Biology subjects

Lee, S.

Publications and source records attributed to Lee, S..

At least 109 records · Page 6Linked to original sources

bioRxiv2026

C9ORF72-derived polyGR polypeptides disrupt passive nucleocytoplasmic transport by tuning protein affinity for the nuclear pore barrier

Nucleocytoplasmic partitioning is an essential determinant of eukaryotic cellular function, governed by the nuclear pore complex, a molecular portal filled by a disordered phenylalanine-glycine (FG)-rich phase that governs selective entry and exit. Disruption of nucleocytoplasmic partitioning is seen across many different diseases, including viral infection, cancer, and neurodegeneration. However, what determines whether a given protein mislocalises when nucleocytoplasmic transport is disrupted remains unknown. This question is central to amyotrophic lateral sclerosis and frontotemporal dementia (ALS/FTD), where cytosolic mislocalisation of the nuclear RNA-binding protein TDP-43 is a defining pathological feature. The most common genetic cause of these diseases is a G4C2 repeat expansion in the gene C9ORF72, which produces aberrant neurotoxic polypeptides that induce nucleocytoplasmic transport defects. Here, we show how the highly toxic poly(glycine-arginine/GR) polypeptide engages the nuclear pore FG-rich selectivity barrier and retunes passive nucleocytoplasmic transport according to client surface chemistry. Using coarse-grained simulations, in vitro FG-phase reconstitution and human cell lines and neurons, we find that polyGR produces a non-linear, biphasic modulation of nuclear pore passage. Proteins with low affinity for the FG phase are unaffected, whereas proteins with higher affinity due to solvent-exposed hydrophobic residues exhibit enhanced transport up to a critical threshold, beyond which highly hydrophobic proteins experience transport suppression, cytoplasmic accumulation and aggregation. Together, these findings establish how disease-associated polypeptides retune the physicochemical rules governing passive nuclear pore transport, leading to biphasic outcomes determined by protein surface chemistry that alter protein compartmentalisation and aggregation. This provides a biophysical mechanism by which polyGR drives selective protein vulnerability to nuclear pore dysfunction in C9ORF72-associated ALS/FTD.

cell biology↗

bioRxiv2026

CDS-BART: A BART-Based Foundation Model for mRNA Sequence Analysis

Summary: Recent advancements in artificial intelligence (AI) have led to the development of foundation models that interpret mRNA as a language. Notable examples include CodonBERT, hydraRNA, EVO2, and Helix-mRNA. These models demonstrate significant potential as powerful tools for mRNA research. However, to best of our knowledge, there is currently no publicly available AI model that is both easy to use and capable of analyzing mRNA sequences up to about 4kb, a length scale typical of many therapeutic mRNAs, including those encapsulated within lipid nanoparticls (LNPs). Thus, we propose CDS-BART, a user-friendly, open-source tool that integrates SentencePiece sub-word tokenization with the denoising sequence-to-sequence training of Bidirectional and Auto-Regressive Transformers (BART). CDS-BART was pre-trained on mRNA data from nine taxonomic groups provided by the NCBI RefSeq database. This comprehensive pre-training, coupled with BARTs denoising capability, ensures effective learning of codon usage, mRNA structure, evolution, and regulation. Thus, CDS-BART can ultimately deliver robust performance across a wide range of mRNA prediction tasks. Availability and ImplementationCDS-BART is released under the MIT License. Latest code is available via Github at https://github.com/mogam-ai/CDS-BART.

bioinformatics↗

bioRxiv2026

Alternative 3' Polyadenylation Responses to Acute Ethanol Exposure Differ Between Drosophila Populations

Alternative polyadenylation is a pervasive post-transcriptional mechanism that generates mRNA isoforms with distinct 3 untranslated regions (3 UTRs), which can influence transcript stability, localization and translation. Ethanol is an environmental stressor for Drosophila melanogaster, and natural populations show distinct responses to ethanol exposure, suggesting that the genetic background can modulate phenotypic response. Yet, how the natural genetic variation modulates ethanol-responsive APA remains understudied. We profiled 3 end isoforms following acute ethanol exposure in three French (cosmopolitan) and three Zambian (ancestral range) D. melanogaster genotypes and identified numerous significant APA events across all genotypes. Ethanol-induced APA was widespread but highly genotype - and population - dependent: the French genotypes showed an enrichment of proximal shifts (3 UTR shortening), while Zambian genotypes were biased toward distal shifts (3 UTR lengthening). Most of the ethanol-responsive APA events were private to a single population, with relatively few genes exhibiting conserved directional changes across both backgrounds. We also classify genes showing opposite-direction shifts between populations, and others with opposite-direction shifts between various classes of APA. For two-site events, inter-poly(A) site distance did not predict response direction; however, spacing strongly predicted the magnitude of remodeling, with larger separations enabling greater absolute 3 UTR length changes. Our findings support a model in which ethanol-responsive 3 end regulation is a flexible genetically contingent, and that population-specific APA remodeling represent a plausible molecular substrate contributing to adaptation.

