Çevresel etkiler altında konformasyon değişikliği gösteren peptiler için aktarılabilir düşük çözünürlüklü model geliştirilmesi

Proteins structure and dynamics is strongly coupled to their functions. In order to study the dynamics and consequentially functions of these biomolecules, in addition to experimental and other theoretical approaches, molecular dynamics can be employed as a powerful tool. However, MD simulations are limited by the system size and simulation time. Processes like folding to native structures or aggregation take place on microsecond to second time scale, which is not accessible in MD simulations. In order to overcome the time and length scale obstacles coarse grained model has been employed as an alternative technique. When it comes to coarse-graining, transferability of coarse-grained (CG) model is one of the key challenges. Most of the state of the art CG models are parameterized to represent a single state point, and their validity upon changing the state point (different temperature, concentration, conformational change, or aggregation) is questionable and has to be carefully analyzed before utilizing. In this thesis, we have developed a transferable coarse-grained model for proteins and peptides that can reproduce folding, aggregation and partitioning behaviour observed at hydrophobic/hydrophilic interfaces. To this aim a model synthetic peptide composed of leucine and lysine residues (with the sequence LKKLLKLLKKLLKL) has been used. LK peptide is designed to have built-in secondary amphiphilicity upon forming an alpha-helix. Recent studies have revealed that LK adopts several different secondary structures in solution (alpha-helix, beta hairpin, etc.), however none of them can dominate the conformational phase space. At a macroscopic (e.g air/water interface) or molecular (hydrophobic/hydrophilic interfaces created by neighboring molecules) interface, on the other hand, alpha-helix conformation dominates eliminating all other possible alternatives. Our model is an implicit water model where the backbone of the peptide is represented by two beads and sidechains are represented by separate beads. The bonded interactions are tuned according to the Boltzmann distributions obtained from vacuum and tetramer simulations. The three main nonbonded interactions that play a key role in the model are: backbone hydrogen bonding, electrostatic interactions due to lysines, and leucine hydrophobic attraction in water and at the air/water interface. We have parametrized these interactions by using structural references derived from atomistic simulations. This model is verified by comparing the CG and atomistic potential of mean force curves for extension of the alpha-helix conformation in bulk water, adsorption and folding at the air/water interface, and finally aggregation of LK peptides in bulk water. Hence, our transferable coarse-grained model is capable of representing the behavior of the peptide in three different states with a single set of parameters.