In-depth analysis of tandem mass spectrometry data from disparate instrument types

Mass spectrometric analyses of protein digests produce large numbers of fragmentation spectra that are not identified by routine database searching strategies. Some of these spectra could be identified by development of improved search engines. However, many of these spectra represent fragmentation of peptide components bearing modifications that are not routinely considered in database searches. Here we present new software within Protein Prospector that allows comprehensive analysis of data sets by analyzing the data at increasing levels of depth. Analysis of published data sets is presented to illustrate that the software is not biased to any instrument types. The results show that these data sets contain many modified peptides. As well as searching for known modification types, Protein Prospector permits the detection and identification of unexpected or novel modifications by searching for any mass shift within a user-specified mass range to any chosen amino acid(s). Several modifications never previously reported in proteomics data were identified in these standard data sets using this mass modification searching approach.     

Universal and confident phosphorylation site localization using phosphoRS

An algorithm for the assignment of phosphorylation sites in peptides is described. The program uses tandem mass spectrometry data in conjunction with the respective peptide sequences to calculate site probabilities for all potential phosphorylation sites. Tandem mass spectra from synthetic phosphopeptides were used for optimization of the scoring parameters employing all commonly used fragmentation techniques. Calculation of probabilities was adapted to the different fragmentation methods and to the maximum mass deviation of the analysis. The software includes a novel approach to peak extraction, required for matching experimental data to the theoretical values of all isoforms, by defining individual peak depths for the different regions of the tandem mass spectrum. Mixtures of synthetic phosphopeptides were used to validate the program by calculation of its false localization rate versus site probability cutoff characteristic. Notably, the empirical obtained precision was higher than indicated by the applied probability cutoff. In addition, the performance of the algorithm was compared to existing approaches to site localization such as Ascore. In order to assess the practical applicability of the algorithm to large data sets, phosphopeptides from a biological sample were analyzed, localizing more than 3000 nonredundant phosphorylation sites. Finally, the results obtained for the different fragmentation methods and localization tools were compared and discussed.     

Modification site localization scoring integrated into a search engine

Large proteomic data sets identifying hundreds or thousands of modified peptides are becoming increasingly common in the literature. Several methods for assessing the reliability of peptide identifications both at the individual peptide or data set level have become established. However, tools for measuring the confidence of modification site assignments are sparse and are not often employed. A few tools for estimating phosphorylation site assignment reliabilities have been developed, but these are not integral to a search engine, so require a particular search engine output for a second step of processing. They may also require use of a particular fragmentation method and are mostly only applicable for phosphorylation analysis, rather than post-translational modifications analysis in general. In this study, we present the performance of site assignment scoring that is directly integrated into the search engine Protein Prospector, which allows site assignment reliability to be automatically reported for all modifications present in an identified peptide. It clearly indicates when a site assignment is ambiguous (and if so, between which residues), and reports an assignment score that can be translated into a reliability measure for individual site assignments.     

Pinpointing phosphorylation sites: Quantitative filtering and a novel site-specific x-ion fragment

Phosphoproteomics deals with the identification and quantification of thousands of phosphopeptides. Localizing the phosphorylation site is however much more difficult than establishing the identity of a phosphorylated peptide. Further, recent findings have raised doubts of the validity of the site assignments in large-scale phosphoproteomics data sets. To improve methods for site localization, we made use of a synthetic phosphopeptide library and SILAC-labeled peptides from whole cell lysates and analyzed these with high-resolution tandem mass spectrometry on an LTQ Orbitrap Velos. We validated gas-phase phosphate rearrangement reactions during collision-induced dissociation (CID) and used these spectra to devise a quantitative filter that by comparing signal intensities of putative phosphorylated fragment ions with their nonphosphorylated counterparts allowed us to accurately pinpoint which fragment ions contain a phosphorylated residue and which ones do not. We also evaluated higher-energy collisional dissociation (HCD) and found this to be an accurate method for correct phosphorylation site localization with no gas-phase rearrangements observed above noise level. Analyzing a large set of HCD spectra of SILAC-labeled phosphopeptides, we identified a novel fragmentation mechanism that generates a phosphorylation site-specific neutral loss derived x-ion, which directly pinpoints the phosphorylated residue. Together, these findings significantly improve phosphorylation site localization confidence.     

