Biology subjects

Liu, Q.

Publications and source records attributed to Liu, Q..

At least 145 records · Page 8Linked to original sources

bioRxiv2025

Representation, Alignment, and Generation: A Comprehensive Survey of Foundation Models for Non-Invasive Brain Decoding

The ability to decode human thoughts, intentions, and perceptions directly from non-invasive brain recordings holds transformative potential for healthcare, communication, and human-computer interaction. However, translating the safety and scalability of methods like Functional Magnetic Resonance Imaging (fMRI), Electroencephalography (EEG), and Magnetoencephalography (MEG) into real-world utility has traditionally been hindered by low signal-to-noise ratios, limited spatial-temporal resolution, and the difficulty to collect large-scale high-quality data from an individual user. Recently, the emergence of Foundation Models (FMs)--large-scale, pre-trained architectures--has expanded the feasible operating region of non-invasive brain decoding under controlled protocols, primarily through representation learning, neuro-semantic alignment, and strong generative priors. However, evidence for robust cross-subject and real-world deployment remains uneven, as many results are still demonstrated on limited cohorts or highly controlled settings. This survey provides a comprehensive overview of how FMs are redefining the boundaries of non-invasive brain decoding. We propose a unified methodological framework that synthesizes recent advancements into a coherent process: extracting robust, transferable representations from noisy neural signals; aligning these signals with the rich semantic spaces of pre-trained vision and language models; and leveraging powerful conditional generative priors to reconstruct high-fidelity outputs. We systematically review state-of-the-art applications across three key domains: visual reconstruction, language and speech decoding, and auditory processing. Furthermore, we critically examine the persisting challenges of computational efficiency, cross-subject generalization, and privacy governance. By mapping the current landscape and identifying key gaps, this work outlines a strategic research agenda aimed at transitioning FM-driven neurotechnology from laboratory proofs-of-concept to reliable, real-world applications.

neuroscience↗

bioRxiv2025

Energy Landscape Analysis Reveals Thalamic Modulation of Brain State Transitions During Movie Watching

Understanding how large-scale brain networks dynamically coordinate during complex, real-world experiences remains a crucial topic in contemporary neuroscience. Leveraging naturalistic paradigm, this study employed energy landscape modeling to characterize the temporal dynamics of brain state transitions and their functional significance. Using the Sherlock fMRI dataset, we identified two dominant brain states: a perception- and attention-oriented state, and an introspective, integrative state, each associated with distinct canonical brain networks. State transitions were differentially associated with the recruitment of these networks and manifested as either the "easy" or "hard" route, depending on the energy required for the transition. Notably, we demonstrated that the probability of specific state transitions robustly predicted inter-subject correlation within three clusters: subcortical-rich, salience/attention-rich, and DMN-rich regions. This indicates that flexible reorganization of cortical networks underlies shared neural engagement during ecologically valid experiences. Critically, the thalamus emerged as a central modulator, displaying activity changes and dynamic nucleus-specific thalamocortical connectivity changes tied to distinct transition profiles. Together, our findings shed light on the energetic and network mechanisms that orchestrate brain dynamics during naturalistic cognition and underscore the pivotal role of cortico-thalamo-cortical circuits in governing flexible, collective neural responses to richly structured real-life stimuli.

neuroscience↗

bioRxiv2025

GPTAnno: Ontology-tree-guided hierarchical cell type annotation based on GPT models for single-cell data

Cell type annotation is critical for interpreting single-cell transcriptomic data but remains challenging due to uncertain cellular clustering granularity and inconsistent labeling across studies. Here we present GPTAnno, an automated, ontology-tree-guided, uncertainty-aware, hierarchical cell type annotation method based on GPT models. GPTAnno directly handles gene expression matrices, integrates multi-resolution clustering with large language model reasoning constrained by the cell ontology to produce standardized, ontology-aware, and reproducible annotations with automatic resolution selection. GPTAnno selects optimal clustering resolutions based on the annotation distance on the ontology tree and quantifies annotation uncertainty to flag ambiguous clusters for expert review. Benchmarking across twelve large-scale datasets demonstrates GPTAnnos superior accuracy on annotating cell types across various species, tissues, and disease contexts against existing methods. Implemented in R and Python with Seurat and Scanpy compatibility, GPTAnno allows simple inputs to streamline the reproducible annotation, considerably reducing human efforts in repeated reclustering, assigning and examining the labels.

genomics↗

bioRxiv2025

Modifications to the gut microbiome alter bone matrix proteomics and fracture toughness at the cellular scale

