Biology subjects

Yang, X.

Publications and source records attributed to Yang, X..

At least 253 records · Page 14Linked to original sources

bioRxiv2025

EXOSC10-mediated pre-tRNA surveillance safeguards neuron survival

tRNA quality control pathways have been identified in yeast, whereby aberrant and hypomodified mature tRNAs are targeted for 5-3 degradation by the rapid tRNA decay (RTD) pathway involving the Xrn1 and Rat1/Xrn2 exonucleases, whereas aberrant precursor tRNAs (pre-tRNAs) are targeted for 3-5 degradation by the nuclear surveillance pathway involving the RNA Exosome. However, the pathways controlling tRNA and pre-tRNA degradation in mammals have not yet been defined and the relevance of pre-tRNA surveillance pathways for normal cell physiology remains largely unknown. The RNA Exosome comprises a core of nine non-catalytic subunits (EXOSC1-9) to which the distinct, DIS3 and EXOSC10, 3-5 exonucleases associate. Here we find that EXOSC10 deficiency leads to accumulation of unspliced precursor tRNAs (pre-tRNAs) in mouse embryonic stem cells (ESCs) and is required for pre-tRNA decay in biochemical assays. Pre-tRNA overexpression causes diminished motor neuronal survival in a mouse ESC differentiation model. Our results identify a pre-tRNA decay pathway that links Exosome deficiency with neuron survival and provides insight into possible pathological mechanisms underlying human neurodevelopmental disorders caused by mutations in Exosome subunits and genes involved in tRNA biogenesis. O_LIExosc10 is required for pre-tRNA surveillance in cells and degradation in biochemical assays C_LIO_LIPre-tRNA expression inhibits survival of neurons C_LIO_LIPre-tRNA decay pathway links exosome deficiency with neurodevelopmental disorders C_LI

molecular biology↗

bioRxiv2025

Pathogen effector forms a hexameric phosphatase holoenzyme with host core enzyme to promote disease

Pathogen effectors play critical roles in pathogenesis by directly manipulating host cellular processes. The Ser/Thr protein phosphatase 2A (PP2A) is commonly targeted by pathogens. Effectors from devastating plant pathogen Phytophthora hijack the PP2A core enzyme in plant hosts, altering the host phosphoproteome. In this study, we present a series of cryo-electron microscopy structures of the Phytophthora effector PSR2 in complex with the host PP2A core enzyme to form a functional holoenzyme. The PSR2-PP2A complex adopts a unique hexameric architecture, driven by PSR2-mediated dimerization of two heterotrimers. This hexamer exhibits greater stability and features a more exposed catalytic pocket compared to the canonical trimeric form of PP2A, likely enhancing its virulence activity. Mutational analyses underscore the importance of this hexameric structure for PSR2s virulence function. These findings provide mechanistic insights into pathogen-mediated manipulation of a key host phosphatase and offer targets for developing disease resistance strategies.

biophysics↗

bioRxiv2025

Nanopore Sequencing Unveils Somatic Structural Variations as Biomarkers in Laryngeal squamous cell carcinoma Genomes

Laryngeal squamous cell carcinoma (LSCC) is an aggressive cancer with poor quality of life, largely due to the absence of reliable molecular biomarkers for early diagnosis and prognosis. While somatic structural variations (SVs) have been documented in LSCC, conventional short-read sequencing approaches are inherently limited in detecting of large-scale somatic SVs, which are critical to understanding tumorigenesis. Here, we presented SomaGauss-SV, a somatic SVs detection workflow leveraging nanopore long-read sequencing data. Benchmarking against five paired tumor cell line datasets showed SomaGauss-SV consistently achieves a balanced high precision and recall. SomaGauss-SV applied to 15 paired LSCC tumor-blood samples uncovered a comprehensive SVs landscape and a significant positive correlation between somatic deletion burden and smoking intensity. Furthermore, a high-frequency somatic simple repeat expansion was identified in 20/27 (74.1%) of LSCC patients, upregulating the expression of genes TP53BP2 and FBXO28 through spatial proximity. These findings underscore the potential of long-read sequencing and SomaGauss-SV in uncovering previously unexplored genomic rearrangements in LSCC, providing valuable resources for biomarker discovery.

bioinformatics↗

bioRxiv2025

Preparation of Scutellariae Radix and Corn-Soy Peptides Complex and Its Hepatoprotective Mechanisms

