Human Mismatch Repair Gene, MLH1, Is Transcriptionally Repressed by the Hypoxia-Inducible Transcription Factors, DEC1 and DEC2

Home , MLH1, RAD51

Oncogene (2008) 27, 4200–4209 & 2008 Macmillan Publishers Limited All rights reserved 0950-9232/08 $30.00 www.nature.com/onc ORIGINAL ARTICLE Human mismatch repair gene, MLH1, is transcriptionally repressed by the hypoxia-inducible transcription factors, DEC1 and DEC2

H Nakamura1,5, K Tanimoto1, K Hiyama1, M Yunokawa1, T Kawamoto2, Y Kato2, K Yoshiga3, L Poellinger4, E Hiyama5 andM Nishiyama 1,6

1Department of Translational Cancer Research, Research Institute for Radiation Biology and Medicine, Hiroshima University, Hiroshima, Japan; 2Department of Dental and Medical Biochemistry, Hiroshima University, Hiroshima, Japan; 3Department of Molecular Oral Medicine and Maxillofacial Surgery, Graduate School of Biomedical Sciences, Hiroshima University, Hiroshima, Japan; 4Cell and Molecular Biology, Medical Nobel Institute, Karolinska Institutet, Stockholm, Sweden; 5Natural Science Center for Basic Research and Development, Hiroshima University, Hiroshima, Japan and 6Saitama Medical University International Medical Center, Saitama, Japan