The gut microbiome can regulate the strength of bone matrix but the specific changes in matrix function and composition are not yet understood. Here we introduce micropillar splitting to determine the fracture toughness of matrix at the cellular scale in concert with proteomic analysis of bone matrix when the gut microbiome was altered by oral antibiotics (ampicillin+neomycin). Male mice were divided into four groups (n = 3-4/group): (1) Unaltered (no alteration), (2) Continuous (alteration from 4-24 weeks), (3) Delayed (alteration from 16-24 weeks) and (4) Reconstituted (alteration from 1-16 weeks following by reconstitution). Micropillars (5 {micro}m diameter) were fabricated using focused ion beam milling on femur cross-sections in regions of matrix formed either before or after changes in the microbiome (16 weeks) (n = 4/group). Proteomics was used to identify differences in matrix protein composition. Micropillar fracture toughness differed by group (p < 0.001) and region (p < 0.001). Fracture toughness in the Unaltered group (1.34 {+/-} 0.32 MPa{surd}m, mean {+/-} SD) was substantially greater than the Continuous group (0.95 {+/-} 0.20) and the Delayed group (0.90 {+/-} 0.21) but not different from the Reconstituted group (1.22 {+/-} 0.25). Bone matrix formed from 16-24 weeks of age had lower fracture toughness than matrix formed before 16 weeks of age in all groups. Notably, micropillar splitting was substantially more precise than whole bone testing; whole bone notched 3-point bending tests did not detect differences in fracture toughness. Proteomics identified 46 extracellular matrix proteins that were differentially abundant between groups, including decreased abundance of Periostin (q < 0.001) and Emilin-1 (q < 0.001) in groups with impaired bone matrix. These findings demonstrate that modifications to the gut microbiome lead to changes in bone matrix throughout cortical bone volume and establish micropillar splitting as a high-precision approach for characterizing matrix material properties. Plain Language SummaryThis study investigated the effect of the gut microbiome on the brittleness of bone as a material. We used a new technique for measuring the brittleness of bone that involves making microscopic pillars on the bone surface and splitting them down the middle to measure a material property called fracture toughness. Mice with altered gut microbiomes had bone with reduced fracture toughness (by 35-40%). Restoration of an altered gut microbiome for two months reversed the effects. The effect of the microbiome on bone occurred throughout the whole bone and was not limited to regions of the bone formed after a change in the gut microbiome.

bioengineering↗

bioRxiv2025

The Human-In-the-Loop Drug Design Framework with Equivariant Rectified Flow

Advancements in AI-based drug design often face obstacles due to incomplete datasets, hindering progress to clinical trials. Human experts bring invaluable expertise and nuanced contextual understanding to drug design. There are two main difficulties in integrating human knowledge into the drug development process. First, human annotations are costly, and traditional machine learning algorithms require a large number of samples to be effective. Second, human experts are unable to accurately describe their expertise using natural language, nor can they precisely specify what kind of molecule is needed before seeing the generated molecules. To address these problems, we propose a new platform, called HIL-DD, for experts to infuse their experience by selecting molecules generated by AI that meet their criteria, or discarding those that do not. The core generative technology utilizes an Equivariant Rectified Flow Model (ERFM), which offers faster generation speeds than conventional diffusion models, enabling efficient human-AI collaboration. More importantly, we provide a user-friendly interface to ensure smooth and effective collaboration between human experts and AI systems. Rigorous experiments demonstrate that our system can produce 3D molecules that align with expert expectations in minimal interactive sessions. These generated molecules maintain drug-like qualities comparable to those created by current state-of-the-art models.

bioinformatics↗

bioRxiv2025

Multimodal AI Decodes Extreme Environment Functional Dark Matter Beyond Homology

Functional annotation of proteins from extreme environments represents a major bottleneck for bioresource discovery, as a vast reservoir of functional dark matter defies existing homology-based methods. We demonstrate that environmental pressures impart conserved physicochemical energy signatures that co-determine protein function with sequence and structure. Here we developed ACCESS, a multimodal graph neural network employing hierarchical contrastive learning with a tailored label-sample co-embedding to fuse energy, sequence, and structural information and overcome homology scarcity. ACCESS surpasses state-of-the-art methods including BLASTp and CLEAN in annotating low-identity enzymes. Applied to extreme environmental metagenomics, we constructed a function map of extremophile enzymes to expand the biocatalyst library, pinpointed functionally critical residues to guide rational design, and enabled large-scale, function-based macro-evolutionary analyses. This paradigm transcends the limitations of homology, illuminating protein dark matter and accelerating the exploration of the biospheres functional diversity for applications in biotechnology and therapeutic development.

