Blood-2011-Schödel-e207-17 | Regulation Of Gene Expression | Promoter (Genetics)

From bloodjournal.hematologylibrary.org at JAWAHARLAL NEHRU UNIVERSITY on May 27, 2013. For personal use only.

2011 117: e207-e217 Prepublished online March 29, 2011; doi:10.1182/blood-2010-10-314427

High-resolution genome-wide mapping of HIF-binding sites by ChIP-seq
Johannes Schödel, Spyros Oikonomopoulos, Jiannis Ragoussis, Christopher W. Pugh, Peter J. Ratcliffe and David R. Mole

Updated information and services can be found at: http://bloodjournal.hematologylibrary.org/content/117/23/e207.full.html Articles on similar topics can be found in the following Blood collections e-Blood (107 articles) Red Cells, Iron, and Erythropoiesis (441 articles) Vascular Biology (404 articles) Information about reproducing this article in parts or in its entirety may be found online at: http://bloodjournal.hematologylibrary.org/site/misc/rights.xhtml#repub_requests Information about ordering reprints may be found online at: http://bloodjournal.hematologylibrary.org/site/misc/rights.xhtml#reprints Information about subscriptions and ASH membership may be found online at: http://bloodjournal.hematologylibrary.org/site/subscriptions/index.xhtml

Blood (print ISSN 0006-4971, online ISSN 1528-0020), is published weekly by the American Society of Hematology, 2021 L St, NW, Suite 900, Washington DC 20036. Copyright 2011 by The American Society of Hematology; all rights reserved.

10. University of Oxford. Expression arrays indicate that many hundreds of genes are regulated by this pathway. 2011. Oxford. in addition to regulating erythropoietin. occurrences of this motif across the genome are too numerous to permit prediction of HIF binding. accepted March 16.117(23):e207-e217) Introduction Oxygen is essential to the survival of all eukaryotic cells. However.7 that in turn modulate stability and transactivating activity of the HIF-␣ subunit. United Kingdom. the extent of direct versus indirect transcriptional regulation of gene expression by the HIF cascade was unclear. Indeed.1. such as erythropoietin. we identify HIF-binding sites across the genome. this article is hereby marked ‘‘advertisement’’ in accordance with 18 USC section 1734. The online version of this article contains a data supplement. NUMBER 23 e207 . proliferation. we demonstrate robust associations with the regulation of gene expression by HIF. For personal use only. and adaptation to altered oxygen tension occurs at both the systemic level and at the cellular level through regulation of both oxygen utilization and delivery. the extent to which HIF exerts direct versus indirect control over gene expression together with the factors dictating the range of HIF-regulated genes remains unclear. Across the entire genome. and ischemic vascular disease. We describe the use of ChIP coupled to next-generation high-throughput sequencing to define HIF-binding sites across the whole genome irrespective of whether they coincide with a promoter. as well as the pathophysiology of inflammation. 2010. March 29. (Blood. stabilizing HIF-␣ and activating target genes.20 this approach is limited.11 However. including the erythropoietin enhancer.27-30 However. and solely to indicate this fact.7-9 Small-molecule analogs of 2-oxoglutarate (eg. 2011. 2013. and 2The Wellcome Trust Centre for Human Genetics. only a small proportion of these potential binding sites are bound by HIF.2 Jiannis Ragoussis.2 Christopher W. by definition. apoptosis. by 2-oxoglutarate-dependent dioxygenase enzymes6. compared with newer techniques of chromatin immunoprecipitation coupled to next-generation high-throughput sequencing (ChIP-seq).From bloodjournal. although occupancy of potential sites was enhanced approximately 20-fold at normoxic DNAse1 hypersensitivity sites (irrespective of distance from promoters). Analysis of HIF-binding motifs demonstrates sequence preferences outside of the core RCGTG-binding motif but does not reveal any additional absolute sequence requirements. these techniques have a relatively low signal-to-noise ratio and are intrinsically biased. to short range interactions. Prepublished online as Blood First Edition paper. suggesting that epigenetic regulation of chromatin may have an important role in defining the response to hypoxia. genome-wide association studies in Han and Tibetan Chinese populations have linked the EPAS1 gene locus with altered hemoglobin concentration.2 and mutations in HIF-2␣ and its regulatory proteins lead to disorders of erythropoiesis. Oxford.3-5 Composed of an ␣/␤-heterodimer.1182/blood-2010-10-314427. Using chromatin immunoprecipitation linked to high throughput sequencing. and is important in embryonic development and adaptation to altitude. indicating that these sites operate over long genomic intervals. It controls processes as diverse as growth factor production. and angiogenesis. being restricted to sequences used in the tiled array. Therefore. Erythropoietin production is regulated in response to tissue oxygenation by the transcription factor hypoxia-inducible factor (HIF). dimethyloxalylglycine [DMOG]) inhibit these enzymes.1 and David R.15-20 In addition to direct gene regulation. Ratcliffe.hematologylibrary. several recent studies have used chromatin immunoprecipitation coupled to tiled microarrays (ChIP-chip) to examine genome-wide binding of HIF-1␣ and HIF-2␣ isoforms to gene promoters. However. Mole1 1Henry Wellcome Building for Molecular Physiology. 9 JUNE 2011 ⅐ VOLUME 117. United Kingdom Hypoxia-inducible factor (HIF) regulates the major transcriptional cascade central to the response of all mammalian cells to alterations in oxygen tension. HIF is primarily controlled through oxygen-regulated modification of the HIF-␣ subunit. HIF also regulates many other transcription factors indicating that HIF controls a network of responses to hypoxia. The work demonstrates the existence of large numbers of HIF-binding sites that are remote from known promoters. University of Oxford. energy metabolism.1 Spyros Oikonomopoulos. has identified a core response element 5Ј-RCGTG-3Ј. controlling diverse processes that in turn orchestrate both oxygen delivery and utilization.org at JAWAHARLAL NEHRU UNIVERSITY on May 27. The publication costs of this article were defrayed in part by page charge payment. 2011. Robust Submitted October 20. HIF drives a major transcriptional cascade responsible for orchestrating the cellular response to oxygen availability in many different tissues.21-25 Analysis of validated HIF-dependent regulatory elements at approximately 50 HIF-target genes. does not provide any mechanistic insight into the factors restricting HIF binding at the molecular level and requires experimental verification.12-14 Expression arrays indicate that many hundreds of genes are regulated by the HIF pathway. cancer. The maintenance of adequate levels of red blood cells and hemoglobin is essential to the delivery of sufficient oxygen to the tissues and is controlled through erythropoiesis. differentiation.1 Peter J. until recently. Using gene set enrichment analysis. VASCULAR BIOLOGY e-Blood High-resolution genome-wide mapping of HIF-binding sites by ChIP-seq Johannes Scho ¨ del. independently of gene architecture. Pugh.26 However. Although attempts have been made to identify potential HIF-binding sites based on mammalian conservation and distance from the transcriptional start site (TSS). © 2011 by The American Society of Hematology BLOOD. To identify direct HIF target genes. DOI 10.