Tumor hypoxia has been reported to cause a functional Introduction loss in DNA mismatch repair (MMR) system as a result of downregulation of MMR genes, although the precise Hypoxia is a common feature in many solidtumors and molecular mechanisms remain unclear. In this study, we the microenvironment is now recognizedas a key factor focused on the downregulation of a key MMR gene, linkedto the biologically aggressive phenotypes and MLH1, and demonstrated that hypoxia-inducible their resistance to chemotherapeutic agents andirradia- transcription repressors, differentiated embryo chondro- tion therapies (Teicher, 1994; Brown andGiaccia, 1998; cytes (DEC1 and 2), participated in its transcriptional Cairns et al., 2006). Extensive studies of molecular regulation via their bindings to E-box-like motif(s) in mechanisms have shown that transcription factor MLH1 promoter region. In all cancer cell lines examined, hypoxia-inducible factor-1a (HIF-1a) is a key regulator hypoxia increased expression of DEC1 and 2, known as of hypoxic reaction; these studies have led to a better hypoxia-inducible genes, but decreased MLH1 expression understanding of the mechanisms of HIF-1a activation in an exposure time-dependent manner at both the mRNA andthe subsequent alteration of gene expressions under and protein levels. Co-transfection reporter assay revealed hypoxic conditions (Harris, 2002; Denko et al., 2003; that DEC1 and, to greater extent, DEC2 as well as Semenza, 2003; Poellinger andJohnson, 2004). hypoxia-repressed MLH1 promoter activity. We further Recently, revealing findings have reported that found that the action was remarkably inhibited by hypoxia can reduce expression of several DNA repair trichostatin A, and identified a possible DEC-response genes—MLH1, RAD51, BRCA1 and MSH2—resulting element in the MLH1 promoter. In vitro electrophoretic in genomic instability in several cancer cell lines gel mobility shift and chromatin immunoprecipitation (Mihaylova et al., 2003; Bindra et al., 2004, 2005; assays demonstrated that DEC1 or 2 directly bounds to Koshiji et al., 2005; Bindra and Glazer, 2006, 2007). the suggested element, and transient transfection assay Since the human mismatch repair (MMR) system plays revealed that overexpression of DEC2 repressed endo- a critical role in the maintenance of genomic integrity, genous MLH1 expression in the cells. Hypoxia-induced the mechanisms of transcriptional repression, especially DEC may impair MMR function through repression of in MLH1 and MSH2 genes, are of key importance in MLH1 expression, possibly via the histone deacethylase- tumor biology: germ line mutations in MLH1 (B50%) mediated mechanism in cancer cells. and MSH2 (B40%) exist in approximately half of all Oncogene (2008) 27, 4200–4209; doi:10.1038/onc.2008.58; hereditary non-polyposis colorectal cancer patients publishedonline 17 March 2008 (Hoeijmakers, 2001; Peltoma¨ ki, 2001). Under hypoxic conditions, the cellular DNA repair function becomes Keywords: MLH1; DEC1; DEC2; hypoxia; HIF-1 impaired, which causes hypermutability to DNA damage (Reynolds et al., 1996; Yuan et al., 2000). These findings strongly suggest that tumor hypoxia probably causes loss of genomic stability through suppression of MMR functions, andthat defects of MMR function may dramatically increase mutation rates. Correspondence: Dr K Tanimoto, Department of Translational These studies have also suggested that a transcription Cancer Research, Research Institute for Radiation Biology and factor, E2Fs, p130, HIF-1a, SP-1 or Myc/Max system Medicine, Hiroshima University, Kasumi 1-2-3, Minami-ku, may participate in the mechanisms of downregulation of Hiroshima 734-8551, Japan. E-mail: [email protected] BRCA1, RAD51, MSH2 or MLH1, but details remain Received4 September 2007; revised17 December 2007; accepted15 unclear. Among numerous hypoxia-inducible genes, February 2008; publishedonline 17 March 2008 differentiatedembryo chondrocyte(DEC) 1 and2 may hMLH1 transcriptional regulation by DEC H Nakamura et al 4201 be the most likely candidates (Ivanova et al., 2001; MLH1 (from À1653 to À4) into a luciferase reporter Miyazaki et al., 2002). DEC1 and2 have been reported plasmid, pGL3-Basic vector, designated as a pGL- to participate in the transcriptional repression MLH1Pro1.65 (Figure 2a). Transient transfection into of PPARG, PER, STAT1 andthemselves via E-box or HepG2 revealedthat pGL-MLH1Pro1.65 has strong other motifs in their promoter regions, which results in promoter activity in comparison with an empty plasmid the regulation of adipogenesis, circadian rhythm, vector pGL3-Basic under normoxic conditions. Since immune system or carcinogenesis (Honma et al., 2002; the pGL3-Basic vector itself has a lot of hypoxia Yun et al., 2002; Ivanova et al., 2007). DEC1 (also response element consensus sequences (according to known as BHLHB2/STRA13) was originally identified Promega, Madison, WI, USA), background reporter as the gene expressed in cyclic AMP-dependently activities were increased under hypoxic conditions (data differentiated embryo chondrocytes that encode a basic not shown). Therefore, MLH1 promoter was swapped helix-loop-helix (bHLH) transcription factor (Shen into pGL4.10 plasmidvector, in which consensus et al., 1997); DEC2 (also known as BHLHB3/SHARP1) sequences for transcription factors were reduced from was identified from a human expression sequence tag backbone sequences (according to Promega), and database as a member of DEC subfamily (Fujimoto transient transfection experiments were performed. As et al., 2001). we expected, MLH1 promoter reporter was down- In the present study, we focused on the mechanisms of regulated under hypoxic conditions, suggesting that downregulation of MLH1, anddemonstratedfor the the promoter region containedhypoxia response repres- first time that the hypoxia-inducible transcription sion sequences (Figure 2b). Interestingly, this promoter repressors DEC1 and2 participatedin the transcrip- activity was repressedby co-transfection with DEC1 or tional regulation through their bindings to E-box-like 2 expression plasmid vector in a dose-dependent motif(s) in MLH1 promoter region. These findings may manner, andthe repression of MLH1 promoter activity contribute to a better understanding of the biological was notable when DEC2 was co-transfected(Figure 2c). functions of tumor hypoxia, basedon the novel As histone deacetylase (HDAC)-dependent mecha- proposal that hypoxia-inducible DEC can impair nisms hadbeen previously suggested(Sun andTaneja, MMR function through repression of MLH1 expres- 2000), trichostatin A (TSA) treatments remarkably sion, andmay subsequently cause genomic instability in canceledthe repression of MLH1 promoter activity by cancer cells. DEC in a treatment dose-dependent manner without any detectable cytotoxicity (Figure 2d). Moreover, mutant type of DEC1 which had DNA-binding domain but lackedmost of functional domains(Li et al., 2003; Results Sato et al., 2004) failedto repress MLH1 promoter activities, even enhancedthem, suggesting that just a MLH1 expression at both protein and mRNA levels competitive occupancy on the promoter was not under hypoxic conditions sufficient to explain the repression (Figure 2e). We first examinedwhether hypoxia decreasedMLH1 expression in cancer cells. HepG2 cells were collected after incubation under normoxic or hypoxic conditions Response element to DEC on MLH1 promoter region for various periods. Immunoblotting using whole cell To identify a response element to DEC in the MLH1 extracts revealedthat hypoxia decreasedMLH1 protein promoter region, we constructeda series of deletion up to 48 h in an exposure time-dependent manner, mutants of MLH1 promoter reporter (Figure 3a). The unlike the stable expression levels of b-actin (Figure 1a). luciferase reporter assays for co-transfection with The hypoxic induction of HIF-1a was confirmedat well- pcDNA (vector only) revealedthat MLH1 promoter detected protein levels as well as that of DEC1, a known hadseveral putative positive (from À556 to À274)- hypoxia-inducible transcriptional repressor, whereas andnegative (from À893 to À557)-regulatory regions. aryl hydrocarbon receptor nuclear translocator (Arnt), We also foundthat promoter activity of pGL- also known as HIF-1b, and b-actin constitutively MLH1Pro0.27 was almost identical to that of expressed(Figure 1a). Next, MLH1 mRNA levels were pGL-MLH1Pro1.65, indicating that the region from evaluatedalong with hypoxia-inducible genes, DEC1 À273 to À4 probably contains critical regulatory 2 and . Real-time RT-PCR analyses demonstrated that regions. Furthermore, co-transfection with DEC-expres- MLH1 mRNA level alone decreased from 6 to 48 h with sing pcDNAs showedthat all of the reporter activities hypoxic treatment (Figure 1b). In contrast to MLH1, were significantly repressedby DEC, suggesting that the expressions of DEC1 and 2 increasedunder hypoxic region from À273 to À4 is the most likely site containing conditions, despite of the relatively short duration of the DEC-response element (Figure 3a). DEC2 upregulation (Figure 1b). We further examined We therefore constructedfour mutant reporters in mRNA expression of these genes in the other cell lines which several nucleotides were substituted in the andfoundthe similar expression patterns (Figure 1c). putative E-box motifs (Figure 3b): Three mutants (MT1–3) showedstronger activity than that of the Promoter activitiesof MLH1 and DEC wild-type promoter reporter (Figure 3c). Co-transfec- To clarify the mechanisms of the decreased MLH1 tion experiments with DEC indicated that MT1 and mRNA level, we next subclonedthe 5 0 region of human MT2 showedresistance to the repression causedby

