Journal of Proteome Research最新文献

Large-scale proteomicsrelies heavily on target–decoycompetitionfor false discovery rate estimation in peptide identification, andthe performance of this strategy depends strongly on the design ofthe decoy database. Classical generators such as reversal and shufflingremain widely used. Here, we introduce the first protein languagemodel-based (PLM) decoy generation for peptide identification andbenchmark it against classical strategies. We evaluate these approachesusing three complementary quality-control layers: sequence-based separability,search-engine-agnostic spectral-space diagnostics, and end-to-endmass spectrometry benchmarks, including pipelines with rescoring.Across these analyses, PLM-based decoys are harder for sequence-onlyneural networks to distinguish than most classical generators, suggestingfewer obvious sequence-level artifacts. However, this signal is onlyweakly informative for search performance. Spectral diagnostics furthershow that short peptides occupy a particularly crowded target–decoyspace and are therefore especially prone to local collisions acrossall generators. In full search pipelines, reverse decoys remain astrong baseline, and current PLM-based generators do not yet providea clear overall advantage. We therefore view PLM-based decoys notas universal replacements for reverse decoys but as tunable toolsfor benchmarking, diagnostics, stress testing, and future adaptivedecoy optimization, with increasing value as search models becomemore expressive.

Moyamoya disease, a rare chronic cerebrovascular disorder,requiresinvasive digital subtraction angiography (DSA) for diagnosis. Thisstudy employed high-throughput proteomics to identify plasma biomarkersfor Moyamoya disease diagnosis. We conducted immunopanel analysisusing the Olink platform to evaluate 92 immune-related proteins inplasma samples from 88 Moyamoya disease patients and 88 healthy controls.Key proteins were identified through differential expression analysis,GO, and KEGG enrichment analysis. A diagnostic model was constructedusing LASSO regression, Boruta algorithm, and machine learning modelsincluding random forest and XGBoost. Validation of these proteinswas performed using GEO external data sets, followed by predictionof potential therapeutic drugs and molecular docking validation throughpharmacogenomic databases. A total of 44 differentially expressedproteins were identified through the Olink immunopanel, with 12 downregulatedand 32 upregulated. GO and KEGG analyses revealed significant enrichmentof these proteins in innate immune responses and signaling pathwayssuch as NF-kB and MAPK. Through LASSO, random forest, and proteinunder-area analysis, four potential biomarkers for Moyamoya disease(MGMT, SIT1, PRDX1, TRAF2) were identified. A diagnostic model usingthese proteins showed the highest AUC value with the XGBoost model.Additionally, TRAF2 and PRDX1 exhibited significant expression differencesin Moyamoya disease patients within the GEO data set. Our study revealedthe immune landscape of Moyamoya disease, identified four biomarkers,and established a variety of diagnostic models.

Alcohol-associated liver disease (ALD) has imposed asubstantialpublic health burden in the United States, and alcohol consumptioninduces systemic lipid dysregulation in ALD. To investigate this,the lipidomes of liver and brain tissues from long-term high fat diet(HFD) feeding and HFD with single ethanol (EtOH) binge (HFD + EtOH)mice were characterized by comprehensive two-dimensional (2D) liquidchromatography–mass spectrometry. Multiple database matchingwas used for high-confidence lipid identification. Univariable andmultivariable analyses were employed to identify lipids with significantlyaltered abundance between HFD + EtOH and HFD mice. Furthermore, differentialcorrelation analysis and cross-tissue codysregulation mapping wereperformed. Our data indicated that alcohol feeding induced profoundtissue-specific lipidomic alterations. Triglycerides increased significantly,and phospholipids decreased in the livers of alcohol-treated mice,whereas their brains exhibited elevation of oxidized lipids with subclass-divergentchanges in glycerolipids, glycerophospholipids, and sphingolipids.Differential correlation analysis revealed extensive remodeling oflipid interaction networks in the alcohol-exposed brain. Cross-tissueanalysis identified 18 codysregulated lipids demonstrating a stronglycorrelated lipid regulation shift in HFD + EtOH mice. Collectively,alcohol rewires organ-specific lipidomes through distinct mechanismsand induces correlated multitissue lipid dysregulation, suggestingpotential lipid-centric communication along the liver-brain axis.These findings provide novel insights into ALD pathogenesis and revealpotential biomarkers for multiorgan involvement.

