E2F6 Negatively Regulates BRCA1 in Human Cancer Cells without Methylation of Histone H3 on Lysine 9
2003; Elsevier BV; Volume: 278; Issue: 43 Linguagem: Inglês
10.1074/jbc.m307733200
ISSN1083-351X
AutoresMatthew J. Oberley, David R. Inman, Peggy Farnham,
Tópico(s)RNA modifications and cancer
ResumoE2F6 contains a DNA binding domain that is very similar to that of the other members of the E2F family of transcriptional regulators. However, E2F6 cannot bind to all promoters that contain consensus E2F-binding sites. Therefore, we used a combination of chromatin immunoprecipitation and genomic microarrays to identify promoters bound by E2F6 in human cells. Although most of the identified promoters were bound by multiple E2F family members, one promoter was bound only by E2F6. To determine which of the newly identified promoters were regulated by E2F6, we reduced the level of E2F6 by using RNA interference technology. We found that mRNA transcribed from promoters bound by E2F6 was increased after reduction of the amount of E2F6 protein in the cell. Interestingly, many of the E2F6-regulated genes encoded functions involved in tumor suppression and the maintenance of chromatin structure. Specifically, our results suggest that E2F6 represses transcription of the brca1, ctip, art27, hp1α, and the rbap48 genes. E2F6 has been postulated to mediate transcriptional repression by recruiting a histone H3 methyltransferase to the DNA. However, we found that the E2F6-regulated promoters did not contain histone H3 methylated at lysine 9. To determine the mechanism by which E2F6 regulates transcription, we performed chromatin immunoprecipitation before and after the introduction of small inhibitory ribonucleic acids specific to E2F6. We found that depletion of E2F6 resulted in the recruitment of E2F1 to the target promoters. In summary, we have identified 48 endogenous target genes of E2F6 and have shown that E2F6 can repress target promoters in a manner that does not require histone H3 methylation at lysine 9. E2F6 contains a DNA binding domain that is very similar to that of the other members of the E2F family of transcriptional regulators. However, E2F6 cannot bind to all promoters that contain consensus E2F-binding sites. Therefore, we used a combination of chromatin immunoprecipitation and genomic microarrays to identify promoters bound by E2F6 in human cells. Although most of the identified promoters were bound by multiple E2F family members, one promoter was bound only by E2F6. To determine which of the newly identified promoters were regulated by E2F6, we reduced the level of E2F6 by using RNA interference technology. We found that mRNA transcribed from promoters bound by E2F6 was increased after reduction of the amount of E2F6 protein in the cell. Interestingly, many of the E2F6-regulated genes encoded functions involved in tumor suppression and the maintenance of chromatin structure. Specifically, our results suggest that E2F6 represses transcription of the brca1, ctip, art27, hp1α, and the rbap48 genes. E2F6 has been postulated to mediate transcriptional repression by recruiting a histone H3 methyltransferase to the DNA. However, we found that the E2F6-regulated promoters did not contain histone H3 methylated at lysine 9. To determine the mechanism by which E2F6 regulates transcription, we performed chromatin immunoprecipitation before and after the introduction of small inhibitory ribonucleic acids specific to E2F6. We found that depletion of E2F6 resulted in the recruitment of E2F1 to the target promoters. In summary, we have identified 48 endogenous target genes of E2F6 and have shown that E2F6 can repress target promoters in a manner that does not require histone H3 methylation at lysine 9. The E2F family consists of six