Biology subjects

Liu, X.

Publications and source records attributed to Liu, X..

At least 379 records · Page 21Linked to original sources

bioRxiv2025

Phosphatidylcholine Metabolism Controls Alveolar Progenitor Renewal and Pulmonary Fibrosis

Idiopathic pulmonary fibrosis (IPF) is a progressive, fatal lung disease marked by alveolar type 2 (AT2) stem cell dysfunction and excessive matrix deposition, with no effective treatments. Recent advances have recognized that AT2 cells act as stem cells, in addition to their role in the production of pulmonary surfactants in the distal alveolar space. We and others have reported a failure of AT2 regeneration and a loss of AT2 cells in IPF. We recently further reported that there is a defect in lipid metabolism in IPF AT2 cells and we discovered a selective loss of lysophosphatidylcholine acyltransferase 1 (LPCAT1) in AT2 cells from IPF, as well as in AT2 cells from bleomycin-injured mice. Pharmacological and genetic experiments confirm that LPCAT1 is required for AT2 cell renewal in 3D organoid assays. AT2 cell-specific Lpcat1 deletion resulted in reduced AT2 renewal, spontaneous lung fibrosis, and heightened susceptibility to bleomycin-induced fibrosis in mice in vivo. Expression-based high-content drug screening with an LPCAT1 knock-in cell line identified several drug families that upregulated LPCAT1 expression. We further confirmed that anti-malarial artesunate and PLA2 inhibitor ONO-RS-082 increased LPCAT1 mRNA expression, promoted AT2 renewal, and attenuated bleomycin-induced lung fibrosis in mice in vivo. Our findings establish LPCAT1 as a critical regulator of AT2 renewal and lipid metabolism in IPF, suggesting that reactivation of LPCAT1 could offer a novel therapeutic strategy for restoring alveolar progenitor function and mitigating lung fibrosis.

cell biology↗

bioRxiv2025

Vascular and Lymphatic Dysregulation via Non-EndoMT Col2a1 Signaling in Bisphosphonate-Related Osteonecrosis of the Jaw

Bisphosphonate-related osteonecrosis of the jaw (BRONJ) is one of the most severe complications of antiresorptive therapy. Although impaired angiogenesis is commonly observed in BRONJ, the underlying vascular mechanisms remain poorly understood, impeding the development of effective treatments. In this study, we integrated whole-organ tissue-clearing imaging, single-cell transcriptomics, and proteomics to construct a multimodal atlas of the mandibular microenvironment in both mouse and human BRONJ. Our findings revealed gross vascular and lymphatic rarefaction in BRONJ, along with 4 endothelial subtypes uniformly exhibiting aberrant upregulation of Col2a1. Moreover, fibroblasts and macrophages emerged as key endothelial interactors, collectively underscoring dysregulation of the endothelium-matrix-immune axis. Unlike homeostatic Dmp1/Tfap2a fibroblasts and Stab1 macrophages, BRONJ lesions featured pathological Lrrc15/Chad fibroblasts and Il6 macrophages, which promote ectopic chondrogenesis and inflammation. Mechanistically, BRONJ activated the COL2A1-CD44 axis (EC-to-fibroblast/macrophage signaling) and COL2A1-SDC4 axis (EC-to-fibroblast signaling), whereas lineage tracing excluded endothelial-to-mesenchymal transition (EndoMT), implicating extracellular matrix remodeling as the driver of dysfunction. Cross-species validation in human mandibles further confirmed vascular-lymphatic dysregulation, inflammatory activation, and aberrant chondrogenesis. Overall, this study establishes vascular and lymphatic dysfunction as a central pathological hallmark of BRONJ, driving matrix remodeling dysregulation and immune-inflammatory imbalance, and identifies the COL2A1-CD44/SDC4 axis as a potential therapeutic target for BRONJ.

cell biology↗

bioRxiv2025

XomicsDB: a database of paired multi-omics datasets for systems biology

Life science research is primarily driven by data integration on the basis of theoretical development. Broad applications of multiomics (xomics) data calls for efficient integration of xomics data to infer cellular operation and regulation. To that end, it is urgently needed good-quality xomics data should be curated and shared for the community. Here we present XomicsDB, a database curating paired xomics datasets at both bulk and single-cell levels in model organisms. It contains 183 datasets from 13 model organisms, with a total of seven xomics data types. Users can download these datasets individually or in a batch from the ftp server provided. Users can also upload their paired xomics datasets for sharing. Incoporation of xomics data intom GSMs are demonstrated in model organisms Escherichia coli and human. Future efforts will be on collecting more single-cell xomics data and curating non-model oragnisms. The XomicsDB can be found at: http://xomicsdb.com.

