Journal of Proteome Research最新文献

The proteomics community has fully embraced data sharing, yet data set metadata provision remains limited, especially at the level of the biological samples and experimental design. This hampers large-scale data reuse, as comprehensive and structured sample context and study design information are often essential for confident, automatic reuse, and (re)interpretation. Although standards such as Sample and Data Relationship Format for Proteomics (SDRF-Proteomics) and supporting tools are already available, their adoption remains limited. Many researchers lack incentives, and enforcement by journals and repositories remains challenging in practice. Still, metadata defines a data set’s long-term value. We propose a coordinated plan to dramatically improve metadata annotation of publicly disseminated proteomics data. Funders can drive progress by investing in a sustainable, scalable metadata infrastructure. HUPO-PSI plays a central role in setting community standards and enabling validation. ProteomeXchange repositories are key to implementing and supporting metadata adoption. Data producers must treat metadata as a part of their scientific output. Instrument vendors can contribute by enabling the automatic capture of technical metadata. Software developers should embed SDRF-Proteomics metadata into analysis workflows. Finally, journals and reviewers are well positioned to shape expectations and enforce compliance. By aligning efforts across these stakeholders, we can build the road to large-scale, context-aware reuse and unlock the full value of public proteomics data sets.

Protein quantification is a crucial data processing step that combines quantitative values at the peptide or fragment level into protein levels in mass spectrometry-based proteomics. However, many of the current algorithms, including the state-of-the-art method MaxLFQ, do not scale well with the increasing number of samples, because of the limited system memory and algorithmic complexities. Here we introduce the iq format, a novel data structure designed to support very large data sets. We optimize existing quantification methods for both speed and memory usage. In particular, the new algorithms maxlfq-bit and rlm-cd significantly improve the base methods, MaxLFQ and the robust linear model, respectively, achieving orders of magnitude speed improvements for a large number of samples. The experimental result shows that the MaxLFQ algorithm achieves the highest accuracy, despite its comparatively higher computational cost. We also introduce a generic algorithm to boost the quantification accuracy of all methods by reducing the effect of noisy ion intensity traces. The experimental results show that the weighting approach improves the performance of all tested methods on a spike-in data set and a mixed species data set. The software implementation is publicly available in the R package iq from version 2.

Small extracellular vesicles (sEVs) are lipid-bilayer-enclosed vesicles secreted by cells into the extracellular environment, carrying a variety of biomolecules, including proteins that reflect the molecular profile of their cell of origin. In particular, cancer-derived sEVs hold significant potential for cancer diagnosis due to their unique biomolecular content. In this study, we utilized mass spectrometry to profile the protein expression in plasma-derived sEVs from both healthy controls (HCs) and breast cancer (BC) patients with the aim of discovering new protein biomarkers for potential BC diagnosis. Through the cross-validation of differentially expressed proteins between the two independent cohorts, we identified nine proteins that were significantly upregulated in BC-derived sEVs. Further validation using online gene expression data sets, Western blot, and ELISA revealed that OIT3 was upregulated in BC tissue compared to HC tissue, suggesting its potential as a novel BC biomarker. These findings contribute to advancing the knowledge of proteins within sEVs, as well as offering promising avenues for the use of sEVs as biomarkers in future cancer diagnostic applications.

Rhabdomyosarcoma (RMS), the most common pediatric soft tissue sarcoma, exhibits marked clinical heterogeneity driven by poorly understood molecular mechanisms. Identifying the molecular characteristics of different RMS subtypes and the molecular pathways influencing the RMS treatment response and recurrence is an urgent clinical need. Here, we perform deep proteomic profiling of 19 RMS tumors (8 alveolar [ARMS], 11 embryonal [ERMS]) and matched normal tissues, integrating bioinformatics with functional validation to delineate subtype-specific pathways, therapy resistance drivers, and actionable targets. ARMS tumors are characterized by ubiquitination pathway activation (UBE2R2, UBE2J2), while ERMS exhibits spliceosome dysregulation. Chemo- and radio-resistant tumors both show significant enrichment in the ribosome pathway. Relapsed cases show phosphonate and phosphinate metabolism pathway enrichment, suggesting metabolism reliance. Unsupervised clustering reveals ribosome- and glycolysis-driven subtypes with distinct metabolic dependencies. Functional studies implicate MED18─a core component of the Mediator complex─in mediating therapy resistance possibly via promoting DNA damage repair. Our study establishes proteomics as a tool to decode RMS heterogeneity, proposing subtype-tailored strategies targeting ubiquitination, splicing, and metabolism.

In living systems, protein function relies on many intra- and intermolecular interactions within a network called the interactome. The majority of available interactome data has been acquired with isolated proteins and complexes, but visualization of interactome changes in living systems is crucial to advance understanding of functional changes with diseases and for the development of improved therapies. With model animal systems, quantitative cross-linking mass spectrometry has been successfully applied to uniquely reveal interactome changes with mitochondrial dysfunction both in heart failure and with age-related muscle function decline. In this study, we investigated the feasibility of qualitative cross-linking mass spectrometry for mitochondrial interactome studies with clinically relevant human muscle biopsy samples and amounts. Analysis of biopsy samples from two volunteers resulted in the identification of 1350 nonredundant peptides from 177 mitochondrial proteins from all mitochondrial subcompartments. Many of the identified human biopsy cross-linked peptides were derived from protein complex and supercomplex assemblies that exhibited altered levels in model systems of heart failure and aging. The findings demonstrate the initial feasibility that these and other cross-linked species can be detected in human muscle biopsy samples to enable future studies of age- and disease-related changes in mitochondrial structure–function relationships.

