Biology subjects

Wu, Y.

Publications and source records attributed to Wu, Y..

At least 307 records · Page 17Linked to original sources

bioRxiv2025

A High-Resolution N-Glycoproteome Atlas Reveals Tissue-Specific Glycan Remodeling but Non-random Structural Microheterogeneities

The mouse is a widely used model organism in biomedical research, yet a comprehensive understanding of its tissue-specific glycoproteome has been limited due to the structural complexity and microheterogeneity of glycans. Here, we present the most extensive high-resolution N-glycoproteomic atlas across 24 mouse tissues, which comprises of 3,045 N-glycans with distinct structures attached at 8,681 glycosites on 74,277 glycopeptides and 5,026 glycoproteins. Overall glycan structural patterns show enormous tissue-specific diversities, acting as superior molecular signatures of tissue identity and system origins. Notably, even commonly expressed glycoproteins undergo tissue-dependent glycan remodeling, suggesting that glycosylation may fine-tune protein functions to meet specialized biological demands. These patterns are further shaped by subcellular localization, which constrains glycan variabilities across compartments. Co-occurrence network analyses also expose substructural biases and non-random microheterogeneities among glycans attached at the same glycosites. The dataset serves as a valuable database resource for advancing the structural and functional understanding of glycoproteins.

biochemistry↗

bioRxiv2025

Clade V MLO proteins are bona fide host susceptibility factors required for powdery mildew pathogenesis in Arabidopsis

Obligate biotrophic powdery mildew (PM) fungi strictly require living host to survive. To search for host factors or processes essential for PM pathogenesis, a tailored genetic screen was conducted with the immuno-compromised eds1-2/pad4-1/sid2-2 (eps) triple Arabidopsis mutant. This led to the identification of five allelic disruptive mutations in Mildew Locus O 2 (MLO2) to be responsible for the compromised-immunity-yet-poor infection (cipi) mutant phenotype upon challenge from an adapted PM isolate. Moreover, the eds1/pad4/sid2/mlo2/mlo6/mlo12 (eps3m) sextuple mutant display near complete immunity to the adapted PM fungus without sign of defense activation, demonstrating that these three clade V MLOs in Arabidopsis are bona fide host susceptibility factors of PM fungi. Confocal imaging revealed focal accumulation of MLO2-GFP in the peri-penetration peg membranous space, implicating MLO2 in repairing and stabilizing the damaged host plasma membrane, which may be co-opted by PM fungi for haustorium differentiation. Results from domain-swapping analysis between MLO1 and MLO2 suggest a bipartite functional configuration for MLO2: its C-terminus determines where and when MLO2 functions, while its N-terminal seven transmembrane domain region executes the cellular function that is critical for PM pathogenesis. Genetic studies further demonstrate that, unlike MLO7 in synergids, focal accumulation of MLO2 does not depend on FERONIA (FER) and its five other family members, nor does it require phosphatidylinositol 4,5-bisphosphate produced from phosphatidylinositol 4-phosphate 5-kinase 1 (PIP5K1) and PIP5K2. Together, these findings define clade V MLOs as host factors co-opted by obligate biotrophic PM fungi for successful host colonization.

plant biology↗

bioRxiv2025

Arabidopsis RabGDIs are essential for the asymmetric division of zygotes and embryonic patterning

An asymmetric division of zygotes sets up the apical-basal body axis and starts the life cycle of angiosperms. In addition to specific expression of transcription factors and polar distribution of auxin, vesicular trafficking-mediated endomembrane dynamics are critical for zygotic division and embryonic patterning. As key regulators of vesicular trafficking, whether Rab GTPases play a role in these processes is unclear. We report that Arabidopsis guanine nucleotide dissociation inhibitors for Rab (RabGDIs) are essential for the asymmetric division of zygotes and embryonic patterning likely through regulating the dynamic targeting of canonical Rab GTPases, especially Rab5. Arabidopsis RabGDIs specifically interact with canonical Rab GTPases. Functional loss of Arabidopsis RabGDI1 and RabGDI2 compromises the asymmetry of zygotic division, cell fate determination, and embryonic patterning. Polar distribution of vacuolar dynamics was disrupted in zygotes and 1-cell embryo of gdi1 gdi2, suggesting a key role of RabGDIs in vacuolar dynamics. Furthermore, genetic interference of canonical Rab5, a key regulator of vacuolar trafficking and biogenesis, in zygotes leads to similar defects, consistent with the disturbed localization of Rab5 in gdi1 gdi2 embryos. Results presented demonstrate the key role of RabGDIs through Rab GTPases in asymmetric division of zygotes and embryonic patterning.

