Published OnlineFirst March 14, 2014; DOI: 10.1158/1078-0432.CCR-13-2407 Clinical Cancer Human Cancer Biology Research FOXM1PromotestheWarburgEffectandPancreaticCancer Progression via Transactivation of LDHA Expression JiujieCui1,2,6,MinShi3,6,DachengXie1,2,DaoyanWei6,ZhiliangJia6,ShaojiangZheng5,YongGao4, SuyunHuang7,andKepingXie6 Abstract Purpose: The transcription factor Forkhead box protein M1 (FOXM1) plays critical roles in cancer development and progression. However, the regulatory role and underlying mechanisms of FOXM1 in cancer metabolism are unknown. In this study, we characterized the regulation of aerobic glycolysis by FOXM1anditsimpactonpancreaticcancermetabolism. ExperimentalDesign:TheeffectofalteredexpressionofFOXM1onexpressionofglycolyticenzymes andtumordevelopmentandprogressionwasexaminedusinganimalmodelsofpancreaticcancer.Also,the underlying mechanisms of altered pancreatic cancer glycolysis were analyzed using in vitro molecular biology. The clinical relevance of aberrant metabolism caused by dysregulated FOXM1 signaling was determinedusingpancreatictumorandnormalpancreatictissuespecimens. Results: We found that FOXM1 did not markedly change the expression of most glycolytic enzymes exceptforphosphoglyceratekinase1(PGK-1)andlactatedehydrogenaseA(LDHA).FOXM1andLDHA were overexpressed concomitantly in pancreatic tumors and cancer cell lines. Increased expression of FOXM1upregulatedtheexpressionofLDHAatboththemRNAandproteinlevelandelevatedLDHactivity, lactateproduction,andglucoseutilization,whereasreducedexpressionofFOXM1didtheopposite.Further studies demonstrated that FOXM1 bound directly to the LDHA promoter region and regulated the expressionoftheLDHAgeneatthetranscriptionallevel.Also,elevatedFOXM1–LDHAsignalingincreased thepancreaticcancercellgrowthandmetastasis. Conclusions: Dysregulated expression and activation of FOXM1 play important roles in aerobic glycolysis and tumorigenesis in patients with pancreatic cancer via transcriptional regulation of LDHA expression.ClinCancerRes;20(10);1–12.(cid:1)2014AACR. Introduction annually,especiallyinindustrializedcountries(1).Inthe UnitedStates,researchersestimatedthat45,220newpan- Pancreaticcanceristheseventhleadingcauseofcancer- creatic cancer cases would be diagnosed and that 38,460 related deaths worldwide, and its incidence is increasing patientswoulddieofthediseasein2013(2).Althoughearly diagnosis of and surgery and systemic chemotherapy for pancreaticcancerhaveimproved,theoverall5-yearsurvival Authors'Affiliations: 1ShanghaiKeyLaboratoryofPancreaticDiseases ratein patients withpancreatic cancerremains below5% Research;2DepartmentofOncology,ShanghaiJiaotongUniversityAffil- iatedFirstPeople'sHospital;3DepartmentofGeneralSurgery,Shanghai (3).Thus,abetterunderstandingoftheunderlyingmechan- JiaotongUniversityAffiliatedRuijinHospital;4DepartmentofOncologyand ismsthatpromotethepathogenesisofpancreaticcanceris TumorInstitute,ShanghaiEastHospital,TongjiUniversitySchoolofMed- icine,Shanghai;5HainanProvincialKeyLaboratoryofCarcinogenesisand urgentlyneeded(4). Intervention,PathologyDepartmentofAffiliatedHospital,HainanMedical TheWarburgeffect,alsoknownasaerobicglycolysis,isa College,Haikou,People'sRepublicofChina;Departmentsof6Gastroen- shiftfromoxidativephosphorylationtoglycolysis,afeature terology,Hepatology,andNutrition;and7Neurosurgery,TheUniversityof TexasMDAndersonCancerCenter,Houston,Texas of which is increased lactate production even at normal oxygenconcentrations,andisconsideredtobetherootof Note:SupplementarydataforthisarticleareavailableatClinicalCancer ResearchOnline(http://clincancerres.aacrjournals.org/). cancer development and progression (5, 6). In pancreatic cancer cases, investigators found that glycolytic flux was J.Cui,M.Shi,andD.Xiecontributedequallytothiswork. elevated and that uptake of glucose increased; thus, 18F- Corresponding Author: Keping Xie, Department of Gastroenterology, labeledfluorodeoxyglucosepositronemissiontomography Hepatology,andNutrition,Unit1466,TheUniversityofTexasMDAnderson can be usedfor diagnosis of and prognosis for pancreatic CancerCenter,1515HolcombeBoulevard,Houston,TX77030.Phone: 713-794-5073/713-834-6685;Fax:713-745-3654/713-745-1163;E-mail: cancer (7, 8). However, the molecular mechanism of [email protected] increased glycolysis in pancreatic cancer cells is largely doi:10.1158/1078-0432.CCR-13-2407 unknown. Recent studies demonstrated that many key (cid:1)2014AmericanAssociationforCancerResearch. oncogenicsignalingpathwaysandfactorsplaycriticalroles www.aacrjournals.org OF1 Downloaded from clincancerres.aacrjournals.org on November 17, 2018. © 2014 American Association for Cancer Research. Published