Biology subjects

Jin, R.

Publications and source records attributed to Jin, R..

At least 19 recordsLinked to original sources

bioRxiv2026

JAK1/2 Inhibition Restores Tissue Integrity by Rewiring Type 1/2 Immunity Toward Targetable IL-17 Responses

Effective immune responses require coordinated communication between immune and epithelial cells to eliminate pathogens while limiting tissue damage. Disruption of this balance drives chronic inflammation and progressive tissue destruction. The Janus kinase (JAK) pathway coordinates cytokine signaling across these compartments and has emerged as a therapeutic target in inflammatory diseases. However, the consequences of JAK inhibition within a complex inflammatory environment remain incompletely understood. Here, we dissect the molecular, cellular, immunological and phenotypical effects of topical JAK1/2 inhibition in a chronic model of tissue-destructive skin inflammation driven by barrier dysfunction and bacterial superinfection. Single-cell transcriptomics and functional profiling revealed that JAK1/2 blockade selectively suppressed pathogenic CD8 T cells and their IFN-{gamma} production while shifting Th2-polarized CD4 T cells toward a Th17 phenotype, preserving type 3 immunity consisting of IL-17A-producing {gamma}{delta}-T cells and neutrophil responses. This immune rebalancing occurs rapidly and concomitantly with the restoration of epithelial barrier integrity through tight junction reinforcement, followed by recovery of antimicrobial defenses and reduced bacterial burden. Combination of JAK1/2 and IL-17A blockade further improved tissue homeostasis by targeting residual type 3-driven pathology. Thus, JAK1/2 inhibition acts as selective immunomodulator rather than a broad immunosuppressant, targeting pathogenic type 1/2 responses while preserving type-3-associated antimicrobial defense programs. These findings reveal how JAK1/2 inhibition coordinates immune rebalancing with epithelial repair and support its therapeutic potential in chronic type 1/2-driven inflammatory disease, while identifying residual IL-17A-mediated pathology as a target for combination therapy.

immunology↗

bioRxiv2026

Conversion of a Viral Glycan Shield into a Binding Anchor via Causal-Driven Antibody Optimization

Therapeutic antibodies are challenged by rapidly evolving pathogens that exploit glycosylation to shield epitopes. SARS-CoV-2 JN.1 exemplifies this, escaping antibodies through the N354-linked glycan. However, targeting glycosylated epitopes remains vacant, as scarce and heterogeneous glycan structures render existing approaches ineffective. Here, we introduce the Antibody Evolution Nexus with Causal-Driven Simulation (AENCS), integrating molecular simulation with causal inference. Applying AENCS to restore S309 efficacy against JN.1, we identified ACC01, exhibiting [~]24-fold improved neutralization. With limited prior knowledge of the N354 glycosylation site, ACC01 stabilized this glycan conformation, facilitating the determination of its cryo-EM structure. Causal dissection revealed how this glycan shield is functionally inverted into a binding anchor through multi-layered interactions. This mechanistic conversion, combined with the conservation of N354 glycosylation, enabled ACC01 to maintain potent activity against the latest variant NB.1.8.1. Collectively, AENCS demonstrates causal-driven antibody engineering can illuminate cryptic glycosylated epitopes, providing viable paradigms for exploring this vacant frontier.

bioengineering↗

bioRxiv2026

Lack of ancestral SARS-CoV-2 imprinting promotes BA.3.2.2 infection in children

The ongoing evolution of SARS-CoV-2 is heavily constrained by population-level immune imprinting. Recent genomic surveillance reveals an unexpected demographic shift: the highly mutated BA.3.2.2 sublineage is significantly enriched in paediatric populations globally, unlike concurrent variants XFG and NB.1.8.1. Here, the evidence demonstrates that this paediatric susceptibility arises from the absence of ancestral Wuhan-strain immune imprinting. Serological analyses reveal a profound failure to neutralise BA.3.2.2 in young children with no ancestral strain exposure. Concurrently, single-B-cell sequencing analyses demonstrate that BA.3.2.2 completely evades Omicron-specific Class 1/4 neutralising antibodies, encoded by IGHV2-5 and IGHV5-51, which dominate the repertoires of weakly imprinted individuals such as young children. Conversely, BA.3.2.2 remains uniquely susceptible to broadly cross-reactive Class 1 antibodies, encoded by IGHV3-53/66, typically enriched in adults with strong ancestral imprinting, such as mRNA vaccine recipients. Critically, sustained transmission of BA.3.2.2 in paediatric populations may catalyse the emergence of secondary variants that combine the paediatric-evading features of BA.3.2.2 with adult-evading Class 1 mutations. This could allow the lineage to breach the imprinted immunity of the adult population, potentially driving widespread global transmission. Together, these findings show that legacy immune imprinting paradoxically provides superior protection against a highly divergent saltation variant, directly explaining the age-biased transmission of BA.3.2.2, and informing future paediatric vaccination strategies.

immunology↗

bioRxiv2026

Closed-loop neurofeedback reshapes preparatory brain states to bias subsequent pain processing

