HIF-1α Induces Genetic Instability by Transcriptionally Downregulating MutSα Expression

Molecular Cell,Vol.17,793–803,March18,2005,Copyright©2005by Elsevier Inc.DOI10.1016/j.molcel.2005.02.015HIF-1␣Induces Genetic Instabilityby Transcriptionally DownregulatingMutS␣ExpressionMinori Koshiji,1Kenneth K.-W.To,1pressor p53or the apoptosis inhibitor Bcl2substantiallyreduced cell death(Graeber et al.,1996).Hypoxia also Stefanie Hammer,1Kensuke Kumamoto,1induces gene amplifications(Rice et al.,1986)by driv-Adrian L.Harris,2Paul Modrich,3,4ing fusion and subsequent reintegration of double min-and L.Eric Huang1,*1Laboratory of Human Carcinogenesisutes into chromosomal fragile sites,generating homo-geneously stained regions(Coquelle et al.,1998). National Cancer InstituteMoreover,hypoxia enriches DNA mismatch repair-defi-National Institutes of Healthcient cells,thereby augmenting microsatellite instability Bethesda,Maryland20892(MIN)in human colon carcinomas(Kondo et al.,2001). 2Molecular Oncology LaboratoryGenetic instability is regarded as a hallmark of most Weatherall Institute for Molecular Medicinehuman cancers.The resulting genetic changes include John Radcliffe Hospitalthose at the nucleotide level(such as base substitu-Cancer Research UKtions and deletions or insertions of a few nucleotides) Oxford OX39DSand those at the chromosomal level involving chromo-United Kingdomsome loss or gain,translocations,and rearrangements 3Howard Hughes Medical Institute(Lengauer et al.,1998).Most human tumors are geneti-Duke University Medical Centercally heterogeneous,presumably resulting from a mu-Durham,North Carolina27710tator phenotype early during their evolution(Loeb et al., 4Department of Biochemistry2003).The molecular mechanisms underlying these ge-Duke Universitynetic alterations have been the subject of continued in-Durham,North Carolina27710vestigation.The human DNA mismatch repair(MMR)systemplays a crucial role in safeguarding genomic integrityby correcting DNA replication errors and by blocking Summaryrecombination events between divergent DNA se-quences(Kolodner,1996;Modrich and Lahue,1996). Hypoxia promotes genetic instability by undefinedMMR has also been implicated in the cellular response mechanisms.The transcription factor HIF-1␣is crucialto certain classes of DNA damage(Modrich and Lahue, for the cellular response to hypoxia and is frequently1996;Li,1999)and may have some function in nucleo-overexpressed in human cancers,resulting in thetide excision repair(Kolodner,1996).Mammalian MMR activation of genes essential for cell survival.Here,proteins include homologs of Escherichia coli MutS we demonstrate that HIF-1␣is responsible for genetic(MSH2,MSH3,and MSH6)and MutL(MLH1and PMS2). instability at the nucleotide level by inhibiting MSH2They form MutSα(MSH2-MSH6),MutSβ(MSH2-MSH3), and MSH6,thereby decreasing levels of the MSH2-and MutLα(MLH1-PMS2)heterodimers(Modrich and MSH6complex,MutS␣,which recognizes base mis-Lahue,1996;Buermeyer et al.,1999;Kolodner and Mar-matches.HIF-1␣displaces the transcriptional ac-sischky,1999).Whereas the MutSαheterodimer recog-tivator Myc from Sp1binding to repress MutS␣nizes both base-base and insertion and/or deletion expression in a p53-dependent manner;Sp1servesmismatches,MutSβspecificity is restricted to the latter as a molecular switch by recruiting HIF-1␣to the gene(Genschel et al.,1998).In addition,MutSαforms a ter-promoter under hypoxia.Furthermore,in human spo-nary complex with MutLαand presumably plays a role radic colon cancers,HIF-1␣overexpression is statis-in the initiation of mismatch repair(Blackwell et al., tically associated with the loss of MSH2expression,2001).Mutations in the MMR genes are associated with especially when p53is immunochemically undetecta-the development of both hereditary and sporadic can-ble.These findings indicate that the regulation ofcers,and germline mutations in