMSO-treatedcontrols.Themean fluorescenceintensityofH2DCFDA, indicatingROSlevel,wasevaluatedinDAPI-negative,AnnexinV-negativecells andnormalizedtothatofDMSO-treatedcells.

Niche-basedassayforpre-LSCs.Theniche-basedassaywasdesignedtosustain optimalpre-LSCsurvivalexvivobymimickingphysiologicalNOTCH1signal strengthandbyaddingKITandFLT3ligandsandinterleukin(IL)-7.Mouse MS5stromalcellsexpressingornotexpressingoptimallevelsofhumanDL4were seededin96-wellplatesinthecompleteT-cellmediumcontainingFLT3ligand (5ng/mL),KITligand(20ng/mL),andIL-7(5ng/mL)(R&DSystems).Thefollowingday,purifiedCD4 CD8 thymocytesfrom5–6-week-oldpreleukemic transgenicmice LMO1tg or SCLtgLMO1tg wereco-culturedonMS5orMS5-DL4 monolayersforanother24hpriortotheadditionofIACS-010759attheindicated concentrations.Pre-LSCviabilitywasdeterminedafter72hofdrugtreatment usingmultiparameter flowcytometry(Thy1+CD4 CD8 CD25+CD44 )withthe VivaFixcellviabilityassay(BioRad)andtheLSRII flowcytometerequippedwitha high-throughput-screeningmodule(BDBiosciences)57.DataacquisitionwasperformedwithaFACSCelesta flowcytometerequippedwithahigh-throughputscreeningmodule(BDBiosciences).(Thy1+CD4 CD8 CD25+CD44 )Pre-LSC viabilitywasnormalizedtothatofcontrolculturescontainingthesolventalone (DMSO)47

DLL4assay.HumanDLL4FcChimeraRecombinantProtein(ThermoFisher), wasdilutedinchilledphosphate-bufferedsaline(PBS)at10 μg/mLas recommended58,and50 μL/wellwascoatedinstandardtissue-culture96-well platesovernightat4°C.WellswerewashedoncewithPBSbeforeseedingcellsto removeanyunboundligandfromthewells.HealthydonorT-lymphocytesisolated fromPBMCsasdescribedin “Methods” werethensuspendedatthedensityof 30,000/wellandseededintheDLL4coatedplateorjustwashedwithsterilePBS platesandfurthersubmittedtotreatmentwithIACS-010759attheconcentration range0-123nM.

Thesamecoatingapproachscaledupto24wellformatwasusedfurtherfor viablecellnumber/apoptosisevaluationbyFLandforseahorseassay.Hereshortly T-lymphocyteswereseededincompleteRPMI1640eitheronDLL4coatedwellsor oncePBSwashedwellsandsubjectedtotreatmentwithIACS-010759atthe concentrationof10nMintriplicated(forFL).Theviabilitychangeswerethen evaluatedafter72hrofincubationbyFL.TheimpactofcombinedDLL4and IACS-010759oncellmetabolismwasevaluatedafter24hincubationbyseahorse assayasdescribedinthe “Method” sectionforSeahorseassay.

Colony-formingunitassay.Healthydonor-derivedfreshbonemarrowcellswere subjectedtotreatmentwith:DMSO,10nMIACS-010759,1uMCB839,VXLora combinationofIACS-010759/CB839orIACS-010759/VXLfor24h.Cellswere thencollectedandcounted,followedbypreparingthesuspensionof1×105 cellsin 400µlofcompleteRPMI-1640(SIGMA),supplementedwith10%FBSin15ml Falcontubes.4mlofMethoCult GFM3434(StemCellTechnologies;cat.3434) wastransferredtoeachFalcontubeusinga5mlsyringewith18g1½ gaugeneedle andsuspendedinthemethylcellulosebyvortexingfor1min.After15minofrest, 3mlofmethylcelluloseweredistributedbetween3bottom-grindedcultureplates withaimeddensityof2.5 × 104 cells/1mlMethylcellulose/well)usinga3mlsyringewith18g1½ gaugeneedle.Cellswereallowedtogrowfor12–14daysbefore coloniesweremanuallycounted.

AnalysisofOxPhosactivity,glycolysisactivity,andATPusingSeahorse technology.TocharacterizethemetabolicphenotypeofT-ALLcells, T-lymphocytesorbonemarrowcellsderivedfromhealthydonors,andtoevaluate theimpactofinhibitionofcomplexIonmitochondrialmetabolism,wemeasured OCRandECARinsetsofT-ALLcelllines,patientsamples,healthyT-lymphocytes,hBMcellsandPDXsamplesexvivousingtheSeahorseMitoStressTest, GlycolysisStressTest,andATPRateAssay(AgilentTechnologies).Eachofthe followingcelllines:ALL-SIL,CCRF-CEM,DND-41,HPB-ALL,JURKAT,LOUCY, MOLT-3,MOLT-4,SUP-T1,PF-382,andTALL-1,aswellasCD3+ T-lymphocytes andtreatedwithIACS-010759atdosesof0–123nMfor4hincompleteRPMI1640media,followedbytheappropriateSeahorsetestforthespeci ficscientific question.

Inbrief,cellswerewashed2timeswithphosphate-bufferedsaline(PBS)and resuspendedinprewarmed(37°C)SeahorseBasalMediumsupplementedwith 1mMpyruvate,2mMglutamine,and5mMglucose,pH7.4(forMitoStressTest); inprewarmedSeahorseBasalMediumsupplementedwith2mMglutamine,pH7.4

(forGlycolysisStressTest);orinbufferedRPMI-1640SeahorseBasalMedium supplementedwith1mMpyruvate,2mMglutamine,and5mMglucose(forATP Real-TimeRateAssay).Next,175 μLofcellssuspendedatadensityof2million/ mLwereplatedinatleast4replicateson96-wellSeahorseCellCultureplates.To allowcellstoattachtothebottomoftheplatesandcreatecellmonolayers,the plateswereprecoatedwithCell-Tak(Corning)accordingtothemanufacturer’ s instructions.Onceplated,cellsweresubjectedtogentlecentrifugationatRTfor 4min,thenat1500rpmwithoutabreak,andthenthecentrifugationwasrepeated intheoppositedirection.

OCRandECARweredeterminedusingaSeahorseXFe96analyzeraccordingto themanufacturer’sinstructions.TheOCRandECARvalueswereobtainedat baseline(3initialmeasurements)andaftertheinjectionsofSeahorseXFMito StressTestKitreagents:oligomycin(finalconcentration1.5 μM),FCCP(final concentration1.0mM),andantimycin/rotenone(finalconcentration0.5 μM); GlycolysisStressTestKitreagents:glucose(finalconcentration10mM), oligomycin(finalconcentration1.5 μM),and2-deoxy-D-glucose(final concentration50mM);orATPReal-TimeRateAssayKitreagents:oligomycin (finalconcentration1.5 μM)androtenone/antimycin(finalconcentration0.5 μM).

AllmeasurementswerecarriedoutbytheMitoStressTestGenerator,Glycolysis StressTestGenerator,orATPReal-TimeRateAssayGenerator(Agilent Technologies),asappropriate,andnormalizedtocellnumberaftertheassays.The datacollectedforbasalandmaximalOCRandECARafterexposuretotherangeof drugconcentrationswereplottedtocalculatetheIC50 values.AUCvalueswere generatedforeachexaminedcelllineandT-lymphocytesby NOTCH1-status. FurtherexaminationofOxPhosmodulationbyCB-839andVXLwascarriedout byMitoStressTestassay.

