Genome-wide mapping of Hif-1α binding sites in zebrafish

Abstract

Background: Hypoxia Inducible Factor (HIF) regulates a cascade of transcriptional events in response to decreased oxygenation, acting from the cellular to the physiological level. This response is evolutionarily conserved, allowing the use of zebrafish (Danio rerio) as a model for studying the hypoxic response. Activation of the hypoxic response can be achieved in zebrafish by homozygous null mutation of the von Hippel-Lindau (vhl) tumour suppressor gene. Previous work from our lab has focused on the phenotypic characterisation of this mutant, establishing the links between vhl mutation, the hypoxic response and cancer. To further develop fish as a model for studying hypoxic signalling, we examine the transcriptional profile of the vhl mutant with respect to Hif-1α. As our approach uses embryos consisting of many cell types, it has the potential to uncover additional HIF regulated genes that have escaped detection in analogous mammalian cell culture studies.

Results: We performed high-density oligonucleotide microarray analysis of the gene expression changes in von Hippel-Lindau mutant zebrafish, which identified up-regulation of well-known hypoxia response genes and down-regulation of genes primarily involved in lipid processing. To identify the dependency of these transcriptional changes on HIF, we undertook Chromatin Immunoprecipitation linked next generation sequencing (ChIP-seq) for the transcription factor Hypoxia Inducible Factor 1α (HIF-1α). We identified HIF-1α binding sites across the genome, with binding sites showing enrichment for an RCGTG motif, showing conservation with the mammalian hypoxia response element.

Conclusions: Transcriptome analysis of vhl mutant embryos detected activation of key hypoxia response genes seen in human cell models of hypoxia, but also suppression of many genes primarily involved in lipid processing. ChIP-seq analysis of Hif-1α binding sites unveiled an unprecedented number of loci, with a high proportion containing a canonical hypoxia response element. Whether these sites are functional remains unknown, nevertheless their frequent location near transcriptional start sites suggests functionality, and will allow for investigation into the potential hypoxic regulation of genes in their vicinity. We expect that our data will be an excellent starting point for analysis of both fish and mammalian gene regulation by HIF.

Fig. 1

vhl mutants shows specific enrichment for the epo-HRE. Graph showing the result of a ChIP-qPCR experiment. In vhl mutants HIF is stabilised and is expected to bind to HREs. Strong enrichment (22-fold) for the region spanning the epo-HRE is seen in vhl mutants, whereas the wild-type HRE, and flanking regions both up- and down-stream of the HRE are much less enriched. A control region ~2Kb distally was used for comparison. Details for the sequences and relative positions of primers as well as the fold change calculation can be found in the Additional file 14 and the Methods section

Fig. 2

The primary motif found in vhl mutant ChIP-peaks. The most significant 1000 peaks from the sequencing of the vhl mutant ChIP material were analysed for the presence of sequence motifs in the 100 bp surrounding the centre of the peak using MEME. This identified a motif that closely resembles the known HRE consensus [41]. The y-axis shows the Relative Entropy in bits, a measure of the probability that the letter will be found at that position relative to the total information content of the stack, the x-axis shows the width of the motif in base pairs. More information can be found on the MEME website (http://meme-suite.org/doc/examples/memechip_example_output_files/index.html?man_type=web)

Fig. 3

HREs are enriched in the most significant peaks. The peaks from our HIF ChIP-seq experiment were ranked according to p-value, a measure of their height above the (local) background of mapped sequence tags. They were then binned into groups of 1000, the percentage of peaks containing an HRE was then calculated, this showed that the most significant peaks show enrichment for HRE’s

Fig. 4

The RCGTG/HRE motif is enriched in regions in close proximity to the peak summit. The position relative to the peak summit of the total number of HRE’s in the vhl mutant data set was assessed. The X axis displays the distance from the summit in basepairs, and the y-axis the number of HREs that fall into a given window. The majority of peaks are found with ±100 bp of the summit when the surrounding 1 kb was analysed

Fig. 5

The distribution of Hif-1α binding sites in vhl mutant zebrafish. a: Definitions of genomic regions. TSS: If peak summit is 1Kb up or down-stream of the TSS of the closest gene (−1Kb to +1Kb). Promoter: From -5Kb upstream to upstream limit of the TSS (−1Kb). Inside: Inside the gene, including introns and exons but excluding areas covered by the TSS. Proximal: 5 Kb Up- and down-stream of the limits of the promoter and inside region. Distal: 90Kb up- and down-stream of the proximal regions. Desert: Any region up- or down-stream of the distal regions. b: The proportion of peaks containing HRE’s that fall within a given region for vhl mutants. c: The proportions of the genome that are described by the regions defined in a