MSH2or MLH1are the DNA repair is an integral part of the hypoxic re-most prevalent cause of hereditary nonpolyposis colo-sponse,providing molecular insights into the mecha-rectal cancers(Buermeyer et al.,1999).Hereditary nisms underlying hypoxia-induced genetic instability.MSH6defects have also been implicated in the devel-opment of late-onset familial colon tumors(Akiyama et Introduction al.,1997;Miyaki et al.,1997).Despite the functional importance of MMR in main-Hypoxia—oxygen deficiency in tissues—is a key factor taining genetic stability,little is known about the regula-of tumor microenvironment that has been tightly asso-tory mechanisms underlying MMR gene expression. ciated with tumor progression,metastasis,and resis-Hypermethylation of the MLH1promoter has been ob-tance to chemotherapy and radiation therapy(Young et served in colon cancer with MIN(Kane et al.,1997).p53 al.,1988;Brizel et al.,1996;Hockel et al.,1996).Hyp-binds to the MSH2promoter for transcriptional activa-oxic microenvironment is linked to DNA base substitu-tion(Scherer et al.,1996;Warnick et al.,2001).Myc has tions and deletions in tumors(Reynolds et al.,1996).been implicated by serial analysis of gene expression Severe hypoxia selects for tumor cells that have de-as a positive regulator of MSH2expression(Menssen fects in apoptosis;genetic alterations of the tumor sup-and Hermeking,2002).In addition,several functionalSp1sites exist in the MSH6promoter(Gazzoli and Ko-lodner,2003).However,how hypoxia induces MIN re-*Correspondence:huange@Molecular Cell794Figure1.Hypoxia Induces Genetic Instabilityby Specifically Inhibiting MutSαExpressionin a p53-Dependent Manner(A)HCT116cells were either subjected tohypoxia or infected with recombinant adeno-viruses expressing GFP or⌬ODD for48hr.The dinucleotide-repeat marker D17S250was used for MIN analysis.Arrows denoteinduced MIN.(B)HEC59cells were treated and analyzedas above.(C)In-frame(IF)and out-of-frame(OF)β-galactosidase reporters were used to ex-amine hypoxia-induced genetic instability incell types as indicated.The results were pre-sented as means+standard deviations fromthree independent experiments in duplicate.*p<0.05;**p<0.01;RBU,relativeβ-galacto-sidase activity;N,normoxia;and H,hypoxia.(D)HCT116cells were subjected to hypoxiafor16hr and analyzed for hypoxic repressionof MSH2and MSH6protein levels by West-ern blot analysis.HIF-1α,p53,andβ-actinwere included as controls.(E)Normal human small airway epithelialcells(SAEC)and fibroblasts(MRC5)weresubjected to hypoxia as above.The mRNAlevels of mismatch repair genes,as indi-cated,were determined with real-time PCR.PGK1served as a positive control for hyp-oxic induction.Representative results fromthree independent experiments in triplicatewere presented as means±standard errors.(F)HCT116(TP53+/+),its TP53−/−mutant,andHeLa cells were treated as above and exam-ined with real-time PCR.mains unclear.Recently,MLH1expression was shown markers(Boland et al.,1998).Figure1A shows that to be specifically inhibited by hypoxia,resulting in func-hypoxia reproducibly increased MIN in MLH1-deficient tional MMR deficiency(Mihaylova et al.,2003),even HCT116cells,as assayed with the dinucleotide-repeat though the proposed mechanism of hypoxia-induced marker D17S250.However,no such change was de-histone deacetylation apparently cannot account for tected in MSH2-deficient HEC59cells(Figure1B),sug-the specific downregulation of MLH1.To ascertain the gesting the requirement of MSH2for hypoxia-induced mechanisms of hypoxia-induced genetic instability,we e of the mononucleotide-repeat marker BAT26 examined the role of HIF-1α,a key regulator of hypoxia-yielded a similar result in HCT116cells(see Figure S1A responsive genes(Giaccia et al.,2003;Huang and in the Supplemental Data available with this article Bunn,2003;Pugh and Ratcliffe,2003;Semenza,2003;online).Poellinger and Johnson,2004),in