Sequences were mapped to NCBI build 36 (hg18) using ELAND software. Primer sequences are given in supplemental Table 3 (available on the Blood Web site. Correlation of HIF-binding data with genome-wide DNAse1 hypersensitivity data obtained by the ENCODE consortium in normoxic cells31.15 Correlation of ChIP-seq data with microarray analysis of gene expression was performed using GSEA software Version 2 (www.org/gsea/). Finally.jhsph.36. loci with more than 4 binding sites were excluded as manual analysis revealed these to be regions of high background or repeat elements. 2mM L-glutamine. and mapping of RCGTG motifs were undertaken using the CisGenome software suite (www. strongly suggesting that these sites promote transcriptional responses to hypoxia over very long genomic intervals. The microarray analysis of MCF-7 cells exposed to 1% ambient oxygen for 16 hours in the presence of either control or combined HIF-1␣ and HIF-2␣ siRNA has been previously described. HIF-2␣ (PM9). indicating that epigenetic factors must play a major role in defining HIF binding. but not sole. Refined analysis of the HIF-recognition motif based on more precise localization of HIF-binding regions defines sequence preferences beyond the core RCGTG motif. predictor of hypoxiainducible HIF binding to an RCGTG motif.5% ambient oxygen for 16 hours before harvest. associations with the regulation of distant target genes were defined by gene set enrichment analysis (GSEA). ␤-actin was used for normalization and fold enrichment at each site was determined using the ⌬Ct method. these additional primary sequence preferences do not greatly restrict the number of potential binding sites in the genome.32 demonstrates that “resting-state” chromatin accessibility is a strong. the estimated false discovery rate was less than 0. However. 36-bp single end sequence analysis was performed on the Illumina GAII platform. First. and 10% FBS (Sigma-Aldrich). Subsequent assessment. High-throughput sequencing Libraries were prepared from immunoprecipitated chromatin using the Illumina ChIP-Seq kit.27 HIF-1␤ (Novus NB-100-110).37 . e208 ¨ DEL et al SCHO BLOOD. of reads per window 314 180 288 160 160 136 128 126 126 123 118 116 109 103 102 98 96 95 95 94 94 89 88 87 86 86 83 81 81 87 Covered by human promoter 1. 9 JUNE 2011 ⅐ VOLUME 117. and RNA Pol2 (Santa Cruz Biotechnology SC-899).33-36 In addition.org at JAWAHARLAL NEHRU UNIVERSITY on May 27. 2-sample peak finding. using rabbit polyclonal anti-sera to HIF-1␣ (PM14). For personal use only.edu/ϳhji/ cisgenome). see the Supplemental Materials link at the top of the online article). no. ChIP ChIPs were performed on MCF-7 cells as previously described. 3 additional filters were applied to define high-stringency peaks. Second.broadinstitute. 2013. Top 30 HIF-1 binding sites HIF-1 binding sites 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 Peak height.0R array No No Yes Yes No No No No Yes Yes Yes No Yes No Yes No No No Yes No Yes Yes Yes No Yes Yes No Yes No Yes Nearest gene locus DNAJB6 NDRG1 RSBN1 ALDOA CHD2 C20orf175 BHLHB2 SNAPC1 PFKFB3 GPI QSCN6 PKM2 CGA FOS BCL2L2 EGLN3 MRPS18A SLC2A3 PFKFB4 EGFR PDK1 ANKRD37 C3orf28 ZFAT1 ENO1 NARF GPR37L1 JMJD1A SEMA4B LOC401152 RefSeq number NM_058246 NM_006096 NM_018364 NM_184041 NM_001042572 NM_080829 NM_003670 NM_003082 NM_004566 NM_000175 NM_002826 NM_002654 NM_000735 NM_005252 NM_004050 NM_022073 NM_018135 NM_006931 NM_004567 NM_201283 NM_002610 NM_181726 NM_014367 NM_020863 NM_001428 NM_001038618 NM_004767 NM_018433 NM_020210 NM_001001701 Chromosome 7 8 1 16 15 20 3 14 10 19 1 15 6 14 14 14 6 12 3 7 2 4 3 8 1 17 1 2 15 4 Start 156986147 134456859 114156539 29984123 91153222 48720371 5003621 61287408 6283489 39541937 178403857 70306014 87921667 74791306 22840340 33477245 43743806 8014473 48569558 55216284 173128811 186554370 123585527 135913721 8861349 78009247 200320766 86521270 88509909 120441100 End 156987039 134457261 114157002 29984574 91153655 48720786 5003966 61287886 6283830 39542408 178404013 70306475 87922197 74791781 22840855 33477551 43744203 8014830 48569870 55216762 173129189 186554738 123585631 135913971 8862115 78009655 200321048 86521896 88510331 120441561 HIF indicates hypoxia-inducible factor. Expression arrays Methods Cell culture MCF-7 (breast cancer) cells were grown in DMEM. Independent confirmation of chromatin enrichment by real-time quantitative PCR SYBR Green-based real-time quantitative PCR using a StepOne thermocycler was used.hematologylibrary. Subconfluent cell cultures were exposed to 2mM DMOG (Frontier Scientific) or 0.05.From bloodjournal. NUMBER 23 Table 1. the number of reads in each direction was at least one-third of the total. motif analysis.biostat. Preimmune serum or IgG were used as negative controls.