members, E2Fs 1-6, and two obligate heterodimeric partners, DP1 and DP2, which are required for binding to DNA (1Trimarchi J.M. Lees J.A. Nat. Rev. Mol. Cell. Biol. 2002; 3: 11-20Crossref PubMed Scopus (970) Google Scholar). All of the E2Fs contain a conserved DNA binding and dimerization domain, and the different E2F-DP heterodimers can bind to the same consensus sequence. E2Fs 1-5 each contain a C-terminal transactivation domain that can interact with a variety of transcriptional coactivators such as CREB 1The abbreviations used are: CREB, cAMP-response element-binding protein; Eu-HMTase, euchromatin-specific histone methyltransferase; ChIP-chip, chromatin immunoprecipitation-microarray analysis; ChIP, chromatin immunoprecipitation; siRNA, small inhibitory ribonucleic acid; RT, reverse transcriptase; myc, myelocytomatosis oncogene; art27, androgen receptor trapped clone 27; brca1, breast cancer 1; ctip, CTBP-interacting protein; rbap48, retinoblastoma-associated p48; hp1α, heterochromatin protein 1; htf6, human teratocarcinoma finger; gapdh, glyceraldehyde-3-phosphate dehydrogenase; RNA pol II, RNA polymerase II; GFP, green fluorescence protein.-binding protein and TFIIH (2Fry C.J. Pearson A. Malinowski E. Bartley S.M. Greenblatt J. Farnham P.J. J. Biol. Chem. 1999; 274: 15883-15891Abstract Full Text Full Text PDF PubMed Scopus (46) Google Scholar, 3Pearson A. Greenblatt J. Oncogene. 1997; 15: 2643-2658Crossref PubMed Scopus (50) Google Scholar). The C-terminal domain also contains sequences required for binding to the pocket protein family of transcriptional repressors (retinoblastoma, p107, and p130). E2Fs 1-3 interact preferentially with retinoblastoma, whereas E2F4 and -5 mainly associate with p107 or p130 (4Classon M. Harlow E. Nat. Rev. Cancer. 2002; 2: 910-917Crossref PubMed Scopus (609) Google Scholar). Depending upon exactly which proteins associate with the C-terminal domain, E2Fs1-5 can therefore mediate either activation or repression. E2F6, the most recently identified E2F family member, lacks the C-terminal transactivation domain found in the other E2Fs. Therefore, E2F6 likely cannot serve as a transcriptional activator or bind to the pocket protein family. However, E2F6 has been shown to be a potent transcriptional repressor (5Morkel M. Wenkel J. Bannister A.J. Kouzarides T. Hagemeier C. Nature. 1997; 390: 567-568Crossref PubMed Scopus (100) Google Scholar, 6Trimarchi J.M. Fairchild B. Verona R. Moberg K. Andon N. Lees J.A. Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 2850-2855Crossref PubMed Scopus (192) Google Scholar, 7Gaubatz S. Wood J.G. Livingston D.M. Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 9190-9195Crossref PubMed Scopus (153) Google Scholar, 8Cartwright P. Muller H. Wagener C. Holm K. Helin K. Oncogene. 1998; 17: 611-623Crossref PubMed Scopus (164) Google Scholar). Because E2F6 lacks the pocket protein interaction domain, transcriptional repression may be mediated via interaction with other proteins. Yeast two-hybrid studies have shown that E2F6 can be found in a multiprotein complex with members of the Bmi1-containing Polycomb group complex of transcriptional repressors (9Trimarchi J.M. Fairchild B. Wen J. Lees J.A. Proc. Natl. Acad. Sci. U. S. A. 2001; 98: 1519-1524Crossref PubMed Scopus (218) Google Scholar). Other studies have shown that E2F6 is also a component of a complex that contains Polycomb group proteins such as RING1, RING2, MBLR, h-l(3)mbt-like protein, and YAF2 (10Ogawa H. Ishiguro K. Gaubatz S. Livingston D.M. Nakatani Y. Science. 