plant biology↗

bioRxiv2025

UFMylation orchestrates chromatin engagement of core NHEJ components to promote DNA double-strand break repair

DNA double-strand breaks (DSBs) are highly cytotoxic lesions whose misrepair can lead to genomic instability, cancer and developmental disorders. Through systematic screening of understudied ubiquitin-like modifiers (UBLs), we identify UFM1 as a previously unrecognised regulator of non-homologous end-joining (NHEJ). Using a structure-guided chemical biology strategy, we develop a photo-crosslinkable UFM1 probe and, together with high-resolution NMR, uncover non-canonical UFM1-binding regions in core NHEJ components, including XRCC4. Mechanistically, proximity-dependent proteomics reveals Ku70 as a key UFMylation substrate, establishing a functional axis in which XRCC4 engages UFMylated Ku70 to promote the chromatin assembly of NHEJ factors. Perturbation of UFM1 signalling, via UFSP2 depletion or a hypomorphic UBA5 allele in patient-derived fibroblasts, impairs these processes, linking UFMylation defects to altered regulation of DSB repair. Our findings define a complete UFM1 signalling module in genome maintenance and uncover a molecular connection between hereditary UFMylation disorders and dysregulated DSB repair pathways.

molecular biology↗

bioRxiv2025

Global hypoperfusion leads to a mismatch in oxygen delivery and consumption in the cerebral watershed area

Despite the pivotal role of pial collaterals in maintaining cerebral blood flow during focal brain ischemia, it is largely unexplored how the microvascular blood flow and oxygenation in the watershed "pial-collateral territory" differ from those in the territory supplied by the major arteries during chronic global hypoperfusion. To answer this question, we applied 2-photon microscopy and Doppler optical coherence tomography to investigate the changes in cerebral microvascular blood flow and partial pressure of oxygen (PO2), induced by bilateral common carotid artery stenosis (BCAS). The measurements were performed in the somatosensory cortex that is supplied by the middle cerebral artery (MCA), and in the adjacent watershed area in the awake, head-restrained C57BL/6 mice, via the chronic cranial window. The results showed that the BCAS induced a larger decrease in capillary red blood cell (RBC) flux in the watershed area than in the MCA territory, especially in the subcortical white matter. Besides, PO2 in the pial collaterals was significantly lower than that in the upstream MCA segments under control conditions. However, the PO2 changes in the arteries and veins under global hypoperfusion displayed different trends in the two interrogated regions, resulting in a significant increase in oxygen extraction fraction in the watershed area. These findings suggest a mismatch between oxygen supply and demand in the watershed area due to global hypoperfusion and increased subcortical white matter vulnerability. We have also observed dilation of the pial collaterals after BCAS, which might suggest a compensatory mechanism to improve the blood flow in the watershed under hypoperfusion.

neuroscience↗

bioRxiv2025

A connectomic resource for neural cataloguing and circuit dissection of the larval zebrafish brain

