Biology subjects

Lee, K.

Publications and source records attributed to Lee, K..

At least 37 records · Page 2Linked to original sources

bioRxiv2026

Lightning Pose 3D: an uncertainty-aware framework for data-efficient multi-view animal pose estimation

Multi-view pose estimation is essential for quantifying animal behavior in scientific research, yet current methods struggle to achieve accurate tracking with limited labeled data and suffer from poor uncertainty estimates. We address these challenges with a flexible framework that can operate with or without camera calibration, combining novel training and post-processing techniques with an uncertainty-aware pseudo-labeling distillation procedure. Our multi-view model processes all camera views jointly using a pretrained vision transformer backbone, and a simulated occlusion technique encourages the model to learn robust cross-view correspondences without requiring camera parameters. When camera parameters are available, 3D data augmentations and a triangulation-based loss further encourage geometric consistency. We extend the Ensemble Kalman Smoother (EKS) post-processor to the nonlinear case, leveraging camera geometry, and introduce a variance inflation technique that detects cross-view inconsistencies and corrects overconfident predictions. We validate our approach on five datasets spanning three species (fly, mouse, bird), including a multi-animal dataset with two visually distinct individuals; the proposed pipeline consistently outperforms existing methods across datasets. We demonstrate how these improvements translate to downstream scientific analyses using data from the International Brain Laboratory, showing improved unsupervised behavioral clustering and neural decoding of paw kinematics with just 200 labeled frames. To facilitate adoption, we developed a browser-based, cloud-compatible user interface that supports the full life cycle of multi-view pose estimation, from labeling and model training to post-processing with EKS and diagnostic visualizations.

neuroscience↗

bioRxiv2026

Differential acquisition of cocaine and heroin self-administration in a rat model of internalizing versus externalizing temperament

Substance Use Disorders (SUDs) constitute a major and rising public health concern. In addition, there is a growing appreciation that different classes of addictive substances are likely to lead to qualitatively different types of SUDs requiring differing treatment and relapse prevention strategies to be most effectively managed. Biological temperament, particularly on the internalizing - externalizing axis, is well established to influence addiction susceptibility. Externalizing behavior has long been understood to predispose individuals to addiction through novelty-seeking, sensation-seeking and impulsivity, while internalizing behavior provides an alternate pathway into addiction via increased occurrence of comorbid disorders (anxiety, depression). Here, we utilize a selectively bred rat model of internalizing vs externalizing temperament (bred High Responders, representing genetically mediated externalizing behavior and bred Low Responders, representing internalizing behavior) to examine differences in the acquisition of self-administration of the prototypical psychostimulant cocaine and the prototypical opioid heroin (diacetylmorphine). We found that, as predicted, cocaine and heroin drove different patterns of acquisition in the two different bred lines of rats. Further, this was influenced by temperament in complex ways. Notably, in females the "telescoping effect" for opioid addiction-like behavior was primarily specific to externalizing temperament. These findings highlight the impact and interaction of many factors, including drug class, temperament, and sex, on the acquisition of drug-taking behavior. Additionally, these findings indicate that sex differences in addiction vulnerability may be influenced in part by biological temperament.

animal behavior and cognition↗

bioRxiv2026

Structural Basis of Electron Transfer by the Human Nitric Oxide Synthase Holoenzyme Complex

Mammalian nitric oxide synthase (NOS) generates nitric oxide (NO), an essential signaling molecule in neurotransmission, inflammation, and cardiovascular regulation. NOS dysregulation contributes to neurodegeneration, septic shock, and ischemia. The structural basis of NOS function remains unclear, likely due to its dynamic, multi-component architecture. Catalysis requires electron transfer across five cofactors, with the oxygenase (Oxy) domain converting L-arginine to NO after receiving electrons from the reductase (Red) domain via calmodulin (CaM) activation. Here, we report cryo-electron microscopy structures of the human inducible NOS:CaM holoenzyme, revealing a previously unresolved architecture in which the Red:CaM arm spans the Oxy dimer in an apparent active state for cross-monomer electron transfer. The FMN subdomain is rotated [~]90{degrees} relative to the Red core, positioning its cofactor [~]9 [A] from the heme, adjacent to a conserved aromatic residue within a narrow Oxy domain tunnel. Red-Oxy flexibility in the active state may facilitate electron shuttling among cofactors, defining how NOS cofactors are transiently aligned to enable catalysis.

