Improved Cross-Linking Mass Spectrometry Algorithms for Probing Protein Structures and Interactions

Abstract

Proteins are the most active molecules in living bodies. They catalyze chemical reactions, provide structural support for cells and allow organisms to move. Their function is intrinsically linked to their folded structure. Resolving the structures of proteins and protein complexes is crucial for our understanding of basic biological processes and diseases. Cross-Linking Mass Spectrometry (XL-MS) is a method to gain structural insights into protein complexes. The field of XL-MS data analysis software is not yet as established as many other methods in proteomics. XL-MS analysis software has significant room for improvement in terms of sensitivity, efficiency and standardization of file formats and workflows to facilitate interoperability and reproducibility. In this thesis we present a new XL-MS search engine, OpenPepXL. We develop an algorithm that scores all candidate cross-linked peptide pairs and is efficient enough to be used on a standard desktop PC for most applications. OpenPepXL supports the standardized XL-MS identification file format defined as a part of the MzIdentML 1.2 specifications that were developed in collaboration with the Proteomics Standards Initiative. We benchmark OpenPepXL against other state-of-the-art XL-MS identification tools on multiple datasets that allow cross-link validation through structures or other means. We show that our exhaustive approach, although not the quickest one, is superior in sensitivity to other tools. We suggest this is due to some tools improving their processing time by discarding too many candidates in early steps of the data analysis. We apply XL-MS analysis with OpenPepXL to multiple protein complexes related to meiosis and the type III secretion system. The first project involved several proteins with unknown structures, some of which are expected to be at least partially intrinsically disordered and therefore difficult to investigate using most traditional structural research methods. Unfortunately, we could not find cross-links between the interaction sites we were interested in the most, but we were able to identify many others in these complexes and gained some structural insights. In the second project we used the photo-cross-linking amino acid pBpa to test very specific hypotheses about interactions within the type III secretion system. We were not able to gain any new structural information yet. However, we could confirm that this is a viable approach. It is possible to identify cross-links between a pBpa residue incorporated into a protein sequence and a residue it cross-links to on a residue level resolution.