Quantitative PCR analysis to independently assess the veracity of a random selection of these peaks confirmed a very high specificity (18 of 20 HIF-1␣ sites and 9 of 12 HIF-2␣ sites.From bloodjournal.5% hypoxia for 16 hours. For personal use only.and HIF-1␤-binding sites. This identified a threshold of approximately 20 reads per window for HIF-1␣ (and 25 reads per window for HIF-2␣). Top 30 HIF-2 binding sites HIF-2 binding sites 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 Peak height. CTGCAAAGACCTTTGAGTTCC). 9 JUNE 2011 ⅐ VOLUME 117. to determine the pan-genomic overlap of HIF-␣.and HIF-1␤-binding sites An important challenge with large ChIP-seq datasets is to identify an appropriate threshold to define binding peaks. Overlap between HIF-␣. Using gene loci identified by previous ChIP-chip analysis27 as a “seed” dataset. 2013. below which the overlap with the ChIP-chip data fell to background (supplemental Figure 1A-B). we used these thresholds.27 not meeting all inclusion criteria for high-stringency definition by ChIP. we undertook a comparable ChIP-seq analysis of HIF-1␤ DNA-binding in MCF-7 cells that were exposed to hypoxia for 16 hours. we examined the effect of altering the threshold used to define ChIP-seq peaks. U2OS cells. no. albeit with low levels of enrichment in some cases (supplemental Figure 2C). together with 3 additional filters (see “High-throughput sequencing”).hematologylibrary. BLOOD. This identified 400 highstringency HIF-1␣-binding sites at 356 gene loci (Table 1. supplemental Figure 2A-B). yielded signals at these sites. indicating that this approach to identifying suitable thresholds for ChIPseq analyses should be generalizable (data not shown). supplemental Table 2). Luciferase vectors and reporter assays HIF-binding sites were cloned into pGL3 promoter (Promega) using BglII (primers arrestin domain containing 3 [ARRDC3]: ACAGGAGTAGCGCGAGCAG. To define high-stringency ChIP-seq-binding sites more rigorously.38 Therefore. Luciferase activity (Promega Luciferase kit) was normalized to ␤-galactosidase activity. A comparable threshold was defined when previously identified Although HIF-␣ proteins were originally defined as part of a dimer with HIF-1␤ that bound DNA at hypoxia response elements. supplemental Table 1) and 425 high-stringency HIF-2␣-binding sites at 357 gene loci (Table 2. ChIP–quantitative PCR analysis of sites identified previously by ChIP-chip. NUMBER 23 MAPPING OF HIF-BINDING SITES BY ChIP-seq e209 Table 2.0R array No No No No No Yes No Yes No No No No No No No Yes No No Yes No No Yes No No Yes No No No No No Nearest gene locus DNAJB6 C20orf175 SEMA4B TFF1 NDRG1 HIG2 TNS1 BHLHB2 BHLHB2 SLC2A3 SNAPC1 DNAJB6 P2RY2 KRT80 EVL ALDOA GPR37L1 EGFR C8orf58 PXDN INHBA LOC401152 IRX1 Kua GPI ELF3 SCARB1 FOS PKP2 DAPK2 RefSeq number NM_058246 NM_080829 NM_020210 NM_003225 NM_006096 NM_013332 NM_022648 NM_003670 NM_003670 NM_006931 NM_003082 NM_058246 NM_002564 NM_182507 NM_016337 NM_184041 NM_004767 NM_201283 NM_001013842 NM_012293 NM_002192 NM_001001701 NM_024337 NM_199129 NM_000175 NM_004433 NM_005505 NM_005252 NM_001005242 NM_014326 Chromosome 7 20 15 21 8 7 2 3 3 12 14 7 11 12 14 16 1 7 8 2 7 4 5 20 19 1 12 14 12 15 Start 156986122 48720413 88509900 42669493 134456898 127882809 218226277 4995548 5003471 8014461 61287435 156890452 72583304 50823667 99628951 29984146 200320701 55216313 22511476 1614654 41579152 120441169 3979121 48215760 39541912 200258697 123888666 74791304 32988395 62024988 End 156987013 48720792 88510359 42669912 134457355 127883250 218226673 4996080 5003914 8014804 61287871 156890887 72583693 50824115 99629206 29984554 200321105 55216811 22511809 1615038 41579497 120441511 3979463 48216285 39542320 200259111 123889062 74791735 32988693 62025281 HIF indicates hypoxia-inducible factor. Conversely. interactions with other transcription factors have been defined that might result in DNA-binding without HIF-1␤ at other sites. Primer sequences are given in supplemental Table 3. of reads per window 268 142 130 119 117 105 103 95 91 91 90 89 88 87 86 85 86 79 79 77 77 75 74 73 72 71 71 70 70 70 Covered by human promoter 1. cotransfected with the pGL3 reporter and ␤-galactosidase reporter. CTGCAAAGACCTTTGAGTTCC. . on the degree of overlap between the 2 datasets.org at JAWAHARLAL NEHRU UNIVERSITY on May 27. Expression of quantitative PCR cDNA was synthesized using the High Capacity cDNA Kit (Applied Biosystems). ERBB receptor feedback inhibitor 1 [ERRFI1] CATGTGACGTCAGAAAGGCT. Expression analysis was performed using the SYBR Green method and the ⌬Ct method. Results Definition of HIF-␣-binding sites HIF-regulated genes15 were used as the “seed” dataset. were cultured in normoxia or 0.

This revealed the existence of 356 overlapping HIF-1␤ peaks. overlap with HIF-1␤ was 95% and 94. when the top 200 HIF-1␣ and HIF-2␣ sites were considered. Distribution of HIF-binding sites. In each case. for instance. Canonical pathways statistically over-represented among HIF-binding loci are shown together with the respective Ϫlog10 (P value) for each isoform. suggesting that these sites are indeed mainly composed of false positive signal (data not shown). Furthermore.5% respectively. very strong concordance between HIF-binding sites. For personal use only. after which overlap fell to a background level. The very high level of concordance between HIF-␣ and HIF-1␤ binding is consistent with very accurate definition of heterodimeric binding complexes by ChIP-seq and suggests that under these conditions the large majority of HIF-␣ binding is complexed with HIF-1␤ and viceversa. independently of the coverage afforded by tiled .D).hematologylibrary. 9 JUNE 2011 ⅐ VOLUME 117. in cells treated with either DMOG or hypoxia. HIF-1␤-binding sites were ranked according to peak height. these findings are consistent with the large majority of HIF-␣ DNA-binding being in complex with HIF-1␤ and vice-versa. e210 ¨ DEL et al SCHO BLOOD. Consequently. NUMBER 23 Figure 1.From bloodjournal. Frequency distribution of the nearest-neighbor distance between (C) HIF-1-binding gene loci. we again plotted the degree of overlap against rank (supplemental Figure 3B. 2013. Distribution of HIF-binding and HIF-regulated gene loci Distribution in the vicinity of the TSS. we based all subsequent analyses on HIF-␣ sites at which HIF-1␤ was also present. the frequency distribution of nearest-neighbor distances for an equal number of randomly selected control gene loci is shown for comparison. Conversely the top 200 HIF-1␤ sites manifest 95% overlap with HIF-␣ sites (supplemental Figure 3E-F). (F) The Ingenuity Systems Inc (IPA Version 8. We therefore retested a selection of apparently exclusive HIF-␣-binding sites by ChIP– quantitative PCR and found enrichment at only 25%.8-3204) software tool was used to examine the HIF-1 and HIF-2-binding loci for functional pathways. (D) HIF-2-binding gene loci. To test this. and the number that overlapped with high-stringency HIF-1␣-binding sites was plotted against this rank. Similar analysis of HIF-2␣-binding sites revealed 301 overlapping HIF-1␤ peaks. suggests that at the level of DNA-binding these stimuli have very similar effects. The position of HIF-1 and HIF-2-binding sites with respect to the nearest TSS was plotted as a frequency distribution using (A) 5-kb windows and (B) 100-bp windows. The association of apparently nonoverlapping sites with lowerranked sites suggested that nonoverlap might largely reflect false assignment in one or the other dataset. This revealed even higher levels of association for the higher ranked peaks. Taken together. The ability to identify HIF-binding sites. and (E) HIF-regulated gene loci.org at JAWAHARLAL NEHRU UNIVERSITY on May 27.