biochemistry↗

bioRxiv2026

Deep-learning-enabled morphodynamic analysis of drug responses in a biomimetic fibrin-based 3D glioblastoma invasion model

Glioblastoma (GBM) lethality arises from aggressive invasion and diffuse infiltration of brain tissue. Conventional GBM preclinical models often fail to predict clinical therapeutic efficacy because they do not recapitulate the pathological extracellular matrix (ECM) cues that drive tumor invasion. Here, we present an ECM mimetic 3D platform using a fibrin scaffold to recapitulate the hemorrhagic, pro-thrombotic tumor microenvironment characteristic of high-grade gliomas. This fibrin scaffold induces a pro-invasive phenotype in GBM spheroids by upregulating proliferation/cell cycle- (MYC, FOXOM1, CCND1) and invasion-associated-(CTSS, FOXM1, CCND1) genes. Traditional cell morphology quantification methods (e.g., circularity) distil complex shapes into singular metrics and cannot capture the nuances of invasion. To address this limitation, we have applied a deep-learning segmentation pipeline (MARS-Net) and high-content morphodynamic descriptors. By using the Preserving Heterogeneity (PHet) algorithm, the 3D platform accurately classifies invasiveness levels and captures the invasion-inhibitory effects of potential repurposable drug candidates. We demonstrate that our model can predict a spheroids long-term invasive fate with high accuracy using only partial image sets from early time-points, rather than the complete time-course images. Our work presents an in vivo-like, scalable 3D platform integrated with a quantitative high-throughput pipeline to elucidate GBM invasion mechanisms and to evaluate anti-invasive compounds.

bioengineering↗

bioRxiv2026

DNA Damage Driven Viability Loss and Transcriptional Reprogramming in Chinese Hamster Ovary Cell Perfusion Culture

Intensified perfusion cultures promise higher yields and consistent product quality, yet extended runs frequently stall due to declining Chinese hamster ovary (CHO) cell viability at high cell density. Here, we identify DNA damage accumulation as a central, previously underappreciated driver of this limitation. Using two antibody-expressing CHO cell lines operated at a control and high cell density in perfusion bioreactors, we combined time-resolved transcriptomics with molecular, biophysical and super-resolution imaging analyses. We observed a progressive, global downregulation of DNA damage response (DDR) pathways accompanied by a time-dependent accumulation of DNA lesions. Notably, {gamma}H2AX signalling declined despite DNA damage, indicating impaired damage sensing and repair. Concomitantly, RNA polymerase II protein levels and transcriptional hub organisation were markedly reduced, consistent with widespread transcriptional dysfunction preceding loss of viability. An extended 21-day perfusion run confirmed continued viability decline beyond day 14, supporting a cumulative damage model. Comparison with HEK293 cells revealed substantially more efficient DNA repair, underscoring an intrinsically attenuated DDR in CHO cells. Together, these findings demonstrate that unrepaired DNA damage fundamentally constrains perfusion culture longevity, exposing limits of CHO genomic plasticity and highlighting DDR pathways as promising targets for host cell engineering to enhance perfusion performance.

bioengineering↗

bioRxiv2026

MORPHE: Bridging Image Generation and Spatial Omics for Tissue Synthesis

Spatially resolved omics technologies reveal tissue organization at single-cell resolution but remain limited by the cost of the assays, incomplete spatial coverage, 2D-only imaging, and experimental artifacts. These factors motivate the need for in silico methods that can reconstruct or extend tissue context beyond what current spatial measurements provide. We present MORPHE (MOdeling of stRuctured sPatial High-dimensional Embeddings), an AI framework that learns to synthesize biologically faithful tissue architecture directly from spatial-omics data. MORPHE introduces a graph-informed probabilistic embedding that maps discrete cell identities and their spatial relationships into a continuous RGB-like latent space compatible with diffusion modeling. This representational bridge enables spatial cellular maps to leverage large pre-trained image-generative models while preserving biological interpretability upon decoding. By modeling cells as the fundamental units of generation and learning how their identities and spatial relationships collectively give rise to large-scale tissue structure, MORPHE enables generation and reconstruction of tissue architecture at single-cell resolution. We applied the method across large-scale single-cell proteomic datasets from the intestine and single-cell transcriptomic datasets from the brain, showing computational scalability acrosss millions of cells. We used MORPHE on these datasets to outpaint beyond experimentally restricted fields of view, inpaint missing or experimentally damaged tissue regions, and perform cross-tissue imputation, connecting separated tissue regions into a single contiguous sample in both 2D and 3D. MORPHE represents a new class of tissue generation algorithms that will help solve current limitations and challenges with single-cell spatial-omics datasets.