DNA mismatch repair To provide additional evidence for the role of MSH2 and demonstrated that HIF-1αis responsible for hyp-in hypoxia-induced MIN,we employed reporter con-oxia-induced genetic instability by inhibiting MutSαstructs containing multiple copies of CA dinucleotide gene expression.repeats that precede theβ-galactosidase reporter geneeither in frame(IF)or out of frame(OF)(Parsons et al., Results1993).Hypoxia not only decreased the expression ofthe IF reporter by>40%,but also increased the OF re-porter activity by w3-fold in HCT116cells(Figure1C). Hypoxia Induces Genetic Instability by InhibitingAgain,no such changes were observed in HEC59cells. MutS␣ExpressionUnlike its parent cells,the MSH2-proficient HEC59+Ch2 To demonstrate hypoxic effects on genetic instability,we first examined microsatellite DNA with specific cells(Aebi et al.,1996)exhibited an w25%decrease ofHIF-1αRepresses MutSαExpression795Figure2.HIF-1αIs Necessary and Sufficient to Induce Genetic Instability by Downregulating MutSαExpression(A)HCT116cells were transfected with siRNAs targeting HIF1A or HIF2A and then subjected to hypoxic treatment.Luciferase siRNA(luc) was used as a negative control.MSH2,MSH6,and PGK1mRNA levels were determined with real-time PCR.Representative results are shown as in Figure1E.(B)Cells were treated as in(A)and assayed for specific protein levels,as indicated,by sequential probing of the same blot with the corre-sponding antibodies.(C)HCT116cells were infected with adenoviruses expressing⌬ODD(Ad-⌬ODD)or transcriptionally inactive⌬ODD(Ad-LCLL)for16hr.Cells without treatment(−)or infected with Ad-GFP served as controls.Transcripts of MSH2and MSH6,along with HIF1A,D ODD,hypoxia-inducible BHLHB2,and ACTB were determined with conventional PCR and quantified with densitometry.The numerical results in reference to ACTB levels are indicated.(D)HCT116cells were subjected to the same treatment as above,and protein levels of specified genes were sequentially probed with specific antibodies.Molecular weights are indicated on the left.(E)HCT116+Ch3cells were analyzed for MMR gene expression with real-time PCR after adenoviral infection as indicated.(F)Effects of Ad-⌬ODD on genetic instability were examined with IF and OF reporters in HCT116and HEC59cells.The results were presented as in Figure1C.**p<0.01.the IF reporter activity and an w30%increase in the MLH1remained essentially unchanged(Figure1E).In OF reporter activity in response to hypoxia.Similar re-contrast,the hypoxia-inducible gene PGK1exhibited a sults were obtained with MLH1-proficient HCT116cells4-fold increase.Similar results were obtained in the (HCT116+Ch3,Figure S1B).Furthermore,we tested thehuman MRC5fibroblasts(Figure1E),as well as in IF and OF reporters in MMR-proficient MCF-7and MMR-proficient MCF-7and HepG2cells(Figure S2B).As expected,hypoxia also decreased MSH2and MSH6 HepG2cells.Again,the hypoxic effects were also sta-tistically significant(Figure S2A).Taken together,these protein levels in HCT116cells(Figure1D).findings indicate the involvement of MSH2in hypoxia-However,in human cervical carcinoma HeLa cells, induced genetic instability.the expression of MLH1(Figure S3A),but not MSH2 To determine whether hypoxia alters MSH2expres-and MSH6(Figure1F),was slightly downregulated, sion,we investigated a panel of MMR genes in the nor-consistent with a previous report(Mihaylova et al., mal human diploid small airway epithelial cells.Results2003).In addition,no significant hypoxia-induced MIN from Taqman assays showed that hypoxia markedly re-was detected with the BAT26marker(Figure S3B).p53 pressed MSH2and MSH6expression by>50%,butis inactivated in HeLa cells(Scheffner et al.,1990),andMolecular Cell796Figure3.Requirement of Myc and p53forMutSαExpression(A)HCT116cells were transfected with MYC-specific siRNA and subsequently assayedfor MutSαexpression in normoxia and hyp-oxia with real-time PCR.Representative re-sults are shown as in Figure1E.