Parmbsc1: Parameterization and Validation of a new State-of-the-art Force Field for DNA Simulations

Abstract

[eng] Classical force fields are the core of classical simulations, particularly of molecular dynamics (MD), a technique that is changing our view on the structure, flexibility and function of biological macromolecules. Originated from the pioneering work of Lifson’s group in the sixties, force fields have been in continuous evolution, improving in each generation the accuracy in the representation of proteins and nucleic acid. Force field development is tightly connected to the refinement of simulation procedures and to the extension of simulation time scales. Thus, as simulation time passed the microsecond barrier, MD simulations have revealed the existence of some errors in the default force field for DNA simulations, parmbsc0 (developed in the group). The goal of this thesis is to address these problems by a reparameterization of AMBER force field that aims to represent a wide range of DNA structures under physiological and non-physiological conditions. Keeping α/γ parmbsc0 corrections and parm99 non-bonded parameters, we systematically reparameterized sugar puckering, ε, ζ and χ torsions using high level QM calculations both in gas phase and solution. The refined force field has been tested for more than 3 years to an unprecedented level of detail, considering a large variety of DNAs, and analyzing structural, mechanical and dynamical properties of the DNAs resulting from the corresponding MD simulations. The refined force field parameters have been also subjected for more than 1 year of β-testing by different groups, finding to our knowledge no major drawbacks. In the world of RNA simulations, despite the recent efforts to improve the description of RNA in MD simulations, RNA force fields are still far in accuracy from those of DNA. A probable cause could be the incomplete understanding of the mechanism of 2’-OH orientation, which in big extent determines the RNA conformation and most probably serves as the molecular switch. THESIS ORGANIZATION This thesis is compiled of five publications (or in the process of publication) works; first three consider DNA force field development and following validation and benchmark while the last two are focused on RNA efforts. For better understanding of this work Chapter 1 introduces the central concepts related to nucleic acids, their structures and ways to study them. Chapter 2 goes into more details of the methodology employed here, briefly explaining basic QM formalism and MD simulations with an emphasis on force fields. Chapter 3 is a small handbook of methods employed in the analysis in this work. All together first three chapters should provide a solid ground to better understand the details and the relevance of the five publications in the following two chapters. Chapter 4 is based on the development of new force field, called parmbsc1, its further testing on the Drew-Dickerson sequence and benchmarking. Chapter 5 focuses on efforts to understand the mechanism of complexity of RNA structures studying 2’-OH rotation, and computational design of a new RNA dumbbell structure. A summary of the major results and a general discussion that reflects on the five projects and future work are presented in Chapter 6, with the main conclusions at the end of this work.