Application of mass spectrometry data in protein structure modeling

Author(s)

Amanda Miquilini Cordibello Marques

• Document type: Master's Dissertation
• Press: Campinas, SP.
• Institution: Universidade Estadual de Campinas (UNICAMP). Instituto de Química
• Defense date: 2021-02-26

Examining board members

Fábio Cesar Gozzo; Leandro Martínez; Francisco Gomes Neto

• Advisor: Fábio Cesar Gozzo

Abstract

Proteins are the most versatile and complex macromolecules of biological systems, being a part of basically all the processes occurring in a cell. Because the function of a protein is closely linked to its structure, there is great interest in the conformational study of proteins in order to understand how they develop their functions. The use of mass spectrometry data in the structural characterization of proteins is very attractive due to the intrinsic characteristics of the technique, such as high sensitivity, speed of analysis, and, mainly, its universality of application. Despite the successful application of the cross-linking/mass spectrometry (XL-MS) technique in elucidating the structure of protein-protein complexes, there is still no effective way to use these data in the structural characterization of proteins. In this project, a protocol with an iterative approach that makes efficient use of structural information in protein tertiary structure modeling was developed. As a proof of concept, the protocol was applied to a set of proteins resulting in more accurate modeling than the current methods that use XLMS. In addition, recent advances in protein structure prediction have shown that valuable distance predictions between residues using coevolutionary data can be obtained with deep learning-based methods. Thus, the complementarity between both types of constraints, already proven in recent studies, has also been explored. 24 targets of the CASP13 competition with diverse topologies and structures were used to test the protocol, and the results show that a simple iterative process combined with both types of constraints leads overall to superior modeling performance. In addition, complementary constraints can greatly improve the modeling of targets with an average initial GDT_TS (~ 40 to 60), especially those with larger sequences (> 150 residues), reinforcing the relevance of the method. Finally, the method developed here paves the way for the efficient use of constraints from XL-MS experiments in the structural refinement in protein modeling (AU)

• FAPESP's process: 19/07961-2 - Implementation, evaluation and application of mass spectrometric data in protein structure modeling
• Grantee: Amanda Miquilini Cordibello Marques
• Support Opportunities: Scholarships in Brazil - Master