LuciPHOr: Algorithm for Phosphorylation Site Localization with False Localization Rate Estimation Using Modified Target-Decoy Approach

The localization of phosphorylation sites in peptide sequences is a challenging problem in large-scale phosphoproteomics analysis. The intense neutral loss peaks and the coexistence of multiple serine/threonine and/or tyrosine residues are limiting factors for objectively scoring site patterns across thousands of peptides. Various computational approaches for phosphorylation site localization have been proposed, including Ascore, Mascot Delta score, and ProteinProspector, yet few address direct estimation of the false localization rate (FLR) in each experiment. Here we propose LuciPHOr, a modified target-decoy-based approach that uses mass accuracy and peak intensities for site localization scoring and FLR estimation. Accurate estimation of the FLR is a difficult task at the individual-site level because the degree of uncertainty in localization varies significantly across different peptides. LuciPHOr carries out simultaneous localization on all candidate sites in each peptide and estimates the FLR based on the target-decoy framework, where decoy phosphopeptides generated by placing artificial phosphorylation(s) on non-candidate residues compete with the non-decoy phosphopeptides. LuciPHOr also reports approximate site-level confidence scores for all candidate sites as a means to localize additional sites from multiphosphorylated peptides in which localization can be partially achieved. Unlike the existing tools, LuciPHOr is compatible with any search engine output processed through the Trans-Proteomic Pipeline. We evaluated the performance of LuciPHOr in terms of the sensitivity and accuracy of FLR estimates using two synthetic phosphopeptide libraries and a phosphoproteomic dataset generated from complex mouse brain samples.Phosphorylation is a common and essential form of post-translational regulation that has been extensively studied via mass spectrometry (1–5). However, tandem mass spectra produced from phosphorylated peptides can be difficult to interpret because of their relatively low abundance within the cell and the presence of intense neutral loss peaks in the MS/MS spectra (6, 7). Correctly determining which residue bears the phosphate group is typically a tedious and error-prone process. Most commonly used database search tools for peptide identification from MS/MS spectra are not optimized for site localization of a post-translational modification, nor do they provide any confidence score for the assigned site. In addition, manual verification of the modification sites is a time-consuming process that requires expertise in mass spectrometry. As a result, the challenges of site localization have been acknowledged by the proteomics community, including within the latest version of the data publication guidelines of this journal (8).A number of computational approaches that localize phosphorylation sites have been reported in the literature, enabling automated phosphoproteomic analysis (reviewed in Ref. 9). These tools either rescore the MS/MS spectra to assign confidence measures for individual sites based on site-determining ions (10–15) or derive localization scores directly from the search engine output (16, 17). Ascore, a representative tool in the rescoring category, scores each candidate phosphosite based upon the peaks representing the site-determining ions and subsequently reports a confidence score for the phosphopeptide sequence (11). This algorithm uses the binomial distribution to compute the probability of a random (incorrect) localization for each candidate site in each spectrum. PhosphoRS extends the scoring approach of Ascore by adjusting the probability of random peak matching based on the density of peaks in different regions of each spectrum (18). In contrast, the Mascot Delta score (MD-score)¹ determines the confidence of phosphosite localization on peptides as the difference in Mascot ion scores between the highest scoring phosphopeptide (the peptide reported by the search engine) and the next best scoring phosphopermutation (same peptide sequence, alternative phosphorylation site (17)). Thus, the MD-score represents the second type of approach, which, instead of rescoring MS/MS spectra for the purpose of improved site localization, derives the scores directly from the database search engine output. A similar idea was implemented in the SLIP score using a modified version of the Batch-Tag search engine of the ProteinProspector suite (16) and in the variable modification localization score of the proprietary software Spectrum Mill (9). These tools, however, apply the logic of delta scoring for individual sites, not for the whole peptide; this is an important consideration in the case of multiply phosphorylated peptides.Although these tools have significantly improved the quality of published phosphopeptide identification data, several important issues remain. The level of uncertainty in modification site localization varies significantly across different peptides depending on the total number of candidate sites and the number of phosphorylated residues on the peptide. This, in turn, makes it difficult to compare localization scores between different peptides. Secondly, few algorithms provide a direct estimation of the false localization rate (FLR) in filtered data. Thirdly, most existing algorithms are tied to specific search engines and/or require proprietary libraries (e.g. Ascore and MD-score were developed for SEQUEST and Mascot, respectively; PhosphoRS requires proprietary libraries from Thermo Scientific). This makes it difficult to access these tools and to compare their performance.Here we present LuciPHOr, an alternative approach for site localization and direct FLR estimation. We introduce a novel scoring approach that utilizes both peak intensity and mass accuracy to aid the computation of an objective score for phosphosite determination and dynamically adapts to characteristic peak properties in different types of instrumentation and fragmentation methods. LuciPHOr computes the scores for phosphosite permutations and associated FLR estimates for the best scoring prediction at the peptide level. It also reports site-level scores for multiphosphorylated peptides, with an acknowledgment that it is difficult to rigorously estimate the FLR in such cases. We also highlight the practical utility of LuciPHOr, which is capable of processing the results of any database search tool (including commonly used search engines X! Tandem (19), SEQUEST (20), and Mascot (21)) that is supported by the widely used Trans-Proteomic Pipeline (TPP) (22). We benchmark LuciPHOr using two previously published datasets generated using synthetic phosphopeptide libraries and demonstrate similar or better performance relative to the existing methods. We also demonstrate the high accuracy of the FLR estimated by LuciPHOr obtained using a target-decoy modification site framework. Lastly, the performance of LuciPHOr is further investigated using a complex mouse brain dataset, and we also discuss the issue of site-level scoring in the analysis of multiphosphorylated peptides.     