OnlineFirst March 14, 2014; DOI: 10.1158/1078-0432.CCR-13-2407 Cuietal. adjacentpancreatictissue,and10normalpancreatictissue Translational Relevance specimens(USBiomax)andtheother(TMA-P)containing Weusedtwopancreaticcancertissuemicroarraysand 154 human pancreatic ductal adenocacinoma specimens molecular biologic and animal models to evaluate the obtainedfromthePancreaticCancerTissueBankatShang- activationandfunctionoftheForkheadboxproteinM1 hai Jiaotong University Affiliated First People’s Hospital (FOXM1)/lactatedehydrogenaseA(LDHA)pathwayin (Shanghai,People’sRepublicofChina).Theprimarypan- humanpancreaticcancercells.Ourclinicalandmech- creatic cancers in the patients represented in TMA-P were anisticfindingsindicatedthatLDHAisadirecttranscrip- diagnosedandlaterconfirmedbyatleasttwopathologists, tional target of FOXM1 and that dysregulated FOXM1 andthepatientswereacceptedforthepatientsunderwent expression,whichoccursfrequentlyinpancreaticcancer, surgery at the Affiliated First People’s Hospital, Jiangsu leadstoaberrantLDHAexpression.Moreover,FOXM1 Province Hospital (Jiangsu, People’s Republic of China) positivelyregulatespancreaticcancercellaerobicglycol- andShanghaiEastHospital(Shanghai,People’sRepublicof ysisandgrowth,suggestinganovelmolecularbasisfor China)from2004to2011.Tumorstagingforthespecimens thecriticalroleofFOXM1overactivationinpancreatic was carried out using the American Joint Committee on cancer metabolism and that dysregulated FOXM1– Cancer staging criteria. The use of human specimens was LDHAsignalingisapromisingnewmoleculartargetfor approvedbytheproperInstitutionalReviewBoards.After novel preventive and therapeutic strategies for this hematoxylin and eosin staining of slides containing sec- malignancy. Therefore, our findings have a significant tionsofoptimalpancreatictumor,normaltumor-adjacent effectonclinicalmanagementofpancreaticcancer. pancreatictissueupto2cmfromthetumor,andnormal pancreatictissueupto5cmfromthetumor,TMA-Pslides were constructed (in collaboration with Shanghai Outdo Biotech,Shanghai,People’sRepublicofChina).Twopunch intheregulationofpancreaticcancermetabolism,including cores 2 mm in greatest dimension were taken from non- Kras,phosphoinositide3-kinase/AKT,c-Myc,p53,andhyp- necrotic areas of each formalin-fixed, paraffin-embedded oxia-induciblefactor-1(HIF-1;ref.9). (FFPE)tumor,matchednormaltumor-adjacenttissue,and Forkhead box protein M1 (FOXM1) is an oncogenic normaltissuespecimen.Primarytumorandmatchednor- transcriptionfactorbelongingtotheForkheadtranscription maltumor-adjacenttissuespecimensobtainedfrom34of factor superfamily. Human FOXM1 has three isoforms— these154patientswereincludedinTMA-P;primarytumor FOXM1A,FOXM1B,andFOXM1C—asaresultofdifferen- and matched normal tumor-adjacent and normal tissue tialsplicingofexonsVaandVIIa.FOXM1BandFOXM1C specimensfrom22ofthesepatientswereincluded.Sections aretranscriptionallyactive,whereasFOXM1Aistranscrip- (4-mmthick)ofFFPEtumorspecimenswerepreparedand tionallyinactive(10).FOXM1playsessentialrolesinreg- processed for immunostaining using anti-FOXM1 (Santa ulationofthecancercellcycle,angiogenesis,invasion,and CruzBiotechnology)andanti-LDHA(SantaCruzBiotech- metastasisbydirectlypromotingitstargetgeneexpression nology)antibodies.Thestainingresultswerescoredbytwo andnetworkingwithotherfactors(11).Inpreviousstudies, investigatorsblindedtotheclinicaldataasdescribedpre- we found that FOXM1 was overexpressed in pancreatic viously(15). cancercellsandpromotedpancreaticcancerdevelopment andprogression(12,13).Researchersalsofoundevidence Celllinesandreagents thatFOXM1participatesinregulationofmetabolism.For The human pancreatic adenocarcinoma cell lines example, in leptin-knockout mice, overexpression of PANC-1, MiaPaCa-2, AsPC-1, BxPC-3, and PA-TU-8902 FOXM1wasrelatedtoelevatedconsumptionofglucose, werepurchased from theAmerican TypeCultureCollec- whereaslowexpressionofitwasnot(14).However,the tion(ATCC).ThepancreaticcancercelllineMDAPanc-28 role of FOXM1 expression in pancreatic cancer aerobic wasagiftfromDr.PaulJ.Chiao(TheUniversityofTexas glycolysisandtheunderlyingmechanismsareunknown. MDAndersonCancerCenter,Houston,TX).Thehuman Therefore, in the present study, we sought to determine pancreatic adenocarcinoma cell line FG was obtained theroleofFOXM1inregulationoftheWarburgeffectand from Michael P. Vezeridis (The Warren Alpert Medical itsmechanismaboutthisroleinpancreaticcancercases. SchoolofBrownUniversity,Providence,RI;ref.16).The We discovered that FOXM1B transcriptionally upregu- human metastatic pancreatic adenocarcinoma cell line lated LDHA expression; increased LDH activity, lactate COLO357anditsfast-growinglivermetastaticvariantin