We present a correlated light and electron microscopy (CLEM) dataset from a 7-day-old larval zebrafish, integrating confocal imaging of genetically labeled excitatory (vglut2a) and inhibitory (gad1b) neurons with nanometer-resolution serial section EM. The dataset spans the brain and anterior spinal cord, capturing >180,000 segmented soma, >40,000 molecularly annotated neurons, and 30 million synapses, most of which were classified as excitatory, inhibitory, or modulatory. To characterize the directional flow of activity across the brain, we leverage the synaptic and cell body annotations to compute region-wise input and output drive indices at single cell resolution. We illustrate the datasets utility by dissecting and validating circuits in three distinct systems: water flow direction encoding in the lateral line, recurrent excitation and contralateral inhibition in a hindbrain motion integrator, and functionally relevant targeted long-range projections from a tegmental excitatory nucleus, demonstrating that this resource enables rigorous hypothesis testing as well as exploratory-driven circuit analysis. The dataset is integrated into an open-access platform optimized to facilitate community reconstruction and discovery efforts throughout the larval zebrafish brain.

neuroscience↗

bioRxiv2025

Small tau aggregates exhibit disease-specific molecular profiles across tauopathies

Tauopathies are neurodegenerative diseases marked by pathological tau aggregation. While disease-specific folds of insoluble tau filaments have been established, it remains unclear whether the smaller, earlier species also differ across tauopathies. Here, we characterise these small tau aggregates from post-mortem brain of individuals with Alzheimers disease (AD), progressive supranuclear palsy (PSP), corticobasal degeneration, Picks disease, and healthy controls. Using two complementary single-molecule assays, we confirm that small tau aggregates vary in abundance, morphology, and post-translational modifications. AD features specific long, fibrillar-shaped aggregates enriched in phospho-epitopes, while PSP aggregates are shorter, round, and selectively phosphorylated at serine-356, a site we identify as correlating with markers of inflammation and apoptosis. Aggregate properties co-vary with cellular stress signatures and align with disease-specific seeding profiles, suggesting distinct pathological mechanisms. These findings suggest that small tau aggregates are not a shared intermediate, but instead encode disease-specific mechanisms, with potential as both biomarkers and therapeutic targets. Key pointsSmall tau aggregates show disease-specific signatures across tauopathies, differing in abundance, morphology, and post-translational modifications. Tau aggregates in AD show enhanced phosphorylation density and structural heterogeneity, including a distinct population of long fibrillar species detectable in the soluble fraction. Alzheimers disease is characterised by specific long, fibrillar-shaped tau aggregates enriched in disease-relevant phospho-epitopes. PSP features round pSer356-positive aggregates that correlate with apoptotic and inflammatory markers, suggesting a distinct mechanism of toxicity. Isoform-specific biosensor assays reveal divergent seeding behaviour: CBD shows strong 4R seeding, while PSP lacks seeding activity. Features of small aggregates co-vary with distinct patterns of gliosis and cell stress, suggesting disease-specific mechanisms of tau-mediated toxicity.

neuroscience↗

bioRxiv2025

SMART: An Integrated Platform Revolutionizing Cryo-Electron Microscopy Workflow from Data Acquisition to Atomic Modeling

Recent advancements in cryo-electron microscopy (cryo-EM) have revolutionized structural biology, yet persistent workflow fragmentation between data acquisition, reconstruction, and modeling limits its full potential. This paper presents SMART, the first fully integrated cryo-EM platform unifying three critical AI-driven modules: DataSmart (automated data collection), CryoSmart (3D reconstruction), and ModelSmart (end-to-end atomic modeling). The platform embeds more than 10 distinct AI algorithms across its workflow, achieving unprecedented automation levels. These advancements not only lower the expertise barrier for cryo-EM adoption but also establish a new paradigm for AI-empowered structural biology, with broad implications for high-throughput drug discovery. This protocol provides a comprehensive delineation of the operational procedures for the SMART platform. As an example, this protocol enabled the determination of the 2.1 [A] resolution structure of TRPML1, a Ca2+-permeable, nonselective, six-transmembrane tetrameric cation channel found in late endosomes and lysosomes (LELs) of mammalian cells. This integrated workflow enables users, including those with minimal cryo-EM experience, to efficiently complete the entire pipeline from data collection to atomic model building.

bioinformatics↗

bioRxiv2025

Low-density lipoprotein receptor-targeting chimeras for membrane protein degradation and enhanced drug delivery