systems biology↗

bioRxiv2025

Programmable Aptamer-Embedded Circular RNAs for Targeted Antigen-Presenting Cells Immunotherapy

The targeted delivery of mRNA to specific cell types in vivo remains a major unmet need in mRNA-based therapeutics. Conventional strategies rely on nanocarriers engineered with various modifications. However, these exogenous materials could cause adverse reactions, and their complex manufacturing processes often hinder clinical application. By leveraging the innate programmability of RNA sequences, we herein report a carrier-free, scalable, and programmable aptamer-embedded circular RNAs (Apt-circRNAs) design that confers targeting capability through the strategic incorporation of different aptamers at varying densities, while retaining circular RNA stability and protein production capability. Experimental results show that antigen-loaded Apt-circRNAs enable targeted delivery to antigen-presenting cells (APCs), drive endogenous antigen expression, activate antigen-specific T cells, and elicit potent immune responses that mediate clearance of both early-and late-stage tumors in mouse models. In humanized models of colorectal cancer, the combination of Apt-circRNA and PD-1 blockade produced synergistic antitumor effects and significantly reshaped the tumor microenvironment. In a first-in-human (FIH) clinical trial, Apt-circRNA vaccine encoding KRAS G12D/G12V neoantigens exhibited a favorable safety profile and induced robust immune activation, T cell mobilization, and neoantigen-specific immune responses. With this study, we have established a new paradigm in carrier-free, APC-targeted RNA vaccine design, offering a safe and effective platform for next-generation cancer immunotherapy.

bioengineering↗

bioRxiv2025

Late treatment initiation leads to reduced antiviral potency

The timing of initiation is critical in antiviral treatment and viral dynamic (VD) modeling is a powerful tool to study the within-host viral load changes and evaluate antiviral treatment effects using mathematical equations. Previous simulation studies have shown that early treatment initiation is critical to maximize the therapeutic response in antiviral treatment in an acute viral infection such as influenza and SARS-CoV-2. A recent experimental study demonstrated that late therapy initiation can lead to diminished antiviral potency. However, most VD model simulations with varying treatment initiation time accounted only for the effect of initiation condition (i.e., state of different cell populations when the therapy started), the loss of drug potency has been under-investigated. This may overestimate the antiviral effect, potentially resulting in suboptimal dose selection. To this end, we aimed to characterize relationship between the drug potency (EC50) and the timing of drug addition, using nirmatrelvir and GS-441524 against SARS-CoV-2 as an example. Viral load data were obtained from in vitro experiments with various drug concentrations and treatment initiated between 0 to 3 days post infection. EC50 values were fitted for each treatment initiation group and were found to vary with the timing of treatment initiation in both drugs. Also, a VD model with time-varying EC50 provided better fits than a constant EC50 model (BIC = 1667.90 vs. 1677.84). Further simulations also indicated that a constant EC50 model overestimated the antiviral efficacy when treatment started late. These findings highlighted the importance of considering EC50 shift when optimizing dosage regimens for patients presenting late.

pharmacology and toxicology↗

bioRxiv2025

Single Cell Foundation Models Evaluation (scFME) for In-Silico Perturbation

Foundation models pre-trained on large single-cell RNA atlases offer a compelling alternative to in-vitro experimentation for understanding gene regulatory networks and conducting gene perturbation analyses, with significant implications for target identification. Numerous foundation models have been developed, building upon early efforts such as Geneformer and scGPT. Hyperparameter optimization also results in multiple variants which require comparative analysis. Current benchmarking approaches focus on feature-based assessments or intuitive biological and statistical tasks, which may not align with the models training objectives. A recent study proposed a systematic benchmarking framework; however, its scope was limited to pre-trained (zero-shot) models. To address these limitations, we propose Single-Cell Foundation Model Evaluation (scFME)--a systematic method designed to benchmark fine-tuned foundation models for insilico perturbation (ISP). scFME ensures comprehensive and robust assessment by requiring sufficient separation between control and perturbed cells at the outset and by quantifying ISP accuracy against zero and random perturbation baselines. Furthermore, scFME enables exploration of model performance across distinct gene categories, facilitating biological interpretation and functional relevance. Using this framework, we evaluated several commonly used models (and some of their variants) and demonstrated that the methodology effectively characterizes their performance in ISP studies. Our results position scFME as a versatile and rigorous methodology for evaluating and comparing current and future foundation models.