production, and glucose utilization; and promoted pan- nude mice, L3.7, were described previously (12). All of creatic tumorigenesis and tumor metastasis. thesecelllinesweremaintainedinplasticflasksasadher- ent monolayers in Eagle’s Minimal Essential Medium Materials and Methods supplemented with 10% FBS, sodium pyruvate, nones- Humantissuespecimensandimmunohistochemical sentialaminoacids,L-glutamine,andavitaminsolution analysis (FlowLaboratories).PANC-1,MiaPaCa-2,AsPC-1,BxPC- ExpressionofFOXM1andLDHAwasanalyzedusingtwo 3, and PA-TU-8902 were characterized or authenticated tissue microarrays (TMA): one (TMA) containing 57 pri- by the ATCC using short tandem repeat profiling and marypancreaticductaladenocacinoma,10normaltumor- passagedinourlaboratoryforfewerthan6monthsafter OF2 ClinCancerRes;20(10)May15,2014 ClinicalCancerResearch Downloaded from clincancerres.aacrjournals.org on November 17, 2018. © 2014 American Association for Cancer Research. Published OnlineFirst March 14, 2014; DOI: 10.1158/1078-0432.CCR-13-2407 FOXM1RegulatestheWarburgEffect receipt.AcompetitiveLDHAinhibitor,oxamatesodium, ConstructionofLDHApromoterreporterplasmids was purchased from Sigma-Aldrich (17). andmutagenesis A1.48-kbfragmentcontainingLDHA50sequencesfrom PlasmidsandsiRNAs (cid:3)1,330toþ150bprelativetothetranscriptioninitiation TheplasmidpcDNA3.1–FOXM1(pcDNA3.1–FOXM1B) sitewassubclonedintothepGL3-basicvector(Promega). and control vector pcDNA3.1 were described previously The final full-length reporter plasmid, which contained (18).AnsiRNAtargetingFOXM1(siFOXM1)consistedofa multiple FOXM1-binding sites, was designated pLuc– poolofthreetarget-specific20-to25-ntsiRNAsdesignedto LDHA-1480.Thedeletionmutationreporterforthisplas- knockdown FOXM1 expression was obtained from Santa mid,pLuc–LDHA-493,wasthengenerated.Bothconstructs CruzBiotechnology. wereverifiedbysequencingtheinsertsandflankingregions oftheplasmids. Transienttransfection Transfection of plasmids and siRNAs into pancreatic Promoterreporteranddualluciferaseassay cancercellswasperformedusingthetransfectionreagents Pancreaticcancercellsweretransfectedwiththeindicated Lipofectamine LTX and Lipofectamine 2000 CD, respec- LDHApromoterreporter,siFOXM1,orexpressionplasmid. tively (Invitrogen). For transient transfection, cells were TheLDHApromoteractivitywasnormalizedviacotransfec- transfectedwithplasmidsorsiRNAsatdifferentconcentra- tionofab-actin/Renillaluciferasereportercontainingafull- tionsasindicatedfor48hoursbeforetheperformanceof lengthRenillaluciferasegene(20).Theluciferaseactivityin functional assays. Pancreatic cancer cells treated with the thecellswasquantifiedusingadualluciferaseassaysystem transfectionreagentsalonewereincludedasmockcontrols. (Promega)24hoursaftertransfection. Reversetranscription-PCR Thechromatinimmunoprecipitationassay TotalRNAwasextractedfromtumorcellsusingaPure- Tumorcells(2(cid:1)106)werepreparedforthechromatin LinkRNAMiniKit(LifeTechnologies).Next,2mgoftotal immunoprecipitation(ChIP)assayusingaChIPAssayKit RNA was reverse transcribed using a First Strand cDNA (Millipore)accordingtothemanufacturer’sprotocol.The Synthesis Kit (Promega) to synthesize cDNA specimens. resultingprecipitatedDNAspecimenswereanalyzedusing Subsequently,2mLofthecDNAproductwassubjectedto PCRtoamplifya341-bpregionoftheLDHApromoterwith PCR amplification with Taq DNA polymerase (Qiagen) the primers 50-TATCTCAAAGCTGCACTGGGC-30 (for- using a thermal cycler with PCR primers to detect each ward) and 50-TGCTGATTCCATTGCCTAGC-30 (reverse) factor (Supplementary Table S1). b-Actin and a-tubulin and a 282-bp region of the LDHA promoter with the were used as internal controls. The PCR products were primers 50-CTGCAGGAAGCCATGATCA-30 (forward) and loadedonto2%agarosegelsandvisualizedusingethidium 50-TCCCACTCACAGTGAAGCCT-30(reverse).ThePCRpro- bromideunderultravioletlight. ductswereresolvedelectrophoreticallyona2%agarosegel andvisualizedusingethidiumbromidestaining. Westernblotanalysis StandardWesternblottingwascarriedoutusingwhole- Animals cellproteinlysatesandprimaryantibodiesagainstFOXM1 Pathogen-freefemaleathymicnudemicewerepurchased andLDHA(SantaCruzBiotechnology)andLDHB(Abcam) fromtheNationalCancerInstitute.Theanimalsweremain- andasecondaryantibody[anti-rabbitIgG(immunoglob- tainedinfacilitiesapprovedbytheAssociationforAssess- ulin G) or anti-mouse IgG; Santa Cruz Biotechnology]. ment and Accreditation of Laboratory Animal Care Inter- Equal protein specimen loading was monitored using an national in accordance with the current regulations and anti-a-tubulinantibody(Oncogene). standardsoftheU.S.DepartmentofAgricultureandDepart- mentofHealthandHumanServices. LDHactivity,lactateproduction,andglucose