Analysisofgenome-widechromatinoccupancybyNOTCH1inpre-LSCs Genome-widechromatinoccupancybyNOTCH1wascomputedfromchromatin immunoprecipitation-DNAsequencingdatareportedbyWang etal.forthe murineT-ALLcelllinesG4A2andT6E35.Thegenelistwasfurtherrestrictedto thoseinwhichNOTCH1-boundpeakswerepresentwithin2kboftheirtranscriptionstartsitesaspreviouslydescribed36.TheexpressionofNOTCH1-bound geneswithintheOxPhospathwayduringnormalthymocytedifferentiationwas analyzedusingultra-lowinputRNA-seqdatafromtheImmunologicalGenome ProjectIMMGEN(http://www.immgen.org/)59 Notch1-inducedgeneexpressionin DPT-cellswasobtainedbyanalysisofmicroarraydatafrompurifiedmurine CD4+CD8+GFP+ cellsexpressingornotexpressingthe Notch1 oncogene(ICN, GSE12948)atthepreleukemic(2weeks)andleukemic(6weeks)stages(https:// www.ncbi.nlm.nih.gov/geo/geo2r)60.Analysisofdifferentiallyexpressedgenesfor GSEAandheatmaps(http://bioinformatics.sdstate.edu/idep/ , https://www. gseamsigdb.org/gsea/index.jsp)wasperformedasdescribedpreviously36

RNAvariantcallingpipeline.TodetectshortvariantsinRNA-seq,weranthe GATKRNA-seqvariantcallingpipeline(https://gatk.broadinstitute.org/hc/en-us/ articles/360035531192-RNAseq-short-variant-discovery-SNPs-Indels -).We obtainedFastq filesofRNA-seqdatafromthepublicT-ALLdataset (PRJNA572580;GEOAccession:GSE137768)33.Qualitycontrolanddatapreprocessingweredonewithfastp(version0.20.0)61.WeusedtheSTARaligner (version2.6.0b)intwo-passmodealignedtothereferencegenomeb37 (Homo_sapiens_assembly19.fasta)62.Duplicatereadsweremarkedandremoved byPicardMarkDuplicates(version2.9.0).GATK(version3.7)SplitNCigarReads wasusedtosplitandtrimreadstoremoveanyoverhangsintointrons.Basequality recalibrationwasperformedwithGATK(version4.1.0.0),BaseRecalibrator,and APPlyBQSR.Tocorrectlyhandlethesplicejunctions,theGATKHaplotypecaller wasusedtoperformvariantcallingwithaminimumphred-scaledconfidence thresholdof20.0.Weremovedknownsingle-nucleotidepolymorphismsusing dbSNP(version138).We filteredvariantcallqualitybasedonFisherstrandvalues greaterthan30.0andQualbyDepthvaluesbelow2.0withGATKVariantFiltration.Toannotategenevariants,weusedANNOVARandcombinedeach outputtable.

DepMAPCrispranalysis.Genome-scaleCRISPR-Cas9screenresultwasobtained attheDepMapPortal(https://depmap.org/portal/achilles)withalldata(rawread countstoprocessedgeneeffectscores)availableinthe19q4DepMapdataset (FigShare: https://figshare.com/articles/dataset/DepMap_19Q4_Public/11384241/ 3)63–65.TodeterminethegeneticdependenciesthatwereenrichedinT-ALLcell lines(HSB-2,PF-382,andSUP-T1),weutilizedlinear-modelanalysesfromthe limmaRpackage66 toperformatwo-tailedt-testforthedifferenceinthedistributionofgeneeffect(dependency)scoresinT-ALLcelllinescomparedtoother celllinesscreened.Statisticalsignificancewascalculatedasaq-valuederivedfrom thep-valuecorrectedformultiplehypothesistestingusingtheBenjamini& Hochbergmethod67

RNA-Seq.PF-382andSUP-T1cellswerecollectedforRNA-seqwhencellswere harvestedformassspectrometry(MS)analysis.CellsfromPDX80werecollected followingshorttreatmentstudieswithIACS-010759andVXLasdescribed.Total RNAwasextractedfrom4 × 106 cellsusingtheRNeasyMiniKit(Qiagen;cat. 74104).RNAconcentrationwasdeterminedbyaNanoDrop

2000spectrophotometer(ThermoFisherScientific),andRNAqualitywasassessed withanAgilentBioanalyzer.Approximately100millionreadspersamplewererun onanIlluminaNextseq500systemusing75 × 75pairedend(PE)readsfor strandedwhole-transcriptomesequencing.Rawsequencingreadswerepseudo alignedusingKallistov0.44.059with30bootstrapsamplestoatranscriptomeindex basedonthehumanGRCh38.92release(Ensembl).Abundancedatawerefurther analyzedwithSleuthv0.30.060usingmodelswithcovariatesforcondition.Genelevelabundanceestimateswerecalculatedasthesumoftranscriptspermillion (TPM)mappedtoagivengene.ThedatahavebeendepositedintheNCBI’sGene ExpressionOmnibus(GEO)database(GSE167305)(https://www.ncbi.nlm.nih. gov/geo/query/acc.cgi?acc=GSE167305).

NetBIDanalysis.TPM-normalizedcountmatricesfrom3celllines(PF-382,SUPT11,andPDX80),werelog2transformedand filteredforgenesthatwerenot detectedinallsamples(i.e.,rowsums>0).Toreducethenoiseintroducedby geneswithlowcounts,weusedaninterquartilerangemethodtorankthegenes andselectedthetop50%ofthehighlyvariablegenestorunNetBIDanalysison eachcelllineindependently.Thisresultedin~18Kgenesfordownstreamanalysis across39samplesfromthe2celllinesPF382andSUPT1and3singledrug-treated and2combination-treatedcelllines(Vehicle,IACS-010759,CB-839,VXL,IACS010759 + CB-839,andIACS-010759 + VXL)and2organs(spleenandBM)of PDXmodelT-ALLPDX80treatedwith(Vehicle,IACS-010759,VXL,andIACS010759 + VXL).Next,wemergedallthegenesetsfromtheMolecularSignatures Database(MsigDB)(http://www.gsea-msigdb.org/gsea/msigdb/)andcalculated pathwayactivityineachsampleusingthemeanexpressionlevelofgenesinthe genesets;wethenperformeddifferentialactivityanalysisatthepathwaylevel. Significantpathwayswereselectedat p <0.05or p <0.1.Wealsoappliedthe NetBIDalgorithm(https://github.com/jyyulab/NetBID )totheT-ALLRNA-seq datafromtheTARGETprojectandtheCOGAALL1231trialtocompare NOTCH1-mutantto-wildtypesamples.

