Studying macromolecular complexes by enhanced cross-linking mass spectrometry and modeling approaches

Abstract

The ability of mass spectrometry (MS)-based proteomics to identify and quantify thousands of proteins has fundamentally altered the way in which biological systems are studied today. Whether individual proteins, protein-protein interactions (PPIs) or system-wide studies of whole organisms, all levels of biological complexity can be addressed using different MS-based approaches. Chemical cross-linking coupled to mass spectrometry (XL-MS) is an increasingly relevant method designed for detecting PPIs and can provide useful information for modelling 3D structures of proteins and protein complexes. The general approach of XL-MS fundamentally relies on covalent bond formation facilitated by cross-linking reagents between proximal functional groups, typically amino acid residues within proteins. Identification and quantification of cross-linked peptides from MS are performed using specialized software tools such as xQuest and pLink. The resulting cross-linking data can subsequently be leveraged to localize protein domains within structural modeling platforms, such as the Integrative Modeling Platform (IMP). Nevertheless, effective identification and characterization of PPIs via XL-MS in a cellular context remains challenging. Significant hurdles include limited membrane permeability of conventional cross-linkers, the complexity and potential errors associated with peptide assignment, and insufficient integration between experimental datasets and artificial intelligence (AI)-based protein structure predictions for modeling of protein complexes. To address these problems, the approach of XL-MS was extended in multiple directions in this cumulative thesis, i) by the synthesis and application of a new cross-linker, ii) an optimized filter to enhance accuracy of XL-MS data and iii) by combining cross-linking data with AI-based structural prediction within IMP to enhance the precise modeling of protein complexes.