HNF1A. HIF-binding sites were also identified in intergenic. together with the position of the HIF-binding gene loci among the genes ranked by fold regulation. BLOOD. and (D) greater than 100 kb from the TSS. NUMBER 23 MAPPING OF HIF-BINDING SITES BY ChIP-seq e211 promoter microarrays.0R microarray).broadinstitute. incubated for 16 hours in 1% ambient oxygen and treated either with control siRNA or siRNAs directed against both HIF-␣ subunits. Using 5-kb windows. and ranked according to the fold regulation by HIF-␣. HIF-1 was more likely to bind close to the TSS. the most frequent position of the binding site laid approximately 100 bp 5Ј to the TSS. distant from promoter regions (as defined by the Human Promoter 1.37. the spread was comparable with the distribution of identical numbers of randomly selected genes. Network analysis identified 3 major networks. HIF-binding loci were then examined using the Ingenuity Pathways Analysis software tool (Version IPA 8. However. Twenty canonical pathways were statistically over-represented among HIF-binding loci with several demonstrating bias toward either HIF-1 or HIF-2 binding (Figure 1F). for HIF-1.8-3204). This gave almost identical distributions to the complete sets of binding sites. Distribution across biologic pathways. permits true genome-wide analysis of HIF-␣ DNA binding. with approximately 40% of HIF-1-binding sites lying within 2. . TGFB1. as opposed to approximately 20% of HIF-2-binding sites (Figure 1A). filtered for present/absent calls. 170 (48%) were novel sites. we repeated the analysis using only the 100 top binding sites for each HIF isoform. GSEA was performed for each subgroup of genes (www. MYC/MAX.39 The cumulative ESs. are plotted for (A) gene loci with the closest HIF-1 binding within 1 kb of the TSS. GSEA stratified by distance between-binding site and TSS. (B) between 1 kb and 10 kb of the TSS. we next examined relationships between the position of HIFbinding sites and effects on gene expression. (E) The log2 fold regulation by HIF-␣ is illustrated. indicating that HIF-1 and HIF-2 target genes are randomly distributed among known transcripts (Figure 1C-E). for either HIF isoform. However. For personal use only. 2013.org/gsea).15 were normalized.5 kb of the TSS.org at JAWAHARLAL NEHRU UNIVERSITY on May 27. thus enabling a complete description of the distribution of HIF-binding sites. HIF-binding and gene regulation Because HIF-binding sites were identified in all regions of the genome. including many at great distances from gene promoters. To test this. To exclude the possibility that an inappropriately low threshold might distort the binding distribution of the HIF-2 sites (by including false positives).and HIF-2-binding loci and HIFregulated genes. and intronic and exonic regions. 5Ј-untranslated region. SP1. However. within the core promoter (Figure 1B). These differing proportions probably account for the greater number of novel HIF-2-binding sites identified by ChIP-seq and suggest differing distributions of the 2 isoforms with respect to the promoter. High-stringency HIF-1-binding gene loci were stratified according to the distance to the nearest TSS into 4 subgroups. The full genome-wide coverage afforded by ChIP-seq also permits analysis of clustering of HIF-␣-binding sites across the genome. TP53. and NOTCH1 indicating interaction with these pathways. we observed clustering of both HIF-1 and HIF-2-binding sites close to the TSS. Of the 356 high-stringency HIF-1-binding sites (HIF-1␣ and HIF-1␤) identified by ChIP-seq. revealed non-Poisson distributions. More detailed analysis of the binding patterns using 100-bp windows indicated that. confirming the more widely distributed pattern of HIF-2 binding (data not shown). Distribution across the genome. Expression array data from MCF-7 cells. with a low frequency of binding extending to much greater distances.hematologylibrary. whereas 212 (70%) of 301 HIF-2-binding sites (HIF-2␣ and HIF-1␤) lay outside these regions. Analyzing nearestneighbor distance. the largest containing 59 focus molecules (supplemental Figure 4) with major nodes centered on ERRB2. The high-stringency Figure 2.From bloodjournal. (C) between 10 kb and 100 kb of the TSS. 9 JUNE 2011 ⅐ VOLUME 117. we examined the frequency distribution of HIF-1 and HIF-2-binding sites about the nearest TSS. 3Ј-untranslated region.

hematologylibrary. fusion of DNA sequences spanning these binding sites to a heterologous luciferase reporter gene conferred hypoxic regulation on luciferase activity. e212 ¨ DEL et al SCHO BLOOD.50. (D) Expression-quantitative PCR confirmation of gene regulation in response to 16 hours at 0. using GSEA. normalized enrichment score ϭ 1. To pursue this further. HIF binding regulates gene expression over long genomic intervals.5% hypoxia. (C) ChIP–quantitative PCR confirmation of the ChIP-seq defined binding peaks. analysis of mRNA levels by quantitative PCR confirmed hypoxia-dependent regulation of these transcripts (Figure 3D). nominal P ϭ . confirming the ability of these sequences to mediate hypoxiainducible gene expression (Figure 3F). These next nearest . To examine the relationship between promoter-distant HIFbinding and transcriptional regulation. For personal use only. Family-wise error rate P ϭ .36. 2013. Finally. HIF DNA binding at both the ARRDC3 and ERRFI1 loci is located more than 100 kb distant from the promoter. ChIP–quantitative PCR analysis of these sites using antibody directed against RNA Pol2 demonstrated interaction with RNA Pol2 after 16 hours at 0. 9 JUNE 2011 ⅐ VOLUME 117.69.0. irrespective of the distance between the HIF-1-binding site and the TSS (Figure 2).5% hypoxia. and HIF-1␤ subunits at the (A) ERRFI1 and (B) ARRDC3 gene loci. indicating that HIF-1 DNA-binding can exert effects on gene regulation over distances of greater than 100 kb. (F) Luciferase reporter assay showing activity of a heterologous luciferase reporter gene fused to sequences spanning each HIF-binding site in normoxia and after 16 hours at 0. HIF-2␣. HIF-binding loci were examined for enrichment among HIFregulated genes using GSEA.15 Two major peaks were present at the ERRFI1 locus and one at the ARRDC3 locus.09.0). loci binding HIF-1 and HIF-2 strongly clustered among the genes that were most highly up-regulated by HIF. although in each case these are the nearest documented genes (Figure 3A-B). indicating hypoxia-inducible interaction between these binding sites and the general transcriptional machinery. no enrichment of HIF-binding loci was observed among the genes down-regulated by HIF (supplemental Figure 5). we next examined for association between HIF binding and regulation of the second-nearest transcript. false discovery rate q-value ϭ 0. At HIF-binding sites where the nearest gene was not regulated. These analyses revealed that HIF-1-binding loci were significantly associated with upregulation of the nearest transcript by HIF. Comparable results were seen for HIF-2 (supplemental Figure 6). In contrast. The association was robust even for loci where HIF-1 binding was greater than 100 kb from the nearest TSS (enrichment score [ES] ϭ 0. The number of reads per 100-bp window and the location of each gene are shown.5% hypoxia. Genomic binding of HIF-1␣. we performed additional analyses on 2 of the most distant HIF-binding sites.From bloodjournal. Each peak was confirmed in ChIP–quantitative PCR analysis with high levels of enrichment (Figure 3C).5% hypoxia (Figure 3E). and both have been reported to be regulated by HIF.37 HIF-regulated genes were identified and ranked using expression array data from MCF-7 cells. NUMBER 23 Figure 3. we stratified HIF-binding gene loci according to the distance between the TSS and the nearest HIF-binding site and ran further tests for association with up-regulation by HIF.15 As expected. (E) ChIP–quantitative PCR demonstrating RNA Pol2 recruitment to the ERRFI1 and ARRDC3 HIF-binding sites in normoxia and after 16 hours at 0. Furthermore.org at JAWAHARLAL NEHRU UNIVERSITY on May 27.