2002; 296: 1132-1136Crossref PubMed Scopus (631) Google Scholar). A general function of Polycomb group complexes is to specify developmental patterning, which is accomplished, in part, by the repression of homeobox gene expression (11Jacobs J.J. van Lohuizen M. Biochim. Biophys. Acta. 2002; 1602: 151-161PubMed Google Scholar). E2F6-null mice display homeotic transformations of the axial skeleton (12Storre J. Elsasser H.P. Fuchs M. Ullmann D. Livingston D.M. Gaubatz S. EMBO Rep. 2002; 3: 695-700Crossref PubMed Scopus (73) Google Scholar) and resemble animals lacking the Bmi1 polycomb group protein (13van der Lugt N.M. Domen J. Linders K. van Roon M. Robanus-Maandag E. te Riele H. van der Valk M. Deschamps J. Sofroniew M. van Lohuizen M. Genes Dev. 1994; 8: 757-769Crossref PubMed Scopus (675) Google Scholar). These studies are consistent with the hypothesis that an E2F6-containing Polycomb group complex is required to silence certain homeotic genes during development. Complexes containing Bmi1 are implicated in the development of cancer through the repression of the p16ink4a and p19arf genes (14Jacobs J.J. Scheijen B. Voncken J.W. Kieboom K. Berns A. van Lohuizen M. Genes Dev. 1999; 13: 2678-2690Crossref PubMed Scopus (541) Google Scholar), suggesting a specific link between E2F6 and the development of cancer. Also, support of a link between E2F6 and carcinogenesis comes from studies demonstrating that overexpression of E2F6 can alter cell growth parameters. For example, overexpressed E2F6 can inhibit entry into S phase of cells stimulated to exit G0 phase (7Gaubatz S. Wood J.G. Livingston D.M. Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 9190-9195Crossref PubMed Scopus (153) Google Scholar) and can also delay the exit from S phase (8Cartwright P. Muller H. Wagener C. Holm K. Helin K. Oncogene. 1998; 17: 611-623Crossref PubMed Scopus (164) Google Scholar). Although the possible links between E2F6 and cell growth control are intriguing, little is known about the exact genes regulated by E2F6, and it is not yet clear exactly how E2F6 mediates transcriptional regulation. Recently, Ogawa et al. (10Ogawa H. Ishiguro K. Gaubatz S. Livingston D.M. Nakatani Y. Science. 2002; 296: 1132-1136Crossref PubMed Scopus (631) Google Scholar) identified a novel euchromatic-specific histone methyltransferase (Eu-HMTase) as part of an E2F6-containing complex from HeLa cells, and this Eu-HMTase was shown to specifically methylate histone H3 at lysine 9 on free histones and mononucleosomes in vitro. This specific histone H3 modification (H3 Me-K9) produced by eukaryotic histone methyltransferases (HMTase), such as SUV39H1, has been shown to recruit HP1 and to create a locally repressive heterochromatic structure (15Neilsen S.J. Schneider R. Bauer U.-M. Bannister A.J. Morrison A. O'Carroll D. Firestein R. Cleary M. Jenuwein T. Herrera R.E. Kouzarides T. Nature. 2001; 412: 561-565Crossref PubMed Scopus (747) Google Scholar). The identification of the E2F6-HMTase complex has led to an hypothesis regarding the mechanism of E2F6-mediated transcriptional repression. It has been postulated that the transcriptional repressive activity of E2F6 begins with the recruitment of the Eu-HMTase, which is followed by the methylation of histone H3 at Lys-9, and the subsequent establishment of heterochromatin via recruitment of HP1. However, it has not been rigorously tested whether E2F6 does indeed convert euchromatin into heterochromatin in vivo. As noted above, all of the E2F family members, including E2F6, can bind in vitro to a consensus element (TTTSSCGC). However, it is not yet clear if the same set of target genes will be regulated by all the E2Fs in vivo. Our group has recently utilized chromatin immunoprecipitation coupled with CpG island microarrays (ChIP-chip) to identify novel in vivo targets of E2F1, E2F4, pRB, and MYC (16Wells J. Yan P.S. Cechvala M. Huang T. Farnham P.J. Oncogene. 2003; 22: 1445-1460Crossref PubMed Scopus (110) Google Scholar, 17Weinmann A.S. Yan P.S. Oberley M.J. Huang T.H.