bioinformatics↗

bioRxiv2025

Prognosis prediction using autophagy gene expression in osteosarcoma

Defining the transition from localized to metastatic osteosarcoma (OS) before overt dissemination is fundamental for improving survival. However, effective early diagnostic tools remain scarce, largely due to limited exploitation of the pre-metastatic tumor microenvironments own record of molecular barcode exposures encoded in cell-intrinsic stress states. We analyzed autophagy-dependent transcriptional datasets from 139 individuals to develop Auto-RS, a computational classifier that can integrate age and sex to deliver individualized risk management. The prognosis-interpretable features of Auto-RS recapitulate established molecular trajectories of metastasis at the single-cell level, capturing tumor cells intrinsic shift from proliferative to invasive states and revealing cooperative programs among cancer-associated fibroblasts and immune cells. More significantly, Auto-RS can expose chemotherapy vulnerabilities of newer drugs, providing a framework to prioritize therapeutics without direct testing in children. This framework brings a potential inflection point, where metastasis prediction and therapeutic stratification may converge to improve OS outcomes.

cell biology↗

bioRxiv2025

Dynamics-enhanced Molecular Property Prediction Guided by Deep Learning

Molecular property prediction (MPP) is a key challenge in computational biology and drug discovery. Traditional approaches mostly rely on feature representations yielded from static structures of molecules, ignoring their dynamic nature. The scarcity of dynamics data in public databases and the complexity of learning such high-dimensional data made it more difficult for dynamics-involved studies. Accordingly, we built a series of dynamics datasets for MPP tasks by performing comprehensive molecular dynamics (MD) simulations on different molecules. In addition, we proposed a dynamically enhanced molecular representation (DEMR) method with multiple sampling strategies for the dynamics frames. Besides, two deep learning pipelines were employed for mapping DEMR to the molecular properties in various tasks. Our models achieved better performance in different MPP tasks, with a practical guidance in efficient frame selection. This study highlights the significance of integrating MD data into MPP tasks and opens new avenues for structure-based drug design. The generated MD datasets are publicly available in a Zenodo repository at https://doi.org/10.5281/zenodo.15788151.

bioinformatics↗

bioRxiv2025

Language may be all omics needs: Harmonizing multimodal data for omics understanding with CellHermes

Decoding cellular systems requires integrating diverse omics data, yet most models are trained from scratch on a single modality, restricting generalization. Here we present CellHermes, a biological language model that leverages pretrained large language models (LLMs) to integrate multimodal forms of omics data, such as transcriptomic profile and PPI network, through natural language for better omics understanding. By reformulating these datasets into question-answer pairs, CellHermes emulates multiple self-supervised learning paradigms within a single, universal, language-based framework with LORA fine-tuning on existing natural language-LLM, achieving comparable or even better performance compared to current single-cell foundation models. Within this framework, CellHermes functions as an encoder, a predictor and an explainer for supporting a range of downstream tasks. For encoder, CellHermes encodes biological entities with embeddings that enable accurate network discovery and cross-dataset generalization; For predictor, CellHermes unifies heterogeneous downstream tasks by translating them into natural language question-answers pairs, allowing a single model to perform multi-task prediction by instruction fine-tuning; For explainer, CellHermes explains molecular mechanisms by combining attention analysis with text-based reasoning, leveraging the interpretability and reasoning capabilities of LLMs. We evaluated the ability of CellHermes as an encoder on representing genes and cells using 5 gene-level downstream tasks and 5 diverse single-cell datasets across different tissues compared with other single cell foundation models. We also evaluated CellHermes as a predictor on BioUniBench, a new benchmark of 10 tasks across 7 databases for LLMs, including perturbation response, cell fitness estimation, gene-disease association, and cell type identification. All benchmarks suggest CellHermes can unify multiple biological tasks within one model without compromising performance. By applying CellHermes as an explainer for a melanoma patients dataset, it can uncover potential key genes on tumor-reactive T cells. Together, CellHermes establishes natural language as a unifying medium for omics, offering a foundation that, in the future, potentially enables a more integrated and interpretable research loop of biological representation, prediction, and interpretability.

bioinformatics↗

bioRxiv2025

MicroRNA156 and its targeted SPL genes interact with the photoperiod, vernalization, and gibberellin pathways to regulate wheat heading time