45. false discovery rate q-value ϭ 0. the observed base frequency was compared with the expected using a ␹2 test.and HIF-2-binding regions contained the core RCGTG motif with a preference for A over G at the R position (Figure 4A-B). BLOOD. HIF-binding loci appear to be functionally similar (ie. Sequences were extended for 100 bp upstream and downstream of the RCGTG motif and aligned. Enrichment of motifs identified in this way was then examined by determining the relative occurrence of each motif in HIF-binding regions compared with a control set of regions (r1). At each position.01 after Bonferroni correction) are highlighted in gray. Similar GSEA analysis showed significant association with up-regulated genes for all subsets (supplemental Figure 7). HIF-binding loci were then grouped according to whether the HIF-1 binding was in the intron. (C) HIF-1 and (D) HIF-2-binding sites containing a single RCGTG motif were analyzed for sequence conservation beyond this core motif. 5Ј-untranslated region.34 As expected. of phylogenetically conserved motifs within phylogenetically conserved portions of HIF-binding regions compared with control regions (r2). normalized enrichment score ϭ 1. However.From bloodjournal. Taken together. The relative occurrence of each nucleotide at each position is indicated as a percentage of the total by the letter height.0.org at JAWAHARLAL NEHRU UNIVERSITY on May 27. the most enriched motif identified in this way for both HIF-1. The height of each stack is then adjusted to signify the information content of the sequences at that position (measured in bits).26 Our unbiased and more precise pan-genomic identification of more than 400 HIFbinding sites allows us to make a more definitive statement regarding sequence preferences surrounding this core motif using 2 independent approaches. The sequence logos40 for motifs with all 3 enrichment ratios greater than 2 are shown. indicating that HIF commonly “skips” the nearest gene to regulate another more distant promoter (data not shown). 2013. nominal P ϭ . the amplitude of transcriptional up-regulation by HIF bore no relationship to the distance of the nearest HIF-binding site from the promoter (supplemental Figure 8). at a genome-wide level.hematologylibrary. Finally.35 which were in turn tested for enrichment by determining the relative number of occurrences in HIF-binding regions compared with control regions. In both . exon. with the most frequent at the top. upstream from the TSS or downstream from the transcriptional end site. (A-B) HIF-1. HIF-binding gene loci also showed significant association with transcripts regulated by HIF (ES ϭ 0. 9 JUNE 2011 ⅐ VOLUME 117. in an unbiased analysis. NUMBER 23 MAPPING OF HIF-BINDING SITES BY ChIP-seq e213 Figure 4.09.0). and of phylogenetically conserved motifs within all portions of the HIF-1 binding and control regions (r3). they exert qualitatively and quantitatively similar effects on hypoxia-inducible gene expression over large genomic intervals irrespective of position with reference to the regulated gene promoter). the minimal cis-regulatory element required for HIF-dependent transactivation has been defined as the RCGTG motif. Analysis of transcription factor binding site motifs at HIF-binding sites Based on analysis of approximately 50 HIF-regulated genes. Characterization of the HIF-binding motif. The height of each letter is proportionate to its frequency. First.45. these motifs also exhibited extended sequence preferences both 5Ј and 3Ј to this core motif. For personal use only. Positions showing nonrandom base frequency (P Ͻ .and HIF-2-binding regions were examined for commonly occurring novel motifs using the Gibbs Motif Sampler module of CisGenome. these results indicate that. HIFbinding sites were examined for common novel motifs using the Gibbs Motif Sampler module of CisGenome. Family-wise error rate P ϭ .

including the HIF-1␤ homodimer. 9 JUNE 2011 ⅐ VOLUME 117. A is preferred over G. The motif identified among HIF-1-binding regions also identified a preference for C or G 2-bp 5Ј to the R position.hematologylibrary. Thus. Apart from excluding some E-box motifs. no significant difference in base frequency was observed at any position flanking the RCGTG motifs either for the 2 isoforms or for promoter proximal versus distant sites. Indeed. with much lower affinity/occupancy. Both HIF-1. Although the unbiased approach will determine novel motifs. Figure 5A shows the plot for all RCGTGs and reveals the expected alignment against a background of “noise” from the non–HIF-binding RCGTGs. even at sites far remote from promoters. This previously described hypoxia response element44. Importantly. only one of the many RCGTG motifs at the egl nine homolog 3 locus coincides with a normoxic DNAse1 hypersensitivity peak. For personal use only. We therefore correlated our HIFbinding results with DNAse1 hypersensitivity peaks from the ENCODE consortium dataset 1 for Hypersensitivity by Digital DNAse1 in MCF-7 cells (wgEncodeUwDnaseSeqPeaksRep1Mcf7). For both isoforms. For example. and is consistent with the functional importance of these remote sites in gene regulation. NUMBER 23 cases. whereas RCGTG motifs that are not binding HIF will be randomly distributed. are considered DNAse1-hypersensitive RCGTG motifs are 19 times as likely to bind HIF-1 and 22 times as likely to bind HIF-2 as insensitive motifs. preference is seen for a CAC motif between positions ϩ13 and ϩ15. this is observed whatever the rank distance of the HIFbinding sites from the nearest TSS.org at JAWAHARLAL NEHRU UNIVERSITY on May 27. An expectation of these analyses is that correctly identified HIF-binding RCGTG motifs will align at the center of the region defined by ChIP-seq. Furthermore. We therefore wished to determine the extent to which epigenetic mechanisms. preference was seen for T immediately 5Ј to the RCGTG and for C immediately 3Ј to this motif. which is remarkably similar for each. This exclusion may result from competition by other proteins that recognize the E-box and reduce their availability to bind HIF. the high degree of concordance observed between HIF-␣ and HIF-1␤ binding (supplemental Figure 3) implies that either HIF-␣ can only bind DNA as a heterodimer with HIF-1␤ or that if HIF-␣ is able to bind DNA with an alternative partner (eg. Importantly. there are 1 114 925 RCGTGs. However.and 462 (0. or at too few sites to be identified in a genome-wide analysis.and HIF-2-binding regions are relatively GC rich. This technique enables true genome-wide coverage and more precise localization of HIFbinding sites than has previously been possible. genome-wide. predicated on the RCGTG motif. The high degree of concordance between HIF-␣ binding in cells exposed to DMOG. Conversely. the frequency of CAC motifs at other positions downstream of the RCGTG motif was not increased above background (supplemental Figure 9). Therefore.42. reflecting their concentration within promoter regions. e214 ¨ DEL et al SCHO BLOOD.1% of hypersensitive RCGTGs bind HIF-1 and 0. Furthermore.43 Correlation of HIF binding with DNAse1 hypersensitivity against the distance from the center of the HIF-binding region (defined experimentally by ChIP-seq) of either all RCGTG motifs or just the RCGTG motifs that coincided with DNAse1 hypersensitivity sites (Figure 5A-B). then a muchimproved “signal-to-noise” ratio is observed (Figure 5B) consistent with the predictive power of DNAse1 hypersensitivity.05%) were identified in HIF-1.04%) in HIF-2-binding regions. the large majority of HIF-1␤ is bound as a HIF-␣/␤ heterodimer. the large majority of RCGTG motifs do not bind HIF.32 Overall. At position 0 (the position of R). despite this.15 and highresolution mapping of DNAse1 hypersensitivity sites in the same cell line. in particular chromatin accessibility. in these cells under these conditions. the proportion of hypersensitive RCGTG motifs that bind HIF remains low (1. it also indicates that.41 Outside the core RCGTG motif. We then plotted this rank Discussion We describe the first use of ChIP coupled to next-generation high-throughput sequencing to define HIF-␣. Mutation of the CAC motif between ϩ13 and ϩ15 has previously been shown to affect functional activity of the erythropoietin 3Ј enhancer.31. Within the reference genome. an extended hypoxia response element consensus motif is identified. might contribute to the definition of HIF-binding sites. In addition. Preference is also seen for T at position Ϫ1. indicating that other epigenetic mechanisms also probably contribute to the definition of a HIF-binding site. The data therefore suggest that genome-wide analyses of DNAse1 hypersensitivity may be used to assist in the identification of functionally important RCGTG motifs at specific loci. sequences binding HIF-1 versus HIF-2. C is strikingly underrepresented at position Ϫ1. This suggests that DNAse1 hypersensitivity in normoxic cells would be an important predictor of HIF binding to its consensus recognition site.and HIF-1␤-binding sites across the whole genome. When all of the 1 114 925 RCGTG motifs. we ranked the HIF-binding loci defined by ChIP-seq according to distance upstream/downstream of the nearest TSS. and has provided the most complete pan-genomic analysis of the transcriptional response to hypoxia to date. as defined by DNAse1 hypersensitivity. implying that the CACGTG E-box motif is inhibited from binding HIF. accessibility as indicated by normoxic DNAse1 hypersensitivity is a major factor in predicting HIF binding to an RCGTG motif. it is poor at distinguishing the relative abundance of bases in 2 related motifs. and hypoxic HIF-1␤ binding indicates that both stimuli produce comparable genome-wide patterns of HIF DNA-binding. including an AP-1-like motif. this approach also identified additional motifs that occur more often in HIF-binding regions than control regions (Figure 4A-B). sequence analysis does not provide any definite rules to define potential binding sites further.9% bind HIF-2). Using highly stringent criteria. for instance. in a second approach. more than 500 HIF-binding sites were identified across the genome. Many of these sites were outside of regions covered by promoter arrays and were remote . Even allowing for false-negative results arising from the high-stringency criteria used to define these binding sites.45 is the only one that binds HIF in all 3 datasets (Figure 5C). To test this and to explore the extent to which coincidence between DNAse1 hypersensitivity and HIF binding was observed away from promoter regions (where DNAse1 hypersensitivity sites may be predicted to be enriched in any case). The use of MCF-7 cells has allowed comparison with previously reported genomewide studies of hypoxia-inducible gene expression. 2013. as part of an Sp-1-binding complex38) that it does so at the same sites as the ␣␤-heterodimer. of which only 606 (0. indicating that normoxic DNAse1 hypersensitivity is a strong determinant of HIF-binding. sequences flanking each unique HIF-binding RCGTG motif were compared (Figure 4C-D). 271 (76%) of HIF-1-binding sites and 246 (82%) of HIF-2-binding sites overlapped with DNAse1-hypersensitive peaks and HIF-binding regions demonstrated a much higher concentration of hypersensitive peaks than the immediate flanking regions (supplemental Figure 10). When the data are restricted to RCGTG motifs that fall within DNAse1-hypersensitive sites.From bloodjournal.

