Book of Abstracts: Albany 2011
June 14-18 2011
©Adenine Press (2010)
The Role of DNA Shape on Protein-DNA Binding Affinity and Specificity
Gene regulation requires highly specific interactions between proteins and their DNA binding sites. This high level of binding specificity in protein-DNA readout is achieved through the recognition of both linear sequence (base readout) and three-dimensional structure (shape readout) (1). DNA shape is specifically recognized by a variety of protein families, and we have identified two different ways of modulating DNA shape. One widely observed phenomenon is the variation of the shape of the double helix as a function of nucleotide sequence. This mode of sequence-dependent DNA shape was first found to affect the DNA binding specificity of the Hox protein Scr (Sex combs reduced) (2). A general study showed the critical role of DNA shape on the binding of other transcription factors to their target sites and on the genome packaging through nucleosome formation (3).
Whereas the findings on the readout of sequence-dependent DNA shape are based on the analysis of crystal structures, we now expanded the description of DNA shape as binding affinity and specificity determinant including systems for which no structural data is available. We used our Monte Carlo algorithm (4) for predicting the shape of various DNA binding sites. In one study we relate binding affinity of the architectural protein Fis (factor for inversion stimulation) to the shape of high- and low-affinity Fis-DNA binding sites (see poster abstract by T. Zhou et al.). In another study the DNA structure predictions of multiple binding sites of the eight Drosophila Hox proteins and their cofactors Extradenticle and Homothorax, as identified in SELEX-seq experiments, reveal intrinsic DNA shape as a selection criterion contributing to binding specificity (see poster abstract by P. Liu et al.).
In addition to the sequence dependence of DNA shape, we identified a different mode for varying the shape of DNA binding sites. In the case of DNA recognition by the tumor suppressor p53, the base pairing geometry at specific positions of the p53 response element deviates from standard-Watson-Crick geometry (5). Certain base pairs assume Hoogsteen geometry, thus affecting local DNA shape and in turn the interaction with arginine residues that were found to be frequently mutated in human tumors. If applicable to other systems, non-Watson-Crick base pairs would effectively extend the four letter genomic alphabet, demonstrating the importance of considering three-dimensional structure (6).
Molecular and Computational Biology Program