Study On Some Problems Of Protein Design And Structure Simulation

In this dissertation we focus on protein design and structural simulation.We have studied a series of problems on computational biology to reveal the realationship of structure and function of proteins,and the mechanism of protein activity.we research some problems in different aspects including design the backbone structure of the pro-tein from scratch,and redesign the protein-ligand binding specificity,and fit the atomic structure model to the cyro EM density map and simulate the paths of allosteric struc-tural changes.The design of the backbone of the protein is a core issue in the current field of pro-tein design and a challenging topic.Although more and more successful examples have emerged in recent years,there is still no systematic solution.One acceptable way to design a backbone structure from scratch is to develop a sequence-independent energy function and use this conventional energy function to generate acceptable backbone structures using 1 conformation sampling and optimization techniques.In this disser-tation,we proposed a statistically-based energy function called tetraBASE,which is a statistical energy model for modeling the stacking of peptide chains in space.We have introduced a novel expression of the backbone stacking interaction,which displays the stacking of the two backbone sites depending on the local conformational information.This indicates that the energy of the stack is orientation-dependent.For this model,we performed two types of tests,including the stability of the backbone of the native pro-tein secondary structural elements in our model and whether the low energy structures produced by our model can reproduce the secondary structural elements in the native protein.Combining rational design with experimental screening is an effective method of engineering protein molecules.Starting from the more and more abundant information on protein structure,we can conduct in-depth exploration of the interaction mechanism of protein-protein,protein-ligand,etc.In this dissertation,we started from the known protein sensor that bind NADH.Through a large number of structural comparisons,we rationalized the binding sites and transform them into NADPH sensors.By analyzing the preference of amino acid residues in the binding pockets of both NADH and NADPH binding proteins,we made reasonable predictions about the binding mechanism of the two types of binding proteins.We analyzed the hydrophobicity of binding specific pockets and the distribution of charged residues at specific positions to determine the binding affinity and specificity of the two molecules.Based on the combination of site structure comparison and analysis,we have designed several mutations that may change the substrate specificity based on the NADH sensor,and verified by the collaborators'experiments.Cryo-EM technology has made great progress in recent years and has become an advantageous tool for the analysis of the structure of biological macromolecules.This technology has produced a large amount of data,providing a vast space for the devel-opment in computational biology.For the macromolecular complex structure obtained by the cryo-EM,fitting the atomic level model to the electron density map may be the most extensive way to study its structure and function mechanism.In this dissertation,our collaborators analyzed the cryo-electron microscopic structure of the ATM/Tell ki-nase that is critical in the process of DNA damage signalling.Based on this,we used a variety of structural modeling methods to build an atomic level structural model.The structural model was flexible fitted to the electron density map of the cryo-EM.For macromolecules with a large number of amino acid residues such as ATM,we have adopted a divide-and-conquer method.We have used the homology modeling tech-nique to build models for those parts that have similar homologous structures,and for those parts that lack homologous structures,we have the primary structure sequence pre-dicts its secondary and tertiary structure.We used the method of molecular dynamics flexible fitting to fit the atomic model with the density map and achieved good results.The use of molecular dynamics simulations to study the large-scale structural changes of proteins is currently facing many difficulties.Force field accuracy and time-scale is-sues have limited us to use molecular dynamics simulation to study such rare events.However,for some specific problems,we can have some special solutions.For the al-losteric regulation of the protein which the initial and final states are known,we can study the mechanism by path sampling.In this dissertation,we use the Finite Temper-ature String(FTS)method to study the structural changes of ribo-binding proteins.We have adopted a series of new methods to ensure the smoothness of the string when ap-plying the FTS method to macromolecular systems.After obtaining the transition path of low free energy in the collective variable space,we have introduced the EDS method to verify the rationality of the path obtained by the FTS method in the high-dimensional space.