utilizationassay Tumorgrowthandmetastasis Pancreatic tumor cells (1 (cid:1) 106) were transfected with Pancreatic tumor cells (1 (cid:1) 106) in 0.1 mL of Hank plasmids and siRNAs and prepared for LDH activity and balancedsaltsolutionwereinjectedsubcutaneouslyintothe lactateproductionassayusingaLDHActivityAssayKitand rightscapularregionofnudemice.Thetumor-bearingmice LactateAssayKit(Sigma)accordingtothemanufacturer’s werekilledwhentheybecamemoribundoronday35after protocol. For glucose utilization assay, tumor cells were inoculation,andtheirtumorswereremovedandweighed. transfected with plasmids and siRNAs, and cultures were Also,tumorcells(1(cid:1)106)wereinjectedintravenouslyinto incubated for 24 hours. The culture media were replaced a separate group of mice via the ileocolic vein. The mice withphenol-redfreeRPMIwith1%FBSorphenol-redfree werekilledonday21afterinoculationorwhentheyseemed RPMI with 1% FBS and 20 mmol/L oxamate sodium in tobemoribund.Theirliverswerethenremoved,andsurface continuous culture for 3 days. Medium specimens were metastases on them were counted after dissection of the collected each day. Glucose concentrations in the media liversintotheirindividuallobes.Everysurfacewasexamined were measured using a Colorimetric Glucose Assay Kit bytwoinvestigatorswhowereunawareoftheexperimental (BioVision)andnormalizedaccordingtocellnumber(19). protocolandscoredseparately(21). www.aacrjournals.org ClinCancerRes;20(10)May15,2014 OF3 Downloaded from clincancerres.aacrjournals.org on November 17, 2018. © 2014 American Association for Cancer Research. Published OnlineFirst March 14, 2014; DOI: 10.1158/1078-0432.CCR-13-2407 Cuietal. Statisticalanalysis Results Thematchingandsignificanceofthepatientspecimen FOXM1-enhancedglycolysisinpancreaticcancercells data were assessed using the Mann–Whitney U test and Aerobicglycolysisistheprimaryaspectofalteredcancer Pearson correlation coefficient. A two-tailed c2 test and metabolism, so we focused on the effect of FOMX1 regu- the Fisher exact test were used to determine the signifi- lationofpancreaticcancermetabolismonglycolysis.Recent cance of the differences among the covariates. The sig- studiesrevealedthatmanykeytypesofoncogenicsignaling nificance of the in vitro and in vivo data was determined regulate cancer metabolism by regulating the expression usingtheStudentttest(two-tailed)orone-wayANOVA. and/oractivityofmetabolicenzymes.Tofurtherinvestigate Inallofthetests,Pvalueslessthan0.05wereconsidered the mechanism of FOXM1 regulation of glycolysis, we statistically significant. The SPSS software program (ver- assessedtheeffectofFOXM1ontheexpressionofglycolytic sion 13.0; IBM Corporation) was used for all statistical enzymes. We treated PANC-1 cells with siFOXM1 and analysis. measuredthelevelsofglycolyticenzymemRNAexpression Figure1. RegulationoftheWarburg effectbyFOXM1.A,RT-PCR analysisofthemRNAexpression forglycolyticenzymesinPANC-1 cellsafterknockdownofFOXM1 expression.ExpressionofPGK-1 andLDHAwassignificantly downregulated.b-Actinand a-tubulinwereusedascontrols.B, RT-PCRanalysisofthemRNA expressionforFOXM1,LDHA, LDHB,andPGK-1inPA-TU-8902 andMDAPanc-28cellsafter knockdownofFOXM1expression. C,WesternblotanalysisofFOXM1, LDHA,andLDHBexpressionin pancreaticcancercells.D,analysis ofthelactateproductionin pancreaticcancercelllinesusinga LactateAssayKit.E,analysisofthe LDHactivityinpancreaticcancer celllinesusinganLDHActivity AssayKit. OF4 ClinCancerRes;20(10)May15,2014 ClinicalCancerResearch Downloaded from clincancerres.aacrjournals.org on November 17, 2018. © 2014 American Association for Cancer Research. Published OnlineFirst March 14, 2014; DOI: 10.1158/1078-0432.CCR-13-2407 FOXM1RegulatestheWarburgEffect A1 A2 PANC-1 FG P = 0.006 P < 0.001 P = 0.042 P = 0.001 140 140 140 140 %)120 %)120 %) 120 %) 120 Fapciagtniuvcrirteeya2at.incdFcOlaaXncMctae1trercepegrllousdl.auAticoatninoodnfBiLn,DH Lactateproduction (108642000000 LDH activity (108642000000 Lactateproduction ( 108642000000 LDH activity ( 108642000000 PtcroaAnnNtsrCfoe-lc1stieaRdnNdwAFi(tAGh),scaieFnlOldsXBwMxeP1reCo-r3and MockContrsoilFOXM1 MockContrsoilFOXM1 MockContrsoilFOXM1 MockContrsoilFOXM1 AsPC-1cellsweretransfectedwith B1 B2 BxPC-3 AsPC-1 pcDNA3.1,pcDNA3.1-FOXM1,or pcDNA3.1-FOXM1andtreated with20mmol/Loxamatesodium 2 P = 0.033 2 P = 0.010 2 P = 0.012 2 P = 0.041 (B).Thecultureswereincubated fwpAaonrcaroatsd4ilvy8uuiztcsyehetoiAtdohusnetrsos(aLl.eyaDAfntKH)aL,iltaayacwnzctedtaaivsttaehitnuyeAsL(lesraDisdcgHathtayott)K.eit Lactateoduction (fold)10..515 H activity (fold)10..515 Lactateoduction (fold)10..155 H activity (fold) 10..155 pr LD pr LD 0 0 0 0 MoCckontprFolOXM1M 1+ Oxa MockConptrFoOlXM1M 1+ Oxa