(B)Effects of MYC siRNA were examined bysequential probing with respective antibod-ies in Western blotting.(C)TP53expression was knocked down inHCT116TP53+/+cells by specific siRNA orrestored in HCT116TP53−/−cells with ade-noviral infection.The mRNA levels of speci-fied genes were determined by real-timePCR.(D)MSH2and MSH6protein levels were alsoexamined in the cells treated as above byWestern blot analysis.p53levels were alsodetermined.the possible role of p53was therefore examined for the hypoxia.Figure2C shows that the infection of HCT116 hypoxic repression of MSH2and MSH6.In contrast to cells with recombinant adenoviruses expressing a sta-the result from wild-type HCT116,the TP53null cells ble HIF-1α(Ad-⌬ODD)(Koshiji et al.,2004)resulted in (Bunz et al.,1998)not only lost hypoxic repression but the downregulation of both MSH2and MSH6and the also exhibited a marked decrease in the basal expres-upregulation of BHLHB2,a known hypoxia-inducible sion of MMR genes(Figure1F).Thus,we conclude that gene(Wykoff et al.,2000).Interestingly,adenoviruses hypoxia promotes genetic instability by specifically in-expressing a stable yet transcriptionally inactive HIF-hibiting MutSαexpression in a p53-dependent manner.1α(Ad-LCLL)(Koshiji et al.,2004)also inhibited MutSαexpression but was unable to activate BHLHB2.Thisfinding suggested that HIF-1αtransactivation domains HIF-1␣,but Not HIF-2␣,Is Critical for Hypoxicare not required for the inhibitory effect.Consistently, Inhibition of Mismatch RepairAd-LLCL decreased protein levels of MutSαsubunits Hypoxia activates HIF-1αas well as HIF-2α/EPAS(Tian(Figure2D).Furthermore,in HCT116+Ch3cells,Ad-et al.,1997),another member of the HIF-αfamily.ToLCLL infection specifically downregulated MSH2and address whether HIF-αis responsible for hypoxic inhi-MSH6expression(Figure2E).These results argue that bition of MMR,we employed small interfering RNAHIF-1αtranscriptional activity is dispensable for inhibit-(siRNA)targeting either HIF1A or HIF2A.As shown ining MutSαexpression.Figure2A,HIF1A,but not HIF2A,siRNA specificallyTo test whether HIF-1αpromotes genetic instability, abrogated hypoxic repression of MSH2and MSH6.Re-we again resorted to the IF and OF reporters and analy-sults in Figure2B confirmed the loss of HIF-1αand HIF-sis of MIN.Ad-⌬ODD infection significantly altered the 2α(Figure S4A)and the unique role of HIF-1αin mediat-reporter activities in HCT116,but not in HEC59(Figure ing the hypoxic repression.2F).Similarly,Ad-⌬ODD infection gave rise to MIN in To substantiate the role of HIF-1αin MutSαexpres-HCT116cells,but not in HEC59(Figures1A,1B,and sion,we asked whether a stable HIF-1αin normoxiawould recapitulate the repression of MMR genes by S1A).Altogether,we conclude that HIF-1αis responsi-HIF-1αRepresses MutSαExpression797To substantiate the role of p53in hypoxic inhibitionof MutSα,we also employed TP53siRNA.Silencing ofTP53expression decreased MSH2and MSH6mRNAlevels by>30%in HCT116cells(Figure3C).Conversely,introduction of p53to TP53null HCT116cells raisedthe basal level similar to that in wild-type HCT116.Moreover,the p53effect on MutSαexpression wasconfirmed by Western blotting(Figure3D).In addition,forced expression of p53in TP53null HCT116cells in-creased the MSH2promoter activity(Figure S5).There-fore,both Myc and p53are required for basal expres-sion of MutSα.To establish the relevance of the above transcriptionfactors to the transcriptional regulation of MutSαgenes,we analyzed their binding to the MSH2andMSH6promoters in vivo with chromatin immunopre-cipitations(ChIP).The MSH2proximal promoter(−1200to+1)was divided into three regions(Figure4A),one ofwhich(−780to−392)showed binding of these tran-scription factors.This region was narrowed down fur-ther to−671to−518.Likewise,the MSH6promoter wasmapped to−644to−300.The ChIP results