Biology subjects

Li, Z.

Publications and source records attributed to Li, Z..

At least 199 records · Page 11Linked to original sources

bioRxiv2026

Discovery of Selective Nrf2 Activators from Natural Products: AComputational Screening Approach to Minimize Off-Target Effects on PXR and CYP2D6

Nuclear factor erythroid 2-related factor 2 (Nrf2) is a central regulator of cellular antioxidant responses and a highly promising therapeutic target for a range of oxidative stress-related diseases. However, the clinical translation of Nrf2 activators has been hampered by significant off-target effects--notably unintended activation of the pregnane X receptor (PXR) and inhibition of cytochrome P450 2D6 (CYP2D6)--which can lead to dangerous drug-drug interactions and metabolic complications. To overcome this critical barrier, we conducted the first large-scale computational screening of 628,898 natural products from the COCONUT database, integrating molecular docking with a rigorous three-tier selectivity strategy designed to prioritize compounds that strongly bind KEAP1 (the primary Nrf2 repressor) while minimizing interactions with PXR and CYP2D6. Our innovative approach identified 10 ultraselective candidates that demonstrate potent KEAP1 affinity, negligible PXR engagement, and only moderate CYP2D6 binding--achieving up to 12.29-fold selectivity for Nrf2 pathway activation. These top hits are structurally novel, enriched in lipid-like and nucleoside-inspired scaffolds, and exhibit promising drug-like properties. By providing both a curated set of chemically diverse, selectivity-optimized leads and a publicly accessible screening dataset, this work establishes a new foundation for the rational development of safer, more precise Nrf2-targeted therapies, bridging a crucial gap between target potential and clinical viability. By prioritizing compounds with minimal off-target effects on PXR and CYP2D6, our approach offers a scalable template for reducing drug development failures and advancing safer therapeutics for oxidative stress-related diseases. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=124 SRC="FIGDIR/small/718057v1_ufig1.gif" ALT="Figure 1"> View larger version (26K): org.highwire.dtl.DTLVardef@13c6c79org.highwire.dtl.DTLVardef@1f5a078org.highwire.dtl.DTLVardef@fa4f4borg.highwire.dtl.DTLVardef@16bc881_HPS_FORMAT_FIGEXP M_FIG O_FLOATNOGraphical abstractC_FLOATNO C_FIG

bioinformatics↗

bioRxiv2026

Expansion Microscopy reveals changes in cellular and extracellular structures between healthy and perturbed tendons

Tendons transmit mechanical forces between muscles and bones through their highly aligned, collagen-rich extracellular matrix. When damaged, resident cells help restore this matrix. However, in tendinopathies, this repair response fails, leading to loss of proper tendon function. How altered mechanical states reshape tendon cell and matrix architecture remains poorly understood because existing methods do not readily capture tendon structure across relevant length scales. Here, we combine Fluorescent Labeling of Abundant Reactive Entities (FLARE) with Expansion Microscopy (ExM) to visualize cellular and extracellular structures in native tendon tissue. FLARE-ExM resolves the dense fibrillar matrix across multiple tendon types, including elastic fibers and glycan-rich cellular protrusions. In a tendon resection model, acute loss-of-tension was associated with increased fibril width and expansion of carbohydrate- and protein-rich regions. In ruptured human Achilles tendon, FLARE-ExM revealed extracellular disorganization and disrupted cellular architecture. These results establish FLARE-ExM as a useful approach for studying how mechanical perturbation remodels tendon architecture across physiological and disease contexts.

physiology↗

bioRxiv2026

Single-crypt evolutionary trajectories of human sporadic and hereditary colorectal precancers

Our understanding of the earliest genomic processes and evolutionary trajectories that distinguish sporadic from hereditary human cancers remains limited, impeding early diagnosis and risk stratification. Here, we performed whole-genome sequencing of 319 single crypts dissected from normal colorectal mucosa, premalignant polyps and malignant colorectal cancers (CRCs) from five sporadic patients and three familial adenomatous polyposis (FAP) patients. We found 66% of premalignant crypts from sporadic polyps did not harbor APC mutations, but frequently accumulated non-canonical but positively selected driver mutations including KMT2C, CACNA1A and FBLN2. In contrast, most premalignant crypts (71%) in FAP indeed acquired one or more somatic APC mutations or 5q loss of heterogeneity (LOH). Sporadic polyps were dominated by clock-like mutational processes at single-crypt level. In contrast, FAP crypts evolved more diverse genomics processes, including reactive oxygen species-associated damages, APOBEC-associated mutagenesis and pks+ Escherichia coli-associated mutagenesis. Premalignant crypts in FAP also exhibited a 2-fold higher burden of LINE-1 retrotranspositions than sporadic ones, despite arising at a substantially younger age. Phylogenetic timing estimations showed that the first somatic APC mutation in FAP often occurs prenatally. Phylogeographic analysis further revealed spatial lineage segregation in early polyclonal polyps, followed by Big-Bang lineage intermixing in later monoclonal ones. Together, these data demonstrate that early colorectal tumorigenesis does not obey a universal APC-first model, where sporadic and hereditary precancers follow distinct evolutionary trajectories. Our study also highlights that polyclonal-to-monoclonal transition in premalignant stages represents the Big-Bang growth of colorectal tumors, offering crucial insight into early diagnosis and risk stratification.