-M. Farnham P.J. Genes Dev. 2002; 16: 235-244Crossref PubMed Scopus (394) Google Scholar, 18Mao D.Y.L. Watson J.D. Yan P.S. Barsyte-Lovejoy D. Khosravi F. Wong W.W.-L. Farnham P.J. Huang T.H.-M. Penn L.Z. Curr. Biol. 2003; 13: 882-886Abstract Full Text Full Text PDF PubMed Scopus (145) Google Scholar). CpG islands are tracts of at least 200 bp C + G-rich sequences that are found in the promoters and first exons of about 70% of human genes (19Ioshikhes I.P. Zhang M.Q. Nat. Genet. 2000; 26: 61-63Crossref PubMed Scopus (224) Google Scholar, 20Davuluri R.V. Grosse I. Zhang M.Q. Nat. Genet. 2001; 29: 412-417Crossref PubMed Scopus (338) Google Scholar). Because E2F-binding sites are most commonly found within proximal promoter regions and because the consensus binding sequence is GC-rich, the CpG arrays have proved to be very useful for identifying E2F target genes. This present study utilizes an improved version of our previous ChIP-chip protocol to identify genomic targets of E2F6 in human cancer cells. We find that E2F6 regulates genes that are involved in the pathogenesis of neoplasia and in regulating chromatin structure and heredity. We also provide in vivo evidence that E2F6 can repress targets via a mechanism distinct from methylation of histone H3 at lysine 9. Chromatin Immunoprecipitations (ChIP)—The ChIP procedure was performed as described previously (21Oberley M.J. Tsao J. Yau P. Farnham P.J. Methods Enzymol. 2003; (in press)PubMed Google Scholar) with HeLa and 293 cells, with the following exceptions (see also mcardle.oncology.wisc.edu/farnham). The chromatin was sheared to an average size of 500-2000 bp. After cross-linking reversal and proteinase K digestion, each individual IP was purified with the use of a QIAquick PCR purification kit (Qiagen, Valencia, CA), and samples were eluted with 30 μl of elution buffer. After elution the IPs were examined by gene-specific PCR on the art27 promoter to ensure that the IP successfully enriched the promoter significantly more than did an IgG control. Antibodies used in the ChIP assays include E2F1 (KH20KH95, Upstate Biotechnology, Inc., Lake Placid, NY), E2F4 (sc866 X, Santa Cruz Biotechnology, Santa Cruz, CA), E2F6 (sc8366X, Santa Cruz Biotechnology), Max (sc197, Santa Cruz Biotechnology), RNA polymerase II (sc-899, Santa Cruz Biotechnology), and histone H3 tri-methylated at K9 (Ab8898, Abcam, Cambridge, UK). E2F6 (sc8366P) and TCF4 (sc8631P) peptides were purchased from Santa Cruz Biotechnology. Amplicon Generation and Labeling—For detailed protocols on these steps see Oberley et al. (21Oberley M.J. Tsao J. Yau P. Farnham P.J. Methods Enzymol. 2003; (in press)PubMed Google Scholar) or mcardle.oncology.wisc.edu/farnham. The generation of amplicons from the individual ChIPs was adapted from Ren et al. (20Davuluri R.V. Grosse I. Zhang M.Q. Nat. Genet. 2001; 29: 412-417Crossref PubMed Scopus (338) Google Scholar). Briefly, two unidirectional linkers oligoJW102 (5′ gcg gtg acc cgg gag atc tga att c 3′) and oligoJW103 (5′ gaa ttc aga tc 3′) were annealed and blunt-end ligated to the chromatin IPs. Amplicons were created by PCR; each sample consisted of 5 μl of 10× Taq polymerase buffer, 7 μl of 2 mm dNTPs, 3 μl of MgCl2, 6.5 μl of betaine, 2.5 μl of oligoJW102 (20uM), 1 μl of Taq (Promega, M1861), and 25 μl of the blunted and ligated chromatin. PCR was run with one cycle at 55 °C for 2 min, 72 °C for 5 min, and 95 °C for 2 min. Twenty cycles were then run at 95 °C for 0.5 min, 55 °C for 0.5 min, and 72 °C for 1 min. Finally the products were extended at 72 °C for 4 min and then held at 4 °C until purified using the QIAquick PCR purification kit according to the manufacturer's instructions. DNA was quantitated and stored -20 °C until labeling. The amplicons were labeled with amino-allyl dUTP using 3 × 2 μg of the E2F6 IP amplicon (or 3 × 2 μg of the IgG IP amplicon) and 3 × 2 μg of the total input chromatin amplicon, as described previously (22Yan P.S. Chen C.