Biology subjects

Fu, Y.

Publications and source records attributed to Fu, Y..

At least 55 records · Page 3Linked to original sources

bioRxiv2026

An oxygen-sensing Polycomb-group protein encodes flooding stress memory in plants

Most organisms, including plants, can encode stress memories that improve resilience to repeated environmental challenges. Flooding events expose plants to recurrent hypoxia, yet whether plants establish an adaptive memory of flooding is unclear. Here we show that somatic flooding stress memory is a conserved feature across multiple angiosperm species. In Arabidopsis, this memory depends on the oxygen-sensitive Polycomb Repressive Complex 2 (PRC2) subunit VERNALIZATION2 (VRN2). Loss of VRN2 impairs epigenetic memory formation and disrupts transcriptional memory at key genes that promote anthocyanin accumulation and repress leaf senescence, adaptive responses that enhance flooding tolerance. Collectively, we reveal a molecular mechanism where VRN2-PRC2 acts as both an oxygen sensor and chromatin effector to establish adaptive flooding stress memory in plants. ShortOur work defines a molecular mechanism for flooding stress memory in plants, actioned via the PRC2 subunit VRN2, that acts as the direct molecular (low) oxygen sensor and transducer. Coupled to its role in cold sensing, this identifies VRN2 as a master sensor subunit of the plant PRC2 that integrates multiple environmental cues to modulate the chromatin landscape and encode environmental memory.

plant biology↗

bioRxiv2026

Integrin-independent Tie2 activation using de novo designed proteins

The Angiopoietin-Tie2 pathway is a key regulator of vascular stability, but therapeutic exploitation has been limited by the poor developability of Angiopoietin-1 (Ang1), and there are key unresolved mechanistic questions. Ang1 binds both Tie2 and 5{beta}1 integrin, and the role of the 5{beta}1 interaction in signaling has been unclear. We used RFdiffusion to design a stable, high-affinity Tie2 minibinder that has no affinity for 5{beta}1. The binder is a selective antagonist in monomeric form, and a potent agonist when assembled into an octavalent architecture (H8mb) which drives Tie2 clustering. H8mb signals as potently as Ang1, indicating that integrin engagement is not required for Tie2 activation. However, the duration of signaling is reduced, and internalization of H8mb-Tie2 complexes is more rapid, suggesting that integrin may function as a co-receptor for Ang1 that prolongs signaling by extending the lifetime of the receptor ligand complex on the cell surface. In a mouse model of acute respiratory distress syndrome (ARDS), H8mb markedly improved survival. These results demonstrate that de novo designed receptor binders can enable dissection of co-receptor control of signaling dynamics, and the more stable and manufacturable H8mb provides routes to more developable therapeutic candidates.

biochemistry↗

bioRxiv2026

JUMPlion improves quantitative DIA proteomics through ion-level recovery of missing values

Incomplete quantification remains a persistent challenge in data-independent acquisition (DIA) mass spectrometry (MS), particularly in low-input and single-cell analyses. In identification-driven workflows, missing protein quantities often arise not from true absence of the corresponding peptides, but from failure to retain low-abundance signals from precursor or product ions for quantification. Here we present JUMPlion (local inference of ion-level missingness), a DIA quantification framework that re-examines MS raw files to recover missing values at the ion level before protein quantification. JUMPlion re-extracts precursor- and production signals directly from raw data, infers ion-level measurements within precursor-specific local quantitative neighborhoods, and combines complementary precursor- and production signals into downstream quantification. Using benchmark datasets acquired on multiple DIA platforms, JUMPlion increased protein-level completeness, improved fold-change accuracy, and enhanced detection of differentially abundant proteins while maintaining low differential-abundance false discovery rates. These gains were most evident in low-input and single-cell DIA datasets. Together, these results show that addressing missingness at the ion level before protein-level summarization can improve DIA quantification in diverse acquisition settings.

bioinformatics↗

bioRxiv2026

CellChem: Cellular transcriptional responses reshape molecular representation space for efficient and multi-scale drug discovery

Despite decades of progress in computational drug discovery, deep learning-based molecular representation models remain largely structure-centric, assuming that chemical similarity approximates functional similarity. However, drug effects in cells are shaped not only by chemical similarity but also by molecular interactions in the cellular context. To capture this complexity, we introduce CellChem, a cellular-chemical representation-learning framework that learns cell-guided molecular representations of small-molecule actions within cells. By incorporating large-scale cellular transcriptional profiles during pretraining, CellChem reshapes molecular representation space from structure-centric to a balanced integration of structure and function. The learned CellChem molecular representations exhibit biologically meaningful geometric organization, such that distances between molecules encode not only structural similarity but also similarity in the cellular responses they elicit, independent of downstream tasks. Using downstream derivative models such as cell-guided compound-protein interaction prediction and drug-induced transcriptional response profile generation, CellChem supports highly efficient, multi-scale drug discovery, achieving significantly better performance than traditional models.