MoCckonptrFolOXM1M 1+ Oxa MockConptrFolOXM1M 1+ Oxa X X X X O O O O F F F F p p p p using reverse transcription-PCR (RT-PCR; Fig. 1A). We WethensoughttoconfirmtheeffectofFOXM1expres- found that knockdown of FOXM1 expression did not siononPGK-1,LDHA,andLDHBmRNAexpressioninPA- change the expression of most of the enzymes except for TU-8902, MDA Panc-28, MiaPaCa-2, and COLO357 cells phosphoglyceratekinase1(PGK-1)andLDHA. (Fig. 1B and Supplementary Fig. S1). LDHA mRNA Figure3. RegulationofLDHA expressionbyFOXM1.AandB, PANC-1andFGcellswere transfectedwithsiFOXM1or controlsiRNA(A),andBxPC-3and AsPC-1cellsweretransfectedwith pcDNA3.1-FOXM1orpcDNA3.1 (B).Thecultureswereincubatedfor 48hours.TotalproteinandRNA lysateswereharvestedtomeasure theFOXM1,LDHA,andLDHB expressionusingWesternblot(left) andRT-PCR(right)analysis. www.aacrjournals.org ClinCancerRes;20(10)May15,2014 OF5 Downloaded from clincancerres.aacrjournals.org on November 17, 2018. © 2014 American Association for Cancer Research. Published OnlineFirst March 14, 2014; DOI: 10.1158/1078-0432.CCR-13-2407 Cuietal. expression decreased significantly, and PGK-1 mRNA inhibition of LDHA expression induced cancer cell oxi- expression decreased, but LDHB mRNA expression did dative stress and suppressed tumor growth (22, 23). In notchange.PGK-1catalyzes1,3-bisphosphoglycerateand the present study, we focused on the effect of FOXM1 ADP to form 3-phosphoglycerate and ATP, a reversible expression on glycolysis by regulation of LDHA expres- reactionprimarilyregulatedbytheratioofATPtoADP.In sion.WethenanalyzedtheassociationofFOXM1expres- comparison, LDHA, a major subunit of LDH, converts sion with LDHA and LDHB expression, lactate produc- pyruvate to lactate, which is the key characteristic of tion,andLDHactivityinpancreaticcancercelllines.We aerobic glycolysis. Furthermore, we observed that LDHA found that expression of FOXM1 was closely correlated was overexpressed in pancreatic cancer cells and that with expression of LDHA, production of lactate, and Figure4. DirectbindingofFOXM1 totheLDHApromoter.A, sequencesandpositionsof putativeFOXM1-bindingelements ontheLDHApromoter(#1,#2,#3, and#4).B,TheChIPassay. Chromatinswereisolatedfrom PANC-1andFGcells,andbinding ofFOXM1totheLDHApromoter wasexaminedusingtheChIP assay.C,TheChIPassayusing chromatinsisolatedfromPANC-1, PANC-1-siFOXM1,FG,andFG- siFOXM1cells.D,TheChIPassay usingchromatinsisolatedfrom BxPC-3,BxPC-3-pFOXM1,AsPC- 1,andAsPC-1-pFOXM1cells. NormalIgGwasusedasacontrol, and1%ofthetotalcelllysateswas subjectedtoPCRbefore immunoprecipitation(input control).E,measurementofthe LDHApromoteractivity.PANC-1 andFGcellswerecotransfected with0.2mgoftheLDHApromoter– luciferaseconstructpLuc–LDHA and0,25,and50nmol/LsiFOXM1 orcontrolsiRNA.Promoteractivity wasexaminedusingadual luciferaseassaykit.F,BxPC-3and AsPC-1cellswerecotransfected with0.2mgoftheLDHApromoter– luciferaseconstructpLuc–LDHA and0,0.3,and0.6mgof pcDNA3.1-FOXM1orpcDNA3.1. Promoteractivitywasexamined usingadualluciferaseassaykit((cid:4)(cid:4), P<0.01). OF6 ClinCancerRes;20(10)May15,2014 ClinicalCancerResearch Downloaded from clincancerres.aacrjournals.org on November 17, 2018. © 2014 American Association for Cancer Research. Published OnlineFirst March 14, 2014; DOI: 10.1158/1078-0432.CCR-13-2407 FOXM1RegulatestheWarburgEffect activity of LDH but was not significantly correlatedwith sion. We found significant decreases in glucose utiliza- expression of LDHB (Fig. 1C–E). tion, lactate production, and LDH activity (Fig. 2A and Supplementary Fig. S2). In comparison, FOXM1 over- FOXM1promotionoftheWarburgeffectinpancreatic expressionmarkedlyincreasedglucoseutilization,lactate cancercells production,andLDHactivityinBxPC-3andAsPC-1cells, Given that, we observed that FOXM1 expression was which had relatively low levels of FOXM1 and LDHA associated with lactate production and LDH activity, to expression(Fig.2BandSupplementaryFig.S2).Further- furtherexploretheimpactofFOXM1ontheWarburgeffect more, elevated glucose utilization, lactate production, in pancreatic cancer cells, we evaluated the correlation of and LDH activity induced by FOXM1 were dependent alteredFOXM1expressionwithglucoseutilization,lactate on LDHA, as specific inhibition of LDHA activity production, and LDH activity. We did so by knocking by oxamate sodium attenuated the increasing effect of downFOXM1expressioninPANC-1andFGcells,which FOXM1 on glucose utilization, lactate production, and had relatively high levels of FOXM1 and LDHA expres- LDHactivity.ThesedataindicatedthatFOXM1mayaffect A AsPC-1 C FG Control pFOXM1 Control siFOXM1 1 M X O F 50 mm 50 m m 0 m 5 m 0 mm 5 A H D L 50 mm 50 m m 0 m 5 m 0 mm 5 B AsPC-1 D FG 0.5 