In-Depth Analysis of Zero-Length Crosslinking for Structural Mass Spectrometry

The completion of the Human Genome Project revealed the sequence identity of essentially every human protein. However, in most cases, amino acid sequences alone convey little implication on the protein static structures, its dynamic conformational changes, and most importantly, its functions. To fully understand the behaviors and properties of macromolecular complexes, solving their 3D structures is necessary and highly critical. Under this rationale, structural genomics collaborations were initiated aiming to determine high-resolution structures of as many proteins and protein folds as possible, relying mostly on X-ray crystallography and NMR spectroscopy. Yet, very large, highly flexible or disordered, and dynamic protein complexes can exceed the capabilities of these high-resolution techniques. Although computational molecular modeling can be utilized, such structures are highly speculative and often inaccurate unless supported by actual experimental data. Structural mass spectrometry recently emerged as an alternative method which can provide medium-resolution spatial information capable of complementing computational approaches, and are applicable to heterogeneous samples with potentially no limit on complex sizes. In particular, chemical crosslinking coupled with mass spectrometry, has recently received considerable interest. Most recent progress focused on developing crosslinkers with special properties such as enrichment tags, isotopic labeling sites, or MS-cleavable bonds along with accompanying data analysis strategies and software packages. These crosslinkers insert their spacer arm between proximal amino acid residues, greatly reducing the stringency of the derived distance constraints. In contrast, zero-length crosslinkers are crosslinks which do not add any extra atoms to the product crosslinked peptides, therefore providing the tightest possible spatial constraints but rendering enrichment and isotopic labeling strategies inapplicable. As a result, zero-length crosslinking received limited attention and no software tools have previously been specifically developed for it. In this thesis project, we developed a multi-tiered mass spectrometry data acquisition and computational data analysis strategy along with a dedicated software tool to enhance identification of zero-length crosslinks in complex samples. Label-free comparison and targeted high-resolution mass spectrometry were utilized to filter out the vast majority of non-crosslinked peptides and increase confidence of crosslink identification, compensating for the lack of enrichment techniques and characteristic MS patterns employed by non-zero-length crosslinking methods. Each step from mass spectrometer acquisition parameters to MS/MS spectra evaluation functions was optimized based on zero-length crosslinking datasets of proteins with known crystal structures. Our pipeline was then applied to probe structures and conformational changes of mini-spectrin, a 90 kDa recombinant protein that closely mimics erythrocyte spectrin\u27s dynamic dimer-tetramer equilibrium. Compared to previous analyses performed in our laboratory, the current strategy more than doubled the number of identified crosslinks and significantly reduced analysis time per experiment from months to just several days. Distance constraints derived from mini-spectrin crosslinks were used as inputs in subsequent homology modeling, allowing development of experimentally-verified medium-resolution structures for wild-type mini-spectrin tetramer and both wild-type and hereditary elliptocytosis (HE) mutant mini-spectrin dimers. The structure models, in combination with independent biophysical experiments, illustrated how such distal HE-related mutations destabilized spectrin dimer-tetramer equilibrium by simultaneously lowering thermal stability of tetramer and giving rise to a more-compact, more-stable closed dimer conformation