molecular biology↗

bioRxiv2026

RVQ-Alpha: Bridging Single-Cell Transcriptomics and Large Language Models via Discrete Tokenization and Verifiable Reinforcement Learning

Single-cell RNA sequencing yields continuous expression profiles, whereas large language models operate over discrete autoregressive sequences, leaving no shared computational interface for language-model reasoning over cell states. Existing approaches either keep cellular information outside the LLM vocabulary, consume context per listed gene, or learn reconstruction codes without gene-level grounding. We introduce RVQ-Alpha, which systematically adapts four stages of LLM training (tokenization, supervised fine-tuning, reinforcement learning, and distillation) to single-cell analysis. Multi-codebook Residual Vector Quantization (RVQ) lexicalizes each profile into a compact, hierarchical cellular alphabet in the models native token stream, while a paired decoder reconstructs the corresponding expression profile. Evidence-First supervision grounds these symbols in named genes and expression-linked evidence; Fact-Aware RLVR then penalizes contradictory claims to support auditable reasoning. Task-specific RLVR yields strong experts, but a single Mixed policy underperforms them across all four task families. To recover specialist competence in a unified model, we adopt Multi-Teacher On-Policy Distillation (MOPD), which consolidates these experts into one All-in-One checkpoint without retaining a separate policy for each task. On the CAPSTONE benchmark, a four-task suite with explicit biological shifts and deterministic ontology-aware graders, the unified checkpoint improves over Mixed on all 12 metrics and remains within 0.005 of the task-routed specialist reference on every primary metric. Overall, RVQ-Alpha provides a unified pipeline for grounded, auditable multi-task single-cell analysis.

bioinformatics↗

bioRxiv2026

Boosting NADP-malic enzyme 1 enhances seed vigor and longevity in Arabidopsis thaliana

Seed longevity is a key determinant of crop establishment, productivity, and germplasm conservation. During storage and germination, reactive oxygen species accumulate and contribute to seed aging through oxidative damage and loss of viability. The maintenance of redox homeostasis therefore relies on NADPH-dependent antioxidant systems, which require a continuous supply of reducing power. NADP-dependent malic enzyme 1 (NADP-ME1), represents a source of NADPH supporting antioxidant defense during seed aging. Here, we show that enhanced expression of NADP-ME1 positively contributes to seed vigor and longevity in Arabidopsis thaliana. NADP-ME1 overexpression lines exhibited faster germination and higher overall germination after accelerated aging, whereas knockout mutants showed markedly reduced germination performance. Enhanced post-aging vigor in the overexpression lines was associated with reduced oxidative damage as indicated by lower malondialdehyde and hydrogen peroxide accumulation, along with preservation of specific polyunsaturated fatty acids, and increased {gamma}-tocopherol levels in aged dry seeds. Enhanced expression of NADP-ME1 reshapes the transcriptome of germinated seeds under fresh conditions compared with the wild type, while only minimal differences between genotypes are detected in aged seeds. These results suggest that NADP-ME1 contributes to the establishment of a transcriptional state associated with enhanced seed vigor and improved post-aging germination. Finally, co-immunoprecipitation coupled to mass spectrometry and bimolecular fluorescence complementation identified aspartate aminotransferase 2 as a NADP-ME1 interactor, pointing to a link between malate metabolism and amino acid-related metabolic adjustment. Together, these results identify NADP-ME1 as a determinant of seed resilience to aging and a potential target for improving seed quality.

plant biology↗

bioRxiv2026

Assessing the Generalizability of Machine Learning and Physics Methods for DNA-Encoded Libraries

