Biology subjects

Lu, W.

Publications and source records attributed to Lu, W..

At least 37 records · Page 2Linked to original sources

bioRxiv2026

Mapping the mammalian dark metabolome by in vivo isotope tracing

Despite decades of biochemical study, a comprehensive map of the mammalian metabolome remains elusive. Mass spectrometry-based metabolomics detects thousands of small molecule-associated signals in mammalian tissues, but it is currently unclear how many of these reflect products of endogenous metabolism. Here, we leverage systematic in vivo isotope tracing to infer the biosynthetic origins of unidentified metabolites. We administered 26 different isotopically labelled nutrients to mice, measured circulating and tissue metabolite labelling by mass spectrometry, and developed a statistical framework to infer the number of carbon atoms incorporated from each of these precursors into more than 4,000 putative metabolites. We show this information can be harnessed for biosynthesis-aware structure elucidation using a multimodal AI model that co-embeds isotopic labelling patterns with chemical structures. This approach revealed several previously unrecognized families of mammalian metabolites, including cysteine-derived alkylthiazolidines, dithioacetal mercapturic acid derivatives, short-chain N-acyltaurines, acylglycyltaurines, and N-oxidized taurines. It further uncovered a family of mevalonate-derived isoprenoid metabolites that includes 2,3-dihydrofarnesoic acid, which is markedly depleted in both mouse and human aging. Age-related depletion of these isoprenoids is driven by impaired coenzyme A synthesis. Our work establishes the biosynthetic precursors for thousands of unidentified metabolites and reveals multiple previously unrecognized branches of mammalian metabolism.

biochemistry↗

bioRxiv2026

UniST: A Unified Computational Framework for 3D Spatial Transcriptomics Reconstruction

Spatial transcriptomics (ST) enables the measurement of gene expression in its native spatial context, yet most ST datasets are acquired as two-dimensional (2D) sections. Consequently, the underlying three-dimensional (3D) organization of tissues is only partially observed, and 3D ST data generated from serial sections are typically sparse and heterogeneous, with substantial tissue loss and missing measurements. These limitations pose major analytical challenges for reconstructing coherent 3D tissue architecture, rather than issues of experimental scalability alone. Here, we present UniST, a unified generative artificial intelligence (AI) framework designed to computationally reconstruct dense and continuous 3D ST landscapes from sparse serial sections, without altering the underlying experimental ST technologies. UniST integrates three complementary modules: kernel point convolution with cross-attention layers for point cloud upsampling, optical flow-based interpolation for continuous slice reconstruction, and a graph autoencoder with implicit neural representations for gene expression imputation. Together, these components densify sparse slices, resolve discontinuities, and map spatial coordinates to high-dimensional transcriptomics. Across multiple ST platforms and tissue contexts, UniST accurately restored structural continuity and biologically meaningful expression patterns. In a mouse embryo dataset, UniST reconstructed a dense 3D heart architecture from sparsely sampled slices. In two 3D human cancer tissues, UniST recovered critical spatial features, including tumor-immune boundaries and tertiary lymphoid structures, that were fragmented in the original data. By providing a generalizable computational solution that complements existing ST acquisition protocols, UniST facilitates cost-efficient and scalable reconstruction of 3D ST landscapes, enabling more faithful investigation of tissue organization and disease biology.

bioinformatics↗

bioRxiv2026

In vivo-compatible spatial multi-omics via hydrogen peroxide-independent APEX2 labeling

Proximity labeling (PL) technologies like APEX2 have transformed spatial multi-omics in live cells, but their long-standing dependence on hydrogen peroxide (H2O2) disrupts redox signaling and prevents use in live animals. Here we introduce H2O2-independent APEX2 (Hi-APEX), which uses a clickable tetrazine-phenol probe, requiring no enzyme engineering. We show that APEX2 directly catalyzes TP radical formation without H2O2 via a mechanism requiring the probes tetrazine group and a key histidine residue. We benchmarked Hi-APEX-based spatial multi-omics by mapping the mitochondrial matrix and dynamic secretomes. Hi-APEX significantly outperforms traditional APEX in capturing redox-sensitive processes such as stress response and ferroptosis, enabling discovering authentic stress granule components and protein interaction networks for mitochondria-localized GPx4. One mGPx4 interactor TRMT61B--known to regulate mitochondrial m{superscript 1}A modifications--promotes ferroptosis. Crucially, Hi-APEX achieves full in vivo compatibility, enabling direct PL in tumor xenografts and hippocampal neurons, thereby expanding PL-based spatial multi-omics from cellular systems to living organisms.