Amidst the growing global burden of alcohol-associated liver disease (ALD) and the limited clinical interventions, this study innovatively developed a complex of Scutellariae Radix and corn-soy composite peptides (SR-CP). SR-CP exhibited significantly enhanced bioaccessible antioxidant capacity, demonstrating a 0.84-2.89-fold increase in DPPH, ABTS, {middle dot}OH, O2-{middle dot}, and FRAP value versus individual components post-gastrointestinal digestion. In ethanol-injured BRL-3A hepatocytes, SR-CP (60 g/mL) significantly ameliorated cellular viability by 38.28% (vs. ethanol model), normalized ALT/AST leakage by 19.92%/87.35%, restored SOD/GSH antioxidant reserves to 24.74 {+/-} 1.71 U/mg prot and 55.35 {+/-} 1.64 mol/g prot respectively, while suppressing lipid peroxidation through 65.93% reduction in MDA production (vs. ethanol model). Mechanistic profiling revealed dual regulatory effects: 1) Inhibiting CYP2E1/NADPH-OX ({downarrow} 51.97%/80.52%) to prevent excessive ROS production and activating the AMPK/SIRT1 axis ({uparrow} 30.70%/6.99%); 2) Potentiation of Nrf2 nuclear translocation ({uparrow} 0.48-fold) and HO-1 induction ({uparrow} 1.63-fold) through AMPK-mediated antioxidant signaling. The superior hepatoprotection of SR-CP over single constituents establishes a novel paradigm for functional food development, bridging traditional herbal wisdom with molecular-targeted ALD intervention via the AMPK/Nrf2/HO-1 axis.

pharmacology and toxicology↗

bioRxiv2025

PB-scope: Contrastive learning of dynamic processing body formation reveals undefined mechanisms of approved compounds

Membraneless organelles (MLOs) formed by liquid-liquid phase separation contribute to intracellular compartmentalization of specific biological functions. The dynamic regulation of MLOs is critical for RNA metabolism, stress response, and signal transduction. While the number and size of MLOs serve as valuable indicators of cellular states, leveraging their spatial distribution and diverse quantitative characteristics to infer cellular conditions has posed significant challenges. Here, we present processing body (PB)-scope: an unsupervised deep learning-based framework designed for imaging-based phenotypic screening of PB, one representative MLO. The model was trained on a substantial dataset comprising over 400,000 single-cell confocal microscopy images from a colon cancer cell line treated with 280 compounds. PB-scope enabled precise drug classification based on their effects on PB characteristics, including number and size. This artificial intelligence framework offers insights previously obscured by the dynamic and subtle nature of PBs. Notably, PB-scope identified drugs with a shared function in the JAK signaling pathway, a previously unrecognized regulator of PB dynamics. PB-scope holds the potential for adapting to the study of other MLOs, offering broad applicability in this emerging field of cell biology.

cell biology↗

bioRxiv2025

Evaluating Large Language Models for Gene-to-Phenotype Mapping: The Critical Role of Full-Text Database Access

Transformer-based large language models (LLMs) have demonstrated significant potential in the biological and medical fields due to their ability to effectively learn from large-scale, diverse datasets and perform a wide range of downstream tasks. However, LLMs are limited by issues such as information processing inaccuracies and data confabulation. These limitations hinder their utility for literature searches and other tasks requiring accurate and comprehensive extraction of information from extensive scientific literature. In this study, we evaluated the performance of various LLMs in accurately retrieving peer-reviewed literature and mapping correlations between 198 genes and six phenotypes: bone formation, cartilage formation, fibrosis, cell proliferation, tendon formation, and ligament formation. Our analysis included three types of models. First, standard transformer-based LLMs (ChatGPT-4o and Gemini 1.5 Pro). Second, specialized LLMs with dedicated custom databases containing peer-reviewed articles (SciSpace and ScholarAI). Third, specialized LLMs without dedicated databases (PubMedGPT and ScholarGPT). Using human-curated gene-to-phenotype mappings as the ground truth, we found that specialized LLMs with dedicated databases achieved the highest accuracy (>80%) in gene-to-phenotype mapping. Additionally, these models were able to provide relevant peer-reviewed publications supporting each gene-to-phenotype correlation. These findings underscore the importance of database augmentation and specialization in enhancing the reliability and utility of LLMs for biomedical research applications.

genetics↗

bioRxiv2025

Neurotrophic Factor-α1/carboxypeptidase E regulates critical protein networks to rescue neurodegeneration, defective synaptogenesis and impaired autophagy in Alzheimer's Disease mice

