CCCTC-binding factor confines the distal action of estrogen receptor

Distal enhancers have recently emerged as a common mode of gene regulation for several transcription factors, including estrogen and androgen receptors, the two key regulators of breast and prostate cancer major subtypes. Despite the rapid success in genome-wide annotation of estrogen receptor-alpha (ERalpha) binding sites in cell lines, the precise mechanism governing the gene-to-enhancer association is still unknown and no quantitative model that can predict the estrogen responsiveness of genes has been hitherto proposed. This article presents an integrative genomics approach to construct a predictive model that can explain more than 70% of estrogen-induced expression profiles. The proposed method combines a recent map of the insulator protein CCCTC-binding factor (CTCF) with previous ER location studies and expression profiling in the presence of the translation inhibitor cycloheximide, providing evidence that CTCF partitions the human genome into distinct ER-regulatory blocks. It is shown that estrogen-responsive genes with a decreased transcription level (down-regulated genes) have a markedly different relative distribution of ER binding sites compared with those with an increased transcription level (up-regulated genes). Finally, Bayesian belief networks are constructed to quantify the effects of ER-binding distance from genes as well as the insulating effects of CTCF on the estrogen responsiveness of genes. This work thus represents a stride toward understanding and predicting the distal activities of steroid hormone nuclear receptors.