Protein-induced bending of DNA plays an important role in regulating its transcription, replication, recombination, and packaging into nucleosomes. In particular, many general and gene-specific transcription factors bend DNA. The TATA box-binding protein (TBP) is essential to promoter recognition, and it provides an especially interesting example of dramatic DNA bending.

Previous work in our laboratory has shown that TBP bound to strained DNA can induce negative supercoiling, proposed to be the result of untwisting without the compensating writhe provided by the Phe stirrups. The structural proposal makes the clear predictions that TBP lacking the Phe stirrups will induce negative supercoiling under all conditions, and that the mutant may require negative supercoiling in order to bind at all.

To test this prediction, we have made the F99A-F116A and F99A-F116A-F190AF207A site-directed TBP mutants that lack the N and C-terminal stirrups. We have characterized the binding of wild type and mutant TBP to linear DNA and to negatively supercoiled minicircles using quantitative hydroxyl radical footprinting, which is the first application for circular DNA, and electrophoretic mobility shift assays (EMSA).

The results of the quantitative hydroxyl radical footprinting and EMSA suggest that mutant TBP binds better to the negatively supercoiled minicircle than to the linear DNA. We also observed quite different footprinting patterns for circular versus linear DNA at the TATA box. This indicates that the structure of TBP bound to minicircles may be different than the linear DNA. The equilibrium dissociation constants (Kd) of wild type TBP derived from hydroxyl radical footprinting titrations for the linear DNA and the -1 topoisomers of 203 bp are 11 nM and 3 nM respectively. This suggests that pre-bending of the TATA box enhanced DNA binding. From these observations, we propose that TBP binding to promoters upon gene activation may be enhanced by the topological strain induced in the DNA upon chromatin remodeling.