Predicting protein-ligand binding is a central challenge in computational drug discovery, and while machine learning (ML) and co-folding methods have advanced rapidly, their ability to generalize beyond training or parameterization regimes remains insufficiently understood. DNA-encoded libraries (DELs) enable ultra-large screening of billions of molecules simultaneously, providing a useful testbed for evaluating these approaches at scale. A recent NeurIPS competition revealed that even top performing ML models trained on DEL data failed at generalizing to out-of-distribution (OOD) chemical space. We investigated whether integrating structural modeling could bridge this generalization gap. We systematically assessed state-of-the-art ML, docking, and co-folding methods including Schrodinger Glide, Rosetta GALigandDock, and Boltz-2 with three biologically diverse protein targets screened against libraries containing multiple DEL synthesis formats. While ML excels in-distribution, OOD hit discrimination is dependent on both the target and ligand context, with no single method consistently dominating. These findings demonstrate that benchmark performance alone is insufficient to predict OOD performance, highlighting the need for system-dependent evaluation of binding prediction methods. We provide an open-source package for assessing protein-ligand prediction methods and analyzing high-throughput screening data: DEL-iver. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=118 SRC="FIGDIR/small/719394v2_ufig1.gif" ALT="Figure 1"> View larger version (25K): org.highwire.dtl.DTLVardef@cd8a17org.highwire.dtl.DTLVardef@252d0aorg.highwire.dtl.DTLVardef@b020f0org.highwire.dtl.DTLVardef@1429e27_HPS_FORMAT_FIGEXP M_FIG C_FIG

biophysics↗

bioRxiv2026

A Haplotype-resolved Telomere-to-Telomere Pig Genome

Haplotype-resolved telomere-to-telomere (T2T) pig genomes have not been reported, but may be essential for xenotransplantation and interspecies chimerism. Here we describe fully phased paternal and maternal T2T assemblies of an Erhualian pig (Ssc_EHL), enabling chromosome-scale reconstruction at base resolution. We uncover 2,498 (paternal haplotype) and 2,577 (maternal haplotype) protein-coding loci previously unresolved in Sscrofa11.1, while maintaining high concordance for anchored gene models between haplotypes. Applying this haplotype-resolved T2T annotation to publicly available pig-to-human kidney xenograft transcriptomes further allows the resolution of xenograft cell states and reveals newly resolved, cell-type-enriched genes. Cross-species synteny and orthology analyses against human identified a subset of high-confidence pig-specific genes with limited human counterparts, thus, refining the genomic basis of interspecies divergence. By integrating these orthology features with organ transcriptomes, we further defined an Interspecies Chimerism Compatibility score to prioritize organ-enriched, highly conserved loci with minimal redundancy imbalance, providing haplotype consistent targets for engineering organ-deficient pigs. Together, this haplotype-resolved T2T pig genome provides a foundational resource for mechanistic studies and rational genome engineering in xenotransplantation and interspecies chimerism.

genomics↗

bioRxiv2026

Leishmaniamajor co-opts IL-7 feedback in monocytes to suppress CD4⁺ T-cell immunity

During immune responses, pro-and anti-inflammatory mechanisms must be balanced to ensure pathogen clearance while limiting tissue damage. Monocyte-derived cells contribute to both processes, yet the underlying regulatory circuits remain incompletely defined. Here, we show that a subset of PDPN+IL-7R+ monocyte-derived cells that impair effector CD4 T cell-mediated control of intracellular pathogens, and thus perpetuate the infection. Fibroblast-derived IL-7 drives this immunosuppressive program, which is up-regulated in response to IFN{gamma}. We thus uncover a cytokine-dependent feedback circuit in which elevated IFN{gamma} induces IL-7 production by fibroblasts, licensing immunosuppressive monocyte-derived cells that restrain CD4 T cell responses. This mechanism links excessive inflammation to immune suppression at the expense of pathogen control. Targeting this feedback loop may enable therapeutic strategies that enhance antimicrobial immunity while preserving tissue integrity.

immunology↗

bioRxiv2026

A rapid, sensitive, and quantitative high plex biomarker digital detection platform enabled by Hypercoding

Low-cost, multiplexed, and automated assays are needed to make omic technologies more broadly accessible in clinical, research, and commercial settings. We present Hypercoding, a scalable technology for detection and quantitation of multi-omic targets. Drawing from data reliability methods in the telecommunications field, Hypercoding uses fluorescent signals from hybridization with an error-correcting code to enable detection of high-plexity targets from biological samples, such as human DNA. In the presence of the target, a linear DNA construct is circularized, immobilized, and amplified to enable single-molecule detection of a target via rapid readout cycles within a 96-well plate. We demonstrate capability for >10,000 code plexity and accurate (98.7%) genotyping of 209 pharmacogenomic variants. Furthermore, we show computation of copy number variation with whole chromosome and sub-gene resolution, as well as quantitation of target abundance down to 10 fM sensitivity with a dynamic range of up to 10 logs.