systems biology↗

bioRxiv2026

Why Boolean network control tools disagree: a taxonomy of control problems

Boolean networks are discrete dynamical models that use Boolean states and logical functions to represent the dynamics of biological systems. A primary application of Boolean networks is to identify controls (e.g., genetic mutations or knockouts) that drive the system toward a desired phenotype. However, existing computational tools often produce inconsistent results because they rely on differing modeling assumptions. To better understand these differences, we survey existing tools and propose a taxonomy of control problems. Our taxonomy unveils hidden coverage relationships among their solutions that arise from these modeling assumptions. We provide a computational framework to empirically assess these relationships by comparing their predicted controls on a suite of artificial and biological models. Finally, we develop a coverage-consistent metric, the mutation co-occurrence score, to prioritize mutations based on their predicted impact on the phenotype. A case study on T-LGL leukemia highlights how an ensemble prediction of the score across multiple tools identifies key mutations associated with apoptosis. Author summaryBoolean networks let us model gene and protein regulation with simple on/off logic. That simplicity makes them useful for asking a practical question: which controls (e.g., genetic interventions, knockouts, or forced activations) can guarantee a desired cellular phenotype such as survival or apoptosis. In practice, different software tools often return inconsistent control sets, creating a practical barrier to reproducibility and reliability of predictions. Here we provide a comprehensive survey to navigate this landscape. We provide a comparison framework to assess coverage relationships, both theoretically and empirically, among the solutions of control tools. We classify the tools based on our new taxonomy that highlights the core differences among them. For many tools, we can theoretically determine their coverage relationships based on which components are fixed together to drive the phenotype. These new findings clearly explain subtle differences among solutions produced by various tools. We then develop a coverage-consistent metric for mutations called the mutation co-occurrence score. This new metric helps prioritize control targets based on their predicted impact on the phenotype. We also demonstrate that averaging scores from multiple tools gives reliable predictions. Our code is extensible to future tools and will facilitate the comparison of Boolean network control tools in biological research.

systems biology↗

bioRxiv2026

Learning functional groups in complex microbiomes

From soil to the gut, communities composed of thousands of microbes perform functions such as carbon sequestration and immune system regulation. Here, we introduce a data-driven approach that explains how community function can be traced to just a few groups of microbes or genes. In gut communities, our neural-network based clustering algorithm correctly recovers known functional groups. In the ocean metagenome, it distills ~500 gene modules down to three sparse groups highlighting survival strategies at different depths. In soils, it distills ~ 4400 bacterial species into two groups that enter a mathematical model of nitrate metabolism. By combining interpretable ML with strain isolation and sequencing experiments, we connect the metabolic specialization of each group to community-wide responses to perturbations. This integrated approach yields simple structure-function maps of microbiomes, allowing the discovery of molecular mechanisms underlying human and environmental health. More broadly, we illustrate how to do function-informed dimensionality reduction in biology. O_FIG O_LINKSMALLFIG WIDTH=158 HEIGHT=200 SRC="FIGDIR/small/709366v1_ufig1.gif" ALT="Figure 1"> View larger version (41K): org.highwire.dtl.DTLVardef@1f76c74org.highwire.dtl.DTLVardef@f13bdcorg.highwire.dtl.DTLVardef@19107c6org.highwire.dtl.DTLVardef@9558f7_HPS_FORMAT_FIGEXP M_FIG O_FLOATNOGraphical AbstractC_FLOATNO An integrated ML and experimental pipeline to discover functional groups and their dynamics in complex microbiomes and beyond. (a) First, our Soft Clustering Function Informed (SCiFI) algorithm identifies functional groups directly from species abundances data using neural networks. Crucially, the learned functional groups are informed by a chosen community function. (b) Descriptions of the system in terms of the original high-dimensional abundances of individual species are prohibitively complex (left). By contrast, a description in terms of the few functional groups identified by SCiFI leads to a simple structure-function map (right). (c) Next, the identified groups can be directly input as variables into predictive mathematical models for the dynamics of the community. (d) The last step of our pipeline relies on the identified groups comprising only a small number of species. This sparsity enables targeted experiments that interrogate isolated species (e.g. with whole-genome sequencing or phenotyping) shedding light on the mechanistic underpinnings of the structure-function map with potential applications beyond microbiomes, from gene expression to neuronal activity. C_FIG