BackgroundThe global aging population is increasingly inflicted with Alzheimers disease (AD), but a cure is still unavailable. Neurotrophic Factor-1/carboxypeptidase E (NF-1/CPE) gene therapy has been shown to prevent and reverse memory loss and pathology AD mouse models However, the mechanisms of action of NF-1/CPE are not fully understood. We investigated if a non-enzymatic form of NF-1/CPE-E342Q is efficient in reversing AD pathology and carried out a proteomic study to uncover the mechanisms of action of NF-1/CPE in AD mice. MethodsAAV-human NF-1/CPE and a non-enzymatic form, NF-1/CPE -E342Q were delivered into hippocampus of 3xTg-AD mice and effects on cognitive function, neurodegeneration, synaptogenesis and autophagy were investigated. A quantitative proteomic analysis of hippocampus of 3xTg-AD mice with and without AAV-NF-1/CPE treatment was carried out. ResultsHippocampal delivery of AAV-NF-1/CPE-E342Q prevented memory loss, neurodegeneration and increase in activated microglia in 3xTg-AD mice, indicating its action is independent of its enzymatic activity. Quantitative proteomic analysis of hippocampus of 3xTg-AD mice that underwent NF-1/CPE gene therapy revealed differential expression of >2000 proteins involving many metabolic pathways. Of these, two new proteins down-regulated by NF-1/CPE: Nexin4 (SNX4) and Trim28 which increase A{beta} production and tau levels, respectively were identified. Western blot analysis verified that they were reduced in AAV-NF-1/CPE treated 3xTg-AD mice compared to untreated mice. Our proteomic analysis indicated synaptic organization as top signaling pathway altered as a response to CPE expression. Synaptic markers PSD95 and Synapsin1 were decreased in 3xTg-AD mice and were restored with AAV-NF-1/CPE treatment. Proteomic analysis hypothesized involvement of autophagic signaling pathway. Indeed, multiple proteins known to be markers of autophagy were down-regulated in 3xTg-AD mice, accounting for impaired autophagy. Expression of these proteins were upregulated in 3xTg-AD mice with NF-1/CPE gene therapy, thereby reversing autophagic impairment. ConclusionsThis study uncovered vast actions of NF-1/CPE in restoring expression of networks of critical proteins including those necessary for maintaining neuronal survival, synaptogenesis and autophagy, while down-regulating many proteins that promote tau and A{beta} accumulation to reverse memory loss and AD pathology in 3xTg-AD mice. AAV-NF-1/CPE gene therapy uniquely targets many metabolic levels, offering a promising holistic approach for AD treatment. Graphic abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=113 SRC="FIGDIR/small/657876v1_ufig1.gif" ALT="Figure 1"> View larger version (35K): org.highwire.dtl.DTLVardef@10e468forg.highwire.dtl.DTLVardef@3a78adorg.highwire.dtl.DTLVardef@c8d62forg.highwire.dtl.DTLVardef@165d118_HPS_FORMAT_FIGEXP M_FIG C_FIG

neuroscience↗

bioRxiv2025

Contrastive Learning-Enhanced Drug Metabolite Prediction Drives the Discovery of an Orally Bioavailable RSK4 Inhibitor for Esophageal Cancer

Drug metabolism liabilities are a major bottleneck in drug development, necessitating accurate metabolite prediction tools. While AI-driven methods have advanced, challenges remain in achieving high-precision, robustness, broad reaction coverage and seamless integration into the drug development workflow. To address this, we first constructed the most comprehensive human-specific drug metabolism database to date, encompassing 11,665 human metabolic reactions and 2,497 high-quality rules. Leveraging this resource, we developed ConMeter (Contrastive learning and feature interaction based Metabolic Reaction prediction), a novel model integrating chemical feature interaction and contrastive learning. ConMeter demonstrates significant improvements, achieving [~]80% accuracy in identifying at least one metabolite within the top 5 predictions and showing an average 25% relative enhancement in ranking true positives compared to state-of-the-art methods. The comprehensive database also expands the range of predictable metabolic reactions. To demonstrate the utility of ConMeter, we integrated it into a metabolite prediction-guided drug design strategy aimed at addressing the low oral bioavailability of a metabolically vulnerable RSK4 lead compound -- a pervasive issue among known RSK inhibitors. This strategy successfully yielded R636, a novel RSK4 inhibitor exhibiting remarkably improved bioavailability (a 63-fold increase, reaching 63% absolute bioavailability) while maintaining its potent activity. Notably, R636 showed favorable safety profile and significant anti-tumor efficacy against esophageal squamous cell carcinoma (ESCC) in vitro and in two distinct patient-derived xenograft (PDX) mouse models. This work introduces ConMeter as a powerful and practical tool for high-accuracy metabolite prediction, demonstrating its potential to accelerate lead optimization, exemplified by the discovery of R636 as a promising orally bioavailable candidate for ESCC.