40 2 40 P = 0.000 P = 0.000 s s P = 0.001 g) 0.4 asi 30 g)1.5 asi 30 ht ( ast ht ( ast eig 0.3 met P = 0.001 eig met r w of 20 r w 1 of 20 mo 0.2 er mo er u b u b T m 10 T0.5 m 10 0.1 u u N N 0 0 0 0 Mock ontrol OXM1 Mock ontrol OXM1 Mock ontrol OXM1 Mock ontrol OXM1 C F C F C F C F p p si si Figure5. InfluenceofFOXM1–LDHAsignalingonpancreaticcancercellgrowthandmetastasisinvivo.AtoD,AsPC-1cellswithFOXM1overexpression(AandB) orFGcellswithknockdownofFOXM1expression(CandD)wereinjectedsubcutaneouslyintotherightscapularregioninnudemice(1(cid:1)106/mouse,n¼5)or intravenouslyintotheileocolicveininnudemice(1(cid:1)106/mouse,n¼5).Thetumor-bearingmicewerekilledwhentheybecamemoribundoronday35 (subcutaneous)orday21(intravenousinjection).Immunohistochemicalstainingofsubcutaneoustumorspecimenswithspecificanti-FOXM1andanti-LDHA antibodieswasperformed.ShownarerepresentativephotographsoftheexpressionofFOXM1andLDHAproteininAsPC-1(A)andFG(C)cells,theweightsof thetumors(BandD,left),andthenumbersofliversurfacemetastases(BandD,right). www.aacrjournals.org ClinCancerRes;20(10)May15,2014 OF7 Downloaded from clincancerres.aacrjournals.org on November 17, 2018. © 2014 American Association for Cancer Research. Published OnlineFirst March 14, 2014; DOI: 10.1158/1078-0432.CCR-13-2407 Cuietal. A Normal Adjacent normal Tumor FOXM1 100 mm 100 mm 100 mm LDHA 100 mm 100 mm 100 mm B 120 FOXM1 120 LDHA Negative 100 P = 0.018 Negative 100 s s P < 0.001 e e Weak as 80 Weak as 80 c c Moderate of Moderate of er 60 er 60 Strong b Strong b Num 40 P = 0.099 Num 40 P = 0.171 20 20 0 0 TT TN NN TT TN NN C D FOXM1 LDHA TMA-P 14 r = 0.674;P < 0.001 12 10 n o 100 mm 100 mm ssi 8 e r p ex 6 A H D 4 L 2 0 0 2 4 6 8 10 12 14 100 mm 100 mm FOXM1 Expression Figure6. ConcomitantexpressionofFOXM1andLDHAinpancreatictumors.PancreatictumorspecimensinTMAandTMA-Pwereimmunostainedwith specificanti-FOXM1andanti-LDHAantibodies.A,representativeimagesofFOXM1andLDHAexpressioninnormalpancreatictissue,normaltumor-adjacent pancreatictissue,andpancreatictumorspecimens(magnification,(cid:1)200).B,FOXM1andLDHAexpressioninTMA-P.(Continuedonthefollowingpage.) OF8 ClinCancerRes;20(10)May15,2014 ClinicalCancerResearch Downloaded from clincancerres.aacrjournals.org on November 17, 2018. © 2014 American Association for Cancer Research. Published OnlineFirst March 14, 2014; DOI: 10.1158/1078-0432.CCR-13-2407 FOXM1RegulatestheWarburgEffect theutilizationofglucoseandproductionoflactate,which DirectimpactofalteredFOXM1–LDHAsignalingon arethemainfeaturesoftheWarburgeffect,viaregulation pancreatictumorgrowthandmetastasisinvivo of the activity of LDH. To determine the effect of altered FOXM1–LDHA signaling on pancreatic cancer biology, we induced TranscriptionalactivationofLDHAexpressionby overexpression of FOXM1 in AsPC-1 cells and knocked FOXM1inpancreaticcancercells down expression of FOXM1 in FG cells and evaluated OurstudydemonstratedadirectassociationofFOXM1 the ability of the cells with altered FOXM1 expression expression with LDH activity and that FOXM1 knock- to grow and metastasize in animal models. Overexpres- down can decrease LDHA mRNA expression. To further sion of FOXM1 significantly elevated the expression of explore the mechanisms of FOXM1 regulation of LDH LDHA in vivo (Fig. 5A), which was consistent with in expression,weexaminedtheeffectsofalteredexpression vitroresults,andpromotedtumorgrowthandmetastasis of FOXM1 on expression of LDHA and LDHB,the main (Fig. 5B). Conversely, downregulation of expression of subunitsofLDH,atboththemRNAandproteinlevelin FOXM1 markedly inhibited the expression of LDHA thepancreaticcancercelllinesPANC-1,FG,BxPC-3,and (Fig. 5C) and tumor growth and metastasis (Fig. 5D). AsPC-1. Silencing of FOXM1 expression significantly These data suggested that FOXM1–LDHA signaling reg- downregulatedLDHAexpressioninPANC-1andFGcells ulated pancreatic cancer development and progression. (Fig. 3A), whereas restored FOXM1 expression markedly upregulatedLDHAexpressioninBxPC-3andAsPC-1cells DirectcorrelationofconcomitantexpressionofFOXM1 (Fig. 3B). The expression of LDHB did not change with andLDHwithpathologicfeaturesofpancreaticcancer eitheroverexpressionorsilencingofFOXM1.Thesedata WeprovideevidencethatFOXM1transcriptionallyreg- revealedthatFOXM1regulatesLDHactivitybyregulating ulatedLDHAexpressionandglycolysisinpancreaticcancer the expression of LDHA rather than LDHB. cells.Tofurtherconfirmourhypothesis,weinvestigatedthe ToidentifythemechanismoftheregulationofFOXM1of expressionofFOXM1andLDHAinpancreatictumorspeci- LDHAexpression,weanalyzedthesequenceoftheLDHA mensusingtheTMAandTMA-Pmicroarrayswithdifferent promoterforthepotentialFOXM1-bindingelements50-A