Biology subjects

Liu, T.

Publications and source records attributed to Liu, T..

At least 145 records · Page 8Linked to original sources

bioRxiv2025

Tracing LYVE1+ peritoneal fluid macrophages unveils two paths to resident macrophage repopulation with differing reliance on monocytes

Mouse resident peritoneal macrophages, called large cavity macrophages (LCM), arise from embryonic progenitors that proliferate as mature, CD73+Gata6+ tissue-specialized macrophages. After injury from irradiation or inflammation, monocytes are thought to replenish CD73+Gata6+ LCMs through a CD73-LYVE1+ LCM intermediate. Here, we show that CD73-LYVE1+ LCMs indeed yield Gata6+CD73+ LCMs through integrin-mediated interactions with mesothelial surfaces. CD73-LYVE1+ LCM repopulation of the peritoneum was reliant upon and quantitatively proportional to recruited monocytes. Unexpectedly, fate mapping indicated that only [~]10% of Gata6-dependent LCMs that repopulated the peritoneum after injury depended on the LYVE1+ LCM stage. Further supporting nonoverlapping lifecycles of CD73-LYVE1+ and CD73+Gata6+ LCMs, in mice bearing a paucity of monocytes, Gata6+CD73+ LCMs rebounded after ablative irradiation substantially more efficiently than their presumed LYVE1+ or CD73- LCM upstream precursors. Thus, after inflammatory insult, two temporally parallel pathways, each generating distinct differentiation intermediates with varying dependencies on monocytes, contribute to the replenish hment of Gata6+ resident peritoneal macrophages.

immunology↗

bioRxiv2025

CHIEF: An Attention-based Ensemble Learning Framework for Functional Protein Design

Protein de novo design is a longstanding challenge in biological field due to the multifaceted physicochemical properties and complexities of proteins. Several deep learning models have emerged for protein de novo design. However, each model exhibits distinct strengths and limitations because of the differences in training strategies or neural network architectures. Therefore, improving the effectiveness of protein design, particularly for functional protein, requires further improvement. To tackle this challenge, we developed CHIEF (Chimera Ensemble Inverse Folding), an attention-based ensemble framework that integrates five pretrained base models (ProteinMPNN-vanilla, ProteinMPNN-soluble, ESM-IF, Frame2seq and PiFold) to leverage their complementary strengths and mitigate limitations. Compared with the base models, CHIEF significantly improved the sequence recovery rate by 16.6-28.0%, while reducing the prediction perplexity by 22.7-34.6%. CHIEF also exhibited superior protein-designing capacity in variable sequence- and structure-based metrics and in large and complex proteins. CHIEF dynamically captured the semantic information from base models in a context-dependent manner and was minimally affected by the ablation of each base model. More importantly, CHIEF demonstrates real-world applicability to design functional malate dehydrogenase (MDH), achieving a 100% success rate. In summary, our study develops an ensemble deep learning model that improves the efficacy of protein sequence design, and will be a valuable platform for protein engineering and drug development.

bioengineering↗

bioRxiv2025

Automated Discovery of Therapeutic Biomaterial for Renally Impaired Hyperuricemia Patients by Natural Language Processing and Machine Learning

The exponential growth of scientific publications presents opportunities for researchers to identify valuable knowledge, especially in the highly interdisciplinary field --- biomaterials, where exploiting possible connections between unmet clinical needs and materials properties from literatures is crucial. However, with traditional literature reading, it is extremely challenging to marry unmet clinical needs with existing materials reported for different applications or other purposes. Here, to provide a not-renally cleared therapeutics for renally impaired hyperuricemia patients, we designed a multi-tiered framework MatWISE that fuses state-of-the-art natural language processing, semantic relationship mapping, and machine learning to automate the complex process of material discovery from a sea of scientific literatures published until December of 2022, and successfully identified and optimized {delta}-MnO2 into an orally administered, nonabsorbable uric acid (UA) lowering biomaterial. {delta}-MnO2 had superior serum and urine UA-lowering effect in three hyperuricemia mouse models, by comparing with a standard of care drug. {delta}-MnO2 is highly promising to serve as a safe and effective UA-lowering drug for renally impaired hyperuricemia patients. We demonstrated a new research paradigm for biomaterials that combining state-of-the-art machine learning techniques and a handful of experiments to discover a translationally relevant material from the massive existing research, for an unmet clinical need.

biochemistry↗

bioRxiv2025

Lamin A/C safeguards replication initiation by orchestrating chromatin accessibility and PCNA recruitment