microbiology↗

bioRxiv2026

Cell body clustering drives gap junction-mediated synchronous activity in command neurons

The nervous system contains densely packed cell bodies, yet the role of neuronal cell body position in circuit function is poorly understood. Here we show that four Drosophila Moonwalker Descending Neurons (MDNs), command neurons for backward locomotion, must maintain cell body contact to allow gap junction-dependent synchronous activity necessary to initiate backward walking. MDNs express the transcription factor Hunchback, which drives expression of the Lar cell adhesion molecule; Hunchback, Lar, and its ligand Dlp promote MDN cell body clustering and backward walking. When clustered, the gap junction protein Inx8 allows synchronous firing of MDNs, which is required to initiate backward walking. These findings reveal a previously unappreciated role for cell body clustering and synchronous firing in neural circuit function.

neuroscience↗

bioRxiv2026

MG53 Protects Against Intestinal Inflammation by Inhibiting NLRP3 Inflammasome Activation

Background and ObjectivesInflammatory bowel disease (IBD) involves dysregulated immune responses and chronic intestinal inflammation. The nod-like receptor pyrin domain-containing 3 (NLRP3) inflammasome plays a critical role in IBD pathogenesis, but regulatory mechanisms remain not fully understood. Mitsugumin 53 (MG53, also known as TRIM72), originally identified as a critical membrane repair protein, has emerged as a novel regulator of inflammatory processes. To investigate the protective role of MG53 in colitis and elucidate its mechanisms in regulating NLRP3 inflammasome activation in IBD. MethodsWe used dextran sodium sulfate (DSS)-induced colitis models comparing MG53 knockout (MG53-/-) and wild-type (WT) mice, assessing disease severity, MG53 tissue uptake, and therapeutic effects of recombinant human MG53 (rhMG53). In vitro studies examined rhMG53s effects on NLRP3 inflammasome activation, caspase-1 cleavage, interleukin-1{beta} (IL-1{beta}) secretion, and MG53-NLRP3 interactions. ResultsMG53-/- mice showed more severe colitis with increased weight loss, higher disease activity scores, shortened colons, and greater inflammation. DSS treatment induced the accumulation of circulating MG53 in inflamed colonic tissue. rhMG53 administration ameliorated colitis severity in MG53-/- mice and dose-dependently suppressed NLRP3 inflammasome activation in vitro. MG53 interacted with NLRP3 and reduced apoptosis-associated speck-like protein containing a CARD (ASC) speck formation and NLRP3 oligomerization without affecting upstream signaling or NLRP3 stability. ConclusionMG53 is a physiological regulator of NLRP3 inflammasome activation that protects against colitis, suggesting therapeutic potential for IBD.

immunology↗

bioRxiv2026

Fast structural search for classification of gut bacterial mucin O-glycan degrading enzymes