Heading time has a large impact on adaptation to different environments and crop productivity. In this study, we characterized the effect of the endogenous age pathway on heading time and its interactions with the photoperiod and vernalization pathways in the leaves of tetraploid wheat (Triticum turgidum ssp. durum). Plants with reduced levels of microRNA156 or increased expression of its downstream targets, the SQUAMOSA PROMOTER BINDING PROTEIN-LIKE genes SPL3, SPL4, and SPL13 exhibited accelerated heading time, with stronger effects under suboptimal inductive conditions. Earlier heading was associated with the upregulation of miR172 and flowering-promoting genes VRN1, FUL2, and FT1 and the downregulation of flowering-repressing genes AP2L1 and VRN2. Additionally, we uncovered complex interactions among SPL, SQUAMOSA (VRN1 and FUL2) and DELLA proteins that modulate wheat heading time. We showed that DELLA proteins, which are negative regulators in the gibberellic acid pathway, can interact with SPL proteins reducing their ability to induce flowering. We also discovered previously unknown interactions between SQUAMOSA and DELLA proteins in wheat that compete with the DELLA-SPL interactions, likely reducing DELLAs ability to repress SPL3 and SPL4 activity. Since SPL3 and SPL4 directly promote VRN1 and FUL2 transcription, these interactions generate a positive regulatory feedback loop that accelerates wheat heading time. Finally, we developed dominant miR156-resistant alleles rSPL3, rSPL4, and rSPL13 that accelerate wheat heading time under both optimal and suboptimal inductive conditions. These publicly available genetic resources can be used to fine tune heading time and improve wheat adaptation to changing environments. Significance statementWheat heading time is critical for adaptation to diverse environments. We generated dominant mutations for the SPL3, SPL4 and SPL13 genes that accelerate heading time. Different combinations of these mutations can be used to modulate heading time and improve wheat adaptation to changing environments.

plant biology↗

bioRxiv2025

Dynamic epigenetic and transcriptional regulatory network in pepper fruit development and ripening

Pepper (Capsicum) is among the most widely cultivated and consumed vegetable crops worldwide. Although extensive studies in model fruit crops such as tomato have provided insights into the genetic and epigenetic regulation of fruit development and ripening, comparable knowledge in pepper remains limited. Here, we employ a multi- omics approach to investigate transcriptome and epigenome dynamics during pepper fruit pericarp development and ripening. Our analyses reveal coordinated changes in chromatin accessibility and histone modifications accompanying transcriptome reprogramming, with changes in H3K27ac closely associated with chromatin accessibility dynamics, and H3K27me3 dynamics potentially contributing to the transition from fruit growth to ripening. Moreover, construction of the transcriptional regulatory network underlying pepper fruit development and ripening suggests that several ripening regulators identified in the climacteric tomato may also play critical roles in governing ripening in the non-climacteric pepper. During ripening, DNA methylation, particularly in the CG and CHG contexts, undergoes global demethylation, especially at promoter regions, which is accompanied by increased chromatin accessibility and likely enhances transcription factor binding activity. We further demonstrate transcriptional and epigenetic regulation of carotenoid and ascorbic acid (vitamin C) biosynthesis pathways. Collectively, this study provides a comprehensive resource for mechanistic dissection and comparative analysis of fruit development and ripening, with practical implications for improving key fruit traits.

plant biology↗

bioRxiv2025

Umbilical Cord Blood Cell Transcriptional and Methylation Signatures at Birth Are Associated with BPD Development.

Bronchopulmonary dysplasia (BPD) is the most common respiratory disease in preterm infants born at less than 28 weeks gestation. Most existing clinical prediction models for BPD show limited accuracy in predicting BPD development when validated using external data, stressing the need for novel biomarkers to identify at-risk infants for early and effective interventions. We leveraged existing frozen umbilical cord blood samples from the Northwestern University Cord Blood Biobank (NUCord) to perform parallel transcriptional and DNA methylation profiling. BPD-associated differentially expressed genes (DEGs) in our cohort included markers previously established in clinical and animal BPD studies, such as genes related to NF-{kappa}B signaling and immune responses. We also identified that BPD development is associated with disrupted methylation signatures in microRNA genes and genes associated with glucose metabolism. Our results suggest that BPD development is associated with distinct transcriptomic and epigenetic signatures when compared with healthy term and preterm infants. These signatures may represent biomarkers measurable at birth that predict BPD development during a time window when preventative or therapeutic interventions could be applied.