MassspectrometryanalysisofcelllinesandPDXderivedcells.The NOTCH1mutatedT-ALLcelllinePF-382andthe NOTCH1-wtT-ALLcelllineSUP-T1were culturedinT75 flasksatadensityof1millioncells/mLandthentreatedwith DMSO,IACS-010759(10nM),CB-839(1 μM),IACS-010759 + CB-839,VXL (1ng/mlofvincristine,10nMdexamethasone,0.1IU/mlof L-asparaginase),or IACS-010759 + VXL,inRPMI-1640media(MPBiomedicals)supplementedwith 10%dialyzedfetalbovineserum(FBS;LifeTechnologies)withGlnandGlc;RPMI1640mediasupplementedwith10%dialyzedFBSwith 13C515N2-glutamine (CambridgeLaboratories)andunlabeledGlc;orRPMI-1640mediasupplemented with10%dialyzedFBSwithunlabeledGlnand 13C6-glucose(Cambridge Laboratories).The NOTCH1-mutatedcelllinePF-382wascollectedafter12hof treatment,andthewild-type NOTCH1 celllineSUP-T1wascollectedafter24h. Eachconditionwasrepeatedintriplicate.Aftertreatment,cellswerecentrifugedin 50mLFalcontubes,andcellpelletswerewashedtwiceincoldPBS.Cellswere recounted,andcellpelletswere flash-frozeninliquidnitrogenforfurtheranalysis. Forexvivometabolomicanalyses,spleenandBMcellscollectedafterorgan harvestingwere firstsubjectedtoengraftmentanalysisbyFL,followedbycounting and flashfreezinginliquidnitrogen.Cellpelletswerethensubjectedtomass spectrometry(MS)analysis.Themetaboliteswereextractedusingamodified Bligh-Dyerprocedurebyaddingwater,methanol,andchloroforminequal volumes(1:1:1).Theresultingsolutionwasvortexedvigorouslyandstirredat 1000g for10min,thencentrifugedat2119g for20minat4°C.Thepolarfraction wascollectedinEppendorftubes,evaporatedtodrynessinaCentriVap refrigeratedvacuumconcentrator(Labconco)at4°C,andresuspendedin180µL ultrapurewaterand20µLofinternalstandardsolution,aspreviously described68,69.The200-µLsolutionwas filteredthroughaNanosep3K ultracentrifugaldevice(PallCo.)at6010rpmfor20minat4°C70.Theresultant filtratewaspouredintoliquidchromatography(LC)vialandstoredat 20°Cuntil LC-MSanalysis.Metabolomicanalysisofpolarfractionswasperformedona Hybridquadrupole-Orbitrapmassspectrometer(QExactive,ThermoScientific) withaThermoScientificAccela1250ultra-high-performanceLCsystemwithan electrosprayionizationsource,simultaneouslyoperatinginpositive/negative polarity-switchingionizationmode.Metaboliteswerechromatographically separatedusingaKinetexC18150 × 2.1mm(2.6 μm,100Å)column (Phenomenex)withgradientelutionof0.2%formicacidinwater(A)and methanol(B)ata flowrateof150µLmin 1 within30min.Thegradientelution wasprogrammedasfollows:0–4min,2%B;4–14min,2–80%B;14–15min, 80–98%B;15–20min,98–98%B;20–25min,98–2%B,equilibrationtime5min. DetectionofmetaboliteswasperformedinfullMSmodeunderthefollowing conditions:sprayvoltage,4.0kV;capillarytemperature,300°C;sheathgas,50 (arbitraryunits);auxiliarygas,10(arbitraryunits);microscans,1;AGCtarget,3e6; maximuminjectiontime,200ms;massresolution,70,000full-widthhalfmaximum; m/z range,50–750.Toensuremassaccuracybelow5ppm,theMS detectorwascalibratedpriortoanalysisusingcommercialcalibrationsolutions. TheLC-MSplatformwascontrolledbyXCalibur2.2software(ThermoScientific). Aqualitycontrolsample,representingtheequivalentconcentrationofallsamples, waspreparedtocontrolpossibleinstrumentalerror(drift)indataacquisitionand wasrunonceevery fivesamples71.AllrawMSdatasetswereprocessedusingSieve 2.2(ThermoFisherScientific),andfeatureswithacoefficientofvariationlower

than25%inthequalitycontrolsampleswereconsideredforfurtheranalysis.Peaks werescaledaccordingtoprobabilisticquotientnormalization,andfeatureswere thenminedagainstanin-housedatabaseofaccuratemassesandretentiontimes generatedinourlaboratoryusingtheIROA300MSMetaboliteLibraryof Standards(IROATechnologies).Inaddition,databasesofaccuratemassestaken fromtheKyotoEncyclopediaofGenesandGenomesandtheHumanMetabolome databasewerealsomined72,73.RawdataaredepositedinZenodoopenaccess repository(https://doi.org/10.5281/zenodo.6442444).

Massspectrometryanalysisofperipheralblood.A10-µLwholebloodsample wascollectedthroughtheretroorbitalsinusofthemiceandimmediatelycombined with100µLextractionsolvent(ice-coldmethanol:water = 80:20with10µMd4citrateinternalstandard),vortex-mixedfor2min,andcentrifugedat20,000g for 10minat4°C.Thesupernatantwastransferredinto1.5-mLsnap-captubes,dried usingaSpeedvacconcentrator,andstoreduntilfurtheranalysis.Uponanalysis,the driedpelletswerereconstitutedin100 μLultra-purewaterandcentrifugedat 20,000 g for10minat4°C.Fromthesupernatants,90 μLwastransferredto polypropyleneautosamplervials.A10-μLvolumeofextractwasinjectedintothe ionchromatography(IC)-MSsystemwithoutdilution.ICgradientandMSsettings wereaspreviouslydescribed74.TheIC-MSplatformincludedaThermoScientific DionexICS-5000+ capillaryICsystemwithaThermoIonPacAS11250 × 2mm4 μmcolumn.DatawereacquiredusingaThermoOrbitrapFusionTribridMass Spectrometerunderelectrosprayionizationnegativeionizationmode.Theraw files werethenimportedintoThermoTraceFindersoftwaretoextractchromatographic peakareasfortargetsofinterest.Peakareaswerenormalizedusinginternal standardpeakareas.RawdataaredepositedinZenodoopenaccessrepository (https://doi.org/10.5281/zenodo.6442444).

Ingenuitypathwayanalysis.Thevaluesmeasuredforeachmetabolitewerecalculatedasexpressionlogratiosofthefollowing:GLS /GLS+,GLS+IACS-010759/ GLS+,andGLS IACS-010759/GLS+ anduploadedtoIngenuityPathwayAnalysissoftware(QIAGEN).Thecomparisonsselectedthemostsignificantlyaffected canonicalpathways.

siRNAtransfection NOTCH1-mutatedT-ALLcelllinesPF-382andJURKAT (1 × 105 cellsperwell)wereseededovernight.Negativecontrol(NT-siRNA),KAG orGACGLS-siRNAwastransfectedwithjetPRIMEtransfectionreagent,according tothemanufacturer’sinstructions(Polyplus).Thefollowingplasmidswereused: pHUSHpuroD3;pHUSHpuroKGAsh5;andpHUSHGACsh3(kindlyprovided byDr.GeorgiaHatzivassiliou,Genentech).Forfurtherstudiespuromycineselected cellswereused.ToinduceknockdownofGLS,doxycyclineattheconcentrationof 2ug/mlwasused.Thedetectionofknockdownwasevaluatedatday5after initiationoftreatmentwithdoxycycline.Thecellswithconfirmedknockdownwere subjectedtoevaluateviability(CTG),cellgrowth,Seahorseanalysisofmetabolic changeswithorwithouttreatmentwithOxPhosinhibitorattheconcentrationof 10nM,andtoimmunoblotting.