cancer biology↗

bioRxiv2026

Long-lived yeast-derived lipids promote longevity via regional and systemic metabolic remodeling

Aging is marked by a progressive breakdown of intestinal integrity and metabolic homeostasis, which together drives systemic decline in physiology and reduces lifespan. Here, we found that dietary lipids extracted from a genetically engineered long-lived yeast strain robustly extend lifespan in Drosophila and further uncovered the mechanisms using deuterium oxide-probed stimulated Raman scattering microscopy to image metabolic dynamics and single nucleus RNA sequencing (scRNA-seq) to unveil the underlying pathways. These yeast lipids are enriched in shorter, more saturated fatty acids and phospholipids as revealed by Raman spectroscopy and lipidomics, contributing to increased membrane order and reduced lipid storage. Functionally, targeted dietary supplementation with these lipid components synergistically prolongs fly lifespan. We show that these lipids reverse age-related declines in gut lipid droplet abundance, enhance membrane lipid incorporation, and increase de novo lipid synthesis, thereby improving epithelial structural integrity and barrier function. snRNA-seq identifies transcriptional remodeling in metabolically active enterocytes, including upregulation of autophagy and protein turnover genes, alongside reduction of unsaturated fatty acid biosynthesis. In the brain, dietary lipids orchestrate a dual metabolic strategy--promoting energy conservation and enhanced signaling across most neuronal and glial populations, while selectively boosting mitochondrial function in memory-critical Kenyon cells. All these leads to the enhancement of gut-to-glia communication, particularly through EGFR and FGFR pathways. Finally, analysis of our data with the Fly Metabolic Analysis Pipeline (FLY-MAP) reveals that yeast lipids restructure gut metabolic modules to coordinate energy production, redox balance, and nutrient flexibility. Our study uncovers a cross-kingdom mechanism of metabolic longevity regulation, paving the way for leveraging yeast-derived nutritional components to support tissue homeostasis and promote healthy aging.

molecular biology↗

bioRxiv2026

Continental-scale multi-omics reveals a distinct microbial-viral biome in sandy beach ecosystems structured by tidal zonation

Sandy beaches are dynamic coastal interfaces shaped by strong physical forcing and intense exchange between marine and terrestrial environments, yet their microbiomes remain poorly resolved at the genomic scale. Here we present a genome-resolved survey of microbial and viral communities across sandy beaches spanning a continental-scale latitudinal gradient along the Chinese coastline. By integrating cross-shore sampling, coastal geochemistry and large-scale multi-omics, we generated 978 metagenomes, 63 viromes and 72 metatranscriptomes, reconstructing 13,337 metagenome-assembled genomes and 38,255 viral populations. Sandy beach microbiomes exhibit exceptionally high genomic novelty, with more than 90% of species-level genomes representing previously undescribed taxa, suggesting that permeable coastal sediments constitute a distinct microbial and viral reservoir. Tidal zonation emerged as a dominant ecological driver structuring microbial diversity, metabolic strategies and virus-host interactions across cross-shore gradients. Genome-resolved analyses revealed systematic metabolic shifts from oxic heterotrophy in supratidal sediments toward increasingly chemolithotrophic and autotrophic pathways toward the low-intertidal and subtidal zone. Sandy beach microbiomes further encode broad potential for hydrocarbon and plastic transformation, together with diverse biosynthetic and antibiotic resistance repertoires that may mediate microbial chemical interactions. Together, these findings identify sandy beaches as a previously under-recognized microbial-viral biome shaped by tidal forcing, providing insight into microbiome evolution and coastal ecosystem resilience under increasing anthropogenic pressure.