bioinformatics↗

bioRxiv2025

A neural circuit for female-specific defensive homeostasis in risk assessment

HighlightsSC CB1R+ neurons encode visual survival threats and initiate risk assessment in both sexes CB1R+ SC-LHb GABAergic circuit maintains defensive homeostasis in risk assessment in females eCBs gate this circuit by disinhibiting presynaptic GABA release exclusively in females Disruption of this mechanism female-specifically impairs risk assessment and stress coping In BriefLiu et al. identify a female-specific subcortical circuit in which eCB signaling, via CB1R on GABAergic terminals in the SC, suppresses transmission to the LHb to maintain precise defensive homeostasis in risk assessment. These findings uncover the circuit and neuronal mechanisms that contribute to the sexually dimorphic maintenance of defensive homeostasis. Risk assessment in defensive behavior is an adaptive mechanism shaped by natural selection, enabling individuals to evaluate potential threats and thereby maintain defensive homeostasis. However, it remains unknown whether specific neural circuits maintain their behavioral homeostasis in a sex-specific manner. To address this, we investigated its neural basis by hierarchical behavior analysis framework with neural circuit dissection. In mice of both sexes, visual survival threats activated cannabinoid 1 receptor (CB1R)-expressing neurons in the superior colliculus (SC) to initiate consistent risk assessment. Deletion of CB1R in SC GABAergic neurons female-specifically impairs risk assessment by disinhibiting GABA release in SC-lateral habenula (LHb) projections, resulting in shortened risk assessment. This disruption furthermore increases the occurrence of abnormal spontaneous behavior following chronic stress exclusively in females. We identified a female-specific SC-LHb GABAergic circuit gated by CB1R maintains defensive homeostasis in risk assessment. Our findings reveal how a conserved neuromodulatory system sex-specifically gates a subcortical circuit to orchestrate distinct survival strategies.

neuroscience↗

bioRxiv2025

Murine models for triple-negative breast cancer with differential responsiveness to immunotherapy

Breast cancer is the most common cancer diagnosis in women. Clinical studies with triple-negative breast cancer (TNBC) are encouraging for immunotherapy combined with chemotherapy (anti-PD-1 with paclitaxel and/or carboplatin). However, additional clinical advances may be pursued more rapidly with assistance from preclinical TNBC models including syngeneic mammary tumor cell lines. Here, we report two mammary tumor cell lines that exhibit differential responsiveness to immunotherapy in vivo. Spontaneous mammary tumors from C57BL/6J MMTV-Cre Trp53fl/+ animals were passaged serially in cell culture and in vivo in the mammary fat pad of fully wildtype animals. The resulting lines, MM001i and MM008i, lost Trp53 and formed 1000 mm3 tumors in the mammary fat pad within 21-28 days. Despite originating from the same genetic background, these lines exhibit differential responses to immunotherapy. For anti-PD-1 therapy, MM001i is poorly responsive and MM008i is strongly responsive with near-complete tumor regression. In comparison, both MM001i and MM008i respond rapidly to anti-CTLA-4 therapy. Both models express unique tumor antigens as evidenced by immunity to subsequent engraftments. Primary MM008i tumors exhibit greater T cell infiltration, and CD8-positive T lymphocytes are required for anti-PD-1 responses. These TNBC models are promising for further mechanistic studies and testing future single and combinatorial therapies.

cancer biology↗

bioRxiv2025

Longitudinal analysis of bacterial community dynamics in normal hospital ward following patient occupancy

The hospital microbiome significantly influences patient recovery and clinical outcomes. However, the dynamics of microbial colonization and transmission following initial patient occupancy remain poorly understood. Here, we employed 16S rRNA gene amplicon sequencing to investigate bacterial community dynamics on surfaces within neurosurgery ward and patients as a new hospital became operational. Our results showed that bacterial colonization in hospital wards follows distinct site-specific patterns, after hospital opening, alpha diversity was significantly increased on floors and drawer handles but decreased on bedrails and faucet handles compared to preopening. Beta diversity analysis showed that surfaces frequently contacted by patients exhibited the greatest compositional turnover, such as bedrails, drawer handles, and faucet handles, bacterial communities in after-opening were more homogeneous across sites than preopening, indicating potential bacterial transmission. Moreover, we found following patient admission, patient hand-derived microbiomes exert a significant influence on the bacterial communities in hospital wards, with a particularly pronounced impact on bedrails. Additionally, the potential disease risk of bedrails in post-opening was significantly higher than preopening. Taken together, these findings underscore the critical role of human contact in shaping hospital microbiomes and highlight the importance of targeted infection control strategies to mitigate pathogen transmission.