Immunoblotting.Cells(1–2 × 106 perwell)wereseededovernightin6-wellplates. Followingtreatmentforthespecifiedduration,totalproteinwasextractedin Laemmlibuffer,fractionatedbysodiumdodecylsulfate-polyacrylamidegelelectrophoresisandtransferredtopolyvinylidene fluoridemembranes(Millipore Sigma).AfterblockingwithLi-COROdysseyblockingbuffer,themembraneswere incubatedwithprimaryantibodiesovernightat4°C,washed,incubatedfor2h withinfrared fluorochrome-conjugatedsecondaryantibodies:OdysseyIrdye680 RDanti-mouseandOdysseyIrdye800CWgoatanti-rabbit,asappropriate.The proteinsignalwasvisualizedusingaLI-COROdysseyimagingsystem.Thepanel ofantibodiesusedforImmunoblottingissummarizedinSupplementaryTable1.

CYTOF.HumanT-ALLPDX80sampleswerecollectedfromspleens.Samples wereinitiallybarcodedandstainedwithmetal-conjugatedantibodiesaccordingto Fluidigm’sprotocolsasdescribedbelow.Briefly,allantibodieswerelabeledwith heavymetalsusingMaxpar-X8labelingreagentkits(FluidigmDVSSciences) accordingtothemanufacturer’sinstructionsandtitratedtodeterminetheoptimal concentration.Foreachmouseevaluated,3 × 106 cellswerealiquotedintoseparate FACStubesandwashedtwicewithMaxparPBSbuffer(Fluidigm;cat.201058).For live/deadcelldiscrimination,cellswereresuspendedin200µLof5µMcisplatin (Fluidigm;cat.201064)for1minonanorbitalshaker,followedimmediatelyby3 washesinMaxparcellstainingbuffer(CSB;Fluidigm;cat.201068).Cellswerethen subjectedfor10minto fixationin1mLof1× FixIbuffer(Fluidigm;cat.201065), followedby2washeswith1× barcodepermbuffer(Fluidigm;cat.201057).Each uniquepalladiumbarcode(Fluidigm;Cell-ID20-PlexPdBarcodingKit,cat. 201060)wassuspendedin100µLofbarcodepermbufferandimmediately transferredtocellsin800µLbarcodepermbuffer.Cellswereincubatedwith barcodesfor30min,thenwashedtwicewith2mLCSB.Barcodedcellswerethen combinedintoasingletubeforCyTOFstaining.Thestainingfactorwascalculated asthetotalnumberofbarcodedcells/3 × 106 cells.Cellswereblockedwith5 μL× stainingfactorofanti-humanFcreceptorbindinginhibitor(eBioscienceInvitrogen)in45µL × stainingfactorofCSBfor15minatRT.Anappropriateamountof

Table4PanelofantibodiesusedforCyTOFstaining.

surfaceantibodymastermix(Table 4)wasaddeddirectlytothetubeandincubated for1hatRTbefore2washeswithCSB.

Forintracellularstaining,cellsweredissociatedin1mLof fixation/ permeabilizationbufferandincubatedovernightat4°C,thenwashedtwicewith 2mLof1× permeabilizationbuffer.Thesupernatantfromthe finalwashwas discarded,andappropriateamountsofintracellularantibodymastermixwere addedtotheresidualpermeabilizationbufferandincubatedfor1hinthedarkat RT.Forcelldiscrimination,ametallointercalatorworkingsolution(2mL PermeabilizationBuffer,125nMiridiummetallointercalator[Fluidigm;cat. 201103A])wasaddedtothesampleandincubatedovernightat4°C.Cellswere washedwithaminimumof3mLCSB,thenagainwith3mLddH2Owith0.1% bovineserumalbumin.Finally,cellswere filteredthrougha35-µm filterand resuspendedin100µLddH2Owith0.1%bovineserumalbumin.TheMD AndersonFlowCytometryandCellularImagingCoreFacilitypreparedthe antibodycocktailsandacquireddataonaHeliosCyTOFmachine(DVSSciences). Datawere firstdemultiplexedbyFluidigmDebarcodersoftware.Individual masscytometrydata files(.fcs)werethen filteredusingFlowJotoremovethe normalizationbeads,debris,doublets,anddeadcells.Theremaininganalysiswas performedinR(version3.6.1,TheRFoundationforStatisticalComputing)using theRpackages ‘cytofkit’ and ‘flowcore ’75.Processeddataweresubjectedtonegative valueprunedinversehyperbolicsinetransformationandclusteredbasedonthe PhenoGraphalgorithm(k = 22)usingallcellsurfacemarkers76.Dimensionality reductionwasperformedusingtheuniformmanifoldapproximationand projection(UMAP)method77

Animalstudies.Forthestudiesofpre-LSCs,allmouselineswerebackcrossedonto aC57BL/6Jbackgroundformorethan12generationstoproduce pSTIL-SCL (SCLtg), Lck-LMO1 (LMO1tg),and Lck-NotchIC9 (Notch1tg)mice(NIAID/Taconic Repository).Mice(maleandfemale)weremaintainedseparatelyinpathogen-free conditionsaccordingtoinstitutionalanimalcareanduseguidelinessetbythe CanadianCouncilonAnimalCare.Atleast3micepergenotypewereusedfor experimentalwork.ForthestudiesofT-ALL,allexperimentalanimalprocedures wereapprovedbyMDAndersonCancerCenter’sInstitutionalAnimalCareand UseCommittee(IACUC).Thestudywascompliantwithallrelevantethicalregulationsregardinganimalresearch.AnimalstudieswereconductedatMD Anderson’sanimalfacilitiesinaccordancewiththeIACUCguidelines.Micewere maintainedinapathogen-freeenvironmentwithfreeaccesstofood.8-week-old femaleC57BL/6micewerepurchasedfromTheJacksonLaboratoryandallowedto acclimatizefor1weekbeforebeingirradiatedwithasublethaldoseof400cGyon thedaybeforeT-ALLcellinjection.ToexaminetheimpactofcomplexIinhibition inageneticmodelofglutaminaseknockout,weused NOTCH1-inducedT-ALL

murinecellswithtamoxifen-inducibleconditional Gls knockout(Rosa26Cre-ERT2/+Glsf/f), whichweregeneratedandkindlyprovidedbyDrs.DanielHerranzandAdolfoFerrando (ColumbiaUniversity).Atotalof1 × 106 cellssuspendedin100 μLsterilePBSwere injectedintothetailveinsofthefemaleC57BL/6mice29.Diseaseprogressionwas monitoredweeklybyretroorbitalblooddraws,andmeasurementsofGFP+ leukemic cellsinthePBwerenormalizedtothetotalamountofmurineCD45+ cells,after exclusionofDAPI+ (dead)cells.OncetheleukemiacellsweredetectableinPB(day7, FigureS22A,B),micewererandomizedinto2treatmentgroups(n = 10miceper group).OnegroupunderwentconditionalknockoutofGLSbyadministrationof1mg tamoxifenover5consecutivedays(GLS ),andtheotherwastreatedwithcornoil (GLS+).TomeasuretheadditiveorsynergisticeffectsofcomplexIinhibition,GLS and GLS+ miceweretreatedwitheithervehicleorIACS-010759(7.5mg/kg)(n = 5miceper arm)byoralgavageoncedailyonascheduleof5dayson/2daysoffstartingatDay7up toDay68.Diseaseprogressionandresponsetotreatmentweremeasuredbyweekly engraftmentchecks.Micewereeuthanizedwhenvisiblesignsofclinicalillnesswere observedorweightlossof>20%wasrecorded.Tomeasuretheshort-termimpactof complexIinhibitionandglutaminaseknockout,mice(n = 3pertreatmentarm) engraftedatalevelof5%ofcirculatingleukemiacellsinPBweresubjectedtothesame treatmentregimenfor5days,followedbymiceeuthanasiabycervicaldislocation,organ harvestingandmeasurementofcirculatingleukemiacellsinPBandthelevelofleukemic cellengraftmentinthespleen,BM,andliver.Tissuescollectedafterthetreatmentwere subjectedtohistopathologicalexamination,andbloodcollectedaftertreatmentwas furthersubjectedtoMSanalysis.