microbiology↗

bioRxiv2026

RUNX1-deficiency drives immune-active ER+ mammary tumorigenesis through activation of interferon signaling

AbstractRecurrent loss-of-function mutations in RUNX1 occur in estrogen receptor-positive (ER+) breast cancers, yet how RUNX1-loss contributes to breast tumorigenesis remains unclear. Here we used genetically engineered mouse models with luminal mammary epithelial cell (MEC)-restricted gene disruption to investigate its role in breast cancer initiation. Loss of RUNX1 alone, or together with RB1, was insufficient to drive tumor formation. In contrast, combined loss of RUNX1 and p53 induced mammary tumors with full penetrance. These tumors contained ER+ cancer cells and exhibited extensive T cell and macrophage infiltration, indicative of an immune hot microenvironment. Mechanistically, RUNX1-deficiency activated interferon signaling in luminal MECs, associated with derepression of RUNX1 target STAT1 and enhanced inflammatory responses. Consistent with these findings, human ER+ breast cancers with low RUNX1 expression displayed elevated immune signatures and poorer patient survival. Together, our results identify RUNX1-loss as a driver of an immune-active subtype of ER+ breast cancer.

cancer biology↗

bioRxiv2026

Convergent Multimodal Evidence of Cortical Excitation-Inhibition Imbalance in Psychosis

Psychosis is increasingly understood as a disorder of disrupted cortical excitation-inhibition balance, yet robust non-invasive translational biomarkers remain lacking. The resting-state fMRI Hurst exponent (HE) and EEG aperiodic spectral exponent are promising complementary biomarkers, with lower values in each proposed to reflect a shift towards cortical hyperexcitability, but they have not been jointly examined in psychosis, and the spatial and molecular architecture of HE alterations remains poorly defined. We therefore tested for convergent systems-level signatures across independent cohorts and modalities, using resting-state fMRI (107 patients, 53 controls) and EEG (547 patients, 363 controls). Whole-brain and regional HE were estimated using wavelet methods, and EEG aperiodic exponents were quantified using spectral parameterisation. Compared with healthy controls, individuals with psychosis showed reduced whole-brain HE and widespread regional reductions. Regional HE case-control differences were associated with cortical gene-expression patterns, with enrichment for potassium channel and GABA receptor pathways, and correlated with noradrenergic, muscarinic, serotonergic, glutamatergic and dopaminergic receptor density maps, but not with cortical thickness or symptom or cognitive measures. In the independent EEG cohort, psychosis was similarly associated with a reduced aperiodic spectral exponent. Together, these findings provide cross-modal evidence for altered cortical resting-state dynamics in psychosis, consistent with a shift towards cortical hyperexcitability. Integration with receptor-density and transcriptomic maps implicates biologically plausible molecular pathways and supports HE and EEG aperiodic activity as scalable translational biomarkers in psychosis.

neuroscience↗

bioRxiv2026

METTL14-dependent m6A modification restrains interferon signaling to prevent myocarditis and dilated Cardiomyopathy

The impact of inflammation on heart failure is increasingly recognized; but how cardiomyocyte restrains innate immune activation remains poorly defined, and nor does the role of N-methyladenosine (mA) modification in maintaining cardiac immune homeostasis. Here, we demonstrate that cardiomyocyte-specific deletion of the mA methyltransferase METTL14 triggers myocarditis, dilated cardiomyopathy, and premature lethality. Meanwhile, widespread hypomethylation and upregulation of innate immune and necroptosis-related transcripts in Mettl14-deficient hearts exemplified by IFN-1 and STAT1. Mechanistically, METTL14 deficiency promotes RIPK1 accumulation thereby priming cardiomyocytes for necroptosis and inflammatory cell death. Genetic ablation of IFN-I receptor Ifnar1 can largely rescue the processes and improve cardiac function and survival. Furthermore, METTL14 loss disrupts mitochondrial integrity and autophagy/mitophagy flux, suggesting mitochondrial dysfunction-driven innate immune activation upstream of IFN-I signaling. Collectively, these findings identify METTL14-mediated mA modification as a critical safeguard against cardiomyocyte-intrinsic IFN-I signaling and necroptosis and establish an epitranscriptomic-innate immune axis that drives inflammatory heart failure.

pathology↗

bioRxiv2026

eSIG-Net: Accurate prediction of single-mutation induced perturbations on protein interactions using a language model