bioinformatics↗

bioRxiv2025

PEDF peptides rescue defects in neurite morphogenesis and intracellular calcium response in cortical neurons from mice exposed to valproic acid

Pigment epithelial-derived factor (PEDF) is a multifunctional protein produced predominantly by the retinal pigment epithelium and expressed in many tissues, including the brain, highlighting its participation in crucial processes, such as neuroprotection and angiogenesis. Some neurodevelopmental disorders, such as ASD, are characterized by neurodevelopmental abnormalities, including altered neurite formation, spine formation, and neuronal activities. Many efforts have been made to resolve NDDs, but until now, some symptoms remain untargeted. PEDF is involved in many steps of neurodevelopment. The treatment of PEDF peptide might improve the outcome of NDD symptoms by altering neuronal morphologies. We used PEDF peptides that contain different functional domains to study the effect of administering PEDF peptides on neuronal morphology in a prenatal valproic acid (VPA)-exposed mouse model. We identified that treatment with PEDF peptides rectified the abnormalities in neurite formation and spine formation in VPA-exposed cortical neurons. In vitro calcium imaging showed abnormalities in spontaneous activity in VPA-exposed cortical neurons. Treatment of a short PEDF peptide normalized the intracellular calcium response to the control level. Accordingly, PEDF peptides have the prospect of serving as potential treatments for patients with neurodevelopmental disorders, such as ASD.

neuroscience↗

bioRxiv2025

PI-FC: Pre-training Individual-specific Functional Connectome through State-invariant Contrastive Learning

Functional MRI enables non-invasive mapping of brain connectivity, yet its clinical translation remains hindered by uncontrolled state-dependent variability that obscures individual-specific signatures during routine scanning. Here we introduce PI-FC -- a deep learning framework leveraging state-invariant contrastive learning to extract stable individual brain signatures across diverse arousal levels, cognitive states, and temporal scales spanning tens of seconds to hours. PI-FC achieves equivalent phenotypic prediction accuracy using substantially reduced scanning time, and eliminates state-dependent effects varying task demands and brain states. Trained on 36,119 subjects across 8 independent datasets, our model demonstrates superior cross-site generalization and outperforms traditional functional connectome (FC) in predicting neuropsychiatric conditions including schizophrenia, autism, depression, and anxiety. Furthermore, PI-FC enables zero-shot inference of brain age, biological sex, and cognitive ability without site-specific retraining. Overall, PI-FC represents a robust, clinically scalable framework that overcomes fundamental barriers to real-world deployment of precision functional neuroimaging.

neuroscience↗

bioRxiv2025

Interpretable Deep Learning Reveals Biologically Relevant Spatial Gene Expression Patterns in Lung Tumors and their Microenvironment

Lung adenocarcinoma (LUAD), the most common subtype of non-small cell lung cancer (NSCLC) exhibits profound histological and molecular heterogeneity, hindering accurate prognosis and effective treatment. Current approaches to assess this heterogeneity, such as histopathology, molecular profiling, and spatial transcriptomics are constrained by high costs, long turnaround times, and limited tissue availability, making them challenging for widespread prognostic use. To address this gap, we developed XpressO-Lung, an explanatory deep learning model that predicts gene expression heterogeneity, spatially, in tumor and its microenvironment, on hematoxylin and eosin (H&E) based diagnostic (Dx) whole-slide images (WSIs) by learning associations between tissue morphology and the corresponding bulk-transcriptomic data. Utilizing 200 LUAD cases from The Cancer Genome Atlas (TCGA), XpressO-Lung predicted spatial expression patterns of NAPSA, SLC47A1, TP53I3, KLRB1, FAM189A1, TICAM1, CD8A, CXCL13, TTF, CDH3, KRT7 and CDKN2A genes (biomarkers) on the respective Dx-WSIs with AUCs ranging up to 0.92. More importantly, the predicted spatial gene expression patterns aligned with the known morphologic interactions of the tumor and its microenvironment, capturing biological events directly on Dx-WSIs. By coupling predictive performance with spatial interpretability of gene expression on Dx-WSIs, XpressO-Lung bridges histopathology and bulk-transcriptomics, enabling explainable morpho-genomic analyses to advance biomarker discovery and offer prognostic insights to inform precision oncology in LUAD, especially in low-resource settings.