and DNAse1hypersensitive RCGTG (DNAse1-hypersensitive site RCGTG) motifs are shown for comparison.46 Interestingly.. Indeed. and HIF-1␤ subunits at the egl nine homolog 3 gene locus.” Ingenuity pathways analysis (Figure 1F) demonstrated that many canonical pathways are targeted by HIF with various bias toward either HIF-1 or HIF-2 binding. we cannot exclude the possibility that distant chromatin elements cluster in 3-dimensional space to form hypoxia-inducible “transcription factories. However. For personal use only. the position of each RCGTG motif. unpublished . HIF-binding sites were observed many hundreds of kilobases distant from any documented promoter. relative to the center of the HIF-binding region.R. This lack of clustering of both HIF-binding and particularly HIF-regulated gene loci provides evidence against widespread operon-like regulation of genes or genomic clustering of HIF transcriptional targets.45 The ability to interrogate pan-genomic patterns of DNA binding for HIF. pathway-specific association of particular HIF isoforms was observed even though siRNA studies have suggested that in MCF7 cells HIF-2 appears to have rather little transcriptional activity. compared with promoter-proximal sites. Similar enrichment was observed for both the core HIFbinding RCGTG motif (Figure 4) and for DNAse1-hypersensitive sites (supplemental Figure 10) among these promoter-distant binding sites. Consistent with previously published work. from core promoters.15 suggesting that isoform-specific patterns of HIF binding might extend across different cells and is consistent with post–DNA-binding mechanisms limiting transcriptional activity.org at JAWAHARLAL NEHRU UNIVERSITY on May 27. whereas CpG islands were enriched only at promoter proximal sites (supplemental Figure 10). Normoxic DNAse1 hypersensitivity predicts hypoxic binding of HIF. HIF-2␣.17. Nearest neighbor analysis of HIF-binding and HIF-regulated genes15 revealed a distribution of distances comparable with randomly chosen control loci. NUMBER 23 MAPPING OF HIF-BINDING SITES BY ChIP-seq e215 Figure 5.. (C) Genomic binding of HIF-1␣. raises a question as to what extent patterns of binding are cell-type specific.hematologylibrary. was plotted (horizontal axis). Together with independent experimental validation by quantitative PCR. 9 JUNE 2011 ⅐ VOLUME 117. High-stringency HIF-1binding regions were ranked according to distance upstream/downstream of the nearest TSS (vertical axis). Previous studies of HIF-binding to promoter regions have revealed in the region of 60% concordance between U87 and HepG2 cells. these results indicate that the identification of promoter-distant sites was accurate and probably functionally relevant. whereas genes with a role in Oct4-regulated stem cell pluripotency predominantly bind HIF-2. we compared sites identified in the current work with an unpublished analysis in 786-0 renal carcinoma cells (J. BLOOD. D. pathways involved in carbohydrate metabolism are principally targeted by HIF-1. The number of reads per 100-bp window and the location of the gene together with the positions of the RCGTG motifs. 2013. (A) For each HIF-1-binding region. which is universally expressed in human cells.From bloodjournal.28 To gain early insights into this question at a pan-genomic level. (B) The same analysis was repeated for the subset of RCGTG motifs that coincided with a normoxic DNAse1 hypersensitivity peak (ENCODE consortium dataset 1).M.S.