genomics↗

bioRxiv2026

Rapid Orthographic and Delayed Phonological Processing: ERP and Oscillatory Evidence from Masked Priming in Korean

A central question in visual word recognition concerns whether orthographic and phonological codes are coordinated sequentially or in parallel during lexical access. Korean Hangul, an alpha-syllabic writing system with morphophonemic spelling principles, allows independent manipulation of orthographic and phonological syllable overlap. In a masked priming lexical decision task with EEG (N = 30), we contrasted orthographically identical primes (e.g., -), phonologically overlapping primes (e.g., -), and unrelated primes. Event-related potentials and time-frequency representations (theta: 4-8 Hz, lower beta: 13-20 Hz, upper beta: 20-30 Hz) were analyzed to capture both evoked and oscillatory neural dynamics. Orthographic priming produced a cascade of facilitative effects: early fronto-central P200 enhancement (150-250 ms) with upper beta synchronization (30-290 ms), followed by centro-parietal N400 reduction (350-550 ms) with frontal theta suppression (400-730 ms), and behavioral facilitation. Phonological priming, by contrast, elicited an inhibitory behavioral effect and sustained lower beta activity over central regions (310-590 ms) but produced no early electrophysiological modulation, consistent with lexical competition. This spatiotemporal dissociation suggests that orthographic syllable processing can emerge at early stages and cascades into later lexical-level processing, whereas phonological syllable effects are confined to later processing stages. These findings provide support for a sequential or cascaded account of orthographic-phonological coordination, as predicted by dual-route models, while challenging strong forms of parallel activation, and suggest that the alpha-syllabic structure of Korean may enable a processing strategy in which orthographic parsing serves as an efficient entry route to the lexicon.

neuroscience↗

bioRxiv2026

Large-scale Perturbation of Systems Biology-Derived Genes Reveals Modifiers of HD-associated Transcriptomic Networks and Pathology

In Huntingtons disease (HD), disease-driver genes are broadly expressed, but pathogenic players specific to vulnerable neurons remain poorly defined. Leveraging a previously defined mutant huntingtin (mHtt) CAG-length-associated gene network, we perturbed 115 module hub genes with heterozygous knockout (KO-het) to assess their striatal transcriptomic modifier effects in wildtype and Q140 HD mice. We generated 3,592 striatal RNA-seq datasets, mapped 6,517 perturbagen-responder gene pairs, and uncovered regulators of medium spiny neuron (MSN) identity gene expression and DNA- methylase/demethylase-sensitive genes. We developed a bioinformatic pipeline to rank the perturbations with significant impacts on striatal transcriptome and HD-associated gene networks in wildtype or Q140 mice. KO-het for FoxP1 and Scn4b (two MSN-selective genes) exacerbated, whereas Pdp1 KO-het ameliorated, Q140 striatal pathology. Importantly, knockdown of functionally opposing ion channels, SCN4B and KCNH4, dichotomously affected aggregation and neurodegeneration in reprogrammed HD patient MSNs. Together, our study rigorously evaluated systems biology-derived candidates to identify modifiers of HD-associated molecular networks and pathology, providing an in vivo perturbation- transcriptome resource and highlighting genes involved in MSN excitability, transcription, calcium signaling, and mitochondrial metabolism in HD pathogenesis.