Identification of Novel Site‐Specific Alterations in the Modification Level of Myelin Basic Protein Isolated from Mouse Brain at Different Ages Using Capillary Electrophoresis–Mass Spectrometry

Myelin basic protein (MBP) is a multifunctional protein involved in maintaining the stability and integrity of the myelin sheath by a variety of interactions with membranes and other proteins. MBP is subjected to extensive posttranslational modifications (PTMs) that are known to be crucial for the regulation of these interactions. Here, we report capillary electrophoresis–mass spectrometric (CE–MS) analysis for the separation and identification of MBP peptides that incorporate the same PTM at different sites, creating multiple localization variants, and the ability to analyze challenging modifications such as asparagine and glutamine deamidation, isomerization, and arginine citrullination. Moreover, we observed site‐specific alterations in the modification level of MBP purified from brain of mice of different age. In total, we identified 40 modifications at 33 different sites, which include both previously reported and seven novel modifications. The identified modifications include Nα‐terminal acetylation, mono‐ and dimethylation, phosphorylation, oxidation, deamidation, and citrullination. Notably, some new sites of arginine methylation overlap with the sites of citrullination. Our results highlight the need for sensitive and efficient techniques for a comprehensive analysis of PTMs.     

Post-translational modification of genetically encoded polypeptide libraries

The genetic encoding of polypeptides with biological display systems enables the facile generation and screening of very large combinatorial libraries of molecules. By post-translationally modifying the encoded polypeptides, chemically and structurally more diverse molecules beyond linear amino acid polymers can be generated. The first post-translational modification applied to encoded polypeptides, the oxidation of cysteine residues to form disulfide bridges, is a natural one and was used to cyclise short peptides soon after the invention of phage display. Recently a range of non-natural chemical strategies for the post-translational modification of encoded polypeptide repertoires were applied to generate optical biosensors, semisynthetic polypeptides, peptide-drug conjugates, redox-insensitive monocyclic peptides or multicyclic peptides, and these strategies are reviewed in this article.     