biochemistry↗

bioRxiv2026

Med12 and Med13 prevent tumorigenic dedifferentiation of intermediate neural progenitors and premature loss of neural stem cells

Med12 and Med13 are components of the kinase module of the mediator complex. Mutations of Med12 and Med13 have been associated with neurodevelopmental disorders and various cancers. However, their functions in neural development are not well understood. Here we show that in the developing Drosophila brain, Med12 and Med13 are required to prevent tumorigenic dedifferentiation of intermediate neural progenitors (INPs) and maintain neural stem cell (NSC) self-renewal. We further demonstrate that Med12 and Med13 prevent INP dedifferentiation by coordinating with a subset of core mediator complex subunits to mediate the activation of genes required for INP fate commitment. In contrast, during the maintenance of NSC self-renewal, Med12 and Med13 antagonize the function of a different subset of core mediator complex subunits. Together, our findings reveal that Med12 and Med13 perform two distinct functions in neural progenitors by coordinating with one subset of core mediator complex subunits while antagonizing another. HighlightsO_LILoss of Med12 and Med13 causes dedifferentiation of intermediate neural progenitors C_LIO_LIMed12 and Med13 mediate the activation of target genes of PntP1 C_LIO_LILoss of Med12 and Med13 leads to premature loss of neural stem cells C_LIO_LIMed12 and Med13 act with one subset of core mediator subunits but oppose another C_LI eTOC blurbZhu and his colleagues show that Med12 and Med13 promote cell fate commitment of intermediate neural progenitor cells and self-renewal of neural stem cells. Med12 and Med13 perform these two distinct functions by coordinating with one subset of core mediator complex subunits while opposing another to regulate the expression of different target genes.

developmental biology↗

bioRxiv2026

To Be or Not to Be an Oocyte: Msps/XMAP215 Controls Oocyte Cell Fate in the Drosophila Ovary

Cell fate determination, the process by which cells commit to specific identities, is fundamental to development. Oocyte specification in Drosophila provides an excellent model to dissect these molecular mechanisms. In the Drosophila ovary, 16 interconnected germline cells arise through four mitotic divisions with incomplete cytokinesis. Among them, the two oldest cells become pro-oocytes, and one is selected to develop into the oocyte while maintaining diploidy; the remaining 15 differentiate into nurse cells and enter endoreplication. Previously, we showed that Mini spindles (Msps), a microtubule polymerase and homolog of XMAP215, is essential for oocyte maintenance. Here, we report that Msps/XMAP215 is expressed early in development, with both mRNA and protein enriched in the pro-oocytes prior to oocyte specification. Knockdown of Msps prevents oocyte specification, leading to egg chambers with 16 nurse cells. Loss of Msps also disrupts the accumulation of the oocyte marker Orb and the microtubule minus-end binding protein Patronin/CAMSAP, both critical for oocyte specification. Remarkably, optogenetic recruitment of Msps is sufficient to increase microtubule polymerization and promote Orb accumulation in nurse cells, suggesting that Msps activity is sufficient to drive oocyte fate determination. Furthermore, we show that Msps associates with the germline-specific organelles, spectrosome and fusome, and becomes asymmetrically distributed among sister cells, allowing the two pro-oocytes to inherit higher levels of Msps than their siblings. Together, we propose a model in which Msps-mediated microtubule polymerization provides the pro-oocyte with a competitive advantage, initiating a positive feedback loop that involves dynein-dependent transport of msps mRNA to reinforce oocyte specification. TeaserA self-reinforcing cytoskeletal feedback loop selects a single egg cell from a cluster of interconnected germ cells.

cell biology↗

bioRxiv2026

Coronavirus Nucleocapsid Proteins Hijack Host Protein Kinase A Catalytic Subunit α into Nucleus to Evade from STAT1 Signaling