ToexaminetheimpactofcomplexIinhibitionincombinationwithstandardof-carechemotherapy,8-to10-week-oldfemaleNOD.Cg-PrkdcscidIl2rgtm1Wjl/ SzJ(NSG)micewerepurchasedfromTheJacksonLaboratory(cat#005557)and allowedtoacclimatizefor1week.Afterirradiationonthedaypriortoinjection, eachmousewasinjectedviathetailveinwith1.0 × 106T-ALLPDXcells suspendedin100 μLsterilePBS;thePDXmodelsusedwereCU76,PDXD115,and PDX80.Diseaseprogressionwasmonitoredweeklybyretroorbitalblooddraws, andFLwasusedtomonitorthepercentageofhumanCD45 + cellsnormalizedto thatofmurineCD45+ cellsafterexclusionofDAPI+ (dead)cells.Treatmentwas initiatedatdetectionof1%to5%circulatingleukemiacellsintheblood(n = 10 miceperarm).IACS-010759wasadministeredviaoralgavageoncedailyat5mg/ kg(5dayson/2daysoff).Standard-of-carechemotherapy(VXL)wasadministered intraperitonealonceperweek,8hafteradministrationofIACS-010759,for 4weeks.VXLwasusedatareduceddoseof:dexamethasone(5mg/kg),vincristine (0.15mg/kg);and L-asparagine(1000IU/kg,approx.25IU/mouse).E.coli-derived L-asparaginasewaspurchasedfromElspar,Lundbeck,Deerfield,IL,USA. Tomeasuretheshort-termimpactofcomplexIinhibitiononleukemia metabolismandVXLsensitivity,miceengraftedatalevelof30–50%circulating leukemiacellsinPBweresubjectedtothetreatmentregimenabovefor5days

(n = 4miceperPDtreatmentarm).12hafterthelastdoseofIACS-010759,mice wereeuthanizedbycervicaldislocation,followedbyharvestofPB,spleen,BMin ordertomeasurecirculatingleukemiacellsinPBandthelevelofleukemiccellsin spleenandBM.Tissuesofspleen,BMandlivercollectedaftertreatmentwere subjectedtohistopathologicalexamination,bloodcollectedaftertreatmentwas subjectedadditionallytoMSanalysis,andspleencellscollectedaftertreatment were flash-frozeninliquidnitrogenforMSanalysisandforRNAextraction followedbyRNA-seq,orfrozenin10%DMSO/90%fetalcalfserumforfurther CyTOFanalysis.

Histopathologyevaluation.Five-micron-thicksectionsoftissueswerecutusing aLeicaMicrosystemscryostat,transferredontoSuperfrost-Plusslides(Thermo FisherScienti fi c),andstainedwithhematoxyl inandeosin(H&E).BM,spleen, andlivertissuesampleswerefurtherevaluatedbypathologist,blindedtothe treatmentgroups,fordeterminingtheleukemicin fi ltrationoftissuesand morphologicchangesofhematopoiesis. Stainedtissueslideswereexamined microscopicallyusingOlympusBX41microscopeandbyscanningtheslides withAperioAT2scannerandviewingtheimageswithAperioImageScope software.

Statisticalanalysis.Unlessotherwiseindicated,allcalculationsandstatistical analyseswerecarriedoutusingGraphPadPrismsoftwarev.7or8.Figurelegends indicatespecificstatisticalanalysesusedanddefineerrorbars.Statisticallysignificantdifferencesbetween2groupswereassessedbyunpairedStudent’ s t tests. Ordinaryone-wayanalysisofvariance(ANOVA)wasusedtoanalyzemorethan2 groups.Two-wayANOVAwasusedtoanalyzecellproliferationatmultipletime points.Resultsareexpressedasmean±SD(asnotedinthe figurelegends)ofat least3independentexperiments,and p valueslessthan0.05wereconsidered statisticallysignificant.Survivalanalysisforinvivostudieswasdoneusing Mantel–CoxandGehan–Breslow–Wilcoxontests.

Reportingsummary.FurtherinformationonresearchdesignisavailableintheNature ResearchReportingSummarylinkedtothisarticle.

Dataavailability

Alldatasupportingthisstudyareprovidedinthemanuscriptormadepubliclyavailable. ChIP-seqdatapublishedpreviouslybyWangetal.wereusedforcomputationof genome-widechromatinoccupancybyNOTCH1andareavailableunderaccessioncode GSE29600

Ultra-lowinputRNA-seqdatafromtheImmunologicalGenomeProjectIMMGEN areavailableunderaccessioncodeGSE100738(http://www.immgen.org/).

MicroarraydatafrompurifiedmurineCD4+CD8+GFP+ cellsexpressingornot expressingtheNotch1oncogeneatthepreleukemic(2weeks)andleukemic(6weeks) stagesareavailableunderaccessioncodeGSE12948(https://www.ncbi.nlm.nih.gov/geo/ geo2r).

Genome-scaleCRISPR-Cas9screenresultwasobtainedattheDepMapPortal (33T33T https://depmap.org/portal/achilles33T)withalldata(rawreadcountsto processedgeneeffectscores)availableinthe19q4DepMapdataset(FigShare: https:// figshare.com/articles/dataset/DepMap_19Q4_Public/11384241/3).

RNA-seqdatageneratedinthisstudyhavebeendepositedintheNCBI’sGene ExpressionOmnibus(GEO)database(GSE167305).

ThegenesetsfromtheMolecularSignaturesDatabase(MsigDB)(http://www.gseamsigdb.org/gsea/msigdb/)weremergedandusedforpathwayactivitycalculationsusing NetBIDineachsampleusingthemeanexpressionlevelofgenesinthegenesets.

DatamaybeaccessedthroughtheTARGETwebsiteat https://ocg.cancer.gov/ programs/target.ThesequencingBAMandFASTQ filesfromTARGETRNA-seqare accessiblethroughthedatabaseofgenotypesandphenotypes(dbGaP; http://www.ncbi. nlm.nih.gov/gap)underaccessionnumberphs000218(TARGET)andsubstudyspecific accessionphs000464(TARGETALLExpansionPhase2).

Fastq filesofRNA-seqdatafromthepublicT-ALLdatasetfromthefrontline Children’sOncologyGroup(COG)T-ALLclinicaltrialAALL1231aredepositedunder accessioncode(PRJNA572580;GEOAccession: GSE137768)

MetabolomicsdataaredepositedinZenodoopenaccessrepository(https://doi.org/10. 5281/zenodo.6442444)

Received:6April2021;Accepted:25April2022; References

1.Giambra,V.etal.LeukemiastemcellsinT-ALLrequireactiveHif1alphaand Wntsignaling. Blood 125,3917–3927(2015).

2.Vadillo,E.,Dorantes-Acosta,E.,Pelayo,R.&Schnoor,M.Tcellacute lymphoblasticleukemia(T-ALL):Newinsightsintothecellularoriginsand infiltrationmechanismscommonanduniqueamonghematologic malignancies. BloodRev. 32,36–51(2018).

3.Kang,M.H.etal.Activityofvincristine,L-ASP,anddexamethasoneagainst acutelymphoblasticleukemiaisenhancedbytheBH3-mimeticABT-737 invitroandinvivo. Blood 110,2057–2066(2007).

