From this set, we selected regions that are unlikely to be transcriptionally modulated by hypoxia, as judged by no differential expression in any of the 16 hypoxia experiments included in our previously reported genome profiling meta-analysis. Finally, a subset from these sequences was chosen that matched the genomic locations and base composition found in the core set. We thereby obtained a custom set of circa 3500 background sequences containing a RCGTG HIF binding consensus. In order to reduce spurious hits, we only considered as positive hits those motifs that were conserved in mammalian species. Fisher’s exact test was applied to these datasets to identify motif Folinic acid calcium salt pentahydrate predictions enriched in the set of core HIF binding regions. Enriched motifs were consistently found across different stringencies and database sets. In addition to HIF PWMs, we found a significant enrichment for PWMs associated to CREB1, FOS/AP1 and NFY. As an independent assessment of enriched motifs that is less dependent on the composition of the core set, we compared the results of the previous analysis with a variable selection approach implemented in the Weka machine learning software. Benzethonium Chloride Correlation-based feature selection was applied to the complete set of high-stringency predictions to detect non-redundant variables able to distinguish between the core and background sets. As expected, a number of the top-ranked PWMs, such as those for HIF1, AP1/ATF3 or NFY were coincident with the Fisher’s exact testpredictions.However, additional enriched motifswere found, probably reflecting an increased predictive power after stratified cross-validation. We next asked whether the TFs associated to the enriched TFBSs may share any common characteristics. Gene annotation enrichment analysis of these enriched transcription factors pointed at stimulus-responsive transcription factors as significantly enriched in core HIF binding regions, and indeed most of the identified DNA-binding proteins have been reported to function as transcription factors of stress responses, including hypoxia-responsive TFs. On the whole, our results suggest that binding sequences of several additional TFs other than HIFs, and in particular diverse stress-responsive TFs, are enriched in bona fide HIF binding regions. The complete elucidation of the molecular principles governing the translation of genomic information to gene regulation remains a central question in biology. In particular, understanding the mechanisms dictating target selection by HIF transcription factors is of fundamental importance to truly dissect the genes directly modulated by HIFs, and therefore to completely characterize the transcriptional response to hypoxia that these factors orchestrate, and its interactions with other transcriptional pathways. Several mechanisms have been proposed to contribute to selective DNA binding and gene regulation by transcription factors with largely generic DNA binding domains, among them the co-binding of several transcription factor molecules. In order to dissect these mechanisms, high-quality collections of binding sites are an obvious pre-requisite. The recent development of highthroughput chromatin immunoprecipitation experiments has spurred knowledge on the genome-wide DNA binding locations of transcription factors, and these techniques hence constitute an essential tool to explore mechanisms of transcriptional regulation on a global scale. In this work, we employed an integrative approach to identify additional transcription factors that could contribute to HIFs binding and target selectivity. This strategy was based on computational prediction of enriched sequence motifs in a set of core HIF binding regions constructed through selection of HIF1 alpha binding locations derived from genome-wide chromatin immunoprecipitation experiments in HeLa, HepG2, MCF-7 and U87 cells.