from theminimal regions revealed that although p53and Mycbound to the MSH2promoter in normoxic HCT116cells,hypoxic treatment markedly reduced the bindingbut induced HIF-1αbinding(Figure4B).Similar resultswere obtained with the MSH6promoter,reminiscent ofthe changes taking place in the CDKN1A promoter(Koshiji et al.,2004).However,unlike the CDKN1A prox-imal promoter,the MutSαgene promoters showed noMiz-1binding.Moreover,p53binding was greatly inhib-Figure4.Myc Displacement by Hypoxia from MSH2and MSH6Pro-ited by hypoxia in the MSH2and MSH6promoters butmoters Is p53Dependent was unaffected in the CDKN1A distal promoter(Koshiji (A)Schematic representation of the MSH2and MSH6proximal pro-et al.,2004).Furthermore,in hypoxic TP53null cells, moters.Double-arrowed lines indicate regions that were PCR-Myc binding remained unchanged in the MSH2and amplified in ChIP.Representative data of the shadowed regions are MSH6promoters but markedly diminished in theshown in(B).CDKN1A proximal promoter(Figure4B),suggesting the (B)Chromatin immunoprecipitations of normoxic and hypoxicdifferential requirement of p53for the displacement of HCT116cells were performed by using antibodies against indi-Myc from MutSαgenes and CDKN1A promoters.Alto-cated transcription factors.Both TP53+/+and TP53−/−cells wereused.Specific primers flanking regions of the MSH2,MSH6,and gether,we infer that hypoxic repression of MutSαCDKN1A proximal promoters were used to amplify the immuno-genes results from the gain of HIF-1αbinding,in con-precipitated genomic DNA.Sheared genomic DNA prior to immu-comitance with the loss of Myc and p53binding from noprecipitations served as input,and immunoglobulin(IgG)servedthe promoter.as a control.HIF-1␣Competes against Myc for Sp1Bindingble for hypoxia-induced genetic instability by downreg-Previously,we showed that HIF-1αforms a weak com-ulating MutSαgenes.plex with Myc,indicative of an indirect interaction(Ko-shiji et al.,2004).The above ChIP of the MutSαgene HIF-1␣Requires Myc and p53to Inhibit MutS␣promoters identified constitutive binding of Sp1,which Previously,we proposed that the N-terminal HIF-1αin-is known to interact with Myc(Gartel et al.,2001)andto engage in a hypoxic response(Discher et al.,1998). dependently regulates a distinct set of hypoxia-respon-sive genes via antagonizing Myc activity(Koshiji et al.,Figure5A shows that knockdown of SP1with siRNA 2004).MSH2is activated by Myc(Menssen and Her-(Figure S4B)abrogated MutSαexpression in both nor-meking,2002)and is repressed herein by a transcrip-moxia and hypoxia,suggesting an essential role for tionally inactive HIF-1α,characteristic of genes regu-Sp1in the transcriptional regulation of MutSαgenes. lated by the HIF-1α-Myc pathway.To confirm the In addition,SP1siRNA abolished hypoxic induction of involvement of Myc in MSH2and MSH6expression,wePGK1(Pfeifer et al.,1990).To establish the functional knocked down MYC expression with RNA interference.relevance of Sp1,we asked whether HIF-1αcompetes Consequently,MYC siRNA not only markedly reducedfor Sp1binding.Figure5B shows that anti-Sp1immu-MSH2and MSH6basal expression but also abrogated noprecipitation captured Myc only in normoxic extract the hypoxic repression(Figures3A and3B),confirmingand HIF-1αinstead in hypoxic extract.Moreover,knock-the essential role of Myc for maintaining MutSαex-down of HIF-1αlevels by specific siRNA restored Myc pression.binding in hypoxia,arguing for the pivotal role of Sp1Molecular Cell798Figure5.Sp1Is Essential for Recruiting HIF-1αonto the MSH2Promoter(A)HCT116cells were transfected with SP1siRNA,and subsequently the expression ofMutSαgenes was determined with real-timePCR.Representative results are shown as inFigure1E.(B)Normoxic and hypoxic HCT116cell ly-sates were immunoprecipitated with anti-Sp1antibody(α-Sp1IP),followed by Westernblotting with anti-Myc and anti-HIF-1αanti-bodies(bottom).Input levels of HIF-1αandMyc were determined by direct Western blot(top).HIF1A expression was also silenced byspecific siRNA.