genomics↗

bioRxiv2026

Estimating chronological and brain age using risk-taking behavior under uncertainty

Risky decision-making under uncertainty reflects complex cognitive processes supported by distributed brain networks that are vulnerable to aging. However, it remains unclear whether risk-taking behavior can serve as a behavioral marker of brain aging. In the present study, we combined behavioral tasks, computational modeling, and structural magnetic resonance imaging to investigate the relationship between risky decision-making, chronological age, and brain age. A total of 55 young adults and 112 healthy older adults completed the Iowa Gambling Task (IGT) and the Balloon Analogue Risk Task (BART), along with neuropsychological assessments and neuroimaging scanning. Decision processes were quantified using computational models, including the Value-Plus-Perseveration model and Exponential-Weight Mean-Variance. Brain age was estimated from gray matter volume. The results showed significant age-related alterations in parameters reflecting feedback sensitivity, learning rate, and loss aversion in both tasks. Within older adults, several decision parameters were significantly associated with both chronological age and brain age. Regression analyses further showed that computational parameters significantly predicted chronological age and brain age, whereas traditional cognitive screening measures did not show significant predictive effect. Structural brain analyses indicated that IGT-related parameters were primarily associated with the basal ganglia, while BART-related parameters were linked to a broader network including prefrontal, cingulate, and temporal regions. These findings suggest that computational markers of risk-taking behavior capture subtle age-related changes in cognitive processes and brain deterioration. Therefore, risk-taking parameters may serve as reliable functional markers of brain aging, providing critical insights into the mechanisms underlying successful aging.

neuroscience↗

bioRxiv2026

CHPT1-LCAT rewires lipolysis towards ferroptosis

Ferroptosis is driven by iron-dependent lipid peroxidation, yet how metabolic flux through central lipid pathways is selectively routed towards pro-ferroptotic lipid species remains unclear. Here, through an unbiased chemical-genetic screen targeting core metabolic enzymes, we identify diacylglycerol (DAG) as a licensing lipid intermediate whose pro-ferroptotic activity depends on its intracellular routing. Systematic manipulation of lipolytic flux reveals that ferroptotic vulnerability is not determined by bulk lipolytic output or downstream intermediates, but by the selective channelling of DAG into a distinct metabolic fate. Mechanistically, DAG is selectively routed through a previously unrecognized intracellular metabolic axis composed of choline phosphotransferase 1 (CHPT1) and lecithin-cholesterol acyltransferase (LCAT). We uncover an enzymatically active intracellular pool of LCAT (iLCAT) that cooperates with CHPT1 on Golgi-trans-Golgi network membranes to generate polyunsaturated cholesteryl esters that execute ferroptosis. Functionally, enforced DAG routing through this axis suppresses tumour growth via ferroptosis in vivo, whereas hepatocyte-specific inhibition of the CHPT1-iLCAT axis attenuates lipid peroxidation and disease progression in metabolic dysfunction-associated steatohepatitis. Together, these findings establish subcellular lipid routing, rather than lipid abundance per se, as a fundamental determinant of ferroptotic vulnerability.

cell biology↗

bioRxiv2026

Aphid stylin cuticular proteins contribute to turnip mosaic virus (Potyvirus) transmission

Hundreds of plant viruses are transmitted by aphid vectors, among which non-circulative ones are acquired and inoculated from one host to another within seconds. These viruses are retained on receptors located at the surface of the cuticle of aphid mouthparts. Members of the Potyviridae family are the most abundant RNA viruses infecting plants, and they cause significant economic losses. Among them, viruses of the Potyvirus genus are transmitted by aphids in a stylet-borne manner. Their receptors in aphid stylets remain poorly characterized. Using turnip mosaic virus (TuMV, Potyvirus rapae) as a model, we developed complementary approaches to investigate potyvirus-aphid interactions in three vector species. Immunofluorescence detection and transmission electron microscopy revealed the presence of TuMV in the distal part of aphid maxillary stylets, both in the food canal and on the acrostyle. This cuticular organ houses Stylin proteins, including Stylin-01, the receptor for the cauliflower mosaic virus (Caulimovirus tesselobrassicae). Using CRISPR-Cas9-edited Stylin-01 mutant lines in the pea aphid, we demonstrated that this protein plays an important role in TuMV transmission. Complementary RNA interference silencing experiments revealed that Stylin-04/04bis also mediate TuMV transmission. Furthermore, our findings reveal that targeting simultaneously Stylin-01 and Stylin-04/04bis more strongly impaired the aphids ability to transmit TuMV, suggesting that virus transmission relies on a multi-component stylin interface rather than a single receptor. In conclusion, these results highlight that in complex interactions between potyviruses and their aphid vectors, Stylin proteins are key actors, underscoring their importance in the transmission of stylet-borne viruses. Author summaryPotyviruses are the largest group of plant-infecting RNA viruses and major threats to crops worldwide. Aphids can transmit them within seconds retaining virus particles on their needle-like mouthparts through specific interactions at the stylet surface. However, the identity of the aphid proteins involved in this process has remained unknown. Using turnip mosaic virus (TuMV), a potyvirus with an exceptionnally broad host ranges, we provide evidence of virus retention at the acrostyle, a specialized cuticular microstructure at the tip of aphid stylets. CRISPR-Cas9 stable aphid mutant lines and RNAi-mediated gene silencing revealed that the acrostyle cuticular proteins Stylin-01 and Stylin04/04bis promote TuMV transmission. These findings uncover previously unknown molecular components of the aphid-potyvirus interface and identify potential targets to disrupt virus-vector interaction and virus transmission.