Lamin A/C, a nuclear lamina protein, is essential for maintaining nuclear architecture, organizing chromatin and preserving genomic stability. However, its role in regulating DNA replication remains unclear. This study investigates how Lamin A/C regulates proper activation of replication origins by orchestrating the chromatin structure and recruitment of proliferating cell nuclear antigen (PCNA). We demonstrate that Lamin A/C stabilizes replication domains (RDs) by restricting chromatin mobility, preserving spatial organization, and maintaining accessibility. Furthermore, Lamin A/C interacts with PCNA and sequesters a pool of PCNA, to regulate its recruitment to replication machinery. The loss of Lamin A/C leads to an increase in chromatin dynamics, RD accessibility, and PCNA availability at RDs, which together trigger excessive activation of replication origins, leading to replication stress and DNA damage. These disruptions prolong the S phase and compromise genome stability, highlighting Lamin A/C as a critical gatekeeper of balanced replication initiation. Our findings reveal Lamin A/Cs dual role in chromatin organization and replication machinery regulation, offering valuable insights into its involvement in replication-associated diseases and potential therapeutic opportunities through targeting replication dynamics.

biophysics↗

bioRxiv2025

spVelo: RNA velocity inference for multi-batch spatial transcriptomics data

RNA velocity has emerged as a powerful tool to interpret transcriptional dynamics and infer trajectory from snapshot datasets. However, current methods fail to utilize the spatial information inherent in spatial transcriptomics and lack scalability in multi-batch datasets. Here, we introduce spVelo (spatial Velocity inference), a scalable framework for RNA velocity inference in multi-batch spatial transcriptomics data. Our model comparison studies show that spVelo compares favorably to existing methods regarding velocity consistency, transition accuracy, and direction correctness across expression levels, spatial graphs in each batch, and MNN graphs between batches. Furthermore, spVelo supports several downstream applications, including uncertainty quantification, complex trajectory pattern discovery, biologically significant state driver marker identification, gene regulatory network inference and temporal cell-cell communication inference. In conclusion, spVelo has the potential to provide deeper insights into complex tissue organization and underscore their biological mechanisms based on spatially-resolved patterns.

bioinformatics↗

bioRxiv2025

Point-Of-Need One-Pot Multiplexed RT-LAMP Test For Detecting Three Common Respiratory Viruses In Saliva

Respiratory viral infections pose a significant global public health challenge, partly due to the difficulty in rapidly and accurately distinguishing between viruses with similar symptoms at the point of care, hindering timely and appropriate treatment and limiting effective infection control and prevention efforts. Here, we developed a multiplexed, non- invasive saliva-based, reverse transcription loop-mediated isothermal amplification (RT- LAMP) test that enables the simultaneous detection of three of the most common respiratory infections, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), Influenza (Flu), and respiratory syncytial virus (RSV), in a single reaction via specific probes and monitored in real-time by a machine-learning-enabled compact analyzer. Our results demonstrate that the multiplexed assay can effectively detect three target RNAs with high accuracy. Further, testing with spiked saliva samples showed strong agreement with reverse transcription polymerase chain reaction (RT-PCR) assay, with area under the curve (AUC) values of 0.82, 0.93, and 0.96 for RSV, Influenza, and SARS-CoV-2, respectively. By enabling the rapid detection of respiratory infections from easily collected saliva samples at the point of care, the device presented here offers a practical and efficient tool for improving outcomes and helping prevent the spread of contagious diseases. SignificanceThis research presents an innovative approach to respiratory infection diagnostics by combining a one-pot isothermal molecular test with machine learning-based analysis to simultaneously detect SARS-CoV-2, Influenza, and RSV in saliva samples. The battery- powered portable analyzer features novel machine-learning-assisted fluorescence detection for multiplexed reporter quantification, eliminating the need for traditional filter- based optical components and enabling adaptation to new targets without hardware changes. The test demonstrates high accuracy in detecting single and co-infections in spiked saliva samples, providing a rapid, cost-effective point-of-need solution. This tool can expand testing access, improve patient outcomes, and support more effective disease control, particularly in resource-limited or decentralized healthcare settings.

bioengineering↗

bioRxiv2025

Neuropilin 2 stabilises adherens junctions and protects against endothelial activation by promoting the interaction between VE cadherin and p120 catenin