cancer biology↗

bioRxiv2025

Unified nonparametric analysis of single-molecule spatial omics data using probabilistic indices

Spatial omics technologies localize individual molecules at subcellular resolution, shedding light on the spatial micro-organisation of living organisms. Yet the development of analysis methods struggles to keep pace with growing numbers of molecules, features and replicates being measured, and with new scientific questions arising on single molecules localization patterns. To meet this need, we present smoppix, a nonparametric analysis method based on the probabilistic index, which unifies tests for several univariate and bivariate localization patterns, such as aggregation of transcripts or colocalization of transcript pairs, in a single framework. These tests can be performed across tissues as well as within cells, while accounting for nested design structures. The high-dimensionality of the data is exploited for variance weighting and for providing a meaningful background null distribution, unique for every individual molecule. smoppix sidesteps segmentation, warping, edge correction and density estimation, and scales to high numbers of molecules and replicates thanks to an exact permutation null distribution. We demonstrate its power by unearthing spatial patterns in four published datasets from different kingdoms, and validate some findings experimentally on Selaginella moellendorfii roots. Our method is available from Bioconductor as the R-package smoppix.

bioinformatics↗

bioRxiv2025

scDenorm: a denormalisation tool for integrating single-cell transcriptomics data

Integrating single-cell omics data at an atlas scale enhances our understanding of cell types and disease mechanisms. However, the integration of data processed by different normalisation methods can lead to biases, such as unexpected batch effects and gene expression distortion, leading to misinterpretations in downstream analysis. To address these challenges, we present scDenorm, an algorithm that reverts normalised single-cell omics data to raw counts, preserving the integrity of the original measurements and ensuring consistent data processing during integration. We evaluated scDenorms performance on large-scale datasets and benchmarked its impact on data integration and downstream analysis across three datasets.

bioinformatics↗

bioRxiv2025

Targeting C42 in PKGIα alleviates diastolic dysfunction in HFpEF

BackgroundIn our exploration of novel therapeutic strategies for treating heart failure with preserved ejection fraction (HFpEF), we focused on cysteine 42 (C42) in cGMP-dependent protein kinase I (PKGI), which plays a crucial role in regulating blood pressure and diastolic relaxation. Our findings demonstrate that targeting C42 on PKGI with urolithin A effectively limits diastolic dysfunction and cardiac remodeling in a multi-hit mouse model of HFpEF, while also improving the kinetics of relaxation and contraction in human-engineered heart tissue. MethodsWe evaluated various polyphenols, including urolithin A, for their capacity to oxidize and activate PKGI. This assessment involved recombinant protein, isolated rat aortic smooth muscle cells, and myography experiments using isolated mesenteric vessels. Additionally, we investigated the ability of urolithin A to stimulate PKGI-dependent phosphorylation of phospholamban (PLN) in neonatal rat ventricular myocytes (NRVMs). A multi-hit mouse model was employed to induce HFpEF in wild-type and C42S PKGI knock-in (KI) mice. Following this, the mice were given vehicle or urolithin A by gavage, and heart function was evaluated using echocardiography. The relevance to human physiology was assessed using human-engineered heart tissue. ResultsThe polyphenols quercetin, fisetin, and urolithin A were shown to induce C42-dependent activation of PKGI. This mechanism of kinase activation led to vasorelaxation, which was attenuated in mesenteric vessels isolated from KI mice. Additionally, urolithin A treatment promoted PKGI-dependent phosphorylation of phospholamban in isolated cardiomyocytes. In a multi-hit model of HFpEF, urolithin A reversed diastolic dysfunction and cardiac remodeling in wild-type mice, but not in KI mice. These findings were shown to be likely relevant to humans, as urolithin A also enhanced the kinetics of relaxation and contraction in human-engineered heart tissue. ConclusionsTargeting C42 in PKGI with urolithin A presents a promising new strategy for the treatment of HFpEF.

