Other examples include cooperation with Smads, Sp1 or CREB. Additionally, USFs have been shown to complement HIF binding either at neighbouring or identical sites, while collaboration with ETS transcription factors has been proposed to play a role in HIF2a target selectivity. Recent genome-wide approaches relying on experimental and computational identification of HIF binding sites have reported overrepresented transcription factor binding sites in the flanking sequences that might be indicative of transcriptional cooperativity. However, significant differences exist in the overrepresented TFBSs predicted in each study, and the functional significance of these enriched motifs remains unclear. Gene expression profiling indicates that the expression of thousands of genes changes with hypoxia, with vast cell-type differences in the specific genes being regulated. HIF1a ChIP-chip binding Mechlorethamine hydrochloride locations have been reported in cell lines of diverse tissue origin, namely HepG2 hepatocarcinoma cells, MCF-7 breast cancer cells and U87 glioma cells, showing differences in the binding sites identified in each experiment. In previous studies we integrated microarray expression profiling experiments and HIF binding site predictions in a core set of tissue-independent HIF target genes. To further investigate the selectivity of HIF1 binding, in this work we conducted HIF1a ChIP-chip in cervical carcinoma HeLa cells and observed largely non-overlapping binding locations with previous studies. To explore the role of cooperativity in HIF target selection, we integrated HIF1 alpha ChIP-chip binding locations across cell-types with a meta-analysis of gene expression profiles of cells exposed to hypoxia. Computational prediction of enriched transcription factor binding sites in this integrated set suggested several stress-responsive transcription factors as potential HIF1 collaborators. Experimental validation of these predictions in cell-based reporter assays indicates that binding sites for stressresponsive transcription factors other than HIFs, such as CEBPs, contribute to cooperative hypoxic activation of individual targets. To obtain a set of background genomic regions, custom perl scripts were used to exploit the microarray metaanalysis results for the identification of genes harbouring conserved RCGTG motifs but that are unlikely to be modulated by hypoxia. To this end, gene loci that contained conserved RCGTG motifs in their non-coding sequences were first selected. For these genes, each of their probes was examined, and only genes for which all of their associated probes exhibited a mean fold value within 0.25 standard deviations of the global mean in 
each of the 19 datasets employed in the meta-analysis were considered as not induced by hypoxia. The selected coordinates of conserved RCGTG motifs mapping to these loci were extended as previously described for the set of core HIF binding sites. Genomic regions from this collection were further selected to match the frequency of genomic locations found in the core HBR set. Briefly, Perl scripts were used to annotate core HBRs as promoter, 59UTR, intronic or 39UTR genomic locations and to choose, from the whole collection, a random sample according to the proportions of genomic locations found in the core HBR set. Similarly as with the set of core HIF binding regions, multiple sequence alignments corresponding to the selected control regions were retrieved. The length of flanking non-coding sequences was based on evolutionary conservation, as Chlorhexidine hydrochloride indicated by genomic annotation of PhastCons elements. Statistical assessment of sequence motif enrichment in this set of sequences requires comparison with a background set, the election of this set greatly influencing the results of the analysis.