Oncogene hMLH1 transcriptional regulation by DEC H Nakamura et al 4202

Figure 1 Hypoxia decreasedMLH1expression andincreasedhypoxia-inducible factor-1 a (HIF-1a), differentiated embryo chondrocyte (DEC) 1 and 2 expression in cancer cell lines. Protein levels of MLH1, HIF-1a, Arnt, DEC1 and b-actin (a), and mRNA levels of MLH1, DEC1 and DEC2 expressedin HepG2 ( b) andHeLa, MCF-7 andHSC-2 ( c) cells after indicated periods of hypoxic treatment analysedby immunoblotting or real-time RT-PCR method.Relative mRNA levels were calculatedas the ratio to that of ACTB, andeach bar represents the mean þ s.d. for at least three independent experiments. *Po0.05 and X0.01, **Po0.01.

DEC, whereas MT3 andMT4 were signiﬁcantly Direct binding of DEC to the response element containing repressed, as strongly as the wild type was (Figure 3c). E-box motif on MLH1 promoter These results suggestedthat DEC might repress MLH1 To demonstrate that DEC directly binds to the response expression through their bindings to the indicated region elements, we performedan electrophoretic gel mobility containing putative E-box motifs. shift assay (EMSA) with 32P-labeledprobes containing

Oncogene hMLH1 transcriptional regulation by DEC H Nakamura et al 4203

Figure 2 Hypoxia or DEC transcription factors repressedpromoter activities of MLH1 in HepG2 cells. (a) The 50 region (nt À1653 to À4) of MLH1 was subclonedinto pGL3 Basic plasmidvector. ( b) The MLH1 promoter reporter was transiently transfectedinto HepG2 cells, and promoter activities were evaluated under normoxic or hypoxic conditions. (c) Various amounts of DEC1 (hatched bar) or DEC2 (closedbar) expression vectors were co-transfectedwith MLH1 promoter luciferase reporter. Effects of trichostatin A (TSA) treatment (d)orDEC1 mutant (e, stripedbar) on MLH1 promoter were evaluatedby co-transfection assay. Relative luciferase activities were calculatedas the ratio to activity of pRL-SV40. Each bar represents the mean þ s.d. for at least three independent experiments. *Po0.05 and X0.01, **Po0.01.

DEC-response elements in the MLH1 promoter from antibodies for DEC1 or 2. The observed intensities of À69 to À47. DEC1 and2 were synthesizedusing in vitro shifted bands indicated that the binding activity transcription/translation system, andprotein amounts of DEC2 to this probe was much stronger than that of were equally adjusted by calculation of incorporated DEC1. 35S-labeledmethionines. A 32P-labeledprobe was in- Chromatin immunoprecipitation (ChIP) assay was cubatedwith synthesizedproteins andsubjectedto then performedafter incubation of HepG2 cells in electrophoresis. EMSA showedthat both DEC1 and2 normoxia or hypoxia for 24 h to examine the binding of specifically boundto these elements (Figure 4a), DEC1 endogenous DEC to response elements in MLH1 or 2 probe-specific DNA-binding complexes had shifted promoter. Real-time PCR clearly demonstrated that andthe complexes formedwere competedout by immunoprecipitation of the chromatin fragment con- pre-incubation with the nonlabeledprobes or specific taining the DEC-response element in MLH1 promoter

Oncogene hMLH1 transcriptional regulation by DEC H Nakamura et al 4204

Figure 3 DEC repressedpromoter activity of MLH1 via the E-box motifs on its promoter region. (a) Comparative analysis of transcriptional activity using 50 deletion mutants of MLH1 promoter. A series of deletion mutants of MLH1 promoter is shown in the schematic (left). Transcriptional activities of the deletion mutants of MLH1 promoter were evaluatedby luciferase assay after co-transfection with or without DEC-expressing vectors (right). (b) Nucleotide sequence of detailed DEC-response elements near the transcription start site. Substituted nucleotides in mutants are indicated above the wild-type sequence. In the open box, consensus E-box-like motif is indicated. The lower bar shows the sequence of oligo-probe for electrophoretic gel mobility shift assay (EMSA). (c) Comparative analysis of transcriptional activity using nucleotide substituted mutants of MLH1 promoter. Transcriptional activities of MLH1 promoter mutants were evaluatedas describedabove. Each bar represents the mean þ s.d. for at least three independent experiments. **Po0.01.