The mechanosensing properties of endothelial cell-cell junctions are essential for vascular beds to respond to the mechanical forces exerted by blood flow. In states of disturbed flow, endothelial cells (ECs) become activated and transition to a pro-inflammatory, atheroprone phenotype. Here, we investigated the role of transmembrane glycoprotein neuropilin 2 (NRP2) in maintaining adherens junction integrity using cultured immortalised mouse ECs and a genetically modified mouse model to demonstrate the effects of an endothelial-specific deletion of Nrp2 in vivo. We reveal that, akin to its ortholog, Nrp1, Nrp2 exists as a constituent of adherens junctions, maintaining surface availability of VE cadherin by promoting its interaction with p120 catenin. As a consequence, endothelial knockout mice (Nrp2flfl.ECKO) display hyperpermeable retinal vasculature during development. Nrp2 depletion was subsequently found to activate key pro-inflammatory cytokines and adhesion molecules known to participate in the progression of atherogenesis, in addition to increased immune cell attachment aortic plaque development. These findings describe a role for Nrp2 in maintaining junctional signalling in ECs, protecting against endothelial activation during a state of vascular disease.

cell biology↗

bioRxiv2025

Pre-training Genomic Language Model with Variants for Better Modeling Functional Genomics

Sequence-to-function models can predict gene expression from sequence data and be used to link genetic information with transcriptomics data to understand regulatory processes and their effects on complex phenotypes. The genomic language models are pre-trained with large-scale DNA sequences and can generate robust representations of these sequences by learning the genomic context. How-ever, few studies can estimate the predictability of gene expression levels and bridge these two classes of models together to explore individualized gene expression prediction. In this manuscript, we propose UKBioBERT as a DNA language model pre-trained with genetic variants from UK BioBank. We demonstrate that UKBioBERT generates informative embeddings capable of identifying gene functions, and improving gene expression prediction in cell lines, thereby enhancing our understanding of gene expression predictability. Building upon these embeddings, we combine UKBioBERT with state-of-the-art sequence-to-function architectures, Enformer and Borzoi, to create UKBioFormer and UKBioZoi. These models exhibit better performance in predicting highly predictable gene expression levels and can be generalized across different cohorts. Furthermore, UKBioFormer effectively captures the relationship between genetic variants and expression variations, enabling in-silico mutation analyses and eQTL identification. Collectively, our findings underscore the value of integrating genomic language models and sequence-to-function approaches for advancing functional genomics research.

bioinformatics↗

bioRxiv2025

Structured Gamma Spikes in Mouse Anterodorsal Thalamus

Gamma-frequency oscillations ([~]30-160 Hz) are a hallmark of neuronal syn-chronization, yet the fine-scale temporal arrangement of spikes within individual gamma cycles remains poorly understood. Here, we examine head direction (HD) cells in the mouse anterodorsal thalamic nucleus (ADn)--a circuit distinguished by prominent high-gamma activity--to uncover the principles governing gamma coordination. We reveal two fundamental mechanisms: (i) a stable anatomical gradient in which neurons firing earlier in the gamma cycle exhibit longer anticipatory time intervals (ATIs), and (ii) a dynamic rate-phase shift whereby spike timing advances as the animals head aligns with a neurons preferred direction. Together, these results delineate a spatiotemporally structured framework for gamma synchronization, advancing our understanding of the functional roles and circuit mechanisms underlying gamma rhythms in local brain networks.

neuroscience↗

bioRxiv2025

Spatial distribution of the proteome in human body and cancers

A comprehensive spatial distribution of the proteome in human body and cancers is fundamental for understanding human biology and diseases including cancers. Here, we present an anatomically resolved human proteome derived from 1781 benign and malignant samples from 58 major tissue types encompassing 251 specific tissues and 25 carcinomas. Based on a spectral library covering over 75% of the human protein-coding genes with 208 understudied and 82 missing proteins characterized, we quantified over 13,000 proteins in these samples using data-independent acquisition proteomics. This data resource presents the so far most comprehensive quantitative proteomic landscape of human tissues and common carcinomas. It allows systematic evaluation of tissue-specific drug responses, identification of drug candidates that may be repurposed as antineoplastics, and discovery of novel targets for anticancer therapy. This resource, available as an online knowledgebase, refines our knowledge of spatial distribution of the human proteome and tumor-specific protein modulation.

systems biology↗

bioRxiv2025

In harmony? A scoping review of methods to combine multiple 16S amplicon data sets