Mass spectrometry proteomics creates complex data representing the peptide/protein contents of biological samples. Various types of machine learning have been central to computational methods used to identify peptides from tandem mass spectra and numerous other aspects of the data analysis process. As deep learning has emerged as a powerful machine learning method for modeling and interpreting data, computational proteomics researchers have leveraged large publicly available data sets to train machine learning models to predict peptide fragmentation spectra and liquid chromatography retention time. Resources like proteomicsML offer extensive demonstrative tutorials for these learning tasks and are closing the gap between the proteomics and machine learning communities. However, in these and other educational materials on deep learning, the critical step of preparing data for learning is frequently omitted. Prior to learning, peptide strings must be converted into a numeric format─an embedding. There are many different peptide embeddings, and some vastly outperform others. Yet the process for creating an embedding, and also the rationale for choosing a specific embedding, is rarely discussed in our proteomics literature. In this technical note, we introduce four Google Colab notebooks to teach peptide embeddings. The series walks users through five different peptide-embedding strategies─ from simplistic single-number encodings to state-of-the-art pretrained embeddings─ through both code examples and narrative descriptions. The final notebook compares the five embeddings in a head-to-head benchmark. By making these notebooks free, we hope to lower the barrier for researchers who want to bring modern deep learning into their proteomics workflows.

Conventional database search methods for proteomics struggle when tasked with identifying dozens or hundreds of modifications simultaneously. Open or error-tolerant searches can address this limitation but at the cost of increased difficulty in downstream interpretation of the results and quantification. We and others have previously described “mass offset” or multinotch searches that sit in between closed and open searches, allowing simultaneous search for hundreds of modifications with more straightforward downstream interpretation than open search. The original mass offset searches were closer to the open search, lacking the ability to restrict modifications to specific amino acids. Here, we describe a new “detailed” mass offset (DMO) search implemented in the MSFragger search engine, which allows each mass offset to have its own site restrictions and fragmentation rules. The benefits of the DMO search over existing mass offset searches are shown with three example searches of complex modification sets: nearly one hundred post-translational modifications, fast photochemical oxidation of proteins (FPOP)-derived modifications, and amino acid substitutions. The DMO search further improves the interpretability of results by reducing ambiguity in site localization, particularly when modifications have overlapping masses, and provides benefits that scale with the complexity of the search.

Sepsis-induced liver injury (SILI) is a severe complication of sepsis and is strongly associated with adverse clinical outcomes. However, the molecular mechanisms driving SILI pathogenesis remain poorly understood. In this study, we applied data-independent acquisition (DIA)-based quantitative proteomics to characterize protein expression profiles in liver tissues from 7 patients with SILI and 14 control patients. A total of 335 proteins were significantly dysregulated in SILI liver tissues, including 126 upregulated and 209 downregulated proteins. GO and KEGG pathway analyses revealed that the upregulated proteins were predominantly enriched in the cellular response to hypoxia and lysosome pathways, whereas the downregulated proteins were mainly associated with metabolic processes, particularly glutathione metabolism. Six key glutathione metabolism-related enzymes (GCLC, GSTO1, SOD1, GPX4, PRDX6, and IDH1) were selected for validation and were confirmed to be markedly reduced in SILI liver tissues by immunoblotting and qPCR. Correlation analyses further demonstrated that decreased expression of these enzymes was strongly associated with elevated markers of inflammation, coagulation disorders, and hepatic dysfunction, linking impaired antioxidant capacity to disease severity. Collectively, our findings reveal a distinct proteomic signature in SILI, characterized by profound suppression of glutathione metabolism, offering mechanistic insight into redox imbalance during SILI. These results highlight glutathione metabolic pathways as promising therapeutic targets for mitigating hepatocellular dysfunction in sepsis.

Since its emergence in 2019, circulating SARS-CoV-2 has been dominated by waves of genetically distinct variants with varying pathogenicity. Understanding the multidimensional responses to SARS-CoV-2 infection and their associations with pathogenesis is critical for developing therapies to prevent severe illness and death. Here, we applied quantitative proteome and phosphoproteome analyses to compare host responses to infections with an ancestral variant (WA-1/2020), a Delta variant (B.1.617.2), and an Omicron variant (BA.1) of SARS-CoV-2 in Syrian golden hamster tissues at 5 days postinfection, when peak inflammatory responses were observed. As has been observed by others, animals infected with the Delta variant lost more weight than those infected with other variants, and this effect was associated with decreased cilia proteins in the trachea tissue and increased signatures of fibrosis in lung tissue. Phosphoproteome analysis revealed a downregulation of Raf-MEK-ERK signaling across all variants, suggesting a suppressed proliferative response in tissues following SARS-CoV-2 infection. These data provide critical in vivo confirmation of observations from in vitro studies and provide a quantitative tissue- and SARS-CoV-2 variant-specific resource of proteome and phosphoproteome responses.

To identify potential biomarkers and explore the underlying mechanisms of elderly acute kidney injury (e-AKI), we performed integrative plasma proteomics analysis on samples from 20 e-AKI patients and 20 age-matched non-AKI controls. Differential expression gene analysis, GSEA, WGCNA, random forest, and LASSO models were employed to identify hub genes, coupled with immune cell infiltration and clinicopathological correlation analyses. A renal ischemia–reperfusion injury mouse model validated key genes at protein and mRNA levels, while in vitro experiments explored the pathway involvement. We identified 229 e-AKI-associated genes enriched in immune, inflammatory, and coagulation pathways. Machine learning combined with the Nephroseq database yielded three hub genes; in vivo and in vitro experiments confirmed fibrinogen alpha chain (FGA) as the most relevant gene, which may regulate e-AKI progression via the cAMP/PKA/CREB pathway. Collectively, FGA holds promise as a diagnostic biomarker and therapeutic target for e-AKI, laying the theoretical foundation for its mechanistic research.