The role of the cAMP/PKA pathway in antiviral innate immunity, including during coronavirus infections, remains unclear. We discovered coronavirus N proteins initiate cAMP-ADCY10-PKA cascade. Cytoplasmic PKA activates STAT1 independently of canonical JAK/TYR2 signaling. Coronavirus N proteins, via a conserved arginine (e.g., PEDV R58, SARS-CoV-2 R92), directly bind and sequester PKA C into the nucleus to evade STAT1 activation. Using PEDV as a model, mutant viruses with NmutNLS and NR58A were generated. Wildtype PEDV infection suppressed STAT1 activation in cells expressing PKA C/PKA CS339A deficient in STAT1 phosphorylation. However, in rXS0101mutNLS- or rXS0101R58A-infected cells, only PKA CS339A expression inhibited STAT1 activation. Downregulation of STAT1 activation is accompanied with increased viral replication. This study first elaborates that PKA C activates STAT1 in the cytoplasm of infected cells distinctly from canonical JAK/STAT1 signaling, while coronaviruses evade this antiviral response by sequestering PKA C into the nucleus via direct N protein interaction.

microbiology↗

bioRxiv2026

Identification of Novel Extracellular Vesicles Scaffold Proteins for Versatile Cargo Engineering

Extracellular vesicles (EVs) are promising drug delivery platforms that have been engineered to carry various drug modalities. Key strategies for generating those therapeutic EVs involve the direct fusion of protein of interest (POI) to EVs scaffold proteins with inherently high EVs-sorting ability. In this work, we identified raftlin (RFTN1) and poliovirus receptor (PVR) as novel EVs scaffold proteins for loading EVs lumen and surface cargos respectively. Truncation studies revealed that the N-terminus 15 residues from RFTN1 (RFTN1-N15) were sufficient for EVs engineering, as demonstrated by distinct cargos including gene editing tools, cytosolic enzymes as well as type II transmembrane proteins. On the other hand, PVR efficiently displayed secreted proteins including antibodies and serum albumins on EVs surface. Critically, RFTN1 and PVR-engineered EVs demonstrated consistent and efficient cargo delivery in vivo. In summary, the discovery of RFTN1 and PVR can potentially benefit EVs engineering for both fundamental research and clinical translation in the future.

bioengineering↗

bioRxiv2025

Efficient calculation of fluid transport in porous media with moving boundaries

A novel hybrid model combining the lattice Boltzmann (LB) and finite difference (FD) methods is proposed to simulate transport in through a junction of actively contracting lymphatic vessels, while also handling flow of interstitial liquid in the surrounding porous tissue. Details of the dynamically flexing walls and valves in the lymphatic vessel and its near vicinity are modeled using a high-resolution LB method, whereas overall efficiency was significantly improved by using low-resolution FD in the larger tissue domain distant from the vessel. Pressure and velocity conditions at the interface between subdomains of the two numerical methods are matched by imposing a partial bounce-back ratio in LB corresponding to the permeability coefficient{kappa} in Darcys law for flow through porous media. Parameters governing the match between the algorithms at their interface can be estimated from the Kozeny-Carman relationship for porous media and further refined with a simpler, parallel flow geometry that also serves to validate the method. Test calculations show that the hybrid method is roughly four times faster than the LB method and permits computation over significantly larger domains. This method should be applicable to a large range of problems involving fluid flow in porous media with embedded conduits that have non-stationary boundaries. Author summaryIt is generally acknowledged that the finite difference method (FDM) is faster and requires less memory than the lattice Boltzmann method (LBM) for comparable domain sizes. However, LBM performs better for simulating fluid flow near complex, deformable, or moving boundaries. For this reason, it can be beneficial to create hybrid models that combine FDM and LBM. In this work, we use such a hybrid model to simulate a contracting lymphatic bifurcation in fluid. Our goal is to demonstrate the models robustness and high efficiency.

bioengineering↗

bioRxiv2025

SPARP-seq reveals poly(A) tail shortening as a hallmark of the active human translatome