OX3 7BN.O..J.R. Deng L.S.cgi.gov/geo/query/ acc.org at JAWAHARLAL NEHRU UNIVERSITY on May 27. but no evidence of previously unannotated transcripts. These studies revealed preferences at a number of sites outside the RCGTG consensus. J. This confirms that down-regulation of transcript abundance by the HIF pathways is very largely.R..hematologylibrary. 2010.329(5987): 75-78. This would suggest that such motifs. 2.nih. Nat Genet. The Wellcome Trust Centre for Human Genetics. References 1. suggesting that competitive binding by E-box proteins may exclude HIF from such sites even in hypoxic cells.. Correlation of HIF-binding data with functional studies of the HIF transcriptional response using GSEA revealed the existence of robust statistical associations with patterns of hypoxia-inducible gene expression even over genomic distances in excess of 100 kb. again suggesting that epigenetic processes must determine the binding of HIF to DNA and may be particularly important in constraining the transcriptional response to hypoxia. This finding is.47 This contrasts with relatively promiscuous primary sequence preferences for HIF binding. http://genome. 4. many DNAse-hypersensitive motifs both close to and distant from actively transcribed genes did not bind HIF. Cancer Research United Kingdom.49 and that the pattern of HIF-binding at promoters correlates with markers of normoxic transcriptional activity at these loci. and the Higher Education Funding Council for England. March 2011).From bloodjournal.M. United Kingdom. we could find no evidence of an association with the loci of transcripts that were negatively regulated by HIF. Acknowledgments This work was supported by the Wellcome Trust. Proc Natl Acad Sci U S A. these studies indicated that for HIF-1 binding. Oxford.M.M.J.28 However. Natural selection on EPAS1 (HIF2alpha) associated with low hemoglobin concentration in Tibetan highlanders.358(2):162-168. For personal use only. and irrespective of the presence of an intervening gene that was not responsive to the HIF pathway.uk. Ang SO. Beall CM.R. and D. Yi X.S. if they exist. Expression array data.gov/geo/query/acc. “epigenetic” chromatin markers contribute to the differential binding of the 2 isoforms. including the relative exclusion of the E-box motif CACGTG.ac.nlm. respectively).M. Indeed. Lucas GS. Disruption of oxygen homeostasis underlies congenital Chuvash polycythemia. The improved accuracy and unbiased ascertainment of HIFbinding across the genome also permitted significant refinements to be made to the HIF consensus. Datasets are available at the following Web sites: ChIP-seq data: http://www.R.M.W.107(25): 11459-11464.P. et al.cgi?accϭGSE3188.S. D. . et al..W. C. This is compatible with a model in which most or all of the promoter-distant HIF-binding sites are hypoxia-inducible enhancers that connect with transcriptional complexes at the promoters of target genes by DNA looping over potentially very large distances and thus exert similar effects on the transcriptional activity of these complexes irrespective of physical distance along the chromosome..27-30 In particular. consistent with their physical association with distant promoters (J.R. Liang Y. irrespective of the position of HIF-1 binding at the gene locus. including cancer and anemia. Percy MJ. Correspondence: David R. Notably. and ENCODE Digital DNAse1 hypersensitivity data.S. unpublished observations. 2008.P. indirect with respect to the binding of HIF to DNA. even for residues outside the RCGTG consensus. performed the experiments. Science.. are scientific cofounders of ReOx Ltd. analyzed and interpreted the data and performed bioinformatic analyses. Furlow PW.. The current ChIP-seq analysis confirms and extends observations on the architecture of the hypoxia-inducible response that have been derived from ChIP-chip analyses using tiled promoter arrays. Mole. D. Further work will be required to define physical interactions at individual loci and the temporal sequence of events governing HIF binding and the formation of transcriptional complexes after induction of HIF by hypoxia. Cavalleri GL. N Engl J Med. This suggests that additional epigenetic factors determining the distribution of these enhancer sites in different cellular contexts probably have major influence on the transcriptional response to hypoxia. This revealed similar concordance both within promoters and at promoter-distant sites (overlap being observed at 42% and 36%. 2010.ox.32(4):614-621. however. and C.edu/cgi-bin/hgTables. Conflict-of-interest disclosure: P. Chen H. University of Oxford. Our results set a framework for such analyses. our finding that the majority of HIF-binding sites coincided with normoxic DNAse1 hypersensitivity sites was somewhat surprising and argues against widespread opening of chromatin conformation by HIF binding or hypoxia itself.ncbi.from HIF-2binding motif preferences. S. and J. The remaining authors declare no competing financial interests. are either too pleiotropic or too distant to detect in motif analysis of HIF-binding regions or that nonsequence defined.?accϭGSE28352. March 2011). e216 ¨ DEL et al SCHO BLOOD. Given the importance of HIF to oxygen-regulated physiology and to the pathophysiology of disease.P.J. Authorship Contribution: D.. 9 JUNE 2011 ⅐ VOLUME 117. 3.ucsc. Sequencing of 50 human exomes reveals adaptation to high altitude. et al. even when promoter-distant sites are included in the analyses.ncbi. Because HIF-␣ proteins are reduced by proteolysis to very low levels in normoxia. in pan-genomic studies of RNA Pol2 binding in hypoxic cells. J. J. Interestingly. and P.. being substantially more prevalent at sites remote from promoters. consistent with observations that the HIFbinding site at the erythropoietin locus corresponded to a normoxic DNAse1-hypersensitive region.48 that chromatin accessibility can predetermine glucocorticoid receptor-binding patterns. Compared with some other transcription factors that have been analyzed at the pan-genomic level. 2013. Huerta-Sanchez E. although HIF-2-binding sites were differently distributed from HIF-1-binding sites. NUMBER 23 observations. designed the research. wrote the manuscript. or even completely.R. et al. C. it was not possible to distinguish HIF-1. and P.R.R. A gain-offunction mutation in the HIF2A gene in familial erythrocytosis. the number of high-stringency HIF-binding sites defined across the genome was relatively modest. e-mail: drmole@well. 2002. http://www. irrespective of distance at least up to 100 kb.W.nlm.. identification of the epigenetic factors regulating these and other processes that constrain the binding of HIF to discrete sites will be important in understanding the pathophysiology and therapeutic tractability of this highly pleiotropic pathway. we have commonly observed a hypoxia-inducible association of RNA Pol2 with these loci. and D.R. the effects on hypoxia-inducible gene expression were qualitatively and quantitatively similar. Hirota K.