pathology↗

bioRxiv2025

Molecular dynamics simulation of the truncated NMDA receptor in the open state

The GluN1/GluN2A N-methyl-D-aspartate receptor (NMDAR) is a critical ligand-gated ion channel in the central nervous system, playing essential roles in synaptic plasticity, learning, and memory. Understanding its dynamics in the open/active state is paramount for deciphering its physiological functions and for developing targeted therapeutics. Despite many past efforts, the active/open state has not yet been fully resolved at atomic resolutions. To elucidate the molecular mechanism of the NMDAR activation, computer modeling and simulation are instrumental in providing detailed information about the dynamics and energetics of the receptor in various functional states. In this study, we started from a previously built open-state model of a truncated NMDAR, and then explored its energetics and dynamics with extensive molecular dynamics (MD) simulation (total simulation time is 5.6 {micro}s). Based on the MD simulation, we employed an array of analysis tools to study the fluctuations/motions at the levels of individual residues, the channel pore, and the global structure, and identified a dynamic network of polar/nonpolar interactions between residues. Furthermore, we used machine learning to identify key interactions and residues relevant to channel opening/closing and validated them with evolutionary conservation grades and annotations of disease mutations in NMDAR. Taken together, our modeling and simulation provided rich structural and dynamic information which will guide functional studies of the activation of this key receptor.

biophysics↗

bioRxiv2025

Sleep deprivation exhibits age-dependent effect on infra-slow global brain activity

Infra-slow (< 0.1 Hz) global brain activity, quantified by the global mean BOLD (gBOLD) signal in resting-state fMRI, is elevated during sleep and closely associated with cerebrospinal fluid (CSF) dynamics, a key mechanism for the brain waste clearance implicated in neurodegenerative disorders such as Alzheimers disease (AD). However, the effect of sleep deprivation on gBOLD activity and its interaction with aging remain poorly understood. Using a rigorously controlled in-laboratory total sleep deprivation (TSD) protocol, we demonstrate that TSD significantly increases both the amplitude of the gBOLD signal and its coupling with CSF flow, suggesting a compensatory mechanism that may enhance glymphatic clearance following acute sleep loss. Notably, these TSD-induced enhancements exhibit robust age dependency, with diminished responses in individuals at midlife (40-50 yrs). The absence of this compensatory mechanism in these midlife participants may exacerbate age-related impairments in neurotoxic clearance and increase dementia susceptibility, thereby offering mechanistic insights into the nexus between sleep disruption, aging, and neurodegeneration.

neuroscience↗

bioRxiv2025

From ambiguous object to illusory faces: EEG decoding reveals a dynamic cascade of face pareidolia

Face pareidolia, perceiving faces in inanimate objects, provides a unique window into how the brain constructs perceptual meaning from ambiguous visual input. In this work, we investigated the neural dynamics of face pareidolia, testing the hypothesis that it emerges from a temporal cascade that integrates global (low spatial frequency, LSF) and local (high spatial frequency, HSF) visual information. With a pareidolia detection task, we combined event-related potentials (ERP), multivariate pattern analysis (MVPA), and representational similarity analysis (RSA) to map the evolving neural code for illusory face perception. Behaviorally, participants experienced pareidolia from both LSF and HSF stimuli, but the percept was strongest for broadband images containing both frequency bands. Electrophysiological results revealed a clear temporal progression. Early processing (<150 ms) was driven by low-level attributes, with distinct P100/N100 components and neural representations for LSF and HSF signals. A critical shift occurred around 145 ms, after which MVPA decoding could reliably distinguish between images perceived as faces and those that were not. Concurrently, RSA showed that the geometry of neural representations shifted to reflect participants subjective face-likeness ratings, rather than the initial physical properties of the stimuli. This transition to a higher-order, subjective evaluation was marked by a later N250 component, whose amplitude indexed the cognitive load of integrating the visual cues. Together, these findings demonstrate that face pareidolia is a dynamic neural cascade, from early sensory encoding to higher-order perceptual evaluation, driven by the integration of global configuration and local features of visual stimuli. Our results illuminate how the brain transforms ambiguous input into meaningful social perception, offering insight into the temporal dynamics of visual awareness.