Robust evidence on relationships between the human microbiome and health are critical for understanding and improving the human condition. However, there is little information about methodological approaches to combine and analyze multiple microbiome data sets. To address this gap, we conducted a scoping review of studies that combine sequencing data from multiple data sets to understand the study objectives, data sources and selection, feature -table assembly methods, and analyses. References were identified through a systematic search of literature published between 2011 and 2022. Our final review included 60 articles. Despite the wide-spread use of the word "meta-analysis," we found that only 24 studies used a systematic process to select their data sets, suggesting multiple meanings of the term within the field. While more than two thirds of studies used at least one publicly available data source, 19 had to request data from the original authors. Most studies (60%) combined data sets from multiple disjoint hypervariable regions. The number of hypervariable regions combined was not associated with the table construction method, but feature table construction method and the number of hypervariable regions influenced analytical resolution. Our results suggest the microbiome community needs to examine the use of terminology and analytic approaches for combining data sets; that additional work is needed to explore the impact of data source on bias in combined studies; and invite an independent evaluation of the methods used for feature table construction across disjoint regions.

microbiology↗

bioRxiv2025

Arp2/3-mediated bidirectional actin assembly by SPIN90 dimers in metazoans

Branched actin networks nucleated by the Arp2/3 complex play critical roles in various cellular processes, from cell migration to intracellular transport. However, when activated by WISH/DIP/SPIN90 family proteins, Arp2/3 nucleates linear actin filaments. Unexpectedly, we found that human SPIN90 is a dimer that can nucleate bidirectional actin filaments. To understand the basis for this, we determined a 3 [A] resolution structure of human SPIN90-Arp2/3 complex nucleating actin filaments. Our structure shows that SPIN90 dimerises via a 3-helix bundle and interacts with two Arp2/3 complexes. Each SPIN90 molecule binds both Arp2/3 complexes to promote their activation. Our analysis demonstrates that single filament nucleation by Arp2/3 is mechanistically more like branch formation than previously appreciated. The dimerisation domain in SPIN90 orthologues is conserved in metazoans, suggesting that this mode of bidirectional nucleation is a common strategy to generate anti-parallel actin filaments.

biophysics↗

bioRxiv2025

YTHDF3 promotes the progression of gastric cancer by activating Wnt/β-catenin signaling pathway via targeting NEK7

Gastric cancer (GC) is a widespread and deadly malignancy among global digestive tract tumors, alarmingly ranked the fifth most common cancer in terms of incidence and mortality. This study used immunohistochemistry and RT-qPCR to reveal a significant upregulation of YTHDF3 expression in GC tissues compared with that of adjacent normal tissues. This increased expression of YTHDF3 was closely associated with the incidence of lymph node metastasis and the progression to more advanced TNM stages. In vitro experiments demonstrated that YTHDF3 was crucial in promoting the aggressive characteristics of GC cells by enhancing their ability to proliferate, migrate, and invade, while simultaneously inhibiting apoptosis. The role of YTHDF3 in the progression of GC was further confirmed using the BALB/c nude mice subcutaneous tumor model, which showed a slowed tumor growth after YTHDF3 knockdown. This study further confirmed through Co-IF RIP-qPCR, western blotting, and rescue studies that YTHDF3 promotes GC progression by targeting NEK7 to regulate the Wnt/{beta}-catenin signaling pathway. In conclusion, our pivotal findings highlight the YTHDF3/NEK7/Wnt/{beta}-catenin axis as a promising and accessible therapeutic target for the development of new treatments aimed at improving clinical outcome of patients with GC.

cancer biology↗

bioRxiv2025

UNICORN: Towards Universal Cellular Expression Prediction with an Explainable Multi-Task Learning Framework

Sequence-to-function analysis is a challenging task in human genetics, especially in predicting cell-type-specific multi-omic phenotypes from biological sequences such as individualized gene expression. Here, we present UNICORN, a new method with improved prediction performances than the existing methods. UNICORN takes the embeddings from biological sequences as well as external knowledge from pre-trained foundation models as inputs and optimizes the predictor with carefully-designed loss functions. We demonstrate that UNICORN outperforms the existing methods in both gene expression prediction and multiomic phenotype prediction at the cellular level and the cell-type level, and it can also generate uncertainty scores of the predictions. Moreover, UNICORN is able to link personalized gene expression profiles with corresponding genome information. Finally, we show that UNICORN is capable of characterizing complex biological systems for different disease states or perturbations. Overall, embeddings from foundation models can facilitate the understanding of the role of biological sequences in the prediction task, and incorporating multi-omic information can enhance prediction performances.

bioinformatics↗

bioRxiv2025

Psychedelic-mediated Reversal of General Anesthesia and Restoration of Brain Dynamics in Rat