Food residues thatbypass human digestion are furtherdigestedby gut microbes, leading to the production of diverse metabolites,including lipids. To investigate how lipids in our intestine are affectedduring this transition, we used a colon simulator with four distinctvessels (V1–V4) that mimic the proximal to distal part of thehuman colon. In total, 44 samples were collected from the colon simulatorvessels (V1–V4). We observed dynamic shifts in a diverse arrayof microbially linked lipid molecules in the simulated intestinalchyme, including bile acids and N-acyl amides withshort- and odd-chain lipids. Histamine-linked N-acyllipids (histamine-C5) increased from the proximal to the distal colonvessels (pH 5.5–7.0), whereas putrescine-linked ones (putrescine-C19:2),initially abundant in the media, decreased across the colon vessels.We uncovered dynamic associations between in vitro-derived short-chain N-acyl lipids (C4:0, C5, C6, and C7 conjugates) and majorlipid species such as cholesterol esters, phosphatidylethanolamines,ceramides, and sphingomyelins. To determine the broader relevanceof these findings, we applied a reverse metabolomics approach andexamined N-acyl lipid profiles in human small intestineand fecal samples from public data sets. Our results validate thecolon simulator as a dynamic model for studying microbially transformedmetabolites and suggest its potential utility as a platform for discoveringnovel microbial metabolites with relevance to human and animal health.

Type 2 Diabetes (T2D) and hypertension (HTN) are commoncomorbiditiesin severe COVID-19, yet their specific impact on proteomic recoveryremains unclear. This study analyzed plasma protein signatures ofcritical COVID-19 patients with and without these comorbidities (COVID-onlygroup [COG] and COVID comorbid group [CTHG]) on the first and lastdays of ICU stay. Proteomic analysis revealed a systemic shift characterizedby upregulated immune responses and downregulated metabolic processesat admission across all patients. Survival was fundamentally definedby the restoration of homeostasis; liver-derived proteins─includingLPA, TTR, and AHSG─were initially suppressed but reboundedsignificantly in survivors. This homeostatic recovery was impairedin CTHG compared to COG, with CTHG survivors showing attenuated recoveryof metabolic markers. Distinct mortality-associated signatures alsoemerged between groups. COG nonsurvivors exhibited liver failure andsevere hemolysis marked by persistent suppression of haptoglobin (HP).In contrast, CTHG mortality was driven by lipid metabolism dysregulation,with CD5L and APOA2 levels dropping specifically in comorbid nonsurvivors,often accompanied by a paradoxical elevation in APOA4─likelyreflecting impaired renal clearance rather than restored lipid homeostasis.These findings indicate that preexisting T2D and HTN hinder physiologicalresolution of metabolic and lipid dysregulation, providing proteomicevidence for distinct mortality risks associated with failure to restoremetabolic homeostasis in comorbid COVID-19 patients.

Laryngealsquamous cell carcinoma (LSCC) lacks reliable biomarkers,and the roles of lactate metabolism and lysine lactylation (Kla) remainlargely unknown. We profiled the lysine lactylome of LSCC, pairedit with adjacent normal tissues, and integrated the data with quantitativeproteomic and transcriptomic analyses. LSCC exhibited a hyperlactate-associatedphenotype characterized by dysregulated lactate-related genes (LRGs),altered protein abundance, increased tissue lactate, and globallyincreased Kla levels. Data-independent acquisition mass spectrometry(DIA-MS) identified 1616 Kla sites on 1468 peptides from 688 proteins,with most differential sites being upregulated in tumors. Differentiallylactylated proteins were enriched in cell–matrix adhesion,cell migration, chromatin remodeling, and gene-regulatory processesand were clustered into cytoskeletal and nuclear regulatory modules.Multiple Kla sites were also detected on the core histones. Immunoblottingand tissue microarray analyses confirmed increased pan-Kla expressionin the LSCC. Pan-Kla levels were independent of sex and age but positivelycorrelated with the tumor stage and lymph-node metastasis. These findingsprovide a systematic resource for hyperlactate-associated lactylomeremodeling in LSCCs and identify candidate Kla-related molecular featuresassociated with clinicopathological progression for future functionaland clinical evaluation.