4.Szymanska,B.etal.Pharmacokineticmodelingofaninductionregimenfor invivocombinedtestingofnoveldrugsagainstpediatricacutelymphoblastic leukemiaxenografts. PLoSONE 7,e33894(2012).

5.Ferrando,A.A.etal.Biallelictranscriptionalactivationofoncogenic transcriptionfactorsinT-cellacutelymphoblasticleukemia. Blood 103, 1909–1911(2004).

6.Belver,L.&Ferrando,A.ThegeneticsandmechanismsofTcellacute lymphoblasticleukaemia. Nat.Rev.Cancer 16,494–507(2016).

7.Sanchez-Martin,M.&Ferrando,A.TheNOTCH1-MYChighwaytoward T-cellacutelymphoblasticleukemia. Blood 129,1124–1133(2017).

8.Hu,Y.etal.DEPTORisadirectNOTCH1targetthatpromotescell proliferationandsurvivalinT-cellleukemia. Oncogene 36,1038–1047(2017).

9.Chiang,M.Y.etal.HighselectivepressureforNotch1mutationsthatinduce MycinT-cellacutelymphoblasticleukemia. Blood 128,2229–2240(2016).

10.King,B.etal.TheubiquitinligaseFBXW7modulatesleukemia-initiatingcell activitybyregulatingMYCstability. Cell 153,1552–1566(2013).

11.LarsonGedman,A.etal.TheimpactofNOTCH1,FBW7andPTEN mutationsonprognosisanddownstreamsignalinginpediatricT-cellacute lymphoblasticleukemia:areportfromtheChildren’sOncologyGroup. Leukemia 23,1417–1425(2009).

12.Liu,Y.etal.ThegenomiclandscapeofpediatricandyoungadultT-lineage acutelymphoblasticleukemia. Nat.Genet 49,1211–1218(2017).

13.Palomero,T.&Ferrando,A.TherapeutictargetingofNOTCH1signalingin T-cellacutelymphoblasticleukemia. Clin.LymphomaMyeloma 9(Suppl3), S205–210(2009).

14.Hounjet,J.etal.Theanti-malarialdrugchloroquinesensitizesoncogenic NOTCH1drivenhumanT-ALLtogamma-secretaseinhibition. Oncogene 38, 5457–5468(2019).

15.DeFord,C.etal.TheclerodanediterpenecasearinJinducesapoptosisof T-ALLcellsthroughSERCAinhibition,oxidativestress,andinterferencewith Notch1signaling. CellDeathDis. 7,e2070(2016).

16.Tatarek,J.etal.Notch1inhibitiontargetstheleukemia-initiatingcellsina Tal1/Lmo2mousemodelofT-ALL. Blood 118,1579–1590(2011).

17.Kuntz,E.M.etal.Targetingmitochondrialoxidativephosphorylation eradicatestherapy-resistantchronicmyeloidleukemiastemcells. Nat.Med. 23,1234–1240(2017).

18.Lagadinou,E.D.etal.BCL-2inhibitiontargetsoxidativephosphorylationand selectivelyeradicatesquiescenthumanleukemiastemcells. CellStemCell 12, 329–341(2013).

19.Baccelli,I.etal.MubritinibtargetstheelectrontransportchaincomplexIand revealsthelandscapeofOXPHOSdependencyinacutemyeloidleukemia. CancerCell 36,84–99e88(2019).

20.Vangapandu,H.V.etal.BiologicalandmetaboliceffectsofIACS-010759,an OxPhosinhibitor,onchroniclymphocyticleukemiacells. Oncotarget 9, 24980–24991(2018).

21.Molina,J.R.etal.Aninhibitorofoxidativephosphorylationexploitscancer vulnerability. Nat.Med. 24,1036–1046(2018).

22.Yehudai,D.etal.Thethymidinedideoxynucleosideanalog,alovudine, inhibitsthemitochondrialDNApolymerasegamma,impairsoxidative phosphorylationandpromotesmonocyticdifferentiationinacutemyeloid leukemia. Haematologica 104,963–972(2019).

23.DiFrancesco,M.E.etal.DiscoveryanddevelopmentofIACS-010759,anovel inhibitorofComplexIcurrentlyinphaseIstudiestoexploitoxidative phosphorylationdependencyinacutemyeloidleukemiaandsolidtumors. CancerRes. 78, https://doi.org/10.1158/1538-7445.Am2018-1655 (2018).

24.Pollyea,D.A.etal.Venetoclaxwithazacitidinedisruptsenergymetabolism andtargetsleukemiastemcellsinpatientswithacutemyeloidleukemia. Nat. Med. 24,1859–1866(2018).

25.Kong,H.etal.Metabolicdeterminantsofcellular fitnessdependenton mitochondrialreactiveoxygenspecies. Sci.Adv. 6, https://doi.org/10.1126/ sciadv.abb7272 (2020).

26.DeRosa,V.etal.CoordinatemodulationofglycolyticenzymesandOXPHOS byimatinibinBCR-ABLdrivenchronicmyelogenousleukemiacells. IntJMol Sci 20, https://doi.org/10.3390/ijms20133134 (2019).

27.Yuan,F.etal.InhibitionofmTORC1/P70S6Kpathwaybymetformin synergisticallysensitizesacutemyeloidleukemiatoAra-C. LifeSci. 243, 117276(2020).

28.Yap,T.A.etal.PhaseItrialofIACS-010759(IACS),apotent,selectiveinhibitorof complexIofthemitochondrialelectrontransportchain,inpatients(pts)with advancedsolidtumors. J.Clin.Oncol. 37,3014–3014(2019).

29.Herranz,D.etal.MetabolicreprogramminginducesresistancetoantiNOTCH1therapiesinTcellacutelymphoblasticleukemia. Nat.Med. 21, 1182–1189(2015).

30.Jacque,N.etal.Targetingglutaminolysishasantileukemicactivityinacute myeloidleukemiaandsynergizeswithBCL-2inhibition. Blood 126, 1346–1356(2015).

31.Konopleva,M.Y.etal.Phase1study:safetyandtolerabilityofincreasing dosesofCb-839,anorally-administeredsmallmoleculeinhibitorof glutaminase,inacuteleukemia. Haematologica 100,378–379(2015).

32.Chan,W.K.etal.GlutaminaseactivityofL-asparaginasecontributesto durablepreclinicalactivityagainstacutelymphoblasticleukemia. Mol.Cancer Ther. 18,1587–1592(2019).

33.Meyer,L.K.etal.Glucocorticoidsparadoxicallyfacilitatesteroidresistancein Tcellacutelymphoblasticleukemiasandthymocytes. J.Clin.Investig. 130, 863–876(2020).

34.Mingueneau,M.etal.ThetranscriptionallandscapeofalphabetaTcell differentiation. Nat.Immunol. 14,619–632(2013).

35.Wang,H.etal.Genome-wideanalysisrevealsconservedanddivergent featuresofNotch1/RBPJbindinginhumanandmurineT-lymphoblastic leukemiacells. Proc.NatlAcad.Sci.USA 108,14908–14913(2011).

36.Gerby,B.etal.SCL,LMO1andNotch1reprogramthymocytesintoselfrenewingcells. PLoSGenet. 10,e1004768(2014).

37.Li,X.,Gounari,F.,Protopopov,A.,Khazaie,K.&vonBoehmer,H. OncogenesisofT-ALLandnonmalignantconsequencesofoverexpressing intracellularNOTCH1. J.Exp.Med. 205,2851–2861(2008).