patientethnicities.Theclinicopathologiccharacteristicsof (C/T)AAA(C/T)AA-30, 50-TAATCA-30, and 50-AGATT- the two arrays are described in Supplementary Tables S2 GAGTA-30 (20, 24, 25). We identified four putative andS3.WeusedtheMann–WhitneyUtesttoanalyzethe FOXM1-binding elements in the LDHA promoter region matchingoftheclinicopathologiccharacteristicsforthetwo (Fig.4A).WethengeneratedtwoLDHApromoters—pLuc– arrays and found that most of the characteristics did not LDHA-1480andpLuc–LDHA-493—(Fig.4A)andcotrans- match(SupplementaryTableS4).Thus,weanalyzedTMA fected the reporters with FOXM1 expression vectors or andTMA-Pseparately. siFOXM1 into PANC-1, FG, BxPC-3,and AsPC-1 cells. As Using a specific anti-FOXM1 antibody, we observed showninFig.4EandF,overexpressionofFOXM1elevated FOXM1-positivestaininginthenucleiand/orcytoplasm theLDHApromoteractivity,whereasknockdownofexpres- of pancreatic tumor cells and FOXM1-negative or weak sionofFOXM1decreasedtheLDHApromoteractivityinthe FOXM1-positive staining in the cytoplasm of normal pLuc–LDA-1480groups,andalteredexpressionofFOXM1 tumor-adjacent and normal pancreatic cells in TMA-P did not change the promoter activity in the pLuc–LDHA- andTMA(Fig.6A,topplots;Fig.6B,left;Supplementary 493groups,whichdidnotcontainthepotentialFOXM1- Fig. S4A). We then analyzed the relationship between binding site. To provide evidence that FOXM1 bound clinicopathologic characteristics and FOXM1 expression directlytotheLDHApromoter,weconductedaChIPassay inpancreatictumors.WeobservedthatinTMA-P,FOXM1 usingchromatinspreparedfromPANC-1andFGcells.The expressionwaspositivelyassociatedwithpTclassification resultssuggestedthatFOXM1boundtoalloftheputative (P<0.001),regionallymphnodemetastasis(P<0.001), bindingsitesbutmainlydidsotosites#1and#2(Fig.4B). TNM (tumor–node–metastasis) stage (P < 0.001; Sup- Toconfirmtheseresults,weknockeddownFOXM1expres- plementary Fig. S3A1 and S3A2), and tumor differenti- sion in PANC-1 and FG cells, which led to decreased ation (P ¼ 0.005; Supplementary Fig. S3B1 and S3B2; FOXM1recruitmenttotheLDHApromoter(Fig.4C),and Supplementary Table S7). In TMA, FOXM1 expression induced FOXM1 overexpression in BxPC-3 and AsPC-1 was positively associated with regional lymph node cells, which led to increased FOXM1 recruitment to the metastasis (P ¼ 0.035) and TNM stage (P ¼ 0.015) but LDHA promoter (Fig. 4D). These data strongly suggested not other clinicopathologic characteristics (Supplemen- thatFOXM1boundtotheLDHApromoterandtranscrip- taryTableS5),whichmayhaveresultedfromthelimited tionallyregulatedLDHAexpression. number of specimens. However, these data strongly (Continued.)Theexpressionlevelsweresignificantlyhigherintumor(TT)thannormaltumor-adjacenttissue(TN)andnormaltissue(NN)specimens,whereas theexpressionofFOXM1andLDHAdidnotdifferbetweenTNandNN.LDHAexpressionwasassociatedwithFOXM1expression.C,representative photographsofnegative(toptwoplots)andpositive(bottomtwoplots)FOXM1andLDHAstaininginpancreatictumorspecimens.D,directcorrelation betweenFOXM1expressionandLDHAexpressioninTMA-P(n¼154;Pearsoncorrelationcoefficient,r¼0.674;P<0.001).Someofthedotsonthegraph representmorethanonespecimen(overlappingscores). www.aacrjournals.org ClinCancerRes;20(10)May15,2014 OF9 Downloaded from clincancerres.aacrjournals.org on November 17, 2018. © 2014 American Association for Cancer Research. Published OnlineFirst March 14, 2014; DOI: 10.1158/1078-0432.CCR-13-2407 Cuietal. suggested that elevated expression of FOXM1 played a cells, including liver cancer, prostate cancer, lung cancer, critical role in pancreatic cancer development and glioma,breastcancer,gastriccancer,andpancreaticcancer progression. cells,anditsoverexpressionisassociatedwithtumordevel- UsingthesamecohortofTMA-PandTMAspecimens,we opmentandprogression(18,30–35).Inthepresentstudy, analyzed the expression of LDHA and observed LDHA- weusedtwopancreaticcancerTMAswithdifferentpatient positive staining in the cytoplasm of the cancer cells. We ethnicities(TMAandTMA-P)andfoundthatFOXM1was also saw increasingly LDHA-positive staining in normal overexpressedinpancreatictumorsandthatitsoverexpres- pancreatictissue,normaltumor-adjacentpancreatictissue, sionwascorrelatedwithTstage,Nstage,TNMstage,and andprimarypancreatictumorspecimens(Fig.6A,bottom tumordifferentiation,whichwasconsistentwithresultsof plots;Fig.6B,right;SupplementaryFig.S4B).Expressionof ourpreviousstudies(12,13). LDHAwasalsopositivelyassociatedwithpTclassification We have demonstrated the critical role of FOXM1 in (P ¼ 0.001), TNM stage (P ¼ 0.008; Supplementary cancer development and progression, but the literature Fig. S3C1 and S3C2), and tumor differentiation (P ¼ contains little evidence