(5br) Describing DNA at the Nanometer Length-Scale

Although any DNA segment can form nucleosomes in vitro, the relative affinity of histones for different DNA sequences may differ by as much as 1000-fold (> 4 kcal/mol in free-energy). Recent experimental studies of nucleosome positioning have established a link between the affinity of a DNA segment for the histone protein complex and its sequence. Crystal structures of nuclesomes show no direct contacts between DNA bases and the histones; therefore, any influence of sequence on nucleosome positioning should be indirect. It has been suggested suggested that certain sequence patterns could influence nucleosome positioning by facilitating the tight bending required in DNA to wrap around the histones. Sequence could facilitate this bending either by creating permanent kinks in the segment (modification of ?bendedness?) or by reducing the amount of energy required to form a bend (modification of ?bendability?).

Experimental efforts to study the effects of sequence on DNA conformation have provided some understanding of the problem, but the complexity of the experimental systems has hindered attempts to distinguish bendedness from bendability in the experimental data. On a theoretical level, the study of sequence effects on DNA conformation requires the ability to simulate large systems during long timescales and in a diverse set of conditions (different sequences, temperatures, ionic strength, etc.). Even with current technological advances, it still becomes computationally expensive to meet these conditions with the use of atomistic molecular dynamics. Attempts to provide a theoretical underpinning using coarser models have been limited, largely because models capable of describing the mechanical properties of dsDNA beyond the elastic limit and its ability to de-hybridize locally on a sequence-dependent manner were not available until recently.

My research has focused on developing a mesoscale DNA model that can reproduce experimentally observed sequence-dependent flexibility. The model has been parameterized to reproduce the experimentally observed melting temperature of DNA, and it can describe the mechanism of DNA hybridization. As an application for this model I have employed umbrella sampling molecular dynamics simulations to examine the mechanical properties of DNA as a function of sequence. Because the model allows for de-hybridization, it becomes possible to examine the effects of sequence on ?pure'' bending and on the overall stability of the DNA double-helical structure when the molecule is subject to severe bending constraints. Omitting the histones from the simulated systems allows discerning whether the sharp bends observed in crystal structures of nucleosomes are a cause of or an effect from the formation of the complex. Results demonstrate that when bent, DNA has a sequence-dependent preference for hybridization and orientation, even in the absence of a protein. Analysis of the results in the context of previous experimental observations suggests a mechanism for nucleosome positioning in which sequence can determine the affinity of a DNA segment for the histones by altering the direction (via anistropic bending) and lifetimes (via non-native hybridization) of the bends it adopts.