was increasedin the hypoxic samples pre-incubatedwith DEC on MLH1 at each cellular level, we then performed anti-DEC1 antibody, indicating that DEC1 specifically immunostaining in HepG2 transiently transfected boundto the elements (Figure 4b). with DEC2. Double staining with anti-MLH1 and -DEC2 showedthat MLH1 andDEC2 were compensa- tively expressedin each cell: MLH1 expression was Expression of DEC and endogenous MLH1 protein significantly decreased in the DEC2 overexpressed To confirm the function of DEC on MLH1 at the cells, while high expression levels of MLH1 were cellular level, we investigatedendogenous MLH1 maintainedin cells without DEC2 expression protein in cells overexpressing DEC. Immunoblotting (Figure 5b). Next, we performedknockdownassay for analysis using whole cell extract preparedfrom HepG2 HIF1A, DEC1 or DEC2, to estimate how HIF-1–DEC cells transiently transfectedwith DEC1 or 2 demon- pathway contribute to the MLH1 expressions. As a stratedthat MLH1 protein decreasedandinversely result, transient transfection of specific siRNA for associatedwith the expression levels of DEC HIF1A in HSC-2 representedmore than 80% reduction (Figure 5a). To confirm the suggestedfunction of of HIF1A expression comparedto that of nonspecific

Oncogene hMLH1 transcriptional regulation by DEC H Nakamura et al 4205 Discussion

Hypoxic reaction has been clearly shown to involve alterations in gene transcription (Harris, 2002; Denko et al., 2003; Semenza, 2003; Poellinger andJohnson, 2004), andHIF-1 is well known as the pivotal factor that regulates cellular responses to hypoxia via transactivation of a variety of genes. We previously demon- stratedthat DEC1 and2 were transcriptionally activatedby HIF-1, suggesting their crucial roles in HIF-1-mediated cellular hypoxic reaction (Miyazaki et al., 2002). The mechanisms of the activation of HIF-1 andthe subsequent transactivation of various genes have also been intensively studied, which has promoteda better understandingof the genetic and molecular basis underlying intricate hypoxic reactions of cells (Harris, 2002; Denko et al., 2003; Semenza, 2003; Poellinger andJohnson, 2004). However, little is known about the precise mechanisms andthe factors causing transcriptional repression under hypoxia, despite their critical roles in cellular hypoxic reaction. In fact, decreased expression of DNA repair genes under hypoxia anda possible association with genomic instability were recently shown (Mihaylova et al., 2003; Bindra et al., 2004, 2005; Koshiji et al., 2005; Bindra and Glazer, 2006, 2007). The analysis of molecular mechanisms is of key importance in understanding cellular hypoxic reaction and its role in tumor biology, so we attemptedto clarify the molecular mechanisms: we foundthat DEC1 and2 strongly repress the promoter activity of MLH1, possibly via an HDAC-dependent mechanism but not by just a competitive occupancy on the promoter. We further Figure 4 DEC directly bound to the DEC-response elements identified a possible DEC-response element on the containing E-box motif on MLH1 promoter. (a) The electro- MLH1 promoter region, andconfirmedthe direct phoretic gel mobility shift assay (EMSA) was performedas binding of DEC to that element. Forced expressions of described in the ‘Materials and methods’. Specificities of their bindings (asterisk (*) for DEC1 complex) were confirmed by pre- both DEC1 and2 efficiently repressed MLH1 promoter incubation with nonlabeledprobes or specific antibodiesfor DEC1 andexpression, andknockdownof DEC2 by siRNA or 2. NS, nonspecific band. (b) The chromatin immunoprecipita- significantly attenuatedhypoxic repression of the MLH1 tion (ChIP) assay was performedas describedinthe ‘Materials and expression. On the other hand, while knockdown of methods’ using anti-DEC1, anti-DEC2 or anti-immunoglobulin G (IgG). Relative amounts of precipitatedDNA fragments were HIF1A also causeddisappearance of hypoxic repression evaluatedby real-time PCR, andcalculatedusing HepG2 genomic of MLH1, DEC1 knockdown failed to attenuate the DNA as a standard. Each bar represents the mean þ s.d. for at least MLH1 repression under hypoxic conditions, since three independent experiments. *Po0.05 and X0.01. decreased expression of DEC1 resultedin increased DEC2 expression as previously reported(Li et al., 2003). Taken together, these results suggestedthat HIF-1– DEC pathway was one of the important mechanisms. (NS) siRNA as well as significant repression of DEC1 Very recently, several mechanisms were suggestedto and 2, andhypoxic repression of MLH1 disappeared participate in regulation of DNA repair genes, including (H/N ratios of siNS:siHIF1A ¼ 0.62:0.94) (Figure 5c). E2F4/p130, HIF-1a/SP-1 andMyc/Max system. Bindra Interestingly, DEC1 knockdown represented a little and Glazer (2007) demonstrated a dynamic shift in increasedexpression of MLH1 under both normoxic occupancy from activating c-Myc/Max to repressive andhypoxic conditions.Since DEC1 represses Mid/Max and Mnt/Max complexes at the proximal DEC2 expression (Li et al., 2003), DEC1 knockdown promoters of MLH1 and MSH2 by using series of ChIP resultedin increased DEC2 expression andpersistence assays, but did not determine repressive activities of of the hypoxic repression of MLH1 (H/N ratios those complexes on the promoters. Although it is well of siDEC1 ¼ 0.69). On the other hand, DEC2 known that both Myc/Max andDEC bindto E-box knockdown strikingly increased expression of MLH1 motif to regulate gene transcription, our experiments under hypoxic condition, indicating complete attenua- using mutant type of DEC1 that hadDNA-binding tion of hypoxic repression of MLH1 (H/N ratios of domain but lacked most of functional domains failed to siDEC2 ¼ 1.01). repress MLH1 promoter activities, even enhancedthem,