revealing the function of FOXM1 0.018;SupplementaryFig.S3D1andS3D2;Supplementary incancermetabolism.TheWarburgeffect,whichresearch- TableS8)inTMA-PbutpositivelyassociatedonlywithTNM ers proposed more than 90 years ago, is considered a stage(P¼0.020;SupplementaryTableS6)inTMA. hallmark of cancer (36). Compared with normal cells, Furthermore, we found direct correlations between cancercellstakeupandconsumelargeramountsofglucose FOXM1 and LDHA expression in both TMA-P and TMA, and maintain much higher rates of glycolysis and lactate which were consistent with our findings for pancreatic production even at normal oxygen concentrations (37). cancer cell lines. In analyzing consecutive pancreatic One of the explanations for the Warburg effect is over- tumor sections, we observed that pancreatic cancer cells expressionofglucosetransportersandtheenzymesinthe expressed both FOXM1 and LDHA (Fig. 6C) and that glycolyticpathwayasaconsequenceofoncogeneactivation FOXM1 expression was significantly correlated with (9). In the present study, we found that knocking down LDHA expression in both TMA-P (r ¼ 0.674, P < expressionofFOXM1didnotchangetheexpressionofmost 0.001;Fig. 6D) and TMA (r ¼0.985, P <0.001; Supple- glycolytic enzymes except for PGK-1 and LDHA. Because mentary Fig. S4C). LDHAmainlyconvertspyruvatetolactate,whichisthekey characteristicofaerobicglycolysis,wefocusedonthefunc- Discussion tionofFOXM1inglycolysisbyregulatingLDHAexpression Inthepresentstudy,wedeterminedtherolesofFOXM1in inthisstudy. pancreaticcanceraerobicglycolysisandofFOXM1–LDHA LDHAisamajorsubunitofLDH,whichisatetrameric signaling in pancreatic tumorigenesis. We provided four enzymecomprisingthemajorsubunitsAandB,resultsin lines of evidence supporting a critical role for FOXM1 in five isozymes: A4 (LDH5), A3B1 (LDH4), A2B2 (LDH3), regulation of aerobic glycolysis by the transcriptionally A1B3(LDH2),andB4(LDH1).Recentstudiesrevealedthat activated LDHA gene. First, we found that FOXM1 and LDHA favors catalysis of the conversion of pyruvate to LDHA were concomitantly overexpressed in pancreatic lactate, whereas LDHB converts lactate to pyruvate; also, tumor specimens and cell lines. Also, the expression of researchers linked an isozyme shift from LDH1 to LDH5 FOXM1andLDHAinpancreaticcancercellswaspositively with cancer progression (38–40). Authors have reported associatedwithlactateproductionandLDHactivity.Second, thatLDHAisoverexpressedinmanykindsofcancercells, overexpressionofFOXM1increasedtheexpressionofLDHA including breast, gynecologic, colorectal, and non–small at both mRNA and protein level and increased glucose celllungcancercells,andsuggestedthatthisoverexpression utilization, lactate production, and LDH activity, whereas promotes cancer development and progression (39, 41– reducedexpressionofFOXM1didtheopposite.Thesedata 43).Inpatientswithpancreaticcancer,theserumLDHlevel indicatedthatFOXM1regulatedaerobicglycolysisbyregu- iselevated,andthiselevationisanindependentunfavor- latingtheexpressionofLDHA.Third,FOXM1bounddirect- ableprognosticfactor(44).Inthepresentstudy,weusedthe lytotheLDHApromoterregionandregulatedexpressionof samecohortsofpancreatictumorspecimensusedtoana- theLDHAgeneatthetranscriptionallevel.Fourth,increased lyze FOXM1 expression and found thatLDHAexpression FOXM1–LDHA signaling promoted pancreatic tumor cell wascorrelatedwithTstage,TNMstage,andtumordiffer- growthandmetastasisinvivo,whereasdecreasedFOXM1– entiation.TheseresultssuggestedthatLDHAexpressionwas LDHAsignalingdidtheopposite.Thus,thisnovelFOXM1– associated with pancreatic cancer development and pro- LDHAsignalingcriticallycontributedtotheWarburgeffect gression. We then found that expression of LDHA was inpancreaticcancercellsandtopancreaticcancerdevelop- associated with that of FOXM1 in both TMA and TMA-P mentandprogression. pancreatictumorspecimensregardlessofpatientethnicity TheoncogenictranscriptionfactorFOXM1isakeyreg- andtypeofcelllines,whereaswedidnotfindasignificant ulatorofM-phaseprogressionandG –S-andG –M-phase correlation between FOXM1 and LDHB expression. Fur- 1 2 transition (26–29). Besides its essential roles in cell-cycle thermore,FOXM1expressioncorrelatedwithLDHactivity regulation, FOXM1 is involved in tumor angiogenesis, andlactateproductioninpancreaticcancercelllines.Next, epithelial–mesenchymaltransition,invasion,andmetasta- we evaluated the effect of altered FOXM1 expression on sis(11).FOXM1isoverexpressedinmanytypesofcancer LDHA and LDHB expression, glucose utilization, lactate OF10 ClinCancerRes;20(10)May15,2014 ClinicalCancerResearch Downloaded from clincancerres.aacrjournals.org on November 17, 2018. © 2014 American Association for Cancer Research.