The Enzyme Commission (EC) numbering scheme provides a hierarchical way to classify enzymes according to their catalytic functions. While recent protein language model (PLM) based approaches like CLEAN and ProteInter have improved sequence-based EC number prediction, they struggle with fine-grained classification at the deepest hierarchical level. Structure-based approaches for grouping similar proteins using alignment tools excel at finding proteins that share overall global structure, but suffer from high false positive rates when classifying proteins that are globally structurally similar but functional differentiation depends on a localized region. This problem is particularly relevant to EC number prediction, as enzymatic function depends on its catalytic domain, which is a relatively small, specific region of the protein. We introduce Deep Enzyme Function Transfer (DEFT) that harmonizes sequence- and structure-based approaches through the key insight that PLM based annotations of the first two EC number hierarchy levels vastly reduce false positives that are likely to show in purely structure-based EC number prediction. Given an enzyme of interest, DEFT first uses a PLM based method to assign the first two levels of the enzymes EC number, and then uses a structure-based method to predict the remaining two levels of the EC number. Using benchmarking datasets, we demonstrate that DEFT achieves superior accuracy compared with current state-of-the-art tools for EC number prediction. Furthermore we show that DEFTs computational efficiency enables high-throughput, genome-wide annotations of total enzyme repertoires in organisms. We illustrate this capability by experimentally validating DEFT predicted glycoside hydrolase (GH) profiles of intestinal mucus associated bacteria. Author summaryEnzymes are ubiquitous proteins that catalyze chemical reactions of living cells. Enzymes are classified using a hierarchical numbering system called Enzyme Commission (EC) numbers that describe the chemical reactions the enzymes catalyze, from a general reaction type (e.g., breaking bonds, transferring chemical groups, etc.) to more specific aspects such as chemical bonds and substrates involved in the reaction. We present a new machine learning method for predicting EC numbers called Deep Enzyme Function Transfer (DEFT). This method improves on previous methods that use either protein sequence- or three-dimensional (3D) structure-based comparisons between enzymes of known and unknown classification. DEFT combines the strengths of both approaches by first using a protein sequence-based model to predict the general enzyme category and then using protein structure comparisons to predict the finer subcategories. We demonstrate that DEFT achieves superior accuracy compared with current state-of-the-art tools for EC number prediction. We next demonstrate how DEFTs computational efficiency enables us to perform high-throughput, genome-wide annotations of organisms enzyme repertoires. We illustrate this capability by experimentally validating DEFT predicted sugar metabolizing enzyme profiles of intestinal mucus associated bacteria.

bioinformatics↗

bioRxiv2026

Appearance of Amyloid-β Early in Life Initiates Neuronal Hyper-excitability, Mitochondrial Decay, and Loss of Dendritic Complexity in the Hippocampal CA1 Region of 5xFAD Mice

Alzheimers disease (AD) is characterized by progressive cognitive decline and stereotyped neuropathology, yet the earliest cellular events that precede overt plaque burden and measurable behavioral impairment remain incompletely defined. Here, we tested the hypothesis that synaptic hyperexcitability and subcellular metabolic dysfunction emerge early in the 5xFAD mouse model and contribute to region-specific neuronal vulnerability before substantial amyloid plaque deposition. Using the 5xFAD heterozygous mouse, we first established the onset of transgene expression and the timing of plaque accumulation. Robust transgene expression was detected by postnatal day 15 and significant plaque accumulation by 4 months of age. Ex vivo electrophysiology revealed an early hyperexcitable phenotype at 1 month of age, including both increased AMPA receptor-mediated transmission and N-methyl-D-aspartate receptor signaling associated with the GluN2B subunit. Given the tight coupling between glutamatergic hyperactivity, calcium dysregulation, and mitochondrial health, we assessed mitochondrial structure and function at this pre-plaque stage. Mitochondrial abnormalities consistent with impaired bioenergetic homeostasis were evident. Morphological analyses further demonstrated that these early changes were associated with altered dendritic architecture in the CA1 and dentate gyrus regions, revealing hippocampal subregional susceptibility. Finally, spatial transcriptomics supported this anatomical selectivity by identifying regionally enriched molecular signatures consistent with differential vulnerability. The CA1 region exhibited more reductions in mitochondria-related transcripts than CA3 or dentate gyrus and these reductions were specifically associated with CA1 pyramidal cell neurons. Together, these findings define a pre-plaque window in 5xFAD mice marked by GluN2B-linked glutamatergic hyperexcitability, early mitochondrial disruption, and selective dendritic and transcriptional vulnerability across hippocampal subregions. This integrated timeline suggests that synaptic and metabolic dysfunctions arise before substantial plaque deposition and may represent tractable early targets for intervention aimed at delaying or preventing downstream neurodegeneration in AD.

neuroscience↗

bioRxiv2026

Label-free quantitative phenotyping of hepatic stellate cell activation using holotomography with AI-enabled subcellular segmentation