Oncogene hMLH1 transcriptional regulation by DEC H Nakamura et al 4206

Figure 5 DEC decreased endogenous MLH1 expression. (a) Immunoblotting analysis was performedusing whole cell extract preparedfrom HepG2 cells transiently transfectedwith DEC1 or 2. Anti-MLH1, anti-FLAG or anti- b-actin was usedfor specific detection of each protein. (b) Immunostaining analysis with anti-MLH1 andanti-DEC2 was performedusing HepG2 transiently transfected with DEC2. (i) 4-6-Diamidino-2-phenylindole (DAPI), (ii) FITC (anti-DEC2), (iii) Rhodamine-red (anti-MLH1), (iv) mergedpictures (bar 10 mm). (c) Knockdown assays for HIF1A, DEC1 and DEC2 were performedusing HSC-2 cells. Expression levels of HIF1A, MLH1, DEC1 and DEC2 were evaluatedas Figure 1. Statistical significances were calculatedby Student’s t-test between the nonspecific and each knocked down cells under normoxic or hypoxic conditions respectively. Each bar represents the mean þ s.d. for at least three independent experiments. *Po0.05 and X0.01, **Po0.01. (d) Hypothetical model of hypoxic malignant cycles.

suggesting that just a competitive occupancy on the functional roles of hypoxia in malignant phenotypes of promoter was not sufficient to explain the repression, various tumors. but HDAC-dependent repressive activities of DEC Our data also suggested that DEC2 might repress transcription factors were important. Since the loss of MLH1 stronger than DEC1 does, which would be an functions of MLH1 is thought to be a significant cause important evidence of diversification of DEC functions. of the complete inactivation of MMR (Hoeijmakers, It has been suggestedthat DEC participates also in 2001; Peltoma¨ ki, 2001)—which may leadto carcino- adipogenesis, circadian rhythm, immune system genesis, tumor progression andemergence of resistance or carcinogenesis through transcriptional repressions to anticancer therapies—these new findings, we believe, of several genes via E-box or other motifs in their could contribute to a better understanding of the promoter regions (Honma et al., 2002; Yun et al., 2002;