-M. Shi H. Rahmatpanah F. Wei S.H. Caldwell C.W. Huang T.H.-M. Cancer Res. 2001; 61: 8375-8380PubMed Google Scholar). We coupled the Cy5 dye to the E2F6 IP (or the IgG) amplicon and Cy3 dye to the total reference amplicon, using standard methods (22Yan P.S. Chen C.-M. Shi H. Rahmatpanah F. Wei S.H. Caldwell C.W. Huang T.H.-M. Cancer Res. 2001; 61: 8375-8380PubMed Google Scholar). Unincorporated dye was then removed with a QIAquick PCR kit, eluting with H2O. The labeled chromatin was then dried with heat in a speed-vac and stored dry at -20 °C until ready for hybridization. CpG Island Hybridization—The generation and hybridization of the CpG island microarrays have been described elsewhere (17Weinmann A.S. Yan P.S. Oberley M.J. Huang T.H.-M. Farnham P.J. Genes Dev. 2002; 16: 235-244Crossref PubMed Scopus (394) Google Scholar, 22Yan P.S. Chen C.-M. Shi H. Rahmatpanah F. Wei S.H. Caldwell C.W. Huang T.H.-M. Cancer Res. 2001; 61: 8375-8380PubMed Google Scholar). The hybridized microarrays were analyzed using the Genepix Pro 4.1 (Axon Instruments) software package. This provided a set of raw values for each feature on each array. Features of poor intensity (<500) and those that had obvious blemishes were manually flagged and removed from the putative positive list. To identify clones that are selectively enriched during IP relative to the starting population, the Cy5 and Cy3 channels were normalized across the entire array, by taking the ratio of the medians for all quality features and normalizing them to unity. After normalization, a ratio was generated that was the intensity in the Cy5 channel minus background divided by the intensity in the Cy3 channel minus background. The ratio was generated for all features that met these criteria, and features with ratios above 2 were selected for further analysis. The identified CpG island clones were then sequenced by standard methods using primers corresponding to vector sequences. PCR Assays—To analyze the identified CpG islands (and other characterized promoter regions), PCRs were performed. Primer sequences for all the ChIP confirmations are available at our website: mcardle.oncology.wisc.edu/farnham. Each PCR mixture contained 2 μl of immunoprecipitated DNA (or 10 ng of each amplicon) and was performed as described previously (16Wells J. Yan P.S. Cechvala M. Huang T. Farnham P.J. Oncogene. 2003; 22: 1445-1460Crossref PubMed Scopus (110) Google Scholar). PCR products were separated by electrophoresis through 1.5% agarose gels and visualized by ethidium bromide intercalation. RNAi Transfections, Westerns, and RT-PCR Assays—293 cells were plated at a density of 6 × 105 on 60-mm plates 24 h before transfection with the siRNAs. siRNAs were purchased from Dharmacon (Louisville, CO) and included siE2F6-01 (AAGGAUUGUGCUCAGCAGCUG-custom order), siE2F6 smart pool (M-003264-00-05), or siGFP (D-001300-01-05). Transfections were performed with OligofectAMINE (Invitrogen) or TransIT-TKO (Mirus, Madison, WI) according to manufacturers' instructions. Western blots were performed as described previously (16Wells J. Yan P.S. Cechvala M. Huang T. Farnham P.J. Oncogene. 2003; 22: 1445-1460Crossref PubMed Scopus (110) Google Scholar). The same antibodies were used for Westerns as described above for the ChIP assays. RT-PCR analysis was performed as described previously (23Kirmizis A. Bartley S.M. Farnham P.J. Mol. Cancer Ther. 2003; 2: 113-121PubMed Google Scholar). Primer sequences for all the RT-PCR assays are available at our website: mcardle.oncology.wisc.edu/farnham. The Identification of E2F6 Genomic Binding Sites Using CpG Microarrays—As noted above, very few bona fide target genes for E2F6 have been identified. In fact, we have found that many well characterized E2F target genes are not bound by E2F6. As an example, we have characterized the binding pattern of the E2Fs on the myc promoter, an established E2F target gene (24Lavia P. Jansen-Durr P. BioEssays. 