Asa class of immune checkpoint inhibitors (ICIs), programmed celldeath protein-1 (PD-1) blockade has demonstrated remarkable efficacyin the treatment of various malignancies. However, their clinicalapplication is constrained by the high incidence of immune-relatedadverse events (irAEs), which arise from nonspecific immune activationand can affect multiple organ systems, with severe cases posing life-threateningrisks. This study integrated high-throughput proteomic and metabolomicanalyses to systematically characterize the molecular features associatedwith irAEs in cancer patients receiving PD-1 inhibitor therapy. Theresults showed that, following the first treatment, patients who developedirAEs exhibited potential involvement of the NF-κB pathway,along with lower baseline levels of SNRPA and higher expression ofCD63. Metabolomic analyses further revealed that the kynurenine/tryptophanratio was significantly elevated in the irAE group both at baselineand post-treatment compared with patients who did not develop irAEs.In addition, significant differences in the abundance of specificlipids were observed between the two groups prior to the administrationof immunotherapy. Our findings provide exploratory insights into immuneand metabolic alterations associated with PD-1 blockade treatmentand may help generate hypotheses for future studies on early irAErisk assessment in cancer patients undergoing PD-1 blockade therapy.

Delayed graft function (DGF) is a common early complicationofkidney transplantation characterized by immune activation. The durationof DGF may significantly affect long-term graft survival, yet theimmune mechanisms underlying the different DGF durations remain unclear.Using a functional definition of delayed graft function (fDGF), definedas a failure of serum creatinine to decrease by at least 10% per dayfor three consecutive days within the first postoperative week, patientswere stratified into short-term DGF (SDGF) and long-term DGF (LDGF)groups according to recovery periods. In this exploratory study, targetedproteomic analysis indicated that proteins enriched in SDGF were primarilyinvolved in innate immune responses and acute inflammatory processes,including neutrophil chemotaxis and migration, whereas LDGF exhibitedfeatures related to adaptive immune responses and chronic inflammation,such as T-cell differentiation and activation. IL-7 and CCL20 wereidentified as candidate molecules potentially associated with differentDGF durations. Targeted metabolomics revealed disturbances in aminoacid metabolism, particularly alanine, aspartate, and glutamate metabolism,as well as in energy metabolism, including the tricarboxylic acid(TCA) cycle, which may be involved in LDGF. These findings providepreliminary insights into immune metabolic features associated withdifferent DGF recovery durations.

This study profiles the personalized immunopeptidomesof 13 Thaipatients with renal cell carcinoma (RCC), addressing a critical knowledgegap in Southeast Asian populations characterized by distinct HLA alleledistributions. We combined whole-exome sequencing (WES)-based personalizedproteome construction with liquid chromatography–tandem massspectrometry (LC–MS/MS), using both database-driven searchesand de novo peptide sequencing. HLA typing identified several classI allotypes that are underrepresented in publicly available immunopeptidomeresources, including seven alleles not previously represented in thedatabases examined; HLA-A*11:01 was the most frequent allele in thiscohort. Database-based analysis identified a single tumor-specificneoantigen derived from a mutant JADE2 peptide in the patient withthe highest tumor mutational burden, which was validated by a mutant-specificELISPOT response. In contrast, de novo sequencing revealed numerousnoncanonical peptides, a subset of which were supported by proteogenomicvalidation using PepQuery and detected exclusively in cancer proteomesbut not in normal tissue data sets, indicating their potential astumor-associated antigen candidates. Together, these results establishan integrated and scalable framework for identifying HLA-presentedtumor-derived peptides and provide a foundational immunopeptidomeresource to support personalized cancer immunotherapy developmentin Southeast Asia.

Sepsis is a life-threatening syndrome characterized bya dysregulatedhost response to infection and profound metabolic alterations thatcontribute to immune dysfunction and organ failure. This Review synthesizesproteomic evidence on sepsis-associated alterations in proteins involvedin metabolic pathways across circulating biofluids, immune cells,and organs. Across plasma and urine, proteomic studies identify disturbancesin lipoprotein-associated pathways, redox homeostasis, mitochondrialfunction, and substrate metabolism, indicating that protein signaturesof metabolic dysregulation are systemic and detectable across biofluids.In immune cells, monocytes and neutrophils, proteomic analyses reveala shift toward glycolysis with concurrent impairment of mitochondrialpathways alongside phenotype-dependent differences in lipid and redox-relatedprograms. Organ-level studies further show that metabolic responsesare heterogeneous, with distinct trajectories in the kidney, heart,liver, lung, skeletal muscle, and brain. These observations supportthe concept that sepsis involves compartment-specific remodeling ofmetabolism-associated protein networks rather than a single convergentmetabolic state. Proteomics also highlights potential translationalopportunities by identifying metabolism-associated proteins linkedto disease severity, clinical phenotypes, and biologically distinctpatient subgroups, although the current evidence remains largely exploratoryand context-dependent. Overall, proteomics provides a complementaryframework for understanding the molecular regulation of sepsis-associatedmetabolic dysfunction and may refine biological stratification andtherapeutic targeting, particularly when integrated with longitudinalsampling and multiomic data.