Disruption and formation of surface salt bridges are coupled to DNA binding by the integration host factor: a computational analysis.

Revealing the thermodynamic driving force of protein-DNA interactions is crucial to the understanding of factors that dictate the properties and function of protein-DNA complexes. For the binding of DNA to DNA-wrapping proteins, such as the integration host factor (IHF), Record and co-workers proposed that the disruption of a large number of preexisting salt bridges is coupled with the binding process [Holbrook, J. A., et al. (2001) J. Mol. Biol. 310, 379]. To test this proposal, we have conducted explicit solvent MD simulations (multiple ∼25-50 ns trajectories for each salt concentration) to examine the behavior of charged residues in IHF, especially concerning their ability to form salt bridges at different salt concentrations. Of the 17 cationic residues noted by Record and co-workers, most are engaged in salt bridge interactions for a significant portion of the trajectories, especially in the absence of salt. This observation suggests that, from a structural point of view, their proposal is plausible. However, the complex behaviors of charged residues observed in the MD simulations also suggest that the unusual thermodynamic characteristics of IHF-DNA binding likely arise from the interplay between complex dynamics of charged residues both in and beyond the DNA binding site. Moreover, a comparison of MD simulations at different salt concentrations suggests that the strong dependence of the IHF-DNA binding enthalpy on salt concentration may not be due to a significant decrease in the number of stable salt bridges in apo IHF at high salt concentrations. In addition to the Hofmeister effects quantified in more recent studies of IHF-DNA binding, we recommend consideration of the variation of the enthalpy change of salt bridge disruption at different salt concentrations. Finally, the simulation study presented here explicitly highlights the fact that the electrostatic properties of DNA-binding proteins can be rather different in the apo and DNA-bound states, which has important implications for the design of robust methods for predicting DNA binding sites in proteins.