Antibody-based therapeutics encompass diverse modalities for targeting tumor cells. Among these, antibody-drug conjugates (ADCs) and extracellular targeted protein degradation (eTPD) specifically depend on efficient lysosomal trafficking for activity. However, many tumor antigens exhibit poor internalization, limiting ADC effectiveness. To address this, we developed low-density lipoprotein receptor-targeting chimeras (LIPTACs), leveraging the constitutive endocytic and recycling activity of the LDLR to enhance lysosomal delivery. LIPTACs enable efficient and selective degradation of diverse extracellular membrane proteins. Additionally, by coupling LIPTACs with cytotoxic payloads to generate degrader-drug conjugates, we can achieve superior intracellular delivery and enhanced cytotoxicity compared to conventional ADCs. The dual modality addresses key challenges of inadequate internalization in conventional ADCs and cytotoxic potency for current eTPD strategies. Our findings demonstrate that LDLR-mediated trafficking can enhance eTPD and ADCs, providing a hybrid blueprint for developing next-generation antibody therapeutics with broader utility and improved efficacy in cancer treatment.

bioengineering↗

bioRxiv2025

Global profiling of the proteome and acetylome in mice with abdominal aortic aneurysms

BackgroundAbdominal Aortic Aneurysm (AAA) is a life-threatening vascular condition characterized by a progressive dilation of the abdominal aorta and a high risk of rupture. Currently, no effective pharmacological therapies are available, and treatment relies mainly on surgical intervention. The lack of drug targets is largely due to limited understanding of the molecular mechanisms underlying AAA development. This study aims to identify potential therapeutic targets by performing comprehensive proteomic and acetylomic analyses in a mouse model of AAA. MethodsWe conducted global proteomic and acetylomic profiling on abdominal aortic tissues collected from an AAA mouse model. A total of 7,858 proteins and 1,790 acetylated proteins encompassing 4,581 acetylation sites were quantified. Bioinformatics analyses were performed to investigate the regulatory networks and biological pathways associated with these proteins and post-translational modifications. ResultsIntegrated analysis revealed that histone-related proteins were significantly enriched in pathways co-regulated by the proteome and acetylome. These pathways are influenced by histone acetyltransferases and deacetylases, implicating histone acetylation in AAA-related inflammatory and immune processes. Notably, the deacetylases Sirt2 and Sirt5 were identified as potential regulators of neutrophil extracellular trap (NET) formation through suppression of histone acetylation. ConclusionsThis study provides the first comprehensive proteomic and acetylomic dataset for abdominal aortic tissues in a mouse model of AAA, substantially expanding the mouse acetylation database. The results highlight the central role of histone modifications in immune dysregulation and inflammation, and propose Sirt2 and Sirt5 as candidate therapeutic targets that may inhibit NET formation and slow AAA progression. These findings suggest new avenues for drug development targeting the sirtuin family, although further experimental validation and cross-species studies are needed to confirm translational relevance. Graphic Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=145 SRC="FIGDIR/small/657536v1_ufig1.gif" ALT="Figure 1"> View larger version (42K): org.highwire.dtl.DTLVardef@1c397a3org.highwire.dtl.DTLVardef@188180org.highwire.dtl.DTLVardef@de1a59org.highwire.dtl.DTLVardef@193e740_HPS_FORMAT_FIGEXP M_FIG C_FIG

bioinformatics↗

bioRxiv2025

OriGene: A Self-Evolving Virtual Disease Biologist Automating Therapeutic Target Discovery

