Molecular modeling and thermodynamics simulation of nucleic acids

Abstract

Nucleic acids participate in many cellular processes. DNA is responsible for gene heredity and its structure is mainly in double helix, whereas RNA has wide functions in gene transcription and regulation so its structures are varied among species. RNA modifications which are known for their abundance and chemical diversity further increase the conformational variability. Functions of some RNAs closely tie to modifications. For example, modified nucleotides maintain correct tRNA structure so that enzyme and ribosome can recognize the tRNA in protein translation. Few epigenetic modifications are also found in DNA, such as 5-methyl cytidine. More often artificially modified DNA, like locked nucleic acid (LNA), is applied to alter the binding affinity of DNA duplex and triplex. Starting from the structures solved by experiments or modeled by programs, molecular dynamics (MD) simulations are employed to mimic the dynamic process and compute the thermodynamic properties, so that the structure and function of nucleic acids can be better understood. This thesis covers computational studies of both RNA and DNA structures.

In paper I, the naturally modified ribonucleotides are parameterized in an additive CHARMM force field. The parameters are targeted on quantum chemistry data. The charge and dihedral parameters are fine-tuned for some molecules to reproduce the experimental conformation. This force field allows wider computational studies on modifications involved RNA molecules.

In paper II, the new force field is used in the simulations of four tRNAs. The results show with modifications the structural stability, nucleotide conformation and base pair maintenance are almost better than those without modifications, especially in dihydrouridine loop and anticodon loop. The enhanced stability by magnesium ions is also observed.

In paper III, MD simulations combined with electrophoretic mobility shift assay illustrate the LNA effects in DNA helical structures. The results show LNA substitutions in duplex strand or the third strand improve the triplex formation, because LNA pre-organizes the DNA strands to reduce their structural adaption required upon triplex forming.

In paper IV, a method is developed to calculate free energy for LNA. The angle energies are transformed to convert the locked ribose to deoxyribose. The protocol can be in one-step or three-step by transforming bonded and nonbonded energies separately. Both protocols solve the reasonable solvation free energy and are expected to be applied in larger systems.