Pain perception is strongly influenced by neural states preceding stimulus onset, yet whether such preparatory states can be causally reshaped remains unclear. Using a double-blind design, we tested whether closed-loop neurofeedback can reconfigure preparatory brain dynamics to influence subsequent pain processing. Real, but not sham, feedback enabled learning-dependent enhancement of pre-stimulus oscillations in primary somatosensory cortex and their trans-hemispheric propagation. Training also selectively reduced transitions between EEG microstates associated with salience and attentional control, indicating reorganization of preparatory brain states. These neural changes were accompanied by attenuated pain-evoked cortical responses and reduced pain perception. Sequential mediation analyses further revealed that somatosensory enhancement predicted microstate reorganization, which in turn suppressed pain-evoked responses. Together, these findings demonstrate that closed-loop neurofeedback can causally reshape preparatory brain states to bias subsequent sensory processing.

neuroscience↗

bioRxiv2026

Defining the Immunomodulatory Determinants of 3-mer Oligonucleotides on TLR7 and TLR8 sensing

Chemical modifications such as 2'-O-methyl (2'-OMe) are central to the efficacy and tolerability of RNA therapeutics. We recently identified that 2'-OMe RNA fragments as short as three nucleotides can exert opposing effects on Toll-like receptor 8 (TLR8) sensing in a motif-dependent manner. This discovery raises important considerations for degradation products of chemically modified RNA therapeutics, which may generate such immunologically active fragments. Here, leveraging their short length, we systematically map how base and sugar modifications within 3-mer oligonucleotides regulate TLR7 and TLR8 responses, and we resolve the structural basis for both TLR8 potentiation and TLR7/8 antagonism by RNA fragments. Building on these insights, we report the development of a dual TLR7/8 inhibitory oligonucleotide with therapeutic potential in autoimmune disease. Together, these findings provide unprecedented resolution of the immunomodulatory properties of oligonucleotide modifications on TLR7/8 and establish 3-mer oligonucleotides as the shortest functional class of RNA therapeutics described to date.

immunology↗

bioRxiv2026

BioClaw: Human-Bot Research Collaboration Ecosystems in Group Chats

Day-to-day research discussions in group chats often generate hypotheses, analysis requests, and interpretation decisions, yet executing those analyses still requires researchers to leave the conversation and rely on fragmented local tools, databases, visualization software, and literature search engines. In this work, we present BioClaw, a human-bot research collaboration ecosystem that converts natural-language requests in group conversations into tool-grounded analyses executed within isolated Docker containers. Deployed across 8 messaging platforms, BioClaw turns each group chat into a persistent execution workspace. Its design combines multi-channel orchestration, per-group state and workspace management, and isolated containerized execution for reliable shared use over long-lived conversations. To support practical research workflows, BioClaw combines containerized execution with preinstalled 31 biomedical tools and 95+ skills. The application of BioClaw spans various biomedical domains (e.g., genomics, clinics, structural biology) and data modalities (e.g., sequencing data, EHR data, protein structure data). These results establish the viability of embedding executable, tool-rich agent workflows within shared digital workspaces, positioning group chats as a transformative paradigm for collaborative scientific discovery and innovation.

bioinformatics↗

bioRxiv2026

Early electrical stimulation promotes functional recovery after volumetric muscle loss

Volumetric muscle loss (VML) injuries overwhelm the inherent regenerative capacity of skeletal muscle, causing persistent functional deficits with no routinely effective therapies. Electrical stimulation (ES) has been shown to preserve muscle structure in other injury models, but technical barriers have prevented daily delivery during the acute post-injury window when critical regenerative programs are established. Here, we developed a fully implantable bioelectronic system with nanoporous platinum-modified electrodes enabling daily therapeutic stimulation and electromyographic recording without repeated anesthesia in a rat tibialis anterior VML model. Animals receiving ES during the acute post-injury period (10 sessions over days 0-14) showed sustained functional improvement, reaching 90% of baseline torque at 8 weeks compared to 71% in unstimulated controls. This recovery reflected enhanced remodeling of injured muscle rather than synergistic muscle compensation. Histological analysis revealed coordinated early increases in vascularization, pro-regenerative macrophages, and satellite cells. These findings establish early ES as a promising intervention for promoting muscle regeneration after catastrophic injury.

bioengineering↗

bioRxiv2025

SimulScan and Partial Least Squares: Visualizing swallowing through functional and dynamic imaging correlations

PurposeSwallowing is a complex function involving the precise coordination of muscles, nerves, and brain areas, and can be disrupted in a variety of neurological conditions. Current imaging studies to visualize the central control of swallowing cannot examine both the biomechanics of the swallow and the brain activity associated with swallow events. An updated version of SimulScan is introduced that provides high-quality and high-speed dynamic imaging, together with fMRI acquisitions, to enable data-driven analysis of swallowing function through a partial least squares correlation (PLSC) analysis. MethodsIntegrating updated dynamic imaging approaches, SimulScan can achieve dynamic MRI at 23.75 frames per second and BOLD fMRI at a 1.6 s TR. Five subjects were recruited and scanned with SimulScan twice and with videofluoroscopy to compare the preliminary reliability of measuring swallowing biomechanics and the test-retest relationship in correlated functional and dynamic components of PLSC. ResultsHigh reliability of biomechanical measures of swallowing were achieved across the two SimulScan runs. In addition, these showed moderate correlation with videofluorscopy measures. Correlations between dynamic and functional imaging across runs also showed high reliability indicating that SimulScan with PLSC can extract maps of linked correlations between the brain and the oropharyngeal region. ConclusionThe updated version of SimulScan with PLSC analysis enables the study of central control of swallowing, providing simultaneous biomechanical visualization of the swallow along with brain functional signals.

neuroscience↗