(C)In vitro-translated,[35S]methionine-labeled(with asterisks)Myc or HIF-1αwas coimmu-noprecipitated with anti-Sp1antibody againstendogenous Sp1.The addition of unlabeledHIF-1αabrogated Myc coprecipitation,whereas the addition of unlabeled Myc hadno effect on HIF-1αcoprecipitation.Input:10%of lysates without immunoprecipi-tations.(D)An MSH2luciferase reporter was co-transfected into HCT116cells with vectorsexpressing p53,Myc,or⌬ODD.Cotransfec-tion with pcDNA3(−)was used as a control.The reporter activity was presented as rela-tive luciferase activity(RLU)as means+standard deviations.An MSH2mutant re-porter(MSH2-Sp1)with a mutation at a puta-tive Sp1site was also examined.The datafrom three independent experiments in du-plicate were presented as means+standarddeviations.**p<0.01.(E)MRC5cell extract was assayed for Sp1DNA binding with[32P]-labeled oligonucleo-tides containing a consensus Sp1element(Sp1*),a putative Sp1site in the MSH2pro-moter(MSH2*),or a mutated Sp1site(MSH2m*).A100-fold excess of unlabeledoligonucleotides of the consensus Sp1(Sp1),wild-type MSH2(MSH2),or mutatedMSH2(MSH2m)was used for competition.An arrowhead denotes Sp1binding and anarrow denotes Sp3binding.(F)Normoxic and hypoxic MRC5extracts were subjected to supershift with the addition of antibodies against Sp1(α-Sp1),HIF-1α(α-HIF), and Sp3(α-Sp3).Minus symbol,no antibody added.in the HIF-1α-Myc pathway.Consistently,an in vitro re-bona fide Sp1binding to oligonucleotides derived from constitution assay demonstrated that HIF-1αdomi-the MSH2promoter(Figure S6A).Moreover,in MRC5 nated Sp1binding(Figure5C).Taken together,these cell extract,Sp1binding to the consensus Sp1oligonu-findings indicate that HIF-1αdisplaces Myc bindingcleotides or the wild-type MSH2oligonucleotides was from the target gene promoters via its interaction with competed for by either of the oligonucleotides in ex-Sp1.cess,but not by those with the same mutation used in Next,we constructed a luciferase reporter driven by the reporter assay above(Figure5E).Furthermore,we the MSH2minimal promoter region that not only boundutilized specific antibodies for supershift to ascertain Myc,Sp1,HIF-1α,and p53(Figure4),but also exhibited the nature of the Sp1complex in MRC5cell extract hypoxic repression(Figure S5).In accordance with the(Figure5F).The anti-HIF-1αantibody abolished the results aforementioned,expression of p53or Myc hypoxic Sp1binding,but not the normoxic one,indicat-markedly stimulated the MSH2promoter,whereas ex-ing that Sp1recruits HIF-1αto the MSH2promoter in pression of a stable HIF-1α,⌬ODD(Huang et al.,1998),hypoxia.Taken together,these findings corroborate thepivotal role of Sp1in mediating the HIF-1α-Myc significantly repressed the reporter activity(Figure5D).Remarkably,mutation of a putative Sp1binding ele-pathway.ment in the promoter abolished Myc stimulation as wellas⌬ODD inhibition.In contrast,the p53stimulatory ef-The N-Terminal HIF-1␣Is Responsiblefect was only slightly affected.To confirm that Sp1for Inducing Genetic Instabilitybinds to this site,we performed electrophoretic mobil-Previously,we mapped out the N-terminal HIF-1αity shift assays with recombinant Sp1and showed a(amino acids1–329)that weakly interacts with Myc;HIF-1αRepresses MutSαExpression799Hence,we argue that HIF-1αdownregulates MutSαex-pression via its N-terminal interaction with Sp1.An Inverse Association between HIF-1␣and MSH2in Human Colon Cancer Tissueswith Undetectable p53To demonstrate the role of HIF-1αin MSH2expressionin human cancers,we performed immunohistochemicalstaining on80specimens of nonhereditary colon can-cers(Table S1).The four most common types of stain-ing,as shown in Figure7A,fall into two groups,p53-undetectable(indicating p53wild-type or loss of p53expression,left two columns)and p53-detectable(indi-cating p53mutations,right two columns).The formergroup includes specimens stained positive for MSH2orHIF-1α.The latter group depicts those that stained HIF-1αpositive or both HIF-1αpositive and MSH2positive.In total,35%of the cases showed MSH2positive stain-ing and HIF-1αnegative staining,whereas27.5%showed the reverse.