Confident identification of 3-nitrotyrosine modifications in mass spectral data across multiple mass spectrometry platforms

3-nitrotyrosine (3NT) is an oxidative posttranslational modification associated with many diseases. Determining the specific sites of this modification remains a challenge due to the low stoichiometry of 3NT modifications in biological samples. Mass spectrometry-based proteomics is a powerful tool for identifying 3NT modifications, however several reports identifying 3NT sites were later demonstrated to be incorrect, highlighting that both the accuracy and efficiency of these workflows need improvement. To advance our understanding of the chromatographic and spectral properties of 3NT-containing peptides we have adapted a straightforward, reproducible procedure to generate a large set of 3NT peptides by chemical nitration of a defined, commercially available 48 protein mixture. Using two complementary LC-MS/MS platforms, a QTOF (QSTAR Elite) and dual pressure ion trap mass spectrometer (LTQ Velos), we detected over 200 validated 3NT-containing peptides with significant overlap in the peptides detected by both systems. We investigated the LC-MS/MS properties for each peptide manually using defined criteria and then assessed their utility to confirm that the peptide was 3NT modified. This broad set of validated 3NT-containing peptides can be utilized to optimize mass spectrometric instrumentation and data mining strategies or further develop 3NT peptide enrichment strategies for this biologically important, oxidative posttranslational modification.     

Fragments of Functional Proteins: Role in Endocrine Regulation

Systematic analysis of structures, localization, formation and biological activities of endogenous peptides derived from functional proteins, such as hemoglobin, myelin basic protein, immunoglobulins, etc., allowed establishing the basic features of that group of compounds. The sets of these peptides in mammalian tissues, or tissue-specific peptide pools are: (i) tissue specific; (ii) stable at normal conditions; (iii) conservative in the same tissues of different mammalian species; (iv) dependent on the general state of homeostasis of tissue or the whole organism. Formation of such peptides has features of both conformation and site specificity and also involves the action of carboxy- and amino-peptidases. As a result, the families of structurally related families of peptides are generated. The fragments of functional proteins exhibit a wide range of the biological effects, characteristic both for hormones and parahormones, from hormone-releasing to growth-regulatory activity. At the same time, the molecular mechanisms of action of the majority of such peptides are unknown. On the basis of the data obtained the components of tissue-specific peptide pools are considered to form a novel regulatory system, complementary to other peptidergic systems such as hormonal, nervous, immune, etc. The biological role of the fragments of functional proteins in vivo and the patterns of interaction with other regulatory systems are suggested.     

Direct identification of ubiquitination sites on ubiquitin-conjugated CHIP using MALDI mass spectrometry

The study of protein ubiquitination, a post-translational modification by ubiquitin, has emerged as one of the most active areas in biology because of the important role of this type of modification on the regulation of various cellular proteins. Advances in techniques for the determination and site mapping of protein ubiquitination can facilitate the elucidation of molecular mechanisms of this modification. We have recently described a novel method for identifying peptides containing ubiquitinated amino acid residues, based on the MALDI-MS/MS analysis of tryptic peptide derivatives. In particular, we have utilized N-terminal sulfonation of these peptides to provide a unique fragmentation pattern that leads to the direct identification and sequencing of ubiquitin modified peptides. Here we present an application of this new method on the characterization of ubiquitin conjugated C-terminal Hsc70-interacting protein (CHIP), a recently identified U-box containing E3 enzyme. Three peptides bearing ubiquitination sites have been identified from the digest of ubiquitinated CHIP; one of these was a site on CHIP, while the other two were found on the ubiquitin molecules, demonstrating that sulfonation of tryptic peptides is a general and efficient method for characterizing protein ubiquitination.     