1999; 21: 221-230Crossref PubMed Scopus (145) Google Scholar, 25Weinmann A.S. Bartley S.M. Zhang M.Q. Zhang T. Farnham P.J. Mol. Cell. Biol. 2001; 21: 6820-6832Crossref PubMed Scopus (332) Google Scholar). By using the ChIP assay, we examined binding of E2Fs 1-4 and 6 in human 293 cells. Primers were used that spanned the consensus E2F-binding site, which is located just upstream of the transcriptional start in the myc promoter. The antibodies to E2F1-4 enriched the myc promoter sequences, indicating that the myc promoter is bound by these proteins in asynchronously growing 293 cells (Fig. 1A). In contrast the E2F6 antibody did not enrich the myc promoter sequences relative to the IgG control sample. Given that the myc promoter was occupied by the other E2Fs, the chromatin structure must be permissible to E2F family member binding. Thus, the absence of E2F6 binding suggests that E2F6 likely requires determinants for binding other than the presence of a consensus E2F-binding site. This indicates that E2F6 may bind to only a subset of promoters bound by other E2F family members. Our laboratory has recently utilized chromatin immunoprecipitation followed by CpG microarray analysis (ChIP-chip) to identify genomic targets of E2F4 (17Weinmann A.S. Yan P.S. Oberley M.J. Huang T.H.-M. Farnham P.J. Genes Dev. 2002; 16: 235-244Crossref PubMed Scopus (394) Google Scholar) and E2F1 (16Wells J. Yan P.S. Cechvala M. Huang T. Farnham P.J. Oncogene. 2003; 22: 1445-1460Crossref PubMed Scopus (110) Google Scholar). Because of the success of these prior experiments, we reasoned that the ChIP-chip technique would be an excellent method to identify a large set of E2F6 target genes. However, prior to beginning the array experiments, it was essential to have a positive control promoter to ensure specific enrichment of E2F6-bound DNA in the immunoprecipitated chromatin. Therefore, we tested a series of additional E2F target promoters for their ability to bind to E2F6. Examination of well characterized E2F target genes (such as dhfr) revealed that E2F6 did not bind to the promoter regions (data not shown). Because promoters that have consensus E2F-binding sites (such as the myc and dhfr promoters) do not bind E2F6, we next tested promoters that bind E2F family members via non-consensus E2F sites (17Weinmann A.S. Yan P.S. Oberley M.J. Huang T.H.-M. Farnham P.J. Genes Dev. 2002; 16: 235-244Crossref PubMed Scopus (394) Google Scholar). Although the promoter for the art27 (Androgen receptor trapped clone 27) gene, which functions as a coactivator of the androgen receptor (26Markus S.M. Taneja S.S. Logan S.K. Li W. Ha S. Hittelmen A.B. Rogatsky I. Garabedian M.J. Mol. Biol. Cell. 2002; 13: 670-682Crossref PubMed Scopus (70) Google Scholar), does not contain a consensus E2F site (as defined as TTTSSCGC), it does have two near consensus E2F sites located at -280 and -109 relative to the transcriptional start site. We have shown previously that the art27 promoter (which was previously called uxt (27Schroer A. Schneider S. Ropers H. Nothwang H. Genomics. 1999; 56: 340-343Crossref PubMed Scopus (46) Google Scholar)) is bound by several E2F family members in HeLa cells (17Weinmann A.S. Yan P.S. Oberley M.J. Huang T.H.