genetics↗

bioRxiv2026

NINTEDANIB AND PIRFENIDONE AFFECT GROWTH AND DIFFERENTIATION OF HUMAN ALVEOLAR TYPE 2 CELLS

BackgroundIdiopathic pulmonary fibrosis (IPF) is a progressive fibrotic lung disease characterized by epithelial cell senescence. Pirfenidone and nintedanib are approved drugs for the treatment of IPF. They significantly slow disease progression, but their mechanisms of action, especially on alveolar type 2 (AT2) cells, are poorly understood. We addressed this question by evaluating colony formation and growth of human AT2 cells co-cultured with fibroblasts in organoid culture in the presence of pirfenidone and nintedanib. We further evaluated molecular changes induced by these drugs via single cell RNA-seq of treated organoids. MethodsAT2 cells isolated from normal donor lungs or IPF patients were mixed with human fibroblasts in 3D culture and grown in the absence or presence of pirfenidone or nintedanib. After 14 days in culture, the organoids were quantified and cells extracted from Matrigel for single cell RNA-seq. ResultsAT2 cell organoids cultured in the presence of pirfenidone or nintedanib resulted in increased colony formation and, in the case of nintedanib, in larger colonies. We observed that untreated or pirfenidone treated AT2 cells lost surfactant protein C (SFTPC) expression and acquired an expression profile consistent with keratin (KRT)17high/KRT5- basaloid cells, whereas a larger proportion of nintedanib treated cells retained SFTPC expression. In contrast, AT2 cells treated with TGF{beta} inhibitor exhibited intermediate (SFTPC-/KRT17low) gene expression profile. ConclusionThese results suggest that nintedanib maintains an AT2-like expression state in culture and acts proximal to TGF{beta}. Conflict of Interest StatementPJW was supported by grants from Boehringer Ingelheim, Roche, Sanofi, Pliant and Arda Therapeutics and received personal fees from Boehringer Ingelheim and Sanofi. None of these companies had a role in the design or analysis of the study or in the writing of the manuscript. ASM, GP, JRR and DG are employees of Genetech. The other authors have no conflicts of interest to declare.

cell biology↗

bioRxiv2026

Hierarchical membrane-chromatin tethering buffers nuclear envelope assembly against alterations in lipid flux

In open mitosis, a single nuclear envelope (NE) forms around segregated chromosomes despite an excess of membrane-chromatin tethers. Here, we identify a hierarchical relationship between the BAF binding membrane tethers LEM-2 and Emerin, in which LEM-2 preferentially accumulates at BAF binding sites during postmitotic NE assembly and limits Emerin accumulation in C. elegans embryos. When LEM-2 is absent, Emerin occupies these sites and compensates for LEM-2 loss; however, NE formation becomes sensitized to ER membrane abundance - excessive phospholipid production through loss of CTDNEP1/CNEP-1 causes membrane invasions in interchromosomal regions and, across multiple systems, formation of lobulated, unstable nuclei with abnormal Emerin accumulations. We find that human CTDNEP1 regulates the NE-associated enzyme CCT to maintain phosphatidylcholine (PC) homeostasis through preventing delayed PC breakdown. Restoring PC levels rescues Emerin assembly and nuclear morphology defects resulting from the combined loss of CTDNEP1 and LEMD2. Together, these findings link membrane-chromatin tethering to ER lipid content and reveal preferential tethering as a determinant of NE assembly fidelity, with broad relevance to disease-related nuclear defects.

cell biology↗

bioRxiv2026

Archaic translation initiation factor eIF5B supports KSHV late lytic replication and viral oncogenesis by mimicking a hypoxic cellular landscape