Oncogene hMLH1 transcriptional regulation by DEC H Nakamura et al 4207 Ivanova et al., 2007). Their differential effects on MLH1 dish) for 12 h, and then the cells were incubated under couldbe explainedin part by varying specificity to the normoxic or hypoxic conditions for 24 h. Cells were then element sequence identified as the binding site, which harvestedandstoredat À80 1C until use. Total RNA was contains a sequence motif of AACGTG with one preparedfrom frozen cell pellets by using Qiagen RNeasy mini nucleotide difference from canonical E-box motif kit (Qiagen) according to the manufacturer’s instruction. (CACGTG). In this study, we found that mRNA expression of DEC1 increasedfor more than 72 h under Reverse transcription–polymerase chain reaction In total 2 mg of total RNA extractedfrom each cell line were hypoxia, while that of DEC2 only temporarily increased. reverse-transcribedusing High-Capacity cDNA Archive Kit Even so, DEC2 was shown to have much stronger (Applied Biosystems, Foster City, CA, USA). Two-hundredth affinity to the MLH1 promoter. These findings led us to aliquot of the cDNA was subjectedto real-time RT-PCR using hypothesize that DEC2 couldbe the initiator of the TaqMan Gene Expression Assays (AppliedBiosystems) for event, whereas DEC1 might act on the maintenance of HIF1A, BHLHB2 (DEC1), BHLHB3 (DEC2) and MLH1, and the downregulated level of MLH1 expression. This Pre-DevelopedTaqMan Assay Reagents (AppliedBiosystems) hypothesis may be supportedin a part by one report for ACTB as an internal control. More than three independent showing that DEC1 transcriptionally repressed DEC2 measurements were averagedandrelative gene expression expression in an autofeedback system, suggesting their levels were calculatedas a ratio to ACTB expression of each hierarchical functions (Li et al., 2003). In the present cell line. study, we did not detect an endogenous DEC2 protein induction as well as other investigators, and did not Immunoblot analysis observe DEC2 binding to MLH1 promoter in vivo using To analyse protein expression, whole cell extracts were preparedfrom culturedcells with or without hypoxic treat- ChIP assay. On the other hand, knockdown experiments ment as previously described (Tanimoto et al., 2000). In total clearly showeda significant role of DEC2 in regulation 25 mg of protein was blottedonto nitrocellulose filters of the MLH1. Taken together, it might be tough to following SDS–polyacrylamide gel electrophoresis. Anti- detect endogenous DEC2 protein in both experiments FLAG (Sigma), anti-MLH1, anti-HIF-1a, anti-Arnt (BD due to an antibody activity, but DEC2 protein actually Pharmingen, San Diego, CA, USA) or anti-b-actin (Sigma) functions on MLH1 regulations. The diverse roles of were used as primary antibodies, diluted 1:5000, 1:2000, DEC1 and2 are now being intensively investigatedin 1:1000, 1:2000 or 1:5000, respectively. A 1:2000 dilution of our laboratory. anti-mouse immunoglobulin G (IgG) horseradish peroxidase In conclusion, we demonstrated here that the hypox- conjugate (GE Healthcare, Tokyo, Japan) was usedas a ia-inducible transcription repressors DEC1 and 2 secondary antibody. Immunocomplexes were visualized using the enhancedchemiluminescence reagent ECL Plus (GE participate in transcriptional regulation of the MLH1 Healthcare). via their bindings to an E-box-like motif in the MLH1 promoter region. Hypoxia-induced DEC1 and Plasmid constructions 2, we think, probably play very important roles in the The 1.65-kb DNA fragment (nucleotide positions from À1653 transcriptional downregulation of genes under hypoxia, to À4 when transcriptional start site is designated as at þ 1) andthe HIF-1–DEC pathway as well as other pathways including the 50 region of MLH1 gene was amplifiedby PCR may impair MMR function through the repression from a HepG2 genomic DNA andsubclonedinto Nhe I and of MLH1 expression, subsequently causing genomic Xho I sites of a luciferase reporter plasmidpGL3-Basic or instability in cancer cells (Figure 5d). pGL4.10 (Promega), andthe construct was designatedas pGL-MLH1 Pro1.65. A series of 50 deletion mutant of pGL-MLH1 Pro was constructedby PCR methodusing internal specific primer sets with pGL-MLH1 Pro1.65 as a Materials and methods template. Base-exchangedmutants of putative E-box sites in pGL-MLH1 Pro0.27 were generatedby PCR-basedsite- Chemicals directed mutagenesis as previously reported (Tanimoto et al., All chemicals were of analytical grade and were purchased 2003). Details of expression plasmidvectors of DEC1 from Wako Pure Chemicals (Osaka, Japan) or Sigma (pcDNA-DEC1, p3xFLAG-CMV-DEC1 or pcDNA-DEC1 (St Louis, MO, USA). 1–139) andDEC2 (pcDNA-DEC2) were previously described (Kawamoto et al., 2004; Sato et al., 2004). pcDNA-FLAG- Cell lines and RNA preparation DEC2 was constructedby swapping DEC2 cDNA fragment of Human cancer cell lines usedwere as follows: a hepatoma line, pcDNA-DEC2 with the pcDNA-FLAG (kindly provided by HepG2 andan oral squamous cell carcinoma line, HSC-2 Dr Igarashi). (the Japanese Cancer Research Resource Bank); a cervical adenocarcinoma line, HeLa and a breast adenocarcinoma line, Luciferase reporter assay MCF-7 (American Type Culture Collection). For gene Transient transfection was performedas follows: pGL-MLH1 expression analyses, cells (2–4 Â 105 per 10 cm diameter dish) Pro (0.3 mg per 15-mm well) with pcDNA-FLAG, p3xFLAG- were culturedundernormoxic (21% O 2) or hypoxic (1% O2) CMV-DEC1 or pcDNA-FLAG-DEC2 (0.001–0.1 mg per conditions for various incubation times (6, 12, 24, 48 or 72 h) 15-mm well) were mixedwith 0.8 ml of Trans-IT LT1 in a hypoxic chamber (Hirosay Corp., Hiroshima, Japan). For Transfection Reagent (Mirus). Renilla-luciferase vector knockdown analyses, HIF1A, DEC1, DEC2 or NS siRNA (pRL-SV40, 1.0 ng per 15-mm well) (Promega) was usedas a (Qiagen Inc., Valencia, CA, USA) was transfectedwith transfection efficacy control. Cells were incubatedunder TransIT-siQUEST Transfection Reagent (Mirus Corporation, normoxic or hypoxic conditions for 36–48 h after transfection Madison, WI, USA) in HSC-2 (1 Â 106 per 10 cm diameter prior to analysis of luciferase reporter activity. Using the