Here, we present OriGene, a self-evolving multi-agent system that functions as a virtual disease biologist, systematically identifying original and mechanistically grounded therapeutic targets at scale. OriGenes architecture integrates over 600 specialized tools through a Model Context Protocol (MCP), enabling it to reason across diverse data modalities including genomics, protein networks, pharmacology, clinical records and literature evidence, to generate and prioritize target discovery hypotheses. We implemented a strategy combining a knowledge graph-based Tool RAG with an advanced agent selection mechanism to enable dynamic, context-aware tool deployment. Through a self-evolving framework, OriGene continuously integrates human and experimental feedback to iteratively refine its core thinking templates, tool composition, and analytical protocols, thereby enhancing both accuracy and adaptability over time. To comprehensively evaluate its performance, we established TRQA, an original benchmark comprising over 1,900 expert-level question-answer pairs spanning a wide range of diseases and target classes. OriGene consistently outperforms human experts, leading research agents, and state-of-the-art large language models in accuracy, recall, and robustness, particularly under conditions of data sparsity or noise. Critically, OriGene nominated previously underexplored therapeutic targets for liver (GPR160) and colorectal cancer (ARG2), which demonstrated significant anti-tumor activity in patient-derived organoid and tumor fragment models mirroring human clinical exposures. These findings demonstrate OriGenes potential as a scalable and adaptive platform for AI-driven discovery of mechanistically grounded therapeutic targets, offering a new paradigm to accelerate drug development.

bioinformatics↗

bioRxiv2025

An Anaerobic Pathogen Rewires Host Metabolism to Fuel Oxidative Growth in the Inflamed Gut

To colonize their host and cause disease, enteric pathogens must deploy their virulence factors to establish distinct nutrient niches. How obligate anaerobic pathogens construct nutrient niches in the densely populated large intestine remains poorly understood. Enterotoxigenic Bacteroides fragilis (ETBF) is considered an obligate anaerobic bacterium and has been implicated in inflammation-associated diseases, including colitis and colorectal cancer. Here we show that ETBF uses its virulence factor, Bacteroides fragilis toxin, to reprogram colonic epithelial cell metabolism to colonize the inflamed gut. Bacteroides fragilis toxin activates colonic epithelial signaling and hijacks the host bile acid recycling pathway, inducing a metabolic shift in the epithelium from oxidative phosphorylation to glycolysis. This shift increases local concentrations of lactate and oxygen, nutrients that support an oxidative metabolism in ETBF. These findings reveal an unexpected strategy by which a pathogenic organism, previously considered to be an obligate anaerobic bacterium, generates and exploits an oxidative niche in the inflamed gut.

microbiology↗

bioRxiv2025

Structural innovations and neurogenic continuity define avian brain development and evolution

Understanding how neurogenic diversity and brain architecture emerge is crucial for deciphering the mechanisms of brain plasticity and cognitive evolution. Here, we present a comprehensive single-cell and spatial transcriptomic atlas of the budgerigar brain, a highly cognitive avian species. We uncover a striking dorsoventral symmetry in excitatory neuron distribution, precisely organized along the lamina mesopallialis intermedia (LMI), a key boundary structure. Cross-species comparisons further reveal the origins of this unique avian pallial organization, offering new insight into avian brain evolution. Our study also highlights distinct developmental trajectories and asynchronous maturation patterns between telencephalic and optic tectum excitatory neurons, underscoring their contribution to innate circuit assembly. Furthermore, we identify adult neural stem cells (NSCs) with evolutionarily conserved transcriptomic signatures across avian species, emphasizing their role in brain plasticity and adaptation. These findings refine fundamental models of avian brain development and elucidate conserved principles of cognitive plasticity.

evolutionary biology↗

bioRxiv2025

PEDOT:PSS Microparticles for Extrudable and Bioencapsulating Conducting Granular Hydrogel Bioelectronics