Hepatic stellate cell (HSC) activation is a central driver of liver fibrosis, yet its quantitative characterization in living cells remains limited by endpoint assays that rely on fixation and labeling. Here we introduce a label-free framework that combines three-dimensional holotomography (HT) with automated, AI-assisted analysis to study HSC activation dynamics in live cells. Using refractive index tomography, we non-invasively visualize hallmark structural features of HSC activation, including lipid droplet depletion, cytoskeletal remodeling, and changes in cellular morphology. Correlative fluorescence imaging validates the biological relevance of HT-derived features, while longitudinal imaging reveals continuous activation trajectories at single-cell resolution. Automated segmentation of whole cells and subcellular organelles enables scalable extraction of multi-parametric biophysical descriptors, defining a quantitative phenotypic fingerprint that distinguishes quiescent and activated states. Together, this work establishes holotomography-based quantitative phenotyping as a powerful approach for studying HSC activation and, more broadly, dynamic cell-state transitions in living systems.

cell biology↗

bioRxiv2026

In Utero CFTR Modulation Alleviates Disease in G551D Cystic Fibrosis Pigs

Previous studies indicate that pigs with CFTR-null and CFTR-{Delta}F508 mutations develop multiorgan disease similar to that in people with cystic fibrosis (CF). At birth, their airways exhibit host defense defects that predispose to airway infection, inflammation, and mucus accumulation. The CFTR-G551D mutation causes CF by producing CFTR channels that localize correctly but have reduced channel activity. Ivacaftor (VX-770) is a small molecule drug developed to potentiate CFTR activity. To test the phenotype of the CFTR-G551D mutation in pigs and determine whether ivacaftor can rescue CF abnormalities, we developed CFTRG551D/G551D (CF-G551D) pigs through homologous recombination in fetal fibroblasts and somatic cell nuclear transfer. Newborn CF-G551D piglets exhibited phenotypes similar to CF-null piglets, including meconium ileus, exocrine pancreatic destruction, micro-gallbladder, vas deferens destruction, and airway structural abnormalities. Compared to wild-type pigs, CF-G551D pigs had reduced forskolin-stimulated short-circuit current in airway and intestinal tissues. Ivacaftor increased the single-channel open state probability of CFTR-G551D and increased short-circuit current to near wild-type levels. Similar to our other CF pig models, we found that 100% of CF-G551D pigs were born with meconium ileus. To test whether in utero ivacaftor treatment could prevent or alleviate meconium ileus, pregnant sows were treated with ivacaftor beginning at day 35 of gestation and continuing until delivery. This treatment rescued the pancreas, gallbladder, and vas deferens phenotype in the majority of CF-G551D pigs. Animals that were spared from meconium ileus were able to survive without ivacaftor treatment. Airway disease developed similar to other CF pig models. These findings indicate that this model may be useful for studies in which CFTR function can be reversed, for investigating in utero CFTR correction strategies, and for longitudinal studies in CF pigs.

physiology↗

bioRxiv2026

Ionic Regulation of Mechanosurveillance and Metastasis via the MRTFA/KCNMB1 Axis

Cellular stiffness profoundly impacts cancer metastasis at multiple levels, but mechanisms that regulate cancer cells stiffness remain poorly understood. Here, we identified potassium efflux and KCNMB1, an auxiliary subunit of the large conductance potassium efflux (BK) channels, as regulators of cellular stiffness downstream of myocardin related transcription factor A (MRTFA). In primary pericytes, KCNMB1 knockdown increased cellular stiffness, which is consistent with the role of potassium efflux in promoting relaxation during excitation-contraction coupling. In a striking contrast, however, KCNMB1 knockdown decreased cellular stiffness in cancer cells. Softer cancer cells were resistant to NK cell mediated cytotoxicity and the low KCNMB1 expression was associated with worse survival in breast cancer patients. Importantly, pharmacological activation of BK channels reduced metastatic burden in mice and improved lysis of cancer cells by cytotoxic T-lymphocytes. These results highlight the unique ionic regulation of stiffness in cancer cells and point to BK channel agonism as a new therapeutic approach in cancer.

cancer biology↗

bioRxiv2026

Dual Oxygen-Partitioned Co-Culture Uncovers Microbe-Specific Epithelial Stress and Homeostatic Programs