and other pathways. 2010. 48. Ragoussis J. 14. Razorenova O. Koshiji M. 7. 101(33):12114-12119. Ebert BL. et al. Vokes SA. 44. 88(13):5680-5684. Semenza GL. 2005. 2007. The von Hippel-Lindau tumour suppressor protein: O2 sensing and cancer. Whitelaw ML. 37. et al. Huang LE. Jiang H.23(9):1949-1956. Huang LE. Nucleic Acids Res. 5(5):343-354.102(43):15545-15550. FEBS J. 10(10):R113.26(9):3514-3526. Giaccia AJ. 13. 6. Discovery of functional noncoding elements by digital analysis of chromatin structure. Krieg AJ. Lavoie T. Nat Genet. Bioinformatics. Ji H. 22. 2002. Raval RR. Transcriptional regulation of vascular endothelial cell responses to hypoxia by HIF-1.34(3): 267-273. 2004. embryonic development. Ratcliffe PJ. 26. EMBO J.270(5):781790. Tanimoto K. Yarchoan R. Kuehn MS. Dondeti VR. NUMBER 23 MAPPING OF HIF-BINDING SITES BY ChIP-seq e217 5. 38. Oxygen sensing by HIF hydroxylases. Kageyama Y. 12.110(6):2140-2147. 24.284(25):16767-16775. Shirato K. Nature. 20. Blood. Chromatin accessibility pre-determines glucocorticoid receptor binding patterns. Myers RM. Genome Biol. Eriksson KF. Life with oxygen. 9 JUNE 2011 ⅐ VOLUME 117.141(1): 63-76. and tumor growth. 27. Mol Cell. Hawrylycz M. Sabo PJ. 2009. Concordant regulation of gene expression by hypoxia and 2-oxoglutaratedependent dioxygenase inhibition: the role of HIF-1alpha. Thurman RE. Pete EA.17(6):793-803. Nejfelt MK. 2003. Thurman R. CisModule: de novo discovery of cis-regulatory modules by hierarchical mixture modeling. Fernandez S. Elvidge GP. Wong WH. Manalo DJ. Cullen D. Subramanian A. 49. Hypoxia signalling in cancer and approaches to enforce tumour regression. 2006. 318(5847):62-64. Scigalla P. 2008. Gustafson-Brown C. 1991. Genome-wide identification of hypoxiainducible factor binding sites and target genes by a probabilistic model integrating transcriptionprofiling data and in silico binding site prediction. Cancer Cell. Minamishima YA. An integrated software system for analyzing ChIP-chip and ChIP-seq data. Blood. 2004. Moslehi J. 2005. To KK. 2010. Wong WH. Characterisation of functional domains within the mouse erythropoietin 3Ј enhancer conveying oxygen-regulated responses in different cell lines. Amigo J. A comparative analysis of genome-wide chromatin immunoprecipitation data for mammalian transcription factors. Br J Cancer. Zhou Q. Gleadle JM. Tamayo P. 2006. Ortiz-Barahona A. 45. 1994. 31. Somatic inactivation of the PHD2 prolyl hydroxylase causes polycythemia and congestive heart failure. Bertout JA. Gordan JD. Identification of functional hypoxia response elements in the promoter region of the DEC1 and DEC2 genes. Mootha VK. 17. 21. et al. Schneider TD. Cancer Res. erythrocytosis. 2007. DNA binding and protein interactions of the AHR/ARNT heterodimer that facilitate gene activation.14(6):435-446. Barrett JC. Mol Cell Biol. Nucleic Acids Res. Subramanian A. 19. HIF prolyl hydroxylase inhibition results in endogenous erythropoietin induction. Hypoxia induces transcription factor ETS-1 via the activity of hypoxia-inducible factor-1. 2004.462(7269):58-64. Proc Natl Acad Sci U S A. HIF-1alpha induces genetic instability by transcriptionally downregulating MutSalpha expression. Wong WH.21(12):2151-2156. Hida W. 15. Biochem J. Oxygen sensing by metazoans: the central role of the HIF hydroxylase pathway. Ratcliffe PJ. 2004. 2008. 2013.96(8):1284-1292. Antonarakis SE. Genes Dev. Wang GL. 111(6):3236-3244. Sabo PJ. Camenisch G. 2009. Nat Rev Cancer.26(11):1293-1300. Sato Y. Iyer S. Preferential binding of HIF-1 to transcriptionally active loci determines cell-type specific response to hypoxia. Lando D. 2001. For personal use only. Calaoagan JM. Wenger RH. Microarray analyses of hypoxia-regulated genes in an aryl hydrocarbon receptor nuclear translocator (Arnt)-dependent manner. Target gene selectivity of hypoxia-inducible factor-alpha in renal cancer cells is conveyed by post-DNA-binding mechanisms. Lindgren CM. Linde NS. Covello KL. Integrative analysis of HIF binding and transactivation reveals its role in maintaining histone methylation homeostasis. Wang V. 18. Kuehn H. 2008. 2008. Tian YM. Lal P. BLOOD.2005(306):re12. Nat Genet. Ratcliffe PJ. 10.20(5):557-570. 2008. Inhibition of prolyl hydroxylases increases erythropoietin production in ESRD. Liu MH. Oikawa M. Wallace JC. Regulation of the histone demethylase JMJD1A by hypoxia-inducible factor 1 alpha enhances hypoxic gene expression and tumor growth. Hu CJ. Naranjo S. Blood.275(22):5618-5634.23(23):3251-3253. Kaelin WG Jr. Glenny L. Mol Cell Biol. 2010. Miyazaki K. 25. The impact of O2 availability on human cancer. Pouyssegur J. Mazure NM. Stiehl DP. 2003. Johnson DS. HIF-1alpha induces cell cycle arrest by functionally counteracting Myc. Hypoxia-inducible nuclear factors bind to an enhancer element located 3Ј to the human erythropoietin gene. Ratcliffe PJ. 2006.105(2): 659-669. J Am Soc Nephrol. Sci STKE. 35. Lemieux ME.281(22):15215-15226. Bronson RT. Pescador N. Koshiji M. Mol Cell Biol.18(20):6097-6100. Chan D.3(7): 511-518. Chodosh LA. Schofield CJ. Swanson HI. Kaelin WG Jr.8(12):967-975. Nat Rev Mol Cell Biol. J Biol Chem. Rowan A. Kurosawa H. Villar D. Sabo PJ.106(11): 4260-4265. Horikawa I. Sequence logos: a new way to display consensus sequences. 2006. Ma W. Mol Cell. 390(1):189-197. A nuclear factor induced by hypoxia via de novo protein synthesis binds to the human erythropoietin gene enhancer at a site required for transcriptional activation. Ebrahim O. 43. Rankin EB. Haque M. Semenza GL. Park H. Kehler J. Gene set enrichment analysis: a knowledgebased approach for interpreting genome-wide expression profiles. .30(4):393-402. 277(49):47014-47021. Mole DR. 2007. 40. Proc Natl Acad Sci U S A. 1992. Bardeesy N. Sataur A. 9. Tamayo P. Ratcliffe PJ. 2008. Proc Natl Acad Sci U S A. GSEA-P: a desktop application for Gene Set Enrichment Analysis. 47.38(7):2332-2345. 2005. Eur J Biochem. 28. Genomescale mapping of DNase I sensitivity in vivo using tiling DNA microarrays. J Biol Chem. Nishiyama M. et al.hematologylibrary.8(11):865-873. Pescador N. Pan YF. Gould J. 11. PGC-1alpha-responsive genes involved in oxidative phosphorylation are coordinately downregulated in human diabetes. 2007. Stephens RM. Yu H. et al. Kato Y. et al. Nat Rev Cancer. HIF-alpha effects on c-Myc distinguish two subtypes of sporadic VHL-deficient clear cell renal carcinoma. et al. et al. Wiesener MS. 42. Xia X. 2005.289(1): 39-43. Copley RR. Wynter A. 8. Nat Methods.30(1):344353. Nature.65(8):3299-3306. Pugh CW. 23.org at JAWAHARLAL NEHRU UNIVERSITY on May 27. John S. Simon MC. 2009. et al. et al. 2002. 1990. J Biol Chem. Gorman JJ. Patel SA. Ji H. Chi SM. 41. Xia X. Pugh CW. Biochem Biophys Res Commun. et al. Dayan F. Abe M. Nucleic Acids Res. 2009. Oh H. Proc Natl Acad Sci U S A. 39. 32. 2005. Science. 2011. et al. Mesirov JP. Lau KW.From bloodjournal.1217(3):297-306. et al. 2002. 46. Laderoute KR. Choi SM. Biochim Biophys Acta. Covello KL.441(7092):437443. Mol Cell Biol. Davis DA. Cuevas Y. Hsieh MM. Kaelin WG Jr. Li W. Mootha VK.43(3):264268. Nat Biotechnol. Blancher C. 29. HIF-2alpha regulates Oct-4: effects of hypoxia on stem cell function. 2006. Honda H. Peet DJ. Bernhardt WM. Identification of a functional hypoxia-responsive element that regulates the expression of the egl nine homologue 3 (egln3/phd3) gene. 36. 16. Integration of oxygen signaling at the consensus HRE. Appelhoff RJ. 2005. del Peso L. The response of c-jun/AP-1 to chronic hypoxia is hypoxia-inducible factor 1 alpha dependent. HIF-2alpha. and modest fetal hemoglobin expression in rhesus macaques. Semenza GL. 30. An oestrogenreceptor-alpha-bound human chromatin interactome. Genomewide association of hypoxia-inducible factor (HIF)-1␣ and HIF-2␣ DNA binding with expression profiling of hypoxia-inducible transcripts. Differential gene up-regulation by hypoxia-inducible factor-1alpha and hypoxiainducible factor-2alpha in HEK293T cells. Kawamoto T. Hammer S. Oxygen-dependent regulation of hypoxiainducible factors by prolyl and asparaginyl hydroxylation. et al. Kung AL. Differential regulation of the transcriptional activities of hypoxia-inducible factor 1 alpha (HIF-1alpha) and HIF-2alpha in stem cells. 33. 2006.22(8):2515-2523.101(48):16837-16842. 34. Chem Biol Interact.12(12):5447-5454. Fullwood MJ. Proc Natl Acad Sci U S A. 2008. et al. Simon MC.34(21):e146. et al.

Sign up to vote on this title
UsefulNot useful

Master Your Semester with Scribd & The New York Times

Special offer for students: Only $4.99/month.

Master Your Semester with a Special Offer from Scribd & The New York Times

Cancel anytime.