Oncogenic Mutations Rewire Signaling Pathways by Switching Protein Recruitment to Phosphotyrosine Sites.

Tyrosine phosphorylation regulates multi-layered signaling networks with broad implications in (patho)physiology, but high-throughput methods for functional annotation of phosphotyrosine sites are lacking. To decipher phosphotyrosine signaling directly in tissue samples, we developed a mass-spectrometry-based interaction proteomics approach. We measured the in vivo EGF-dependent signaling network in lung tissue quantifying >1,000 phosphotyrosine sites. To assign function to all EGF-regulated sites, we determined their recruited protein signaling complexes in lung tissue by interaction proteomics. We demonstrated how mutations near tyrosine residues introduce molecular switches that rewire cancer signaling networks, and we revealed oncogenic properties of such a lung cancer EGFR mutant. To demonstrate the scalability of the approach, we performed >1,000 phosphopeptide pulldowns and analyzed them by rapid mass spectrometric analysis, revealing tissue-specific differences in interactors. Our approach is a general strategy for functional annotation of phosphorylation sites in tissues, enabling in-depth mechanistic insights into oncogenic rewiring of signaling networks.

Deorphanizing Peptides Using Structure Prediction.

Many endogenous peptides rely on signaling pathways to exert their function, but identifying their cognate receptors remains a challenging problem. We investigate the use of AlphaFold-Multimer complex structure prediction together with transmembrane topology prediction for peptide deorphanization. We find that AlphaFold's confidence metrics have strong performance for prioritizing true peptide-receptor interactions. In a library of 1112 human receptors, the method ranks true receptors in the top percentile on average for 11 benchmark peptide-receptor pairs.

Improving Peptide-Spectrum Matching by Fragmentation Prediction Using Hidden Markov Models.

Tandem mass spectrometry has become the method of choice for high-throughput, quantitative analysis in proteomics. Peptide spectrum matching algorithms score the concordance between the experimental and the theoretical spectra of candidate peptides by evaluating the number (or proportion) of theoretically possible fragment ions observed in the experimental spectra without any discrimination. However, the assumption that each theoretical fragment is just as likely to be observed is inaccurate. On the contrary, MS2 spectra often have few dominant fragments. Using millions of MS/MS spectra we show that there is high reproducibility across different fragmentation spectra given the precursor peptide and charge state, implying that there is a pattern to fragmentation. To capture this pattern we propose a novel prediction algorithm based on hidden Markov models with an efficient training process. We investigated the performance of our interpolated-HMM model, trained on millions of MS2 spectra, and found that our model picks up meaningful patterns in peptide fragmentation. Second, looking at the variability of the prediction performance by varying the train/test data split, we observed that our model performs well independent of the specific peptides that are present in the training data. Furthermore, we propose that the real value of this model is as a preprocessing step in the peptide identification process. The model can discern fragment ions that are unlikely to be intense for a given candidate peptide rather than using the actual predicted intensities. As such, probabilistic measures of concordance between experimental and theoretical spectra will leverage better statistics.

Combining mass spectrometry and machine learning to discover bioactive peptides.

Peptides play important roles in regulating biological processes and form the basis of a multiplicity of therapeutic drugs. To date, only about 300 peptides in human have confirmed bioactivity, although tens of thousands have been reported in the literature. The majority of these are inactive degradation products of endogenous proteins and peptides, presenting a needle-in-a-haystack problem of identifying the most promising candidate peptides from large-scale peptidomics experiments to test for bioactivity. To address this challenge, we conducted a comprehensive analysis of the mammalian peptidome across seven tissues in four different mouse strains and used the data to train a machine learning model that predicts hundreds of peptide candidates based on patterns in the mass spectrometry data. We provide in silico validation examples and experimental confirmation of bioactivity for two peptides, demonstrating the utility of this resource for discovering lead peptides for further characterization and therapeutic development.

RAIN: RNA-protein Association and Interaction Networks.

Protein association networks can be inferred from a range of resources including experimental data, literature mining and computational predictions. These types of evidence are emerging for non-coding RNAs (ncRNAs) as well. However, integration of ncRNAs into protein association networks is challenging due to data heterogeneity. Here, we present a database of ncRNA-RNA and ncRNA-protein interactions and its integration with the STRING database of protein-protein interactions. These ncRNA associations cover four organisms and have been established from curated examples, experimental data, interaction predictions and automatic literature mining. RAIN uses an integrative scoring scheme to assign a confidence score to each interaction. We demonstrate that RAIN outperforms the underlying microRNA-target predictions in inferring ncRNA interactions. RAIN can be operated through an easily accessible web interface and all interaction data can be downloaded.Database URL: http://rth.dk/resources/rain.