Conducting hydrogels are promising materials for forming physiomimetic bioelectronic interfaces to monitor and stimulate biological activity. However, most developed materials are non-microporous and possess fixed shapes, both of which can limit the integration of cells and tissues with devices. In non-conducting biomaterials, materials fabrication strategies imparting microporosity and dynamic mechanical properties have been shown to support cell infiltration and support biointerfaces of various geometries. Specifically, granular hydrogels have enabled encapsulating, conformal, and injectable interfaces through these features. However, granular hydrogels remain largely unexplored as conducting biomaterials. We present methods for fabricating spherical, poly(3,4-ethylenedioxythiophene):poly(styrene sulfonate) (PEDOT:PSS) hydrogel microparticles. When densely packed, these microparticles form a conducting granular hydrogel with microporosity as well as shear-thinning and self-healing dynamic mechanical properties. The PEDOT:PSS granular hydrogel can be extruded and maintain structure post-3D printing. Modulating microparticle PSS content achieves high granular hydrogel conductivity (137 S/m), and microparticles exhibit excellent cytocompatibility (>98% viability). Finally, we demonstrate utility as bioencapsulating electrodes for electrophysiological monitoring. These results highlight the functionality of our PEDOT:PSS conducting granular hydrogel, suggesting its potential as 3D printed bioencapsulating electrodes, 3D tissue engineering scaffolds for monitoring encapsulated cells, and injectable therapies for enhanced cell recruitment and tissue regeneration combined with electronic stimulation.

bioengineering↗

bioRxiv2025

Harnessing distinct tissue-resident immune niches via S100A9/TLR4 improves ketone, lipid, and glucose metabolism

Immunometabolism contributes to the development of metabolic diseases. Yet, how certain metabolic disorders, such as insulin deficiency (ID), influence immune cell function is poorly understood. Here, we observe that ID rearranges the immune landscape of the liver, causing a decrease in T cells and an increase in Kupffer cells, accompanied by a shift in the transcriptome and polarization of the latter. Treating ID mice with the protein S100A9 rescues the polarization and lipid-related changes caused by ID in the KCs, and rescues hypertriglyceridemia and hyperketonemia in a TLR4-dependent manner. Additionally, S100A9 acts on other immune niches to increase glucose uptake in skeletal muscle, improving hyperglycemia. In summary, the S100A9-TLR4 axis is a new tool to harness immune cells for improving ID-related metabolic dysfunction.

physiology↗

bioRxiv2025

Comparative Transcriptomics Reveals Context- and Strain-Specific Regulatory Programs of Agrobacterium During Plant Colonization

Agrobacterium is a genus of plant-associated bacteria capable of transferring DNA into host genomes to induce tumorigenesis. The process has been primarily studied in a few model strains, particularly C58, and developed into Agrobacterium-mediated transformation (AMT) for genetic manipulation. However, the diversity of wild-type strains and their context-specific regulatory responses remain poorly characterized. Here, we evaluated five wild-type strains and identified 1D1108 as superior in tumorigenesis on legumes and transient transformation in Nicotiana benthamiana. Under in vitro virulence induction with acetosyringone (AS), we identified 126 differentially expressed genes (DEGs) in 1D1108. Although the number of DEGs was comparable to those in C58 and the legume isolate 1D1609 under the same condition, only 22 DEGs, primarily within the vir regulon, were conserved, indicating extensive divergence among these Agrobacterium strains. Leaf infiltration of N. benthamiana revealed 1,134 DEGs specifically regulated in planta for 1D1108. These included genes involved in attachment, virulence regulation, type IV pilus, succinoglycan biosynthesis, and diverse nutrient transporters, providing new evidence on expression regulation during colonization. Comparative analyses of in planta transcriptomes with C58 and Pseudomonas syringae DC3000 revealed distinct secretion systems required for pathogenesis, namely type IV for Agrobacterium and type III for Pseudomonas, and only approximately 5-19% of DEGs were conserved. These limited transcriptomic overlaps underscore the importance of studying gene expression in strains and conditions directly relevant to the biological context, rather than relying on model systems. Together, this work reveals how environmental and host-associated cues shape transcriptional responses in plant-associated bacteria. IMPORTANCEAgrobacterium is a key tool for plant genetic engineering, yet much of our knowledge about its biology comes from a few strains studied under artificial conditions. This work combines tumor formation assays on legumes and transient transformation in Nicotiana benthamiana with transcriptomic analyses to explore how diverse Agrobacterium strains, particularly the high-performing wild-type strain 1D1108, respond to host and environmental cues. By comparing gene expression under in vitro and in planta conditions, we found that the bacterial response within plant tissue is more complex than previously appreciated. Cross-strain comparisons further revealed limited conservation in gene expression regulation, even under similar conditions. Comparative analysis with Pseudomonas syringae revealed the activation of distinct secretion systems for pathogenesis and differences in regulatory programs used during plant colonization. These findings underscore the importance of studying bacteria in biologically relevant contexts and highlight the limitations of relying solely on model strains to infer regulatory responses.