(Figure7B).Additionally,25%ofthe cases stained both positive,and12.5%were bothnegative.Statistical analyses showed a two-tailed pvalue of<0.05by Fisher’s exact test and an odds ratioof0.325,suggesting an inverse association betweenHIF-1αand MSH2expression in these specimens.Con-sistent with the requirement of wild-type p53for MSH2inhibition by hypoxia,stratified analysis of the p53-undetectable cases yielded a p value of<0.001with anodds ratio of0.076.By contrast,in the p53-detectablegroup,no statistical significance was observed.There-fore,these results are consistent with our hypothesisthat overexpressed HIF-1αsuppresses MSH2expres-sion in a p53-dependent manner.DiscussionFigure6.The N-Terminal HIF-1αInteracts with Sp1and InducesIn this study,we ascertained the mechanism underlying Genetic Instabilityhypoxia-promoted genetic instability by focusing on (A)HIF-1α1–329,HIF-1α1–167,and⌬ODD and its R27G mutantthe role of HIF-1αin MutSαexpression.HIF-1αhas were translated in vitro and subjected to anti-Sp1immunoprecipi-tation.Input:10%of lysates without immunoprecipitations.been demonstrated as a transcriptional activator that (B)HIF-1α1–329and the above mutants were tested for their effect is responsible for the upregulation of an expanding on wild-type MSH2promoter(MSH2)and the promoter with a mu-group of hypoxia-responsive genes(Harris,2002;Sem-tation at the Sp1binding site(MSH2-Sp1).The results were pre-enza,2003).However,to gain a comprehensive under-sented as in Figure5D.**p<0.01.standing of the hypoxic response,hypoxia-downregu-(C)HIF-1α1–329and its R27G mutant were examined together withlated genes merit an equal amount of attention(Koshiji other HIF-1αmutants as specified for their ability to induce geneticinstability with the IF and OF reporters.The results were presented and Huang,2004),even though until now whether and, as in Figure1C.if so,how HIF-1αplays such a role in gene suppressionwas unclear.We demonstrate that HIF-1αacts as atranscriptional repressor of MSH2and MSH6.Although whereas further C-terminal deletion(HIF-1α1–167)the underlying mechanism conforms to the HIF-1α-Myc eliminates the binding(Koshiji et al.,2004).Consistently,pathway identified in the hypoxic activation of CDKN1A Figure6A shows that Sp1captured HIF-1α1–329,but(Koshiji et al.,2004),the outcome is just the opposite. not HIF-1α1–167.In addition,Sp1interacted with Of particular interest is an absolute requirement of wild-⌬ODD and a R27G mutant unable to bind DNA(Koshijitype p53for MSH2and MSH6repression,but not for et al.,2004).To confirm that HIF-1α1–329is sufficient CDKN1A activation(Goda et al.,2003;Koshiji et al.,2004).Apart from the HIF-1αdisplacement of Myc from to downregulate MSH2expression,we tested effects ofHIF-1α1–329on the MSH2reporter.Results in Figure the MSH2and MSH6promoters,p53binding was also 6B show that HIF-1α1–329,but not HIF-1α1–167,sig-decreased by hypoxia,in sharp contrast to the unal-nificantly inhibited the MSH2promoter activity.As ex-tered binding in the CDKN1A distal promoter(Koshiji et pected,no such effect was seen when the Sp1bindingal.,2004).p53’s crucial role was further substantiated site was mutated.Furthermore,transfection with HIF-by the finding that in TP53−/−HCT116cells,hypoxia 1α1–329also reversed the IF and OF reporter activitiesshowed no effect on Myc binding to the MutSαgene (Figure6C).Similar results were obtained with a HIF-1αpromoters,again distinct from a decreased Myc bind-1–329(R27G)mutant,but not with HIF-1α1–167.ing in the CDK1NA promoter.Therefore,it would be。