The poly(A) tail at the 3 end of mRNAs plays a critical role in the synthesis of proteins, but its dynamic relationship with the translation process has been unclear. Here, using a novel single-molecule approach SPARP-seq that integrates poly(A) assessment with ribosome profiling, we demonstrate that poly(A) tail shortening is a universal hallmark of active translation in human cells. We find that most mRNAs begin translation on single ribosome (monosome) with long tails that progressively shorten as additional ribosomes are recruited. In contrast, highly expressed housekeeping genes are translated efficiently on monosomes and possess intrinsically short poly(A) tails. Furthermore, we find that translationally repressed transcripts, such as those retaining introns, are trapped on monosomes with long tails. Pharmacological inhibition of translation induces global poly(A) tail lengthening, confirming that deadenylation is coupled to ongoing translation. These findings reveal a dynamic system where the poly(A) tail integrates mRNA status with translational fate, defining the monosome as a central hub for post-transcriptional control.

molecular biology↗

bioRxiv2025

Deep Learning Enables Automated Segmentation and Quantification of Ultrastructure from Transmission Electron Microscopy Images

Transmission electron microscopy (TEM) has become an essential technique for observing subcellular ultrastructure, and is widely used in both clinical diagnosis and biomedical research. However, analysis of TEM data remains extremely labor-intensive and often inconsistent across operators due to the lack of dedicated computational methods. Here, we present TEAMKidney, a deep learning framework for accurate and scalable measurement of ultrastructures in TEM images across species, magnifications, and instrument platforms. We collected 12,991 TEM images from patients with multiple kidney diseases and from different animal models. By combining a self-training-based semantic segmentation stage with a TEM-tailored panoptic segmentation model, we address two major challenges in TEM data analysis: the lack of accurately labeled training data and the difficulty of achieving high segmentation accuracy for complex ultrastructure. Application of TEAMKidney to both human and animal images successfully reveals disease-associated changes in two critical glomerular ultrastructures: the glomerular basement membrane and podocyte foot processes. In addition to significantly outperforming existing tools, TEAMKidney shows close agreement with pathological expert measurements used in clinical assessment protocols. By reducing dependence on manual tracing while preserving expert-level accuracy, TEAMKidney demonstrates that deep learning can substantially reduce the burden of image analysis in both clinical pathology and biomedical research settings.

bioinformatics↗

bioRxiv2025

Hi-C informed kernel association test: integrating 3-dimensional genome structure into variant-set association for whole-genome sequencing data

Variant-set association analysis is a powerful strategy for genetic studies of whole genome sequence (WGS) data, especially for rare variants. By aggregating variant signals, variant-set analysis can improve statistical power, result interpretability, and study replicability. Motivated by evidence that three-dimensional (3D) genome architecture plays a critical role in regulating gene transcription, several works have incorporated 3D genome architecture into gene-based association tests and demonstrated great promise. In this work, we extend the idea of 3D-genome guided test from gene-centric to gene-agnostic, whole-genome testing by introducing a Hi-C informed kernel association test. We present a principled procedure that converts Hi-C contact confidence into borrowing weights and integrates these weights into genetic similarity kernels so that higher-confidence interacting loci contribute more to the association test of the target variant set. We use a controlling parameter to adaptively determine the appropriate degree of information borrowing from its interacting loci during association testing. We assess the performance of the Hi-C informed test using simulations and illustrate its advantage in detecting rare-variant sets using WGS data from the ARIC study in the Trans-Omics for Precision Medicine (TOPMed) program.

bioinformatics↗

bioRxiv2025

HLA-DQ2/8 genotype imprints strain-level gut microbiome and constrains pantothenate availability in healthy adults