38.Meyers,R.M.etal.Computationalcorrectionofcopynumbereffectimproves specificityofCRISPR-Cas9essentialityscreensincancercells. Nat.Genet. 49, 1779–1784(2017).

39.Tremblay,M.etal.ModelingT-cellacutelymphoblasticleukemiainducedby theSCLandLMO1oncogenes. GenesDev. 24,1093–1105(2010).

40.Birsoy,K.etal.Anessentialroleofthemitochondrialelectrontransportchain incellproliferationistoenableaspartatesynthesis. Cell 162,540–551(2015).

41.Fendt,S.M.etal.Reductiveglutaminemetabolismisafunctionofthealphaketoglutaratetocitrateratioincells. Nat.Commun. 4,2236(2013).

42.Metallo,C.M.etal.ReductiveglutaminemetabolismbyIDH1mediates lipogenesisunderhypoxia. Nature 481,380–384(2011).

43.Holmstrom,K.M.&Finkel,T.Cellularmechanismsandphysiological consequencesofredox-dependentsignalling. Nat.Rev.Mol.CellBiol. 15, 411–421(2014).

44.Sayin,V.I.etal.Antioxidantsacceleratelungcancerprogressioninmice. Sci. Transl.Med. 6,221ra215(2014).

45.Mansour,M.R.etal.Targetingoncogenicinterleukin-7receptorsignalling withN-acetylcysteineinTcellacutelymphoblasticleukaemia. Br.J.Haematol. 168,230–238(2015).

46.Rytelewski,M.etal.Inhibitionofoxidativephosphorylationreversesbone marrowhypoxiavisualizedinimageablesyngeneicB-ALLmousemodel. FrontOncol. 10,991(2020).

47.Gerby,B.etal.High-throughputscreeninginniche-basedassayidentifies compoundstotargetpre-leukemicstemcells. J.Clin.Invest. 126,4569–4584 (2016).

48.daSilva-Diz,V.etal.Anovelandhighlyeffectivemitochondrialuncoupling druginT-cellleukemia. Blood 138,1317–1330(2021).

49.Nguyen,T.L.etal.Downregulationofglutaminesynthetase,not glutaminolysis,isresponsibleforglutamineaddictioninNotch1-drivenacute lymphoblasticleukemia. Mol.Oncol. 15,1412–1431(2021).

50.Gregory,M.A.etal.Targetingglutaminemetabolismandredoxstatefor leukemiatherapy. Clin.CancerRes. 25,4079–4090(2019).

51.Gallipoli,P.etal.GlutaminolysisisametabolicdependencyinFLT3(ITD) acutemyeloidleukemiaunmaskedbyFLT3tyrosinekinaseinhibition. Blood 131,1639–1653(2018).

52.Lu,X.etal.ThecombinedtreatmentwiththeFLT3-inhibitorAC220andthe complexIinhibitorIACS-010759synergisticallydepletesWt-andFLT3mutatedacutemyeloidleukemiacells. FrontOncol. 11,686765(2021).

53.ZavorkaThomas,M.E.etal.Gilteritinibinhibitsglutamineuptakeand utilizationinFLT3-ITD-positiveAML. Mol.CancerTher. 20,2207–2217 (2021).

54.Krall,A.S.etal.AsparaginecouplesmitochondrialrespirationtoATF4 activityandtumorgrowth. CellMetabolism, https://doi.org/10.1016/j.cmet. 2021.02.001 (2021).

55.Kurtz,S.E.etal.Molecularlytargeteddrugcombinationsdemonstrate selectiveeffectivenessformyeloid-andlymphoid-derivedhematologic malignancies. Proc.NatlAcad.Sci.USA 114,E7554–E7563(2017).

56.DiVeroli,G.Y.etal.Combenefit:aninteractiveplatformfortheanalysisand visualizationofdrugcombinations. Bioinformatics 32,2866–2868(2016).

57.Anonymous.Reproducingstromalmicroenvironmenttoscreenformore effectiveantileukemicdrugs. J.Clin.Invest. 126,4(2016).

58.Shukla,S.etal.ProgenitorT-celldifferentiationfromhematopoieticstemcells usingDelta-like-4andVCAM-1. Nat.Methods 14,531–538(2017).

59.Yoshida,H.etal.Thecis-regulatoryatlasofthemouseimmunesystem. Cell 176,897–912.e820(2019).

60.Barrett,T.etal.NCBIGEO:archiveforfunctionalgenomicsdatasets update. NucleicAcidsRes. 41,D991–D995(2012).

61.Chen,S.,Zhou,Y.,Chen,Y.&Gu,J.fastp:anultra-fastall-in-oneFASTQ preprocessor. Bioinformatics 34,i884–i890(2018).

62.Dobin,A.etal.STAR:ultrafastuniversalRNA-seqaligner. Bioinformatics 29, 15–21(2013).

63.Li,H.etal.Thelandscapeofcancercelllinemetabolism. Nat.Med 25, 850–860(2019).

64.Dempster,J.M.etal.Agreementbetweentwolargepan-cancerCRISPR-Cas9 genedependencydatasets. Nat.Commun. 10,5817(2019).

65.Tsherniak,A.etal.Definingacancerdependencymap. Cell 170,564–576 e516(2017).

66.Ritchie,M.E.etal.limmapowersdifferentialexpressionanalysesforRNAsequencingandmicroarraystudies. NucleicAcidsRes. 43,e47(2015).

67.Benjamini,Y.,Drai,D.,Elmer,G.,Kafkafi,N.&Golani,I.Controllingthefalse discoveryrateinbehaviorgeneticsresearch. Behav.BrainRes. 125,279–284 (2001).

68.Lu,X.etal.Theearlymetabolomicresponseofadiposetissueduringacute coldexposureinmice. Sci.Rep. 7,3455(2017).

69.Lodi,A.etal.Combinatorialtreatmentwithnaturalcompoundsinprostate cancerinhibitsprostatetumorgrowthandleadstokeymodulationsofcancer cellmetabolism. NPJPrecis.Oncol. 1, https://doi.org/10.1038/s41698-0170024-z (2017).

70.Tiziani,S.etal.Optimizedmetaboliteextractionfrombloodserumfor1H nuclearmagneticresonancespectroscopy. Anal.Biochem 377,16–23(2008).

71.Sumner,L.W.etal.Proposedminimumreportingstandardsforchemical analysisChemicalAnalysisWorkingGroup(CAWG)Metabolomics StandardsInitiative(MSI). Metabolomics 3,211–221(2007).

72.Okuda,S.etal.KEGGAtlasmappingforglobalanalysisofmetabolic pathways. NucleicAcidsRes. 36,W423–426(2008).

73.Wishart,D.S.etal.HMDB:theHumanMetabolomeDatabase. NucleicAcids Res. 35,D521–526(2007).

74.Sun,Y.etal.Functionalgenomicsrevealssyntheticlethalitybetween phosphogluconatedehydrogenaseandoxidativephosphorylation. CellRep. 26,469–482e465(2019).

75.Hahne,F.etal. flowCore:aBioconductorpackageforhighthroughput flow cytometry. BMCBioinform. 10,106(2009).

76.Levine,J.H.etal.Data-drivenphenotypicdissectionofamlreveals progenitor-likecellsthatcorrelatewithprognosis. Cell 162,184–197(2015).

77.Becht,E.etal.Dimensionalityreductionforvisualizingsingle-celldatausing UMAP. Nat.Biotechnol, https://doi.org/10.1038/nbt.4314 (2018).