-M. Farnham P.J. Genes Dev. 2002; 16: 235-244Crossref PubMed Scopus (394) Google Scholar). However, binding of E2F6 was not tested in the previous study. We have now demonstrated in vivo binding of E2Fs 1-4 and 6 to the art27 promoter in 293 cells (Fig. 1A). As a negative control, the 3′-untranslated region of the dhfr gene was analyzed. As expected, this region was not enriched by antibodies to any of the E2Fs. Interestingly, E2F6 bound to the art27 promoter in 293 cells but not in HeLa cells (Fig. 1B), despite the fact that E2F6 protein is expressed at significant levels in HeLa cells and that these cells were used to purify the E2F6-containing complex that included the novel Eu-HMTase (10Ogawa H. Ishiguro K. Gaubatz S. Livingston D.M. Nakatani Y. Science. 2002; 296: 1132-1136Crossref PubMed Scopus (631) Google Scholar). These data indicate that E2F6 shows target gene selectivity and cell type specificity. Because E2F6 has been postulated to be a transcriptional repressor, we compared the activity in 293 cells of art27 promoter constructs containing or lacking the sequences that resemble E2F-binding sites (Fig. 2). We transfected each promoter-luciferase construct into 293 cells, along with a cytomegalovirus-Renilla control for transfection efficiency, harvested the cells after 48 h, and measured the promoter activity. Interestingly, we found that deletion of each E2F-like site resulted in increased activity of the art27 promoter, which suggests that E2F6, perhaps in conjunction with other E2Fs (28Croxton R. Ma Y. Song L. Haura E.B. Cress W.D. Oncogene. 2002; 21: 1359-1369Crossref PubMed Scopus (125) Google Scholar), acts to repress the transcriptional activity of the art27 promoter. Based on these studies, we have used 293 cells as an experimental system, and the art27 promoter as our positive control, for the study of E2F6-mediated transcriptional regulation of target genes. Our next step was to perform a high-throughput genomic screen to identify E2F6-binding loci in 293 cells. Our lab has previously screened chromatin immunoprecipitates with CpG island microarrays to identify genomic targets of E2F family members (16Wells J. Yan P.S. Cechvala M. Huang T. Farnham P.J. Oncogene. 2003; 22: 1445-1460Crossref PubMed Scopus (110) Google Scholar, 17Weinmann A.S. Yan P.S. Oberley M.J. Huang T.H.-M. Farnham P.J. Genes Dev. 2002; 16: 235-244Crossref PubMed Scopus (394) Google Scholar). In our previous studies, multiple individual ChIP samples were pooled to obtain enough material to probe a microarray. Here we have adapted several changes to our previously published protocols that have allowed a more rigorous discrimination of robust in vivo targets (Fig. 3A). One critical change was the inclusion of a ligation-mediated PCR step (29Ren B. Robert F. Wyrick J.J. Aparicio O. Jennings E.G. Simon I. Zeitlinger J. Schreiber J. Hannett N. Kanin E. Volkert T.L. Wilson C.J. Bell S.P. Young R.A. Science. 2000; 290: 2306-2309Crossref PubMed Scopus (1594) Google Scholar) that has allowed a considerable reduction in the quantity of cells required. We performed a ChIP assay on 1 × 107 cells with antibodies specific for E2F6 or an IgG control. After extensive washing, the cross-links were reversed, and the DNA was purified. Next, the E2F6 IP, the IgG IP, and 10 ng of total input DNA were blunt-ended, ligated to a unidirectional linker, and amplified to generate enough DNA to probe the CpG microarrays. Equal amounts of amplicons (6 μg of the E2F6 IP and 6 μg of total input) were labeled with amino-allyl dUTP, and then the E2F6 labeled amplicon was conjugated with Cy5, whereas the reference total input amplicon was conjugated with Cy3. These two labeled pools of DNA were mixed with CoT-1 DNA, which binds the repeat DNA in the samples, and applied to a CpG island microarray under stringent hybridization conditions. The arrays were hybridized overnight at 60 °C, washed, and then scanned on an Axon Instrument 4000B scanner, using