biochemistry↗

bioRxiv2025

A synthetic biology toolkit for interrogating plasmid-dependent methylotrophy and enhancing methanol-based biosynthesis of Bacillus methanolicus

Bacillus methanolicus represents a thermophilic methylotroph whose methanol utilization depends on plasmid-encoded genes. It serves as a unique model for deciphering plasmid-dependent methylotrophy and an ideal chassis for low-carbon biomanufacturing using CO2-derived C1 substrates. Despite its evolutionary uniqueness and industrial potential, the lack of synthetic biology tools has hindered both mechanistic understanding and strain engineering. Here, we present a comprehensive synthetic biology platform comprising a high-efficiency electroporation protocol, a CRISPR method enabling robust and multiplex genome editing, diverse neutral loci for gene integration and overexpression, and a cloud-based genome-scale metabolic model iBM822 for user-friendly biodesign. Leveraging this toolkit, we systematically dissected plasmid-dependent methylotrophy, host restriction-modification systems, and functional significance of the chromosomal methylotrophic genes through targeted deletion. To address plasmid loss-induced strain degeneration, we integrated the large endogenous plasmid pBM19 into the chromosome for stable and intact methylotrophic growth. Finally, by integrating metabolic modeling with CRISPR editing, we engineered L-arginine feedback regulation to achieve the first L-arginine biosynthesis from methanol. This study establishes a synthetic biology framework for B. methanolicus, promoting mechanistic exploration of methylotrophy and low-carbon biomanufacturing.

synthetic biology↗

bioRxiv2025

Chromatix: a differentiable, GPU-accelerated wave-optics library

Modern microscopy methods incorporate computational modeling as an integral part of the imaging process, either to solve inverse problems or optimize the optical system design itself. These methods often depend on differentiable optics simulations, yet no standardized framework exists--forcing computational optics researchers to repeatedly and independently implement simulations with limited reusability and performance. These common problems limit the potential impact of computational optics as a field. Here we present Chromatix: an open-source, GPU-accelerated, differentiable wave optics simulation library. Chromatix builds on JAX to democratize fast, parallelized simulation of diverse optical systems and expand the design space in computational optics. Chromatix standardizes a growing collection of optical elements and propagation methods allowing a broad range of applications, which we demonstrate here for snapshot microscopy, holography, and phase retrieval. We demonstrate speed improvements of 2-6x on a single GPU and up to 22x on 8 GPUs.

bioinformatics↗

bioRxiv2025

PepSeek: Universal Functional Peptide Discovery with Cooperation Between Specialized Deep Learning Models and Large Language Model

Recent computational foundation models have revitalized the scientific discovery pipeline. However, developing foundational models for functional peptide discovery is costly due to the scarcity of wet-lab validated data. Meanwhile, conventional deep learning models are hard to generalize to unseen tasks or data distribution. Here, we introduce PepSeek, a universal approach for peptide discovery that synergistically integrates the most advanced large language model (LLM) with specialized small models. PepSeek harnesses the robust reasoning and generalization capabilities of LLM while leveraging the high predictive accuracy of specialized models trained for tasks such as antimicrobial activity regression and functional peptide generation. We have devised multiple collaborative strategies and task-specific modules demonstrating leading performance in peptide identification and generation. Notably, PepSeek achieves remarkable zero-shot prediction accuracy for peptides with diverse functionalities. We used PepSeek to identify a group of broad-spectrum antimicrobial peptide that exhibits low toxicity and high activity against drug-resistant bacteria, with the best surpassing all peptides currently undergoing clinical trials. Our framework establishes a new pipeline for scientific discovery with the help of LLM and specialized models.

bioinformatics↗

bioRxiv2025

Exploring the mechanism of kaji-ichigoside F1 on bacterial pneumonia based on the network pharmacology and transcriptome

