Analysis of the DNA-binding affinity, sequence specificity and context dependence of the glucocorticoid receptor zinc finger region

The glucocorticoid receptor binds with high affinity to glucocorticoid response elements (GREs), which commonly consist of imperfect DNA palindromes with hexameric core binding motifs separated by three base-pairs; the receptor dimerizes upon binding to these sequences, with one monomer occupying each core motif. To examine quantitatively the receptor-DNA interaction, and to characterize the effects of single base-pair mutations and sequence context on this interaction, we have studied the binding of a purified glucocorticoid receptor fragment, T7X556, to a 30 bp oligonucleotide containing a complex arrangement of potential receptor binding sites (RBS): two consensus core motifs, RBSo and RBSa reside in the same orientation and overlap by one base-pair, and a third site, RBSb, a partial match to the consensus separated by three base-pairs and in opposite orientation from RBSa. Using four different footprinting reagents, we detected a receptor dimer bound to one face of the DNA double helix, making close contacts with two adjacent major grooves. One monomer bound with high affinity to RBSa and the other bound with lower affinity to RBSb; transfection studies revealed that both binding sites were necessary for GRE activity in vivo. We measured the affinity of the receptor interaction with RBSa, RBSb and non-specific sites, and showed that protein binding at RBSb was improved approximately 150-fold by cooperativity with RBSa. Remarkably, T7X556 failed to bind to the consensus RBSo sequence, even when it was mutated to match exactly RBSa; the preference for RBSa over RBSo was due neither to the presence of RBSb, nor to occlusion of RBSo by protein bound at RBSa. To examine the relative contribution of each core nucleotide to the binding reaction, we saturated RBSo and RBSa with point mutations. The results implied that four RBSa nucleotides, (or 5'- x G x ACA-3'), may make specific contacts with the protein, as certain mutations at these positions reduced binding drastically. Consistent with the footprinting and transfection assays, equivalent mutations in RBSo had no effect on protein binding. Thus, these findings indicate that the consensus core motif alone is not sufficient to specify a functional RBS, and that flanking sequences create an appropriate context for protein binding.