microbiology↗

bioRxiv2026

NADP-malic enzyme 1 couples ABA signaling to ROS-auxin patterning to restrict Arabidopsis root growth

Abscisic acid (ABA) restricts primary root growth by reshaping reactive oxygen species (ROS) dynamics and hormone signaling at the root apex, yet how cellular reductant supply for redox homeostasis is integrated into this response remains unclear. Here, we show that the cytosolic NADP-dependent malic enzyme 1 (NADP-ME1) is required for full ABA inhibition of Arabidopsis primary root elongation after germination. Three independent me1 loss-of-function mutants retained significant elongation of primary roots under ABA compared with wild type. In wild type, ABA induced an asymmetric auxin response at the root tip, whereas me1 roots failed to establish this auxin asymmetry and instead accumulated superoxide, indicating disrupted ROS balance. Pharmacological perturbation of auxin transport and ethylene biosynthesis/signaling attenuated the mutant phenotype, linking NADP-ME1 function to auxin-ethylene interactions during ABA-regulated growth. NADP-ME1 loss amplifies ABA-dependent transcriptional rewiring, including induction of oxidative stress and depression of growth-associated hormone modules. Co-immunoprecipitation coupled with bimolecular fluorescence complementation identified ascorbate peroxidase 1 (APX1) and major latex protein-like 34 (MLP34) as NADP-ME1 interaction partners, suggesting functional coupling between NADPH production and ROS detoxification. Together, our results support a model in which NADP-ME1 shapes the superoxide/H2O2 balance to control auxin patterning at the root tip and thereby execute ABA-mediated growth inhibition.

plant biology↗

bioRxiv2026

PhosSight: a Unified Deep Learning Framework Boosting and Accelerating Phosphoproteome Identification to Enable Biological Discoveries

Protein phosphorylation is a key regulator of signaling, with mass spectrometry (MS) based phosphoproteomics serving as the premier technology for its analysis. However, phosphorylation profiling is hindered by acquisition biases: Data-Dependent Acquisition (DDA) suffers from stochastic undersampling and missing values, while Data-Independent Acquisition (DIA) faces computational bottlenecks and inefficiencies from vast spectral libraries. We present PhosSight, a unified deep learning framework designed to augment identification depth and accelerate search efficiency. PhosSight features PhosDetect, a model that explicitly encodes phosphorylation specific physicochemical features to accurately predict peptide detectability. For DDA, PhosSight leverages predicted retention time, fragment intensity, and detectability to refine site localization and rescoring, recovering marginal, low-abundance spectra. For DIA, PhosSight utilizes detectability-guided library pruning to remove non-detectable noise, accelerating search speeds without compromising sensitivity. Benchmarking on synthetic and real world datasets confirms PhosSights superior performance in both modes. Applying PhosSight to a large-scale Uterine Corpus Endometrial Carcinoma (UCEC) cohort significantly reduced missing values and expanded the quantifiable phosphoproteome. This enhanced completeness enabled the discovery of novel prognosis associated kinase targets, such as MARK2, underscoring PhosSight as a powerful tool for biological discovery in precision oncology.

bioinformatics↗

bioRxiv2026

Optical Windows for Transcranial Brain Imaging in Living Mice: Skull Thinning, Clearing, and Beyond