genomics↗

bioRxiv2025

BASSL-MI: Batch-Agnostic Self-Supervised Learning Uncovers Clinically Relevant Tumor Niches in Multiplexed Imaging

Multiplexed imaging enables rich, high-resolution characterization of the tumor microenvironment but relies on labor-intensive and error-prone cell segmentation and phenotyping pipelines. We present BASSL-MI, a batch-agnostic, self-supervised framework for discovering tissue niches directly from multiplexed imaging data. BASSL-MI operates directly on image patches, eliminating the need for explicit cell segmentation while mitigating image-, sample-, or batch-specific artifacts. Built on a modified contrastive block disentanglement architecture, BASSL-MI learns dual latent representations that separate biologically informative features from batch-dependent factors through spatially guided augmentations and batch-invariance objectives. Applied to a 56-marker colorectal cancer CODEX dataset, BASSL-MI-trained embeddings markedly reduce image-specific variability and recover biologically interpretable spatial niches. Notably, it uncovers CD20-rich follicular regions associated with improved survival, outperforming published findings from cell segmentation-driven clustering. This work demonstrates that self-supervised, patch-based learning can capture clinically relevant spatial organization within tumor microenvironments, advancing toward automated, non-cell-based analysis of multiplexed imaging data.

bioinformatics↗

bioRxiv2025

StoPred: Accurate Stoichiometry Prediction for ProteinComplexes Using Protein Language Models and GraphAttention

Proteins often function as part of complexes, and the specific stoichiometry of these assemblies is critical for their biological roles, but experimental determination of assembly composition remains challenging and existing computational methods for stoichiometry prediction are limited. Existing approaches rely on template-based searches or require predefined stoichiometry for structure prediction, hampering their applicability to proteins without close homologs or known assembly states. Recent advances using protein language models (pLM) have enabled sequence-based prediction of homo-oligomer stoichiometry, but these methods are not applicable to hetero-oligomeric complexes and do not fully leverage inter-subunit relationships. Here, we present StoPred, a method that predicts the stoichiometry of protein complexes by integrating pLM embeddings with a graph attention network to model subunit-level interactions. StoPred infers stoichiometry directly from sequence or structure features for both homo-and hetero-oligomers, without requiring template assemblies or predefined composition. We benchmark StoPred against deep learning-based and template-based methods, and show that it achieves improved accuracy and efficiency across curated and blind datasets, with up to 16% and 41% higher top-1 accuracy for homomeric and heteromeric complexes, respectively, compared to the strongest prior method on our held-out test dataset. More importantly, StoPred is the first deep learning-based method capable of accurately predicting the stoichiometry of hetero-oligomeric complexes.

bioinformatics↗

bioRxiv2025

ACE2 Phosphorylation Modulates Angiogenesis via the Activator Protein-1

BackgroundAngiogenesis plays a crucial role in organ development. However, aberrant blood vessel growth is involved in various diseases, including tumors and neovascular eye diseases. Angiotensin-converting enzyme 2 (ACE2) is a critical enzyme regulating the health of cardiovascular system, and its post-translational modifications (PTMs) are crucial to determine ACE2 expression level and activity. Here, we studied how the PTM of ACE2 in vascular endothelial cells (ECs) affect pathological retinal neovascularization and tumor angiogenesis. MethodsThe angiogenic capabilities of ECs were assessed by tube formation, sprouting assays, and 5-ethynyl-2-deoxyuridine (EdU) and filopodia staining. EC angiogenesis was examined by poteome profiler array and aortic ring assays in vitro and by the oxygen-induced retinopathy (OIR) and tumor angiogenesis models in vivo. High-throughput screening involving data from RNA-seq, ATAC-seq, and ChIP-seq were used to explore the epigenetic and transcriptional regulations of pro-angiogenic genes regualtged by ACE2 PTMs. ResultsACE2 Ser-680 dephosphorylation in connection with Lys-788 ubiquitination increased EC angiogenic phenotype, which were manifested by aberrant vascularization in mouse OIR and tumor models. ACE2 Ser-680 dephosphorylation led to the activation of activator protein-1 (AP-1), which transactivated multiple genes involved in angiogenesis. AP- 1 inhibition mitigated such angiogenesis in vivo. ConclusionOur findings show a novel PTM mechanism of ACE2 involved in pathological angiogeneis. Specifically, ACE2 Ser-680 dephosphorylation facilitated AP-1 transactivation of the downstream pro-angiogenic genes in ECs.

cell biology↗

bioRxiv2025

Targeting the host factor HGS-viral membrane protein interaction in coronavirus infection