Oncogene hMLH1 transcriptional regulation by DEC H Nakamura et al 4208 HDAC inhibitor, TSA, treatments were started(final concen- amounts were calculatedas a ratio to amplicons using HepG2 trations 10 or 100 ng/ml) 24 h before harvesting cells. genomic DNA. Luciferase luminescence was measuredas previously described (Tanimoto et al., 2003). Immunostaining HepG2 cells grown on cover slips were transiently transfected Electrophoretic gel mobility shift assay with DEC2 expression plasmid, pcDNA-DEC2. After incuba- Double-stranded oligoprobes containing consensus DEC- tion for 24 h, immunostaining was performedwith anti-DEC2 binding sequences in the MLH1 promoter from À69 (1:100) or anti-MLH1 (1:100) as primary antibodies, fluore- to À47 were synthesizedas follows: sense, 5 0-AAGAAC scein isothiocyanate (FITC)-conjugatedgoat anti-rabbit Ig’s GTGAGCACGAGGCACTGGG-30 andantisense, 5 0-CA (1:100) (BioSource, Camarillo, CA, USA) or Rhodamine- GTGCCTCGTGCTCACGTTCTTGG-30 andlabeledwith conjugatedsheep anti-mouse Ig’s (1:100) (Chemicon, [a-32P]-deoxycytidine triphosphate. Adjusted equal amounts Temecula, CA, USA) as secondary antibody. Nuclei of in vitro translatedDEC1 or 2 were incubatedwith 200 pmol were stained with 4-6-diamidino-2-phenylindole (DAPI). of labeledprobe in 20 ml of reaction mixture for 30 min at room Subcellular distribution of fluorescence was examined using a temperature. A 100-foldexcess amount of unlabeledprobes for Zeiss Axiovert 135 microscope with an FITC-filter set, competition or 2.5 ml of anti-DEC1 or 2 polyclonal antibody epifluorescence with illumination from a Gixenon burner (Carl (Kawamoto et al., 2004) for supershift was pre-incubatedfor Zeiss Jena GmbH, Jena, Germany). 30 min at room temperature before the addition of hot-labeled probes. The reaction mixtures were then loaded onto 5% Statistical analysis polyacrylamide gels and were run for 4 h at 4 1C. Resulting gels All of the statistical tests were performedusing StatView were dried and visualized using BAS2000. version 5.0 software (SAS Institute Inc., Cary, NC, USA), and Student’s t-test was usedto determinethe P-value. Chromatin immunoprecipitation assay The ChIP assay was performedusing EZ ChIP Chromatin Acknowledgements Immunoprecipitation Kit (Upstate USA Inc., Charlottesville, VA, USA) according to the manufacturer’s instruction. Anti- We thank Dr H Eguchi (Saitama Medical University), Dr S DEC1 or 2 rabbit polyclonal antibody (Kawamoto et al., 2004) Tashiro, Dr N Oue, Dr K Miyazaki (Hiroshima University), was usedfor a specific precipitation, andanti-IgG mouse Dr Y Makino (Asahikawa Medical College) and Dr K monoclonal antibody was used as a negative control for an Igarashi (Tohoku University) for their helpful contributions immunoprecipitation. The PCR primer set was synthesizedto to this work. We also thank Ms I Fukuba, Ms K Nukata, Ms encompass the candidate DEC-binding sites in MLH1 C Oda, Ms M Wada and Ms M Sasaki for their technical and promoter as follows: forward, 50-ATCAATAGCTGCCGCT secretarial support. A part of this work was carriedout at the GAA-30 andreverse, 5 0-CTCGTGCTCACGTTCTTCCT-30, Analysis Center of Life Science, Hiroshima University. This andthe probe (#42) was selectedfrom Universal Probe work was supportedby Grants-in-Aidfor Exploratory Library (UPL, Roche Diagnostics, Tokyo, Japan). Real-time Research from Japan Society for the Promotion of Science PCR was performedusing the 1/30 volume of precipitates. andGrant-in-Aidfor Young Scientists from the Ministry of Three independent measurements were averaged and relative Education, Culture, Sports, Science and Technology of Japan.