O-glycosylated human MUC1 repeats are processed in vitro by immunoproteasomes

The targeting of epitopes on tumor-associated glycoforms of human MUC1 represents a primary goal in immunotherapeutic anticancer strategies. Effective immune responses to cancer cells certainly require the activation of specific cytotoxic T cell repertoires by cross-priming of dendritic cells either via immunoproteasomal or by endosomal processing of ectodomain epitopes on MUC1-positive carcinomas. Because no evidence is currently available on the capacities of human immunoproteasomes to cleave mucin-type O-glycosylated peptides, we performed in vitro studies to address the questions of whether glycosylated MUC1 repeats are cleaved by immunoproteasomes and in which way O-linked glycans control the site specificity of peptide cleavage via their localization and structures. We show for the first time that mucin-type O-glycosylated peptides are effective substrates of immunoproteasomes, however, the patterns of cleavage are qualitatively and quantitatively influenced by O-glycosylation. The nonglycosylated MUC1 repeat peptide (clusters of oligorepeats AHGVTSAPDTRPAPGSTAPP or AHGVTSAPESRPAPGSTAPA) is cleaved preferentially within or adjacent to the SAP and GST motifs with formation of a complex fragment pattern that includes major nona- and decapeptides. O-GalNAc modified peptides are largely resistant to proteolysis if these preferred cleavage sites are located adjacent to O-glycosylation, whereas peptides even with elongated glycans at more distant sites can form effective substrates yielding major glycopeptide fragments in the class I size range.     

Confident phosphorylation site localization using the Mascot Delta Score

Large scale phosphorylation analysis is more and more getting into focus of proteomic research. Although it is now possible to identify thousands of phosphorylated peptides in a biological system, confident site localization remains challenging. Here we validate the Mascot Delta Score (MD-score) as a simple method that achieves similar sensitivity and specificity for phosphosite localization as the published Ascore, which is mainly used in conjunction with Sequest. The MD-score was evaluated using liquid chromatography-tandem MS data of 180 individually synthesized phosphopeptides with precisely known phosphorylation sites. We tested the MD-score for a wide range of commonly available fragmentation methods and found it to be applicable throughout with high statistical significance. However, the different fragmentation techniques differ strongly in their ability to localize phosphorylation sites. At 1% false localization rate, the highest number of correctly assigned phosphopeptides was achieved by higher energy collision induced dissociation in combination with an Orbitrap mass analyzer followed very closely by low resolution ion trap spectra obtained after electron transfer dissociation. Both these methods are significantly better than low resolution spectra acquired after collision induced dissociation and multi stage activation. Score thresholds determined from simple calibration functions for each fragmentation method were stable over replicate analyses of the phosphopeptide set. The MD-score outperforms the Ascore for tyrosine phosphorylated peptides and we further show that the ability to call sites correctly increases with increasing distance of two candidate sites within a peptide sequence. The MD-score does not require complex computational steps which makes it attractive in terms of practical utility. We provide all mass spectra and the synthetic peptides to the community so that the development of present and future localization software can be benchmarked and any laboratory can determine MD-scores and localization probabilities for their individual analytical set up.     

Demographic model selection using random forests and the site frequency spectrum

Phylogeographic data sets have grown from tens to thousands of loci in recent years, but extant statistical methods do not take full advantage of these large data sets. For example, approximate Bayesian computation (ABC) is a commonly used method for the explicit comparison of alternate demographic histories, but it is limited by the “curse of dimensionality” and issues related to the simulation and summarization of data when applied to next‐generation sequencing (NGS) data sets. We implement here several improvements to overcome these difficulties. We use a Random Forest (RF) classifier for model selection to circumvent the curse of dimensionality and apply a binned representation of the multidimensional site frequency spectrum (mSFS) to address issues related to the simulation and summarization of large SNP data sets. We evaluate the performance of these improvements using simulation and find low overall error rates (~7%). We then apply the approach to data from Haplotrema vancouverense, a land snail endemic to the Pacific Northwest of North America. Fifteen demographic models were compared, and our results support a model of recent dispersal from coastal to inland rainforests. Our results demonstrate that binning is an effective strategy for the construction of a mSFS and imply that the statistical power of RF when applied to demographic model selection is at least comparable to traditional ABC algorithms. Importantly, by combining these strategies, large sets of models with differing numbers of populations can be evaluated.