ObjectiveTo explore the anti-inflammatory mechanism of kaji-ichigoside F1 against bacterial pneumonia based on transcriptomics and network pharmacology. MethodNetwork pharmacological was used to analyse the potential target genes of kaji-ichigoside F1 action on bacterial pneumonia; molecular docking was used to analyse the docking binding energy of kaji-ichigoside F1 with key target genes; RAW264.7 macrophages were treated with klebsiella pneumoniae fluid and given kaji-ichigoside F1 intervention, the expression of inflammatory factors IL-1{beta}, IL-6, TNF- and IL-10 were dectected by qRT-PCR; transcriptomic analysis was performed to obtain differentially expressed genes, and relevant signaling pathways. ResultNetwork pharmacological analysis showed that the five the key target genes for kaji-ichigoside F1-bacterial pneumonia interaction were were TLR4, NFKB1, STAT3, IL1B, and JUN; molecular docking results of kaji-ichigoside F1 with key target genes showed the docking binding energy ranging from -5.9 to -8.4 kcal/mol; kaji-ichigoside F1 can reduce the klebsiella pneumoniae induced inflammatory response of macrophages, manifested with reducing the mRNA expression of pro-inflammatory factors IL-1{beta}, IL-6 and TNF-, and increase the mRNA expression of anti-inflammatory factor IL-10; transcriptomics analysis showed that the signaling pathways involved were mainly the TLR signaling pathway and NFKB signaling pathway. ConclusionThis study showed that kaji-ichigoside F1 alleviates bacterial pneumonia by targeting TLR and NFKB signaling pathways, rebalancing macrophage polarization with suppressing pro-inflammatory cytokines and increasing anti-inflammatory cytokines, highlighting kaji-ichigoside F1 as a novel agent for combating bacterial infections. Graphic abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=144 SRC="FIGDIR/small/649383v1_ufig1.gif" ALT="Figure 1"> View larger version (36K): org.highwire.dtl.DTLVardef@7c96eorg.highwire.dtl.DTLVardef@13977a6org.highwire.dtl.DTLVardef@1de6214org.highwire.dtl.DTLVardef@f91c50_HPS_FORMAT_FIGEXP M_FIG C_FIG

pharmacology and toxicology↗

bioRxiv2025

ShinyCell2: An extended library for simple and sharable visualisation of spatial, peak-based and multi-omic single-cell data

Single-cell technologies now span multiple modalities, generating large, complex datasets that challenge analysis and sharing. We present ShinyCell2, an enhanced R package for interactive visualisation of single-cell multi-omics and spatial transcriptomics data. ShinyCell2 retains the simplicity and lightweight deployment of its predecessor while introducing advanced visualisations, cross-modality comparisons, and statistical tools tailored to spatial and multi-omic data. It enables intuitive, rapid exploration of high-dimensional data without requiring extensive computational expertise.

bioinformatics↗

bioRxiv2025

Microbiota-derived secondary bile acids promote STING activation and antitumor activity

Microbiota metabolism generates diverse bile acids that are associated with health and disease, but the molecular targets and mechanisms of action for these metabolites have not been fully elucidated. Using deoxycholic acid (DCA) photoaffinity probes, with cholic acid (CA) probe as a comparator, we found many protein targets of microbiota-derived DCA in mammalian cells. Of note, we discovered that DCA binds the transmembrane domain of stimulator of interferon genes (STING), and enhances agonist-induced accumulation of higher-order STING species, type I interferon signaling and antitumor immunity. Moreover, DCA-producing microbiota species enhanced STING agonist-induced antitumor immunity primarily through the activation of natural killer (NK) cells. This reverse chemical microbiology approach revealed an unpredicted mechanism of action for DCA on STING activation and suggests that specific secondary bile acids and their associated microbiota species may promote NK activation and impact the efficacy of STING-targeted therapeutics.

immunology↗

bioRxiv2025

Reshaping epistatic network modules enhances wheat yield potential

Increasing wheat yield is a key approach to ensuring global food security and an enduring focus in crop breeding. Here, we assembled a panel of 3,030 wheat lines to dissect the genetic mechanisms of yield improvement. We conducted large-scale field trials across five ecological environments over two consecutive years to evaluate yield performance and constructed a haplotype atlas using the Wheat 660K genotype array. 234 quantitative trait loci (QTLs) and 522 haplotype blocks (HBs) were identified for 14 traits through genome-wide association studies (GWAS), including 10 QTLs and 35 HBs associated with grain yield. Genomic analysis revealed that wheat yield improvement is primarily driven by changes in the epistatic network rather than the introduction of new haplotype segments. SNPs or HBs explain 53.3%-62.9% of the phenotypic variation in yield, whereas epistasis explains 70.4%, highlighting the role of epistasis in yield improvement. Moreover, yield improvement is significantly correlated with the accumulation of favorable epistatic modules. Pedigree-based analysis revealed that the modules remained relatively stable under the pressure of breeding selection, while new favorable modules were also created during hybrid breeding, highlighting that crop design breeding should be based on genetic modules. This study provides insights into genetic mechanism of wheat yield improvement and guidance for designing future wheat in the era of artificial intelligence.

genetics↗