neuroscience↗

bioRxiv2025

stk32a links sleep homeostasis to suppression of sensory and motor systems

Sleep is regulated by a homeostatic process and associated with an increased arousal threshold, but the genetic and neuronal mechanisms that implement these essential features of sleep remain poorly understood. To address these fundamental questions, we performed a zebrafish genetic screen informed by human genome-wide association studies. We found that mutation of serine/threonine kinase 32a (stk32a) results in increased sleep and impaired sleep homeostasis in both zebrafish and mice, and that stk32a acts downstream of neurotensin signaling and the serotonergic raphe in zebrafish. stk32a mutation reduces phosphorylation of neurofilament proteins, which are co-expressed with stk32a in neurons that regulate motor activity and in lateral line hair cells that detect environmental stimuli, and ablating these cells phenocopies stk32a mutation. Neurotensin signaling inhibits specific sensory and motor populations, and blocks stimulus-evoked responses of neurons that relay sensory information from hair cells to the brain. Our work thus shows that stk32a is an evolutionarily conserved sleep regulator that links neuropeptidergic and neuromodulatory systems to homeostatic sleep drive and changes in arousal threshold, which are implemented through suppression of specific sensory and motor systems.

neuroscience↗

bioRxiv2025

Enteric ChAT-expressing neurons as new target for Lactobacillus plantarum ameliorates inflammatory bowel diseases

The gut microbiota plays a crucial role in inflammatory bowel diseases (IBD), yet how specific microbial components influence disease progression remains incompletely understood. Our study reveals that decreased abundance of Lactobacillus species correlates with ulcerative colitis severity in patients. Among tested strains, L. plantarum A736, isolated from traditional fermented foods, demonstrated remarkable efficacy in ameliorating dextran sulfate sodium (DSS)-induced colitis in a mouse model. Strikingly, multi-omics analysis revealed that L. plantarum A736 uniquely enhances choline acetyltransferase-expressing (ChAT+) neurons in the enteric nervous system and elevates acetylcholine (ACh) levels. Further analysis identified that L. plantarum A736 metabolizes tryptophan to produce indole-3-lactic acid (ILA), and supplementation with ILA significantly mimics this strains anti-inflammatory effects by activating ChAT+ neurons. Crucially, selective ablation of colonic ChAT+ neurons completely abolished the therapeutic benefits of both L. plantarum A736 and ILA, establishing their essential role in mediating these effects. These findings demonstrate for the first time that microbial metabolites can directly modulate enteric neuronal circuits to regulate gut immunity. The identified ILA-ChAT+ neuron axis represents a novel neuro-immune pathway that could be targeted for IBD treatment. Graphical abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=200 SRC="FIGDIR/small/673902v1_ufig1.gif" ALT="Figure 1"> View larger version (63K): org.highwire.dtl.DTLVardef@1e0d3c5org.highwire.dtl.DTLVardef@16f5481org.highwire.dtl.DTLVardef@152283org.highwire.dtl.DTLVardef@18cfad0_HPS_FORMAT_FIGEXP M_FIG C_FIG

microbiology↗

bioRxiv2025

Crosslinking of Lipid Nanoparticles Enhances the Delivery Efficiency and Efficacy of mRNA Vaccines

Lipid nanoparticles (LNPs) have enabled the effective delivery of RNA therapeutics and mRNA vaccines. However, their broader applications are limited by the suboptimal stability and endosomal escape efficiency. Here, we present an easy-to-adopt post-assembly crosslinking approach to enhance the structural and functional stability of mRNA LNPs. By leveraging a series of cholesterol derivatives and crosslinking methods, we induce crosslinks of the lipid components following mRNA LNP assembly to form the crosslinked LNPs (cLNPs). We systematically evaluated crosslinking parameters and identified optimal conditions that enhance both the physical stability and transfection efficiency of cLNPs. Our findings demonstrate that cLNPs exhibit improved structural integrity under storage and lyophilization conditions, as well as increased extracellular stability and endosomal escape efficiency, resulting in improved performance of mRNA LNPs both in vitro and in vivo. This crosslinking strategy represents a critical advance in LNP engineering, enabling more resilient LNPs and broadening the applicability of LNP-based therapies for gene therapy and vaccine delivery. Our work lays the foundation for developing next-generation LNPs with superior stability and delivery efficiency, broadening the impact of RNA therapeutics and vaccines.

bioengineering↗