Analysis of Post‐translational Modifications

Abstract

René Zahedi The characterization of post-translational modifications (PTM) is not just one of the major strengths of mass spectrometry (MS)-based proteomics, but also one of its biggest challenges. Whereas enzymatic phosphorylation of proteins was initially discovered in the 1950's, the first reports about MS-based analysis of PTM emerged in the early 80's. Currently there are approximately 300 PTM and numerous in vitro modifications known. However mostly due to their low stoichiometry and often transient character, dedicated workflows and methods for specific large-scale characterization and quantification are established only for some of those. Whereas strategies for large-scale qualitative and quantitative analysis of phosphorylation, glycosylation, acetylation and ubiquitination have been established in recent years, the field certainly is in need for methods to tackle further PTM and to specifically examine the interaction between PTM in the context of cellular signaling. For PTM which are labile under conditions typically used during sample preparation/separation/detection, and for PTM which can be easily induced artificially in vitro, alternative analytical strategies have to be developed. In this context we are proud to present this Special Issue on the analysis of post-translational modifications, which comprises a total of 16 valuable contributions from various experts in the field of PTM research. This issue overs various topics which are most important for MS (and non-MS)-based PTM analysis: (1) standardization and general considerations, (2) sample preparation, (3) fragmentation of PTM peptides and its impact on peptide identification and PTM site-assignment, (4) systematic and structural characterization of specific proteins, (5) PTM analysis of patient samples, (6) non-MS-based approaches for PTM analysis, and (7) novel bioinformatics approaches for improved PTM analysis. Thus, this issue covers all aspects of current PTM research, from methodical development and optimization to the application to real samples for clinical research. One fundamental task – not only with regard to PTM but rather proteomics in general – is to generate and utilize dedicated standards which allow a quality control and benchmarking of analytical methods, as discussed in the introducing viewpoint by Alexander Ivanov and his colleagues from the ABRF Proteomics Standards Research Group (sPRG). The groups of Larsen and Gevaert provide detailed and timely reviews of the current progress in the fields of phosphoproteomics (Engholm-Keller et al.) and redox modifications in plants (Jacques et al.). Albert Sickmann The contribution from Kollipara et al. deals with the unanticipated induction of in vitro modifications, they determined the extent of urea-induced carbamylation during sample preparation: an issue which might be underestimated in current proteomic studies leading to reduced identification rates. The following contributions deal with the fragmentation of modified peptides and the consequences for following data interpretation. Thus, Karl Mechtler and colleagues systematically analyzed the fragmentation behavior of phosphoarginine-containing peptides using different fragmentation modes and the impact on the site-localization by search algorithms. As has been published for other PTM as well, they could demonstrate that electron transfer dissociation (ETD) yields higher spectrum quality and confidence when compared to collisional-induced dissociation (CID) which suffers from severe neutral losses. In the same context, Hansen et al. characterized the MS-based identification of adenylation (also AMPylation), the addition of a 5’ adenosine phosphodiester group to tyrosine or threonine residues. The CID-based fragmentation of AMPylated peptides leads to a variety of neutral losses and reporter ions, which may differ between Thr and Tyr, and considering those in database searches lead to an improved identification of model peptides. In the next contribution, Cui and Reid systematically analyzed the phosphate group rearrangement during CID-MS/MS and CID-MS3 in ion traps and its impact on phosphorylation site localization, an important issue with regard to the ever-increasing amount of phosphoproteomic data. They utilized specifically designed synthetic phosphopeptides of varying composition, identity, number and position of phosphate groups and non-phosphorylated “acceptor” residues and conclude that unanticipated competing fragmentation and rearrangement reactions during ion activation can have a negative impact on the following peptide identification and phosphorylation site assignment. They argue that the results from their study as well as other fundamental gas phase fragmentation studies should be incorporated into search algorithms/software – pointing at an unfortunately still sometimes present lack of communication between experimentalists and computer scientists to further push on the field of proteomics. In the next contribution, Pan and Borchers used Calmodulin as a model system to promote the use of top-down hydrogen/deuterium exchange mass spectrometry (HDX-MS) for the structural analysis of PTM protein isoforms. While the present study focused on methionine oxidation, they are currently extending HDX-MS to further PTM such as phosphorylation or glycosylation. Burlingame and colleagues employed ETD and higher energy collisional dissociation (HCD) to study the PTM pattern of Host Cell Factor C1, identifying a total of 19 O-GlcNAc sites, two phosphorylation sites and two dimethylarginines. They argue that although HCD did not allow reliable site assignment, the generated reporter ions could be used to validate poor scoring ETD spectra of O-GlcNAc peptides. Nicolardi et al. utilized a novel workflow comprising an optimized and automated sample preparation protocol and MALDI-FTICR-MS analysis to study peptide and protein profiles from 96 human serum samples. They demonstrate that this strategy has great potential for screening large patient cohorts with high reproducibility. Pieragostino et al. utilized MS to investigate changes in the oxidation state of Transthyretin in serum and cerebrospinal fluid. They found atypical oxidative PTM in the cerebral Transthyretin of multiple sclerosis patients compared to patients suffering from other neurological diseases and patients with acute disseminated encephalomyelitis. They argue that their results could pave the way for a better diagnostic certainty in the early stages of multiple sclerosis, requiring only small sample amounts. The final two experimental contributions deal with alternative, non-MS-based, technologies for PTM analysis. Thus, Poetz and colleagues established a fast and cost efficient bead-based peptide array containing 384 peptides representing phosphorylated, acetylated, methylated and citrullinated N-terminal regions of histones H2A, H2B, H3, and H4. Using their straightforward approach they profiled the binding of 40 PTM-specific antibodies which are important in epigenetic proteome research with regard to their on- and off-target interaction. In their contribution, Schweigel et al. employed Far Western SH2 domain profiling for sensitive analysis of changes in the phosphotyrosine state of human platelets upon activation with ADP in a time-resolved manner. The last three contributions of this special issue deal with novel and improved bioinformatics tools for PTM research. In this context Vandermarliere and Martens describe a completely novel approach to re-analyze PTM with respect to existing structural data. Using phosphorylation site data from Uniprot/Swissprot, they demonstrate that most sites are found at the protein surface, whereas some are buried in the inner core of the protein in a way that would require unfolding of the complete protein for kinase/phosphatase accessibility. For some of these sites there are alternative phosphorylatable residues in close proximity (i.e. within the same peptide) which would be easier to access and thus a wrong phosphorylation site assignment seems likely. In the same context, the contribution by Vaudel et al. introduces the so-called D-score, which is based on posterior error probabilities and thus allows the assessment of PTM-site assignment for multiple search engine studies. Compared to Mascot alone, this approach yielded up to 25% more correctly localized phosphorylation sites. In the final contribution, Nahnsen et al. present a novel computational pipeline, PTMeta, to dig deeper into large-scale proteomic data. They pre-scan for the most abundant potential PTM in a sample to conduct extensive database searches with different modification settings. They developed a new scoring scheme which allows combining the results from different modification settings to yield better identification rates in proteomic analyses with statistical confidence. We hope that you will not only enjoy this special issue but also gain novel insights for your own research to further push PTM research in life and death. We also would like to thank all the authors for their valuable contributions and everyone else who helped to realize this special issue. René Zahedi Albert Sickmann