Kaposis Sarcoma (KS) Herpesvirus (KSHV) is the etiological agent of KS, an AIDS-defining illness. Recent studies have demonstrated that even during normoxic conditions, KSHV facilitates replication by modulating hypoxia-inducible factors (HIFs), creating a hypoxia-like environment that promotes cellular transformation. KSHV lytic viral genes are favored for protein synthesis during infection by upregulating HIF2 and utilizing the hypoxic eIF4E2 translation initiation complex in oxygen-replete conditions. This translation initiation plasticity (TRIP) links viral replication strategies to angiogenic signaling and oncogenicity. However, the molecular basis of this plasticity remains poorly understood. This study reveals that viral inhibition of eukaryotic initiation factor 2 (eIF2) induces the use of another alternative initiation factor, eIF5B, which canonically mediates delivery of initiator methionine-tRNAiMet during hypoxia. We demonstrate through ribosome density fractionation that eIF2 accumulates in translationally inactive monosome fractions while eIF5B redistributes its translation activity toward polysomes during lytic replication. Progressive dependence on eIF5B during KSHV infection was illustrated by impaired late lytic gene expression, diminished virion production, and altered polysome profiles following eIF5B knockdown. Transcriptomic analyses further reveal that the mRNA landscape in KSHV-infected cells depleted of eIF5B mirrors that of uninfected hypoxic cells. Moreover, silencing of eIF5B in a natural infection model reduces both VEGF secretion and anchorage-independent growth--two hallmarks of KSHV viral oncogenicity. These results demonstrate that eIF5B functions as an essential component of alternative translation initiation machinery activated in response to eIF2 inactivation during KSHV lytic replication, emphasizing its key role in viral pathogenesis and its potential as a novel therapeutic target. IMPORTANCEKaposis Sarcoma Herpesvirus (KSHV) is a human cancer virus that causes severe malignancies in immunocompromised individuals, including the blood vessel cancer Kaposis Sarcoma and a highly aggressive body-cavity lymphoma. Although cells mount a response to infection by shutting down global protein synthesis, how viruses like KSHV continue making viral proteins when this cellular machinery is shutdown remains unclear. This study uncovers that KSHV exploits an ancient cellular factor typically used for protein production under low-oxygen conditions. We identify a previously unrecognized role for the translation factor, eIF5B, in supporting KSHV replication and select production of viral proteins during infection. Our work further demonstrates that KSHVs utilization of eIF5B contributes to a hypoxia-like environment during infection, ultimately contributing to cancer-promoting changes within the cell. These findings highlight a new strategy for tumor-virus reprogramming of host cells, opening an avenue for novel therapeutic targets to eliminate virus-associated cancers.

microbiology↗

bioRxiv2026

Sex- and hepatocyte PPARγ-dependent effects of an obesogenic dietary approach to induce MASH with fibrosis in mice

Mouse models of metabolic dysfunction-associated steatotic liver disease (MASLD) are valuable tools for identifying novel molecular mechanisms that drive progression from MASLD to metabolic dysfunction-associated steatohepatitis (MASH). However, generating a clinically relevant MASLD/MASH mouse model with obesity and peripheral metabolic dysfunction remains a challenge. In this study, we fed two different MASH-inducing diets to male mice with pre-existing high-fat (HF) diet-induced obesity. While a HF diet containing 40% Kcal from fat (mostly corn-oil shortening), 2% cholesterol, and 22% fructose reduced adiposity in these mice, a high-fat diet with 60% Kcal from fat (mostly lard), containing 2% cholesterol and supplemented with 10% fructose in the drinking water (HFC+Fr diet) promoted body weight and fat mass gain. Of note, 24 weeks of the HFC+Fr diet induced obesity, metabolic dysfunction, and liver steatosis in male and female mice, and promoted MASH with fibrosis in male mice. Furthermore, the HFC+Fr diet increased the expression of hepatocyte peroxisome proliferator-activated receptor {gamma} (Pparg), but the knockout of Pparg in hepatocytes (Pparg{Delta}Hep) reduced the development of MASH and fibrosis in male mice. In addition, the expression of key hepatic genes involved in methionine metabolism was downregulated by the HFC+Fr diet and upregulated by Pparg{Delta}Hep only in male mice. Overall, the HFC+Fr diet is obesogenic and promotes MASLD in both male and female mice. However, the HFC+Fr diet promotes MASH in a sex- and hepatocyte Pparg-specific manner, which may be associated with downregulation of hepatic methionine metabolism. New & NoteworthyWe explored how a new dietary intervention with fructose in the drinking water and added cholesterol to a high-fat diet extensively used to induce obesity and insulin resistance, promotes the onset of MASLD with obesity and metabolic dysfunction in male and female mice. This clinically relevant model of MASLD shows increased expression of hepatocyte PPAR{gamma} in both male and female mice, but only male mice have PPAR{gamma}-dependent impaired methionine metabolism and develop MASH with fibrosis.