Most proteins exert their functions in complex with other interactors. Single mutations can exhibit a profound impact on perturbing protein interactions, leading to human disease. However, predicting the effect of single mutations on protein interactions remains a major computational challenge. Deep learning, particularly protein language models or transformers, has become an effective tool in bioinformatics for protein structure prediction. However, the functional divergence of mutations makes it difficult to predict their interaction perturbation profiles. To address this fundamental challenge, we present eSIG-Net (edgetic mutation Sequence-based Interaction Grammar Network), a novel sequence-based "Interaction Language Model" for predicting protein interaction alterations caused by single mutations. eSIG-Net combines various protein sequence embeddings, introduces a mutation-encoding module with syntax and evolutionary insights, and employs contrastive learning to evaluate mutation-induced interaction changes. eSIG-Net significantly outperforms current state-of-the-art sequence-based and structure-based prediction methods at predicting mutational impact on protein interactions. We highlight examples where eSIG-Net nominates causal variants with high confidence and elucidates their functional role under relevant biological contexts. Together, eSIG-Net is a first-in-kind "interaction language model" that can accurately predict interaction-specific rewiring by single mutations with only sequence information, and exhibits generalizability across biological contexts.

bioinformatics↗

bioRxiv2026

TRiC folds the giant ciliary protein IFT172 via a non-canonical open-state mechanism

The eukaryotic chaperonin TRiC/CCT is essential for folding complex proteins, yet how it folds substrates that exceed its closed chamber capacity remains a longstanding paradox. Here, we define the folding pathway of IFT172, the largest subunit ([~]200 kDa) of the intraflagellar transport (IFT) machinery, and uncover a "divide-and-conquer" mechanism. TRiC and HSP70 engage IFT172 concurrently but on distinct domains: TRiC captures the N-terminal WD40 {beta}-propellers within its chamber, whereas HSP70 independently stabilizes the C-terminal TPR domain in the cytosol. To accommodate this oversized client, specific TRiC subunits (CCT4, CCT2, and CCT7) undergo pronounced Z-shaped outward bending, thereby expanding the chamber. Unexpectedly, the first WD40 domain reaches a near-folded state within the open, ATP-bound chamber, and subsequent TRiC ring closure triggers substrate ejection rather than encapsulation. This non-canonical "fold-and-eject" mechanism challenges the classical view that the closed chamber is an obligate folding cage. We further demonstrate that this pathway is essential for ciliary functions in vivo, and reveal a conserved mode of chaperonin recognition among IFT components bearing tandem WD40-TPR architectures. Together, our findings establish a new paradigm for the folding of oversized, multi-domain proteins and identify TRiC as a central proteostasis hub in ciliary biogenesis, with direct implications for ciliopathy pathogenesis.

biochemistry↗

bioRxiv2026

Mitochondrial ATP production promotes T cell differentiation and function by regulating chromatin accessibility

Immune elimination of chronic infection or cancer requires cytotoxic CD8+ T cells that adopt and maintain an effector phenotype. Cytotoxic T cell function is a bioenergetically demanding process and T cells subjected to chronic antigen exposure have compromised effector function despite high rates of glycolysis. Here we report the ability of the short-chain -hydroxy acid, D--hydroxybutyrate, to act as a signaling molecule that increases mitochondrial ATP production and drives the conversion of proliferating T cells into cytotoxic effector cells. DAHB signaling switches ATP production from glycolysis to oxidative phosphorylation supported by fatty acid oxidation, even in glucose-replete media. This conversion suppresses both AMPK phosphorylation and the integrated stress response (ISR) in activated T cells while significantly elevating the level of the phosphagen, phosphocreatine (PCr). Both the PCr bioenergetic reserve and oxidative phosphorylation were required for T cell effector differentiation. DAHB-induction of CD8-effector gene transcription was coupled to bioenergetics by enhanced ATP-dependent remodeling of chromatin accessibility at effector gene loci. DAHB enhanced CD8+ T cell antitumor activity both in vitro and in vivo, and DAHB treatment of transferred T cells led to persistent in vivo antitumor effects. Together, these findings link cellular bioenergetics to the regulation of chromatin accessibility and gene expression required to support effector function.

immunology↗

bioRxiv2026

EthoClaw: An Integrated AI Workflow Platform for Automated Analysis in Neuroethology