GenePix Pro 4.1 software (Axon Instruments) to analyze the array. After manually flagging any obvious blemishes on the resulting image, all features that gave quality signals were included in a data set that was normalized by setting the ratio of all medians to unity. The results of three array experiments are depicted in Fig. 3B, which is a dot plot on a logarithmic scale where the Cy5 intensity of each feature is plotted on the x axis and the Cy3 intensity of the feature is plotted on the y axis. For the first array hybridization, the total input sample was divided in two; half was labeled with Cy5, and half was labeled with Cy3. These labeled amplicons were then hybridized to an array to give a total versus total comparison (Fig. 3B, red squares). As expected, this plot produced a straight line with a good regression value (R 2 = 0.9468) for spotted arrays. Then, as a negative control, an IgG amplicon was labeled with Cy5 and compared with Cy3-labeled total input chromatin (yellow circles); any CpG-islands that were non-specifically enriched by IgG during the ChIP assay were then discarded from further analysis. Finally, the experimental plot is depicted with blue diamonds, which show the results of hybridization of the E2F6 IP amplicon labeled with Cy5 versus the reference total input amplicon that was labeled with Cy3. As is evident from Fig. 3B, there were many features that displayed enriched Cy5 intensities relative to Cy3, and those that were enriched at least 2-fold over total in independent hybridization experiments were chosen for follow up analysis. After repeating each hybridization twice, there were 77 CpG islands that were specifically enriched at least 2-fold higher than starting input chromatin by the E2F6 ChIP; these were not enriched by IgG in either experiment. The clones from the CpG island library (30Cross S.H. Charlton J.A. Nan X. Bird A.P. Nat. Genet. 1994; 6: 236-244Crossref PubMed Scopus (401) Google Scholar) that were spotted onto the arrays have not been sequenced, so all putatively positive clones were sequenced and localized in the human genome using the University of California Santa Cruz Genome Web Browser (genome.ucsc.edu). The majority of the sequenced clones (92.2%) were CpG islands located in the promoters and/or first exons of either characterized genes, mRNAs, ESTs, or predicted genes (Fig. 4). Of the remaining six clones, we were unable to sequence five of them, and the final one did not map to any region of the genome using the April 2003 version of the University of California Santa Cruz Genome Web Browser. Because there is some degree of redundancy of the CpG islands spotted on the arrays, the 77 features that were identified as positive in the two independent hybridizations actually represented 48 independent CpG islands. Of these 48 independent CpG islands, 26 were found in the promoter and 5′ region of characterized genes, and 22 were localized to uncharacterized mRNAs, ESTs, or predicted genes. The redundancy on the arrays provided a measure of the robustness of the hybridization procedures. For example, the hp1α locus was identified as an E2F6 target seven times. Fig. 4 shows the identity of the characterized genes, their function, and whether an E2F consensus binding site exists within 1000 bp upstream of the transcriptional start site. Because Ogawa et al. (10Ogawa H. Ishiguro K. Gaubatz S. Livingston D.M. Nakatani Y. Science. 2002; 296: 1132-1136Crossref PubMed Scopus (631) Google Scholar) identified E2F6 in a complex with Mga and Max, E2F6 may also be recruited to promoters via E boxes (CACGTG), due to protein-protein interactions with Max-Mga heterodimers. Therefore, the presence or absence of a consensus E box (CCACGTGG) is also indicated. Interestingly, ne
Referência(s)