microbiology↗

bioRxiv2025

DeepAden: an explainable machine learning for substrate specificity prediction in nonribosomal peptide synthetases

Microbial non-ribosomal peptides (NRPs) exhibit remarkable structural diversity and serve as valuable sources of lead compounds for clinical drug development. The biosynthesis of NRPs relies on non-ribosomal peptide synthetases (NRPSs), in which adenylation (A) domains play a pivotal role in defining the core structure by selectively recognizing and activating amino acid substrates. Accurately predicting the substrate specificities of A-domains is thus essential for understanding the core structural and biosynthetic logic of NRPs. Here, we present DeepAden, a two-stage deep learning framework. In the first stage, a graph attention network (GAT)-based model localizes 27-residue binding pockets within 6 [A] of bound substrates and convert these into pocket representations. In the second stage, pocket representations are then encoded alongside substrate information using pretrained language models, and aligned using contrastive learning. In addition, we introduce a SHapley Additive exPlanations (SHAP)-guided data augmentation strategy to mitigate class imbalance and improve robustness, particularly for nonproteinogenic substrates. DeepAden achieves competitive performance compared with state-of-the-art tools on a benchmark dataset, and enabled the identification of two Streptomyces NRPS gene clusters through accurate A-domain substrates specificity predictions. DeepAden offers a powerful tool for precise pocket localization and robust substrate prediction, accelerating the discovery and characterization of novel NRP natural products for future work. The DeepAden web server is available at https://deepnp.site/.

bioinformatics↗

bioRxiv2025

Accurate prediction of protein stability changes from single mutations using self-distillation and antisymmetric constraint strategies

Computational approaches for accurately predicting protein stability changes upon residue mutations are crucial for protein engineering and design. Sequence-based methods are easier to apply to large-scale proteins since they do not rely on high-quantity structures. However, existing sequence-based approaches struggle to capture structural changes, resulting in lower performance compared to structure-based methods. In this study, we propose DPStab, a sequence-based deep learning solution that accurately predicts protein stability changes upon single residue mutations. DPStab transfers a protein large language model as a core component and incorporates a cross-attention mechanism to capture the contact changes around mutated positions for {Delta}{Delta}G and {Delta}Tm prediction. To address data imbalance and the antisymmetric nature of mutation effects, DPStab employs a self-distillation inference strategy under the supervision of an antisymmetric constraint. Benchmarking demonstrates that DPStab achieves state-of-the-art performance in both {Delta}{Delta}G and {Delta}Tm prediction. Practical evaluations confirm DPStabs capability in accurately ranking protein stability on large-scale datasets and effectively identifying critical structural contacts impacting stability. More experiments on extensive cDNA display proteolysis data demonstrate the significant contributions of self-distillation and antisymmetric constraint strategies. Significance StatementSingle amino acid mutations significantly influence protein stability, thereby affecting biological function and potential therapeutic uses. Accurately predicting how mutations affect protein stability is fundamental to protein engineering and therapeutic design. However, current sequence-based computational methods fail to capture the structural context changes around mutated residues. To overcome this, we propose DPStab, a sequence-based deep learning approach that combines a protein language model and a cross-attention mechanism with self-distillation and antisymmetric strategies. DPStab effectively captures residue contact changes and predicts stability changes without structural data. Sufficient experiments demonstrate that DPStab significantly outperforms existing methods, providing a fast and practical tool for enhancing protein engineering and biomedical research.

bioinformatics↗