Computational methods have advanced the analysis of animal behavior, yet significant challenges remain in data standardization, analytical reproducibility, and workflow integration. Existing computational solutions often demand extensive programming proficiency or compel users to navigate a highly fragmented ecosystem of disconnected tools for tracking, statistical analysis, and visualization. Here, we present EthoClaw, an open-source, artificial intelligence-driven workflow platform built upon the OpenClaw agentic framework, functioning as a locally deployable AI assistant for behavioral research. EthoClaw provides an integrated computational infrastructure that seamlessly bridges the gap between raw behavioral video acquisitions and publishable scientific results. In this study, we demonstrate the platforms capacity to natively ingest video data via a dual-mode tracking architecture: utilizing ultra-fast image processing for rapid object detection, and leveraging the SuperAnimal methods for precise, markerless postural tracking. To ensure maximal interoperability, EthoClaw automatically converts various tracking data formats into DeepLabCut-compatible formats, enabling high-throughput phenotyping by generating publication-quality visualizations alongside rigorous multidimensional statistical profiling. Furthermore, the platform incorporates a large language model (LLM)-driven reporting module that dynamically synthesizes analytical documents, ensuring methodological transparency. Through an open field test, we validate the practical usability of EthoClaw while accelerating computational throughput by localizing heavy video processing to circumvent cloud bandwidth bottlenecks. Operating via an omnichannel natural language interface that integrates seamlessly with ubiquitous instant messaging software, EthoClaw democratizes advanced computational behavioral analysis, offering a holistic, highly efficient ecosystem that enforces experimental reproducibility and open science principles.

animal behavior and cognition↗

bioRxiv2026

Psilocybin Attenuates Cortical Representations of Aversion in the Mouse Auditory Cortex

Psilocybin can produce sustained benefits in affective and trauma-related disorders, yet if and how it reshapes sensory representations of learned valence associations remains largely unclear. To address this, we used longitudinal two-photon calcium imaging in awake C57BL/6 mice to examine how psilocybin modulates layer 2/3 auditory cortex activity at single-cell and population levels. Evoked responses were measured for tones with or without prior associations with valenced stimuli, as well as for the valenced stimuli themselves. Most responsive neurons were selective for tones alone, while distinct subsets responded exclusively to reward or aversive stimuli, and a smaller population encoded both. Psilocybin selectively reduced responses to aversive stimuli and earlier-established aversive-associated tones, without affecting aversive association, reward responses, or responses to newly aversive-associated tones. At the population level, psilocybin acutely increased coordination across tone-responsive neurons, while later reducing it selectively among neurons encoding the aversive-associated tone. These results demonstrate that psilocybin preferentially dampens well consolidated aversive sensory representations in auditory cortex, rather than fresh associations, without broadly affecting auditory processing or new aversive learning.

neuroscience↗

bioRxiv2026

Influenza A virus membrane fusion is regulated by the balance between receptor binding and cleavage

Interactions between influenza A virus (IAV) and its host receptor, sialic acid, influence multiple stages of infection through the opposing activities of hemagglutinin (HA) and neuraminidase (NA). Multivalent HA binding to low-affinity receptors generates high avidity and attachment specificity, while NA-mediated receptor cleavage promotes penetration through sialylated respiratory mucus and progeny virion release. The role of receptor interactions in post-attachment entry, particularly during genome delivery through HA-mediated endosomal membrane fusion, remains unresolved due to conflicting data, limited control of receptor conditions, and difficulty separating attachment from fusion at low efficiency. Here, we address this question using a flow-cytometry based assay that quantifies time-resolved lipid mixing in hundreds of individual virion-membrane pairs per second, combined with programmable control of receptor density and chemistry on target membranes. We show that HA-receptor interactions regulate membrane fusion efficiency by promoting productive fusion-peptide insertion by HA. This effect depends on receptor context: within a defined regime, lipid-mixing efficiency increases with receptor density whereas NA activity reduces it by depleting receptors. Receptor type and HA-receptor binding avidity further modulate lipid-mixing outcomes, establishing a direct link between HA-receptor interactions and fusion efficiency. Together, these results identify receptor binding as an active regulator of membrane fusion and provide a framework that reconciles prior conflicting observations. More broadly, they extend the functional interplay between HA and NA to the level of membrane fusion, with implications for viral adaptation and host specificity. Significance StatementWe resolve a long-standing question by showing that host receptor binding by HA and cleavage by NA regulate influenza A virus membrane fusion, the entry step delivering the viral genome into the cell. HA-receptor engagement promotes productive HA membrane insertion, leading to fusion. Receptor density, HA-receptor affinity, and NA catalytic activity tune fusion outcomes. These results were enabled by two methodological advances: precise control of receptor presentation on target membranes and sensitive measurement of attachment and membrane fusion at the single-virion level. By defining how the opposing activities of HA and NA regulate fusion, this study extends their functional interplay to a key step in entry and provides new insight into how these proteins coevolve during viral adaptation.