HLA-DQ2/8 haplotypes are established genetic risk factors for autoimmune diseases and are known to influence gut microbiota assembly in early life. However, their impact on the adult microbiome and functional consequences for host physiology remain unclear. Here, we performed a genotype-stratified multi-omics analysis of 60 healthy adults, including 28 HLA-DQ2/8 carriers and 32 non-carriers. We found that host HLA-DQ2/8 genotype was significantly associated with gut microbiome composition, with an effect size exceeding that of sex and BMI. HLA-DQ2/8 carriers exhibited higher gut microbial alpha-diversity, lower virulence factor abundance, and a distinct species profile enriched in butyrate-producing taxa. We identified pervasive intra-species phylogenetic and functional divergence linked to the DQ2/8 genotype. This diversification reflects predicted HLA-restricted microbial peptide-binding specificities, suggesting a possible role for antigen presentation-mediated immune selection, a mechanism further supported by AlphaFold3 structural modeling. We found an enrichment of microbial pathways for pantothenate and coenzyme A (CoA) biosynthesis in carriers, primarily driven by functionally divergent Blautia obeum strains. This functional shift paralleled lower levels of serum pantothenate and HDL-cholesterol in the host. Our findings suggest a potential genotype-microbiome-host axis where antigen presentation-mediated immune selection may modulate microbial adaptation, with possible implications for the host availability of essential cofactor precursors and lipid metabolism.

microbiology↗

bioRxiv2025

Molecular and cellular dynamics of measurable residual disease progression in myelodysplastic syndromes

Cancer relapse after treatment invariably proceeds from the persistence and progression of measurable residual disease (MRD), an ultrasmall malignant population. Despite the poor prognostic impact established by increasingly sensitive MRD detection methods, the molecular pathways defining MRD remain unknown. To identify the unique features of MRD and the molecular forces shaping its progression, we performed single-cell multi-omic profiling on DNA, RNA and protein layers for a longitudinal patient cohort with relapsed myelodysplastic syndromes (MDS) after stem cell transplantation (SCT), the only curative modality for MDS. We provide a comprehensive molecular portrait of MRD cells with novel markers, shared across genetically heterogeneous patients. MDS relapse after SCT manifested universally with marked phenotypic evolution. Genotype and phenotype analyses revealed MRD progression as a dynamic, evolutionary process rather than a static expansionary one, driven by both subclonal sweeping and cell state transitions. Malignant cells adapted to infiltrating T cells by rewiring IFN-{gamma} responses to activate a key immunoevasive pathway. Our study demonstrates the power of longitudinal, single-cell multi-omic analysis for identifying, tracking, and understanding MRD cells, opening new avenues to target MRD persistence and progression.

cancer biology↗

bioRxiv2025

Proteome integral solubility alteration assay for monitoring antibody-antigen cell interactions

The Proteome Integral Solubility Alteration (PISA) assay is commonly used to identify and characterize interactions between small molecules and the cellular proteome. In this study, a modified PISA approach identified tumor necrosis factor alpha (TNF) as the antigen of the monoclonal antibody infliximab in activated monocytes - uniquely among [~]8,000 cellular proteins. These results highlight the potential of PISA-based analysis not only for monitoring antibody-antigen interactions in therapeutic development, but also for identifying new antigens and characterizing the initial effector functions of antibodies within complex proteome environments. TeaserModified PISA pinpointed TNF- as Infliximabs antigen among 8000 proteins, highlighting broad potential for therapeutic antigen studies.

molecular biology↗

bioRxiv2025

Spen and Nito prevent dedifferentiation of progenitors by translationally repressing E(Spl)mγ

Rapid termination of Notch signaling by asymmetrically segregated Numb is essential for specification of differentiating progeny during asymmetric stem cell division. Using Drosophila type II neuroblasts as a model, here we report that specification of the differentiating progeny also requires the expression of the Notch target E(Spl)m{gamma} in the stem cell to be kept at low levels by the SPEN family proteins, Spen and Nito. We show that loss of Spen or Nito leads to a drastic increase in E(Spl)m{gamma} expression in the stem cell and subsequent dedifferentiation of its progeny. Genetic and biochemical studies demonstrate that Spen and Nito maintain the expression of E(Spl)m{gamma} in the stem cell at low levels by directly binding to two identical novel motifs in the 5UTR to repress its translation. The low expression in the stem cell prevents excessive inheritance of E(Spl)m{gamma} proteins by the differentiating progeny and ensures its rapid removal from the progeny. Together, our work uncovers a novel post-transcriptional mechanism regulating Notch signaling that is critical for cell fate specification during asymmetric stem cell division.

developmental biology↗

bioRxiv2025

EED Maintains the Small Cell Lung Cancer Neuroendocrine Phenotype and Drives Lung Cancer Histological Transformation