Serotonergic psychedelics enhance neurophysiological complexity and the repertoire of brain states, whereas general anesthetics produce opposite effects. Serotonergic psychedelics are also known to increase wakefulness and reduce sleep time in rodents. Therefore, we hypothesized that 2,5-dimethoxy-4-iodopamphetamine (DOI), a serotonergic psychedelic, will reverse general anesthesia and restore neurophysiological conditions associated with normal wakefulness. We demonstrate that intravenous administration of DOI in rats under general anesthesia induced wakefulness despite ongoing delivery of the anesthetics, propofol or isoflurane. Behavioral arousal was accompanied by recovery of directional and non-directional high gamma (125-165Hz) functional connectivity and restoration of functional brain network structure. These behavioral and neurophysiological effects were blocked by a 5-HT2A antagonist, volinanserin. Intravenous administration of a non-psychedelic 5-HT2A agonist, lisuride, failed to restore wakefulness or brain dynamics in anesthetized rats. To our knowledge, these results provide the first evidence of psychedelic-mediated reversal of general anesthesia and concurrent restoration of brain dynamics associated with normal wakefulness.

neuroscience↗

bioRxiv2025

spEMO: Exploring the Capacity of Foundation Models for Analyzing Spatial Multi-Omic Data

Recent advances in pathology foundation models (PFMs), which are pretrained on large-scale histopathological images, have significantly accelerated progress in disease-centered applications. In parallel, spatial multi-omic technologies collect gene and protein expression levels at high spatial resolution, offering rich understanding of tissue context. However, current models fall short in effectively integrating these complementary data modalities. To fill in this gap, we introduce spEMO, a novel computational system that unifies embeddings from pathology foundation models and large language models (LLMs) to analyze spatial multi-omic data. By incorporating multimodal representations and information, spEMO outperforms models trained on single-modality data across a broad range of downstream tasks, including spatial domain identification, spot-type classification, whole-slide disease-state prediction and interpretation, inference of multicellular interactions, and automated medical report generation. The outstanding performances of spEMO in these tasks demonstrate its strength in both biological and clinical applications. Additionally, we propose a new evaluation task, known as multi-modal alignment, to assess the information retrieval capabilities of pathology foundation models. This task provides a principled benchmark for evaluating and improving model architectures. Collectively, spEMO represents a step forward in building holistic, interpretable, and generalizable AI systems for spatial biology and pathology.

bioinformatics↗

bioRxiv2025

KAT6A acetylation drives metabolic adaptation to mediate cellular growth and mitochondrial metabolism through AMPK interaction

Diets influence metabolism and disease susceptibility, with lysine acetyltransferases (KATs) serving as key regulators through acetyl-CoA. We have previously demonstrated that a ketogenic diet alleviates cardiac pathology, though the underlying mechanisms remain largely unknown. Here we show that KAT6A acetylation is crucial for mitochondrial function and cell growth. Proteomic analysis revealed that KAT6A is acetylated at lysine (K)816 in the hearts of mice fed a ketogenic diet under hypertension, which enhances its interaction with AMPK regulatory subunits. RNA-sequencing analysis demonstrated that the KAT6A acetylation-mimetic mutant stimulates AMPK signaling in cardiomyocytes. Moreover, the acetylation-mimetic mutant mitigated phenylephrine-induced mitochondrial dysfunction and cardiomyocyte hypertrophy via AMPK activation. However, KAT6A-K816R acetylation-resistant knock-in mice unexpectedly exhibited smaller hearts with enhanced AMPK activity, conferring protection against neurohumoral stress-induced cardiac hypertrophy and remodeling. These findings indicate that KAT6A regulates metabolism and cellular growth by interacting with and modulating AMPK activity through K816-acetylation in a cell type-specific manner.

cell biology↗

bioRxiv2025

Eye metrics are a marker of visual conscious awareness and neural processing in cerebral blindness

Damage to the primary visual pathway can cause vision loss. Some cerebrally blind people retain degraded vision or sensations and can perform visually guided behaviors. These cases motivate investigation and debate on blind field conscious awareness and linked residual neural processing. A key challenge in this research is that subjective measures of blind field visual conscious awareness can be misleading. Alternatively, eye metrics, including pupil size and eye movements are promising objective markers of conscious awareness and brain activity. In this study, we examined stimulus-evoked changes in pupil size, blinking, and microsaccades in the sighted and blind field of cerebrally blind participants. Using standard analysis and innovative machine learning methods, our findings support that eye metrics can infer blind field conscious awareness, even when behavioral performance on a visual perception task indicated otherwise. Furthermore, these eye metrics were linked to blind field visual stimulus-evoked occipital cortical field potentials. These findings support recording eye metrics in cerebral blindness and highlight potential clinical applications, including tracking the recovery of conscious vision and visual neural processing.

neuroscience↗