microbiology↗

bioRxiv2026

Computational aberration-corrected volumetric imaging of single retinal cells in the living eye

The eye offers a unique non-invasive window for accessing single-cell level structures and functions of the central nervous system (CNS) throughout the retina. However, strong and space-varying ocular aberrations, along with limited volume rates, challenge large-scale cellular imaging in living eyes and stymie the full potential of possible biological and pathological studies in retina. Here, we present plenoptic illumination scanning laser ophthalmoscopy (PI-SLO), a 3D fluorescent retinal imaging modality that enables high-speed, widefield, volumetric single-cell imaging with low phototoxicity. By capturing multiple angular images of fluorescence signals from the entire volume, PI-SLO enables digital aberration correction and 3D imaging across a >20{o} FOV with >23 Hz volume rate. We leverage this structural and functional imaging modality to investigate three key aspects of CNS physiology through the living mouse retina, including: microglial process dynamics, vascular perfusion, and light evoked calcium fluxes in inner retinal neurons. PI-SLO is a versatile non-invasive platform for in vivo investigation of retinal and CNS physiology at the cellular level.

bioengineering↗

bioRxiv2026

Two ancient nuclear lineages and divergent reproductive strategies drive global success of the wheat stripe rust pathogen

Understanding the mechanisms for generating and maintaining plant pathogen genetic diversity is essential for predicting disease emergence and safeguarding the food supply. Here, using population-genomic analyses of 507 global isolates of the wheat stripe rust fungus Puccinia striiformis f. sp. tritici, we uncover two nuclear lineages that diverged before modern agriculture and reinforced during wheat domestication. Distinct lineage combinations underpin three major global populations: two predominantly sympatric, homozygous lineages (nuclA-nuclA and nuclB-nuclB) and a globally dominant heterokaryotic lineage (nuclA-nuclB) of hybrid origin. This hybrid population evolves largely clonally, accompanied by recurrent somatic nuclear exchange. Identical virulence traits can arise independently via somatic nuclear exchange in clonal populations or through sexual recombination in recombining populations, revealing convergent evolutionary routes to adaptation. Together, our results provide a transformative nuclear-level perspective linking deep evolutionary history and reproductive strategy to the global success of cereal rust fungi.

genetics↗

bioRxiv2026

DNA-triggered AIM2 condensation orchestrates immune activation and regulation

The innate immune sensor AIM2 detects cytosolic DNA and initiates inflammatory responses, yet its activation mechanism remains incompletely understood. Here, we show that AIM2 undergoes liquid-liquid phase separation upon DNA binding, forming dynamic condensates both in vitro and in cells. These condensates serve as platforms for inflammasome and PANoptosome assembly, promoting immune activation across multiple pathways. Direct structural determination from condensates reveals the assembly of active-form ASC filaments. Mechanistically, liquid-phase condensation is governed by multivalent interactions involving different AIM2 domains, including previously uncharacterized regions and species-specific elements. In vitro and in vivo assays show that mutants specifically disrupting condensation impair immune complex assembly, cell death initiation, antimicrobial defense, and intestinal homeostasis. Moreover, AIM2-DNA condensates function as regulatory hubs targeted by host- and pathogen-derived factors to balance immune homeostasis or facilitate immune evasion. These findings establish liquid-phase condensation as a fundamental mechanism of AIM2 activation and a potential therapeutic target.

immunology↗

bioRxiv2026

Translating Histopathology Foundation Model Embeddings into Cellular and Molecular Features for Clinical Studies

AI-powered pathology foundation models provide general-purpose representations of histopathological images by encoding image tiles into numerical embeddings. However, these embeddings are not directly interpretable in biological or clinical terms and must be translated into biologically meaningful features, such as cell-type composition or gene expression, to enable downstream clinical applications. To bridge this gap, we developed STpath, a framework that integrates histopathology image embeddings derived from existing pathology foundation models with matched, spatially resolved transcriptomics data. STpath consists of cancer-specific XGBoost models trained to infer cell-type compositions and gene expression from histopathology image tiles. We evaluated STpath in colorectal and breast cancer datasets and showed that it provides accurate estimates of the composition of major cell types and the expression of a subset of genes, with further performance gains achieved by combining embeddings from multiple foundation models. Finally, we demonstrated that STpath inferred features that can be used in downstream studies to evaluate their associations with clinical outcomes.

bioinformatics↗