References

Bindra RS, Gibson SL, Meng A, Westermark U, Jasin M, Pierce AJ Harris AL. (2002). Hypoxia-a key regulatory factor in tumour growth. et al. (2005). Hypoxia-induced down-regulation of BRCA1 expres- Nat Rev Cancer 2: 38–47. sion by E2Fs. Cancer Res 65: 11597–11604. Hoeijmakers JHJ. (2001). Genome maintenance mechanisms for Bindra RS, Glazer PM. (2006). Repression of RAD51 gene expression preventing cancer. Nature 411: 366–374. by E2F4/p130 complexes in hypoxia. Oncogene 26: 1–10. Honma S, Kawamoto T, Takagi Y, Fujimoto K, Sato F, Noshiro M Bindra RS, Glazer PM. (2007). Co-repression of mismatch repair gene et al. (2002). Dec1 andDec2 are regulators of the mammalian expression by hypoxia in cancer cells: role of the Myc/Max network. molecular clock. Nature 419: 841–844. Cancer Lett 252: 93–103. Ivanova AV, Ivanova SV, Danilkovitch-Miagkova A, Lerman MI. Bindra RS, Schaffer PJ, Meng A, Woo J, Ma˚ seide K, Roth ME et al. (2001). Regulation of STRA13 by the von Hippel–Lindau tumor (2004). Down-regulation of Rad51 and decreased homo- suppressor protein, hypoxia, andUBC9/ubiquitin proteasome logous recombination in hypoxic cancer cells. Mol Cell Biol 24: degradation pathway. J Biol Chem 276: 15306–15315. 8504–8518. Ivanova SV, Salnikow K, Ivanova AV, Bai L, Lerman MI. (2007). Brown JM, Giaccia AJ. (1998). The unique physiology of solidtumors: Hypoxic repression of STAT1 andits downstream genes by a opportunities (andproblems) for cancer therapy. Cancer Res 58: pVHL/HIF-1 target DEC1/STRA13. Oncogene 26: 802–812. 1408–1416. Kawamoto T, Noshiro M, Sato F, Maemura K, Takeda N, Nagai R Cairns R, Papandreou I, Denko N. (2006). Overcoming physiologic et al. (2004). A novel autofeedback loop of Dec1 transcription barriers to cancer treatment by molecularly targeting the tumor involvedin circadianrhythm regulation. Biochem Biophys Res microenvironment. Mol Cancer Res 4: 61–70. Commun 313: 117–124. Denko NC, Fontana LA, Hudson KM, Sutphin PD, Raychaudhuri S, Koshiji M, To KKW, Hammer S, Kumamoto K, Harris AL, Modrich P Altman R et al. (2003). Investigating hypoxic tumor physiology et al. (2005). HIF-1alpha induces genetic instability by trans- through gene expression patterns. Oncogene 22: 5907–5914. criptionally downregulating MutSa expression. Mol Cell 17: 793–803. Fujimoto K, Shen M, Noshiro M, Matsubara K, Shingu S, Honda K Li Y, Xie M, Song X, Gragen S, Sachdeva K, Wan Y et al. (2003). et al. (2001). Molecular cloning andcharacterization of DEC2, a DEC1 negatively regulates the expression of DEC2 through binding new member of basic helix-loop-helix proteins. Biochem Biophys Res to the E-box in the proximal promoter. J Biol Chem 278: Commun 280: 164–171. 16899–16907.

Oncogene hMLH1 transcriptional regulation by DEC H Nakamura et al 4209 Mihaylova VT, Bindra RS, Yuan J, Campisi D, Narayanan L, helix-loop-helix protein DEC1 expressedin differentiatedhuman Jensen R et al. (2003). Decreasedexpression of the DNA mismatch embryo chondrocytes. Biochem Biophys Res Commun 236: 294–298. repair gene Mlh1 under hypoxic stress in mammalian cells. Mol Cell Sun H, Taneja R. (2000). Stra13 expression is associatedwith growth Biol 23: 3265–3273. arrest andrepresses transcription through histone deacetylase Miyazaki K, Kawamoto T, Tanimoto K, Nishiyama M, Honda H, (HDAC)-dependent and HDAC-independent mechanisms. Proc Kato Y. (2002). Identification of functional hypoxia response Natl Acad Sci USA 97: 4058–4063. elements in the promoter region of the DEC1 andDEC2 genes. Tanimoto K, Makino Y, Pereira T, Poellinger L. (2000). Mechanism J Biol Chem 277: 47014–47021. of regulation of hypoxia-inducible factor-1a by von Hippel–Lindau Peltoma¨ ki P. (2001). Deficient DNA mismatch repair: a common tumor suppressor protein. EMBO J 19: 4298–4309. etiologic factor for colon cancer. Hum Mol Genet 10: 735–740. Tanimoto K, Yoshiga K, Eguchi H, Kaneyasu M, Ukon K, Kumazaki Poellinger L, Johnson RS. (2004). HIF-1 andhypoxic response: the T et al. (2003). Hypoxia-inducible factor-1a polymorphisms plot thickens. Curr Opin Genet Dev 14: 81–85. associatedwith enhancedtransactivation capacity, implying clinical Reynolds TY, Rockwell S, Glazer PM. (1996). Genetic instability significance. Carcinogenesis 24: 1779–1783. induced by the tumor microenvironment. Cancer Res 56: 5754–5757. Teicher BA. (1994). Hypoxia anddrugresistance. Cancer Metastasis Sato F, Kawamoto T, Fujimoto K, Noshiro M, Honda KK, Honma S Rev 13: 139–168. et al. (2004). Functional analysis of the basic helix-loop-helix Yuan J, Narayanan L, Rockwell S, Glazer PM. (2000). Diminished transcription factor DEC1 in circadian regulation. Interaction with DNA repair andelevatedmutagenesis in mammalian cells exposed BMAL1. Eur J Biochem 271: 4409–4419. to hypoxia andlow pH. Cancer Res 60: 4372–4376. Semenza GL. (2003). Targeting HIF-1 for cancer therapy. Nat Rev Yun Z, Maecker HL, Johnson RS, Giaccia AJ. (2002). Inhibition of Cancer 3: 721–732. PPAR gamma 2 gene expression by the HIF-1-regulatedgene Shen M, Kawamoto T, Yan W, Nakamasu K, Tamagami M, Koyano Y DEC1/Stra13: a mechanism for regulation of adipogenesis by et al. (1997). Molecular characterization of the novel basic hypoxia. Dev Cell 2: 331–341.

Oncogene