While current antivirals primarily target viral proteins, host-directed strategies remain underexplored. Here, we performed a genome-wide CRISPRi screening to identify the host protein, Hepatocyte Growth Factor-Regulated Tyrosine Kinase Substrate (HGS), as essential for the pan-coronaviruses infection both in vitro and in vivo. Mechanistically, HGS directly interacts with the viral membrane (M) protein, facilitating its trafficking to the ER-Golgi intermediate compartment (ERGIC) for virion assembly. Conversely, HGS deficiency caused M retention in the ER, blocking assembly. Leveraging this interaction, we designed M-derived peptides and screened over 5,000 FDA-approved drugs, identifying riboflavin tetrabutyrate (RTB). Both the peptides and RTB bind HGS and disrupt its interaction with the M protein, leading to M retention in the ER and subsequent blockade of virion assembly. These agents demonstrated broad anti-pan-coronavirus activity in vitro and in vivo. Collectively, our findings establish HGS as a druggable host target and identify RTB as a promising broad-spectrum antiviral candidate.

microbiology↗

bioRxiv2025

Systematic characterization of existing and novel inducible transgenic systems in human pluripotent stem cells after prolonged differentiation

The ability to control transgene expression both temporally and quantitatively in human-relevant cells and tissues is a cornerstone of biomedical research. Additionally, precise transgene control is crucial for optimizing human cell-based gene therapies. Human pluripotent stem cells (hPSCs) have facilitated major advances in disease modeling and the potential for regenerative medicine. Still, they are significantly limited by the lack of inducible transgenic systems that avoid silencing but maintain robust inducibility after differentiation to defined cell lineages. Here we systematically characterize the leakiness, inducibility, and tunability of multiple existing and novel transgenic systems in hPSCs and differentiated macrophages and microglia. Notably, we report the application of a small molecule-mediated splicing switch (Xon) that allows for tunable transgene expression both before and after differentiation, without the large protein tags required for current state-of-the-art degron-based methods. We use Xon to achieve tight control of reporter genes, overexpression of multiple neurodegeneration-associated genes, and Cas9-mediated genome editing. We also characterize the current limitations of this system and describe approaches that can alleviate some of these limitations. By assessing multiple transgenic systems whose inducibility spans the transcriptional, post-transcriptional, and post-translational levels, we highlight and improve upon a major technical challenge that hinders basic, translational, and clinical research in physiological human-based systems.

bioengineering↗

bioRxiv2025

Lipid peroxidation intrinsically induces mitochondrial iron overload via Bach1-HO-1 signaling to promote cardiac ferroptosis

BackgroundLipid peroxidation and iron accumulation are hallmarks of ferroptosis, a form of cell death characterized by iron-dependent oxidative damage to cellular membranes. However, the molecular link between lipid peroxidation and iron overload in the execution of ferroptosis remains elusive. Moreover, the pathophysiological implications of the interaction between lipid peroxidation and iron overload in cardiac homeostasis and remodeling are also unknown. MethodsWe assessed the role of lipid peroxidation in mediating cardiac iron overload and ferroptosis using genetic mouse models. We also performed molecular and cellular biology studies to elucidate the mechanisms by which lipid peroxidation regulates iron homeostasis and ferroptosis signaling in cardiomyocytes. ResultsCardiomyocyte-specific ablation of Gpx4 (glutathione peroxidase 4), a key suppressor of lipid peroxidation, promoted iron overload and ferroptosis in the heart, leading to dilated cardiomyopathy. Mice with heterozygous Gpx4 knockout were also predisposed to adverse cardiac remodeling and dysfunction following pressure overload. Mechanistically, elevated lipid peroxidation due to GPX4 inactivation intrinsically induced iron overload by promoting the nuclear export of Bach1 and subsequent induction of heme oxygenase-1 (HO-1). Genetic and pharmacologic inhibition of HO-1 markedly attenuated iron overload and ferroptosis in cardiomyocytes and rescued dilated cardiomyopathy associated with Gpx4 deficiency. Moreover, we identified HO-1 mitochondrial translocation as a key mechanism driving mitochondrial iron overload and ferroptosis. Targeted inhibition of mitochondrial iron overload or lipid peroxidation abrogated cardiac ferroptosis and pathological remodeling induced by Gpx4 deficiency. ConclusionsThese findings identified a mechanistic link between lipid peroxidation and iron overload via the Bach1-HO-1 signaling pathway, revealing new regulators and molecular targets for cardiac ferroptosis.

cell biology↗