(638g) Validating a DNA Simulation Model through Hairpin Experiments

We will present a comparison between large-scale melting experiments of DNA hairpins in common biological buffers and the predictions of a coarse-grained Brownian Dynamic computation model (M. Kenward and K. D. Dorfman, Journal of Chemical Physics, 130, 1, (2009)). Most computational efforts to date take a physical approach to modeling single stranded DNA. While this is effective for understanding the fundamental properties of self-interacting polymers (with an eye towards the particular interactions that characterize DNA), such studies are rarely related to laboratory experiments in realistic biological conditions. To address this need, we have studied DNA hairpins with AT or GC stems and relatively long non-interacting loops. This system emphasizes the role of stacking and hydrogen bonding energies, which are characteristics of DNA, rather than backbone bending, stiffness and excluded volume interactions, which are generic characteristics of semi-flexible polymers. On the experimental side, we employed a two-color fluorescence assay where (i) the DNA were end-labeled with Cy5 and Iowa Black in order to obtain a FRET signal which provides a metric of the end-to-end probability distribution function and (ii) Sybr Green I was used as an intercalating dye in order to obtain standard melting curves. Seeking not simply the melting temperature, we generated fluorescence intensity versus temperature curves for both the Cy5 and Sybr channels in a variety of common biological buffers including Tris, PBS, and Buffer A. In conjunction with our experiments, we methodically explored the phase space of our computational model by varying the hydrogen bonding parameters, stacking values, and temperature while fixing the DNA's excluded volume, stiff spring, and bending potentials. The comparison of our simulation and experimental results provides a strong test of our model's suitability for capturing the sensitive behavior of simple hairpin systems. Furthermore, our approach and experimental data can be used to validate other similar coarse-grained simulation models.