Longitudinal, noninvasive in vivo imaging is essential for studying brain physiology and pathological mechanisms. Advances in transcranial optical windows, including thinned-skull and optical clearing techniques, have markedly improved imaging depth and resolution when combined with multiphoton microscopy. However, their optical performance often deteriorates rapidly, and quantitative studies on long-term stability and the causes of image quality loss remain limited. In this work, we systematically investigated current transcranial window approaches using multiphoton excited fluorescence microscopy (MPEFM) and adaptive optics, examining longevity, optical aberrations, and imaging resolution. Our results reveal that progressive skull regrowth is a fundamental limitation across all window types, leading to substantial declines in signal quality and resolution for conventional MPEFM thereby reducing achievable high-resolution imaging depth. To address these challenges, we developed a localized glucocorticoid (GC) delivery strategy that significantly extends window performance for up to one month. Furthermore, we demonstrated that a GC-loaded hydrogel sealing method effectively suppressed skull regrowth while preserving optimal optical properties, offering a potential and practical route to chronic, high-fidelity transcranial imaging. These findings provide mechanistic insight into window degradation and establish a framework for sustained, long-term in vivo brain imaging.

neuroscience↗

bioRxiv2026

Transcranial direct current stimulation combined cognitive training modulates risk-taking behavior in older adults

Risk-taking behavior is a complex cognitive function that often declines with aging, contributing to impaired decision-making and reduced quality of life. While transcranial direct current stimulation (tDCS) has shown promising effects in modulating cognitive function, its influence on ecologically valid, complex behaviors like risky decision-making in older adults remains poorly understood. We investigated whether a single-session stimulation over the medial orbitofrontal cortex (MOFC) combined with cognitive training could enhance decision-making in healthy older adults. In a randomized, sham-controlled study, bilateral MOFC stimulation (left/anode, right/cathode) was applied during a training task based on the Iowa Gambling Task. Pre- and post-intervention assessments utilized conventional behavioral measures and the Values-Plus-Perseveration computational model, alongside task-related fMRI to examine MOFC network changes. Compared to sham, tDCS significantly enhanced the ability to distinguish advantageous from disadvantageous options. Modeling analysis revealed stimulation-induced changes in multiple latent components, such as learning rate, loss aversion, and perseveration decay. Generalized Psychophysiological Interaction analysis showed that tDCS reconfigured the MOFC network by reducing fronto-frontal hyper-connectivity and enhancing fronto-striatal connectivity. These behavioral improvements were specifically associated with the left MOFC network targeted by anodal stimulation. Our results provide causal evidence that tDCS can mitigate age-related impairments in risk-taking by reconfiguring MOFC-related networks. These findings advance the mechanistic understanding of tDCS in aging and highlight its potential as an intervention to prevent age-related decline in decision making.

neuroscience↗

bioRxiv2026

ALKBH5 Promotes Hepatic Gluconeogenesis via m6A-mediated Stabilization of Ogt mRNA

Excessive hepatic glucose production is a hallmark of fasting hyperglycemia in type 2 diabetes mellitus, yet the epitranscriptomic mechanisms that sustain this dysregulation remain incompletely understood. Here, we show that dexamethasone/forskolin-induced gluconeogenic activation triggers widespread remodeling of RNA m6A methylation in mouse primary hepatocytes. Among the m6A regulators examined, the demethylase ALKBH5 is robustly induced both in vitro and in fasted livers. This induction is mediated by the glucocorticoid receptor, which binds to the Alkbh5 promoter and activates its transcription. Functionally, ALKBH5 overexpression enhances gluconeogenic gene expression and glucose production, whereas both global and hepatocyte-specific Alkbh5 knockout mice exhibit suppressed hepatic gluconeogenesis. Mechanistically, we identify Ogt as a primary downstream target, defining an ALKBH5-OGT epitranscriptomic axis in hepatocytes. ALKBH5 demethylates and stabilizes Ogt mRNA, thereby elevating OGT expression and promoting gluconeogenic gene expression. Consistently, the pro-gluconeogenic effects of ALKBH5 are largely abrogated by OGT knockdown. Finally, pharmacological inhibition of ALKBH5 with 18l effectively improves glucose tolerance and suppresses hepatic gluconeogenesis in both wild-type and db/db mice. Together, these findings reveal a glucocorticoid-responsive epitranscriptomic mechanism whereby ALKBH5 stabilizes Ogt mRNA to drive gluconeogenesis and highlight ALKBH5 as a promising therapeutic target for type 2 diabetes mellitus.

molecular biology↗