Analysis layered structure of proteins and their amino acid interactions by constructing residue networks

Official URL: http://risc01.sabanciuniv.edu/record=b1586655 (Table of Contents)

Determining the factors in protein stability and identifying how amino acid types contribute to the protein folding mechanism are hot topics in biomolecular sciences. By determining these factors and amino acid types, protein structures and their functions can be better understood. Our motivation is to detect the importance of the net charge distribution and the amino acid interactions in protein stability and to find important amino acid types that play key roles in protein folding. In the first part of this thesis, the stability of proteins is inspected by analyzing the charge distribution. In the second part, interactions in proteins are evaluated by constructing amino acid networks. Once generated, the redundant part of these networks is deleted by breaking the interactions between amino acids choosen by several sets of predetermined criteria. These subnetworks are compared to each other through analyzing their matrix structures by making use of singular value decomposition. In the first part of the thesis, the structure of the protein is examined according to the charge distribution at different accessible surface areas (ASAs) and depths of amino acids. ASA does not discriminate between atoms just below the protein surface and those in the core of the protein. In order to differentiate the location of such buried residues, the depths of amino acids from the protein surface are calculated. It is inferred from these calculations that there is a layer composition of protein structures. At the innermost and the outermost parts of the protein, there is a net negative charge, while the middle has nearly neutral. This layered composition gives stability to the protein. Also, the ASA and depths are evaluated based on subsets of proteins with different secondary structure types. It is inferred from these analyses that the layered structure does not display any tracktable differences for different protein secondary structure types (α, β, α/β, α+β); i.e. this distribution is universal to all proteins. In the second part of thesis, the amino acid network of all proteins are constructed using each amino acid as node and the interaction between them as an edge. The interactions between amino acids are deleted (either randomly, or according to predetermined rules) to identify the redundancies in the network. We conclude that when seventy percent of the edges is randomly deleted, the path length does not change more compared to the full amino acid network. In addition, the amino acid subnetworks which consist of specific type of amino acids are constructed and these subnetworks are compared according to their similarity to the whole amino acid network. The subnetworks which are composed of long-range interactions between the hydrophobic amino acids only are more similar to the amino acid networks constructed from using all interactions. This result supports a protein folding mechanism where a hydrophobic core is formed, followed by the rearrangement of the rest of the amino acids around this core. Also, all amino acid subnetworks and whole amino acid network are evaluated by projecting their contact maps onto a set of large eigenvectors that are related to the collective motions in proteins. The projection results show that the hydrophobic amino acid subnetwork is more similar to full amino acid network in this projected space. The amino acids in this subnetwork are also found to be more conserved than others.