Acknowledgements

SupportedinpartbygrantsfromtheCancerPreventionandResearchInstituteofTexas (CPRIT)andtheNationalInstitutesofHealth(NIH)R01CA231364,LeukemiaSPORE P50CA100632andCPRITRP180309toM.K.andS.T.,R01GM134382toJ.Y.MD AndersonCancerCenter’sCancerCenterSupportGrant(P30CA016672)fromtheNIH/ NationalCancerInstitutesupportstheFlowCytometryandCellularImagingFacility,the RNASequencingCoreFacility,andtheMetabolomicsCoreFacility.TheMetabolomics CoreFacilityisalsosupportedbyCPRITgrantRP130397andNIHgrantS10OD01230401.ThisworkwasalsosupportedbyresearchgrantsfromtheCanadianCancerSociety (T.H.,Grant#704867),theOncopole-CancerResearchSociety-Fondsderecherchedu Québec(T.H.#265879),theCanadianInstituteforHealthResearch(T.H.,PJT–148943), GenomeCanadaandGénomeQuébec(G.S.),andafellowshipfromtheColeFoundation (D.T.V.).ThisworkbenefitedfromdataassembledbytheIMMGENconsortium.We thankAmyNinetto,ScientificEditor,ResearchMedicalLibrary,MDAndersonCancer Center,foreditingthemanuscript.WethankCraigSmithandJayDunnfromAgilentfor technicalsupportwithSeahorseexperiments.

Authorcontributions

N.B.,T.H.,andMa.Koconceivedanddesignedthestudy;N.B.,A.L.,Y.D.,M.C.,S.S., R.P.,S.T.,Me.K.,S.S.S.,A.S.,Sh.P.,M.T.,V.M.K.,A.C.,K.H.,N.F.,J.G.,M.E.D.F.,J.R.M., M.O.W.,D.D.,P.L.L.,R.E.D.,M.D.,L.M.G.,G.A.-A.,K.R.,Su.P.,H.M.,V.R.,S.R.-S., A.H.,D.T.V.,Y.G.,A.M.,G.S.,T.H.,D.H.,A.F.,M.G.performedexperiments;N.B.,A.L., S.S.,R.P.,M.C.,S.T.,S.H.,Y.D.,J.Y.,D.T.T.,T.M.H.,F.W.H.,S.K.,K.R.,K.F.,Ko.T.,N.F., J.G.,M.E.D.F.,J.R.M.,W.Y.,J.J.Y.,E.J.J.,S.R.-S.,A.H.,D.T.V.,Y.G.,A.M.,G.S.,P.L.L., T.H.,M.G.,Ka.T.,Ma.Ka.,M.A.analyzedandinterpreteddata;N.B.,T.H.,andMa.Ko wrotethemanuscript;N.B.,A.L.,S.T.,R.E.D.,T.H.,D.H.,A.F.,J.R.M.,Ma.Ko.revisedthe manuscript,Ma.Ko.,S.T.,T.H.supervisedwork,directedthestudy,providedadviceon experiments,allauthorscommentedonthemanuscript.

Competinginterests

Theauthorsdeclarenocompetinginterests.

Additionalinformation

Supplementaryinformation Theonlineversioncontainssupplementarymaterial availableat https://doi.org/10.1038/s41467-022-30396-3

Correspondence andrequestsformaterialsshouldbeaddressedtoMarinaKonopleva.

Peerreviewinformation NatureCommunications thanksG.Helgasonandtheother, anonymous,reviewer(s)fortheircontributiontothepeerreviewofthiswork.Peer reviewerreportsareavailable.

Reprintsandpermissioninformation isavailableat http://www.nature.com/reprints

Publisher’snote SpringerNatureremainsneutralwithregardtojurisdictionalclaimsin publishedmapsandinstitutionalaffiliations.

OpenAccess ThisarticleislicensedunderaCreativeCommons Attribution4.0InternationalLicense,whichpermitsuse,sharing, adaptation,distributionandreproductioninanymediumorformat,aslongasyougive appropriatecredittotheoriginalauthor(s)andthesource,providealinktotheCreative Commonslicense,andindicateifchangesweremade.Theimagesorotherthirdparty materialinthisarticleareincludedinthearticle’sCreativeCommonslicense,unless indicatedotherwiseinacreditlinetothematerial.Ifmaterialisnotincludedinthe article’sCreativeCommonslicenseandyourintendeduseisnotpermittedbystatutory regulationorexceedsthepermitteduse,youwillneedtoobtainpermissiondirectlyfrom thecopyrightholder.Toviewacopyofthislicense,visit http://creativecommons.org/ licenses/by/4.0/

©TheAuthor(s)2022

1DepartmentofLeukemia,TheUniversityofTexasMDAndersonCancerCenter,Houston,TX,USA.2DepartmentofNutritionalSciences,Dell PediatricResearchInstitute,DellMedicalSchool,TheUniversityofTexasatAustin,Austin,TX,USA.3St.JudeGraduateSchoolofBiomedical Sciences,St.JudeChildren’sResearchHospital,Memphis,TN,USA.4InstituteforResearchinImmunologyandCancer,TheUniversityof Montreal,Montréal,QC,Canada.5DepartmentofStemCellTransplantationandCellularTherapy,TheUniversityofTexasMDAndersonCancer Center,Houston,TX,USA.6DepartmentofBioinformaticsandComputationalBiology,TheUniversityofTexasMDAndersonCancerCenter, Houston,TX,USA.7TheJacksonLaboratoryforGenomicMedicine,Farmington,CT,USA.8DepartmentofOralandMaxillofacialSurgery, HirosakiUniversityGraduateSchoolofMedicine,Hirosaki,Aomori,Japan.9DepartmentofPharmaceuticalSciences,St.JudeChildren’sResearch Hospital,Memphis,TN,USA.10DepartmentofCancerSystemsImaging,TheUniversityofTexasMDAndersonCancerCenter,Houston,TX, USA.11TRACTIONPlatform,TherapeuticsDiscoveryDivision,UniversityofTexasM.D.AndersonCancerCenter,Houston,USA.12Department ofImmunology,St.JudeChildren’sResearchHospital,Memphis,TN,USA.13DepartmentofTranslationalMolecularPathology,TheUniversityof TexasMDAndersonCancerCenter,Houston,TX,USA.14DepartmentofLymphomaandMyeloma,TheUniversityofTexasMDAnderson CancerCenter,Houston,TX,USA.15RutgersRobertWoodJohnsonMedicalSchool,CancerInstituteofNewJersey,NewBrunswick,NJ,USA. 16IrvingCancerResearchCenter,ColumbiaUniversityIrvingMedicalCenter,NewYork,NY,USA.17InstituteforAppliedCancerScience,The UniversityofTexasMDAndersonCancerCenter,Houston,TX,USA.18PerelmanSchoolofMedicine,TheUniversityofPennsylvania, Philadelphia,PA,USA.19TexasChildren’sCancerCenter,BaylorCollegeofMedicine,Houston,TX,USA.20DepartmentofVeterinaryMedicine andSurgery,TheUniversityofTexasMDAndersonCancerCenter,Houston,TX,USA.21DepartmentofComputationalBiology,St.Jude Children’sResearchHospital,Memphis,TN,USA.22DepartmentofPharmacologyandPhysiology,FacultyofMedicine,UniversityofMontreal, Montreal,QC,Canada. ✉email:mkonople@mdanderson.org