Search Databases and Statistics: Pitfalls and Best Practices in Phosphoproteomics.

Advances in mass spectrometric instrumentation in the past 15 years have resulted in an explosion in the raw data yield from typical phosphoproteomics workflows. This poses the challenge of confidently identifying peptide sequences, localizing phosphosites to proteins and quantifying these from the vast amounts of raw data. This task is tackled by computational tools implementing algorithms that match the experimental data to databases, providing the user with lists for downstream analysis. Several platforms for such automated interpretation of mass spectrometric data have been developed, each having strengths and weaknesses that must be considered for the individual needs. These are reviewed in this chapter. Equally critical for generating highly confident output datasets is the application of sound statistical criteria to limit the inclusion of incorrect peptide identifications from database searches. Additionally, careful filtering and use of appropriate statistical tests on the output datasets affects the quality of all downstream analyses and interpretation of the data. Our considerations and general practices on these aspects of phosphoproteomics data processing are presented here.

Systems Analysis for Interpretation of Phosphoproteomics Data.

Global phosphoproteomics investigations yield overwhelming datasets with up to tens of thousands of quantified phosphosites. The main challenge after acquiring such large-scale data is to extract the biological meaning and relate this to the experimental question at hand. Systems level analysis provides the best means for extracting functional insights from such types of datasets, and this has primed a rapid development of bioinformatics tools and resources over the last decade. Many of these tools are specialized databases that can be mined for annotation and pathway enrichment, whereas others provide a platform to generate functional protein networks and explore the relations between proteins of interest. The use of these tools requires careful consideration with regard to the input data, and the interpretation demands a critical approach. This chapter provides a summary of the most appropriate tools for systems analysis of phosphoproteomics datasets, and the considerations that are critical for acquiring meaningful output.

From Phosphosites to Kinases.

Kinases play a pivotal role in propagating the phosphorylation-mediated signaling networks in living cells. With the overwhelming quantities of phosphoproteomics data being generated, the number of identified phosphorylation sites (phosphosites) is ever increasing. Often, proteomics investigations aim to understand the global signaling modulation that takes place in different biological conditions investigated. For phosphoproteomics data, identifying the kinases central to mediating this response is key. This has prompted several efforts to catalogue the immense amounts of phosphorylation data and known or predicted kinases responsible for the modifications. However, barely 20 % of the known phosphosites are assigned to a kinase, initiating various bioinformatics efforts that attempt to predict the responsible kinases. These algorithms employ different approaches to predict kinase consensus sequence motifs, mostly based on large scale in vivo and in vitro experiments. The context of the kinase and the phosphorylated proteins in a biological system is equally important for predicting association between the enzymes and substrates, an aspect that is also being tackled with available bioinformatics tools. This chapter summarizes the use of the larger phosphorylation databases, and approaches that can be applied to predict kinases that phosphorylate individual sites or that are globally modulated in phosphoproteomics datasets.

CrossWork: software-assisted identification of cross-linked peptides.

The increased interest in chemical cross-linking for probing protein structure and interaction has led to a large increase in literature describing new cross-linkers and search programs. However, this has not led to a corresponding increase in the analysis of large and complex proteins. A major obstacle is that the new cross-linkers are either not readily available and/or have a low reactivity. In combination with aging search programs that are slow and have low sensitivity, or new search programs that are described but not released, these efforts do little to advance the field of cross-linking. Here we present a method pipeline for chemical cross-linking, using two standard cross-linkers, BS3 and BS2G, combined with our freely available CrossWork search program. By this approach we generate cross-link data sufficient to derive structural information for large and complex proteins. CrossWork searches batches of tandem mass-spectrometric data, and identifies cross-linked and non-cross-linked peptides using a standard PC. We tested CrossWork by searching mass-spectrometric datasets of cross-linked complement factor C3 against small (1 protein) and large (1000 proteins) search spaces, and show that the resulting distance constraints agree with the established structures. We further investigated the structure of the multi-domain ERp72, and combined the individual domains of ERp72 into a single structure.