Estudo da aplicação de descritores de densidades eletrônica de proteínas no estudo de suas características químicas e biológicas

Abstract

Proteins are macromolecules composed of aminoacids for which the ?nal geometry, like that of a complex bundle, allows these molecules to be highly versatile and to be occupy a wide array of chemical roles in living cells. The equilibrium geometry of proteins in their native conditions are determined, in great part, by their primary sequence which is itself a product of biological evolution. In this work the Canterakis-Zernike Moments and the Rotationally Invariant Fingerprints have been used to study possible similarity patterns between the electron densities of modern proteins and those obtained from the tertiary structures predicted from their ancestral sequences, obtained from the literature. tRNAAminoacyl Synthetases were chosen as the set of proteins to be analyzed due to the wealth of phylogenetic information about these molecules in the literature, including ancestral sequences obtained through the application of the Ancestral Sequence Reconstruction (ASR) technique. The structures for the modern and ancestral sequences were obtained through the tertiary structure prediction algorithm I-TASSER. Sequences were BLASTed in order to identify possible dimerization sites. The ClusPro tool was used to carry out the molecular docking between the two subunits of the homodimeric proteins.The I-TASSER/ClusPro structure of the proteins were then submitted for Molecular Dynamics runs for the obtention of ?nal structures with conformational stability. The electron densities were obtained from these ?nal strucutures by means of semiempirical single point calculations using the PM7 Hamiltonian implemented on MOPAC2012. The moments of Canterakis-Zernike and their rotationally invariante Fingerprints were obtained via a new algorithm developed in this work for the projection of the analytical densities computed by Gaussian2009 and MOPAC2012 in Canterakis-Zernike space, which is implemented in the DAMQT software. The densities were grouped according to their Fingerprints in order to study whether or not these descriptors could be used to group electron densities according to the evolutionary origins of the proteins from which they had been computed. It was concluded that the electron densities are capable of clustering the proteins according to their dimerization state which, although not directly linked to their evolutionary past, is related enough to warrant further and more detailed investigations.