Lung cancer histological subtypes include lung adenocarcinoma (LUAD) and small cell lung cancer (SCLC). While typically distinct, combined LUAD/SCLC histology tumors occur, and LUAD can transform into SCLC as a resistance mechanism to targeted therapies, especially in EGFR-Mutant LUADs with RB1/TP53-inactivation. Although PRC2 complex expression increases during this transformation, its functional role has remained unclear. Using CRISPR-based autochthonous immunocompetent GEMMs, we demonstrate that inactivation of EED, the core PRC2 scaffolding subunit, impairs SCLC tumorigenesis and drives histological transformation from ASCL1-positive SCLC to LUAD through a transient NEUROD1-positive intermediate state. Mechanistically, EED loss de-represses bivalent genes co-marked by H3K27me3 and H3K4me3, including LUAD oncogenic RAS, PI3K, and MAPK pathway genes, to promote transformation to LUAD. Consistently, these same signaling genes are bivalently repressed in human SCLC patient-derived xenograft (PDX) tumors, suggesting a conserved PRC2-dependent mechanism to repress LUAD lineage oncogenic signaling to maintain the SCLC neuroendocrine identity. In a complementary EGFR-mutant LUAD GEMM with RB1/TP53 inactivation, EED was required for LUAD-to-SCLC transformation and distant metastasis upon EGFR withdrawal. These findings identify the PRC2 complex as a key epigenetic enforcer of SCLC neuroendocrine identity and nominate EED inhibition as a potential strategy to block SCLC transformation in high-risk LUAD.

cancer biology↗

bioRxiv2025

Accurate Protein-Protein Interactions Modeling through Physics-informed Geometric Invariant Learning

AlphaFold has set a new standard for predicting protein structures from primary sequences; however, it faces challenges with protein complexes across species, engineered proteins, and antigen-antibody, where co-evolutionary signals may be sparse or missing. Herein, we present ProTact, a SE(3)-invariant geometric graph neural network that integrates physics-informed geometric complementarity and trigonometric constraints as robust inductive biases to enhance protein-protein contact predictions. ProTact is applicable to both experimental and predicted monomer structures and utilizes a modulated key point matching algorithm to approximate accurate docking poses. Experimental evaluations demonstrate that ProTact consistently outperforms state-of-the-art sequence-based and structure-based methods on benchmark datasets, achieving notable relative improvements of 31.63% in Precision@10 for CASP 13 and 14 targets, and 31.94% for DIPS-Plus datasets. Moreover, when combined with AlphaFold3 as re-scoring functions, ProTact surpasses its default confidence scores, offering over 30.48% improvements in the low-MSA contexts. We anticipate that the proposed framework will advance our understanding of protein interactions, functions, and design.

bioinformatics↗

bioRxiv2025

Microbial Peptidoglycan Engages Autophagy Receptor P62 to Induce Protective Mitophagy in the Liver

Although mitophagy is critical for maintaining mitochondrial integrity and hepatic homeostasis, the microbial-derived signals controlling this process remain unknown. Given the gut microbiotas profound influence on liver pathophysiology, identifying specific bacterial factors that directly regulate hepatocyte mitophagy could unlock novel therapeutic strategies. In this study, we identify bacterial peptidoglycan (PGN)--a conserved cell wall component--as a key activator of mitophagy that protects against hepatocyte death. Through both in vivo and in vitro studies, we demonstrate that either heat-killed Escherichia coli or purified PGN attenuates hepatocyte death. Mechanistically, PGN is internalized by hepatocytes, localizes to mitochondria, and initiates mitophagy via direct interaction with the autophagy adaptor p62/SQSTM1. Genetic ablation of p62 in hepatocytes completely abolishes PGN-induced mitophagy, underscoring the pathways essential role. Strikingly, therapeutic administration of PGN markedly alleviates carbon tetrachloride (CCl4)-induced hepatic fibrosis, reducing collagen deposition and suppressing hepatic stellate cell activation through enhanced autophagic flux. Our work unveils a previously unrecognized host-microbe crosstalk in which PGN acts as a mitophagy inducer, offering a potential therapeutic avenue for liver diseases driven by mitochondrial dysfunction.

cell biology↗