Study sequence rules of estrogen receptor α-DNA interactions using dual polarization interferometry and computational modeling

Estrogen receptor α (ERα) is a ligand-activated transcription factor. In a classical model, ERα regulates gene expression by binding to DNA sequences called estrogen response elements (EREs). A perfect ERE contains a palindromic consensus sequence of 5′-GGTCAnnnTGACC-3′. A slight variation in ERE sequence alters ERα binding affinity and, thus, the gene transcription activity. In this study, all possible singly mutated EREs of 15 sequences (three possible base substitutions at each of one to five positions of one half-site) were created. Dual polarization interferometry (DPI) was used to measure the receptor binding to generate an in vitro binding energy model. A motif discovery algorithm, Thermodynamic Modeling of ChIP-seq (TherMos), was used to compute the binding energy model from in vivo genome-wide ERα binding data. The in vitro affinity model measured by DPI correlates very well with the TherMos prediction (in vivo model), with a rank correlation coefficient of 0.91, which indicates that the DPI-determined model is reliable and powerful in understanding of ERα binding in vivo in the whole genome. This is the first report of DPI study of protein-double-stranded DNA (dsDNA) interactions. The assay protocols developed are efficient for screening a large quantity of DNA sequences with single base variation sensitivity.