Direct epithelial-microbial interactions occur across a steep aerobic-anaerobic interface in the intestine, yet mechanistic analysis has been limited by the difficulty of sustaining oxygen-dependent epithelial cells together with strictly anaerobic microbes in vitro. Here, we establish a dual oxygen-partitioned co-culture system that reproducibly maintains aerobic intestinal epithelial cells and obligate anaerobic gut bacteria, enabling controlled analysis of epithelial responses under physiologically relevant oxygen architecture. Using Mediterraneibacter gnavus and Lacticaseibacillus casei as contrasting microbial partners, we identify distinct epithelial programs. M. gnavus induces epithelial stress characterized by tight-junction disruption, suppression of mitochondrial respiration, and activation of chromatin-associated and innate immune regulatory pathways, whereas L. casei preserves barrier integrity and supports mitochondrial capacity with restrained immune modulation. Integrated transcriptomic and proteomic analyses reveal a shared epithelial energy-conserving response to microbial proximity, upon which microbe-specific stress or homeostatic programs are imposed. In vivo analysis using a dextran sodium sulfate-induced colitis model demonstrates that these epithelial programs become physiologically relevant under inflammatory conditions, with L. casei promoting epithelial repair and immune rebalancing. Together, these findings define oxygen-partitioned epithelial-microbial interactions as a determinant of microbe-specific epithelial states.

microbiology↗

bioRxiv2026

A novel imprinting cluster at the porcine CRSP complex locus defines a species-specific imprinted domain

BackgroundGenomic imprinting is an epigenetic phenomenon that results in parent-of-origin-specific gene expression and has been extensively characterized in mice and humans. However, in pigs, imprinting has been investigated primarily through analyses of orthologs of known imprinted genes in mice and humans. The objective of this study was to examine DNA methylation status and gene expression at a porcine locus containing newly identified imprinted calcitonin receptor-stimulating peptide (CRSP)-encoding genes, to compare orthologous loci in mice and humans, and to investigate a potential underlying mechanism. ResultsAnalyses of differentially methylated regions (DMRs) between porcine parthenogenetic embryos and biparental controls revealed multiple parental DMRs at a locus we term the CRSP complex locus, which harbors CRSP-encoding genes that likely arose through gene duplication. In contrast, orthologous genomic intervals in mice and humans exhibited unmethylated promoters and lacked evidence of imprinting. Consistently, CRSP-encoding genes in pigs showed parent-of-origin-specific monoallelic expression, whereas genes within the orthologous locus in mice and humans were biallelically expressed. Further analysis indicated that porcine CRSP promoters are embedded within oocyte-expressed alternative transcripts and co-occurred with DNA methylation, suggesting a transcription-dependent imprinting mechanism. ConclusionsOur comparative analyses identified CRSP-encoding genes at the porcine CRSP complex locus as novel imprinted genes, indicating species-specific evolution of this imprinted domain. The results further suggest that lineage-specific gene duplication may have contributed to the emergence of imprinting at this locus.

genomics↗

bioRxiv2025

Genome-Level Hierarchical Attention Transformer with Multi-Head Attention Weighted Sum for Broad-Spectrum Antimicrobial Resistance Prediction and Discovery of Resistance-Related Genomic Contexts

Antimicrobial resistance is a growing global health concern, requiring reliable tools for predicting resistance across a wide range of bacteria and antibiotics. In this study, we introduce a genome-level hierarchical attention transformer (GL-HAT) that integrates a pretrained genomic foundation model with hierarchical attention mechanisms to analyze the full protein sequence context of bacterial genomes. GL-HAT is designed to deliver both accurate resistance predictions and interpretable insights into the genomic features associated with resistance using specialized modules. Trained and validated on a comprehensive dataset of bacterial genomes and antimicrobial susceptibility profiles, the model outperformed traditional machine learning method, with an F1-score of 0.845 and AUROC of 0.953. Moreover, GL-HAT could demonstrate the resistance-related contexts in the genome, such as vanA operon and nearby resolvase genes in vancomycin-resistant Enterococcus and tet(M) gene adjacent to conjugation proteins in tetracycline-resistant Streptococcus pneumoniae. By combining high accuracy with interpretability, GL-HAT offers a practical approach for advancing AMR diagnostics and research, supporting more informed clinical and public health decisions.

bioinformatics↗