physiology↗

bioRxiv2026

Effects of Cognitive Demand Reduction on Choice Overload

Choice overload occurs when an ever-growing number of options impairs decision quality, because evaluating options taxes cognitive resources. We investigated whether reducing cognitive demand could mitigate overload by encouraging greater cognitive effort to achieve optimal choice. We conducted two experiments manipulating cognitive demand in complementary ways: Experiment 1 reduced demand by presenting high-attractiveness sets, and Experiment 2 did so by providing a shortlist tool. In both experiments, participants chose from sets of 6-24 options while their eye-gaze and electroencephalographic (EEG) data were recorded. We found that reducing demand made decisions faster, but did not improve choice performance as set-size increased. Under low-demand conditions, eye-gaze measures revealed narrower search and EEG measures showed reduced working memory engagement per option, together indicating less searching and processing efforts. These results suggest that even with reduced cognitive demand, people coast through easier decisions, conserving effort and leaving the choice overload effect largely intact.

neuroscience↗

bioRxiv2026

Quantifying Behavioral Structure and Persistence in Open-Field Assays Using Entropy and Spectral Metrics

Traditional open-field assays quantify rodent behavior using low-dimensional scalar metrics that often overlook dynamical and temporally dependent aspects of spontaneous behavior. Here, we establish a high-throughput framework that integrates multi-camera 3D pose estimation from AVATAR-3D (1) with unsupervised autoregressive hidden Markov modeling using Keypoint-MoSeq (2-4) to transform recordings into time sequences of discrete, recurring behavioral units termed syllables. Drawing inspiration from information theory (5) and spectral analysis of Markov systems (6,7), we compute Shannon entropy and the second-largest eigenvalue (Eigen2) of the syllable transition matrix. Entropy and Eigen2 offer complementary lenses through which spontaneous behavior can be understood--one capturing dispersion of behavioral expression, the other reflecting temporal persistence and mixing dynamics. Together, these metrics quantify the organization of behavior as a stochastic dynamical system rather than a collection of isolated actions.

animal behavior and cognition↗

bioRxiv2026

GFMBench-API: A Standardized Interface for Benchmarking Genomic Foundation Models

The rapid scaling of Genomic Foundation Models (GFMs) has created a critical need for standardized evaluation frameworks. Current benchmarking practices are often fragmented, relying on model-specific preprocessing and inconsistent metric implementations that hinder reproducible comparisons. We present GFMBench-API, a high-level Python interface designed to unify the evaluation lifecycle of GFMs. GFMBench-API provides a modular "middleware" architecture that decouples model-specific tokenization and embedding logic from task-specific data streams and performance metrics. By standardizing the input/output schemas for common genomic tasks, such as regulatory element prediction, variant effect scoring, and long-range interaction mapping, GFMBench-API enables researchers to integrate new models or tasks with minimal "glue code." Our interface ensures mathematical consistency across evaluations, providing a robust foundation for the transparent and systematic benchmarking of GFMs.

genomics↗

bioRxiv2026

Novel single molecule imaging approaches reveal structure-function alterations to the nuclear pore complex in early C9ORF72-associated TDP-43 proteinopathy

Nucleocytoplasmic transport through the nuclear pore complex is essential for the maintenance of cellular homeostasis by regulating the movement of molecules between the nucleus and the cytoplasm. This process becomes dysfunctional in many diseases but has been particularly implicated in amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD) linked to the C9ORF72 mutation and associated aberrantly produced polypeptides. To directly study nucleocytoplasmic transport in intact non-genetically modified cells we have developed two new approaches for single molecule tracking through nuclear pore subdomains and super-resolved imaging of nuclear pore structural organisation. Using these techniques we have examined the early impact of the neurotoxic C9ORF72 polypeptide poly(glycine-arginine) on nuclear pore transport dynamics and structure. We find that soluble poly(glycine-arginine) peptides can disrupt molecular flow of passive cargo predominantly during nuclear export which is associated with altered structural organisation and molecular interaction of nuclear basket and central nuclear pore complex domains. These changes converge with perturbed nucleocytoplasmic homeostasis of the key ALS/FTD pathological protein TDP-43 and begin to explain this initiating step in disease.

cell biology↗

bioRxiv2026

Estradiol modulates neuronal network hyperexcitability in select NDD risk genes

Sex hormone signaling during neurodevelopment may act as a modulator of risk for neurodevelopmental disorders (NDDs), including autism spectrum disorders (ASDs). By conducting a systematic and multi-modal evaluation of estradiol effects across loss-of-function deletions in 36 functionally diverse, large-effect ASD/NDD genes in human induced pluripotent stem cell-derived neurons and larval zebrafish, we uncover convergent and divergent gene-by-estradiol interactions at the transcriptomic, circuit, and behavioral levels. Whereas estradiol partially ameliorates dysregulated expression signatures across all ASD/NDD gene knockouts examined, it selectively and rapidly dampens network hyperexcitability phenotypes in only a subset of ASD/NDD gene knockouts. Two genes, ASH1L and SCN2A, are particularly responsive to estradiol treatment, with dramatic rescue of molecular and cellular effects in human knockout neurons, coupled with behavioral phenotypes in larval zebrafish mutants. The seven most estradiol responsive genes show enhanced protein-protein interaction network connectivity, strong co-expression in specific postnatal time periods, and enrichment in upper-layer mature excitatory neurons. Altogether, we describe distinct estradiol rescue in different modalities, elucidating shared transcriptomic effects as well as gene-specific cellular and behavioral rescue in the context of loss of function mutations of select ASD/NDD genes.

genomics↗

bioRxiv2026

SHANK3-CAMSAP2 interaction links synapses to dendritic microtubule organisation in PV neurons

The microtubule cytoskeleton plays an essential role in establishing and maintaining neuronal polarity. In neurons, microtubules are initially generated at the centrosome which gradually loses this role in development. In mature neurons, microtubules are stabilised by the microtubule minus end-binding protein CAMSAP2. How and where microtubule minus ends are anchored within dendrites of mammalian neurons remains an open question. Here, we show that microtubules are directly attached to the postsynaptic density of excitatory synapses through an interaction of CAMSAP2 with the scaffolding protein SHANK3. This process is particularly relevant in parvalbumin-positive GABAergic neurons, in which excitatory synapses are predominantly located directly on the dendritic shaft in direct proximity to microtubules. Notably, this association is strongly enhanced by the Autism Spectrum Disorder (ASD)-associated SHANK3L68P mutation, leading to an increase of synaptic levels of CAMSAP2. This promotes an increased microtubule association with the synapse and alters microtubule dynamics. Downregulation of CAMSAP2 in SHANK3L68P parvalbumin neurons tips the balance between site-specific microtubule stabilisation and dynamics, reshaping dendritic architecture, connecting SHANK3-CAMSAP2 dependent microtubule regulation to synaptic ASD pathology. TeaserInteraction between synapses and microtubules is enhanced in parvalbumin neurons bearing the ASD-associated SHANK3 L68P mutation.

neuroscience↗

bioRxiv2026

GP130 Y814 SIGNALING IS REQUIRED FOR THE DYNAMIN-MEDIATED ENDOCYTOSIS, MAPK/P38 ACTIVATION AND PERSISTENCE OF CHRONIC SYSTEMIC INFLAMMATION INDUCED BY HIGH FAT DIET

In chronic inflammatory diseases such as obesity, tissue and cellular homeostasis are disrupted by persistent low-grade inflammation, where one of the most prominent pro-inflammatory cytokines that correlates with age and body mass index (BMI) is IL-6. All members of the IL-6 family of cytokines signal through their obligate co-receptor gp130, in whichsignaling tyrosines on the intracellular portion of gp130 activate multiple downstream pathways such as the canonical STAT and MAPK pathways. However, non-canonical gp130 pathways such as SRC family of kinases (SFKs) signaling have emerged as drivers of cellular stress response. Our recently published mutant mouse model carrying a constitutive inactivation of gp130-Y814 (F814 mice), which impairs SFK activation, exhibited an enhanced resolution of inflammatory responses and improved regenerative outcomes in both acute skin wound healing and post-traumatic osteoarthritis models. The current study was designed to explore whether the gp130-Y814 mutation reduces systemic chronic inflammation and multimorbidity in a high-fat diet (HFD)-induced model and to interrogate the possible downstream cellular mechanisms that are affected. In response to HFD, F814 mice showed significantly reduced systemic and tissue-specific inflammatory responses and protection from obesity-induced bone loss and osteoarthritis compared to wild type (WT) mice. After extensive characterization, the role of gp130-Y814 in monocytes/macrophages appeared to be dispensable, but we discovered that the F814 mutation blunts gp130 receptor internalization, p38/MAPK activation and the release of matrix metalloproteinases (MMPs) in chondrocytes. Finally, we showed that F814 chondrocytes had markedly reduced activation of the SFK-dependent GTPase dynamin 2 (Dyn2) in response to catabolic IL-6 cytokines. Introduction of a dominant-negative Dyn2 mutant into WT chondrocytes phenocopied the effects of the F814 mutation, resulting in attenuated p38/MAPK signaling and reduced MMP activation following stimulation with IL-6 cytokines. This study demonstrates, for the first time, that the Y814 residue is directly implicated in Dyn2-mediated internalization of the gp130 receptor, thereby modulating downstream signaling and contributing to pathological outcomes in chronic inflammatory and degenerative diseases in a cell type-specific manner.

cell biology↗

bioRxiv2026

Chromatin architecture and physical constriction cooperate in phenotype switching and cancer cell dissemination

Phenotypic plasticity is a prominent cancer feature that contributes to metastatic potential and resistance to therapy across multiple cancer types. Cancer cell state transitions have been attributed to transcriptional programs, such as the AP1/TEAD-regulated gene network driving the mesenchymal-like (MES) phenotype. In addition, during dissemination, tumor cells are subjected to variable loads of physical mechanical pressure and constriction across transited tissue, which are thought to impact nuclear molecular crowding. How the interplay between mechanical pressure, global 3D nuclear architecture and transcriptional programs contributes to MES identity and metastatic adaptation remains unclear. Using cutaneous melanoma as a model for early dissemination, we integrate in vitro and in vivo epigenomic profiling with nanoscale imaging of cell lines and patient samples to investigate chromatin organization features underlying the MES phenotype. We find that in MES cells, CTCF is relocated from domain boundaries to regulatory regions of EMT-like genes, leading to reduced insulation, extended topological associated domains (TADs) and increased inter-domain contacts, and de novo formation of chromatin hubs. This conformational rewiring, along with loss of heterochromatin, supports nuclear deformability during invasion and dissemination. Conversely, physical constriction of melanocytic cells induces MES-like chromatin features--including CTCF repositioning and heterochromatin loss-- and promotes metastasis in vivo. Similarly, pharmacological inhibition of the heterochromatin mark H3K9me3 triggers MES characteristics and increases invasiveness. These results demonstrate that metastatic competency involves both epigenetic and structural nuclear reprogramming, enabling shifts in gene networks and physical adaptability. Our findings reveal mechanistic links between nuclear architecture and aggressive tumor behavior, identifying potential biomarkers and therapeutic targets to intercept metastatic progression.

cancer biology↗

bioRxiv2026

Model-Driven Hybrid AI Framework for End-to-End Autonomous Decision-Making in Drug Development

Decision-making in drug development spans heterogeneous stages from molecular design to clinical optimization, yet computer-aided workflows across stages remain fragmented, limiting traceability from evidence-derived clinical questions to simulation scenarios and decision endpoints. We present a model-driven hybrid AI framework for end-to-end decision support that treats PICO (Participants, Intervention, Comparison, Outcome) as a machine-actionable specification to define scenarios and align endpoints across modules. Given a PICO-defined clinical question and a drug SMILES, the framework infers PBPK-ready parameters from structure, generates exposure via mechanistic PBPK simulation, and links exposure outputs to clinic-facing therapeutic drug monitoring (TDM) decision support using an open-source NLME-based Clinical Pharmacokinetic Consultant Service (CPCS). We additionally implement an LLM-based PICO extraction layer to structure clinical abstracts into P/I/C/O elements for systematic evidence ingestion. Results include backbone verification for DTI-guided intrinsic clearance prediction under fold-error bounds and PBPK trajectory plausibility, and external validation of the hybrid AI-PBPK pipeline showing substantial dispersion for several endpoints. An input-controlled comparison between PhysioSim and PK-Sim on N = 9 drugs yields similarly degraded accuracy under identical AI-predicted inputs, suggesting upstream parameter quality as a dominant constraint. For evidence ingestion, prompting-only PICO extraction achieves F1 of 0.687 on EBM-NLP and 0.552 on TB-PICO. For clinical utility, CPCS improves TDM performance over a PKS baseline on phenobarbital and vancomycin, reducing MAPE by approximately 26-60% depending on configuration. Overall, the framework provides a modular, traceable blueprint linking evidence-defined questions to exposure simulation and TDM decision support while preserving mechanistic interpretability. Future work will integrate outcome-aligned time-to-event modeling and federated learning to close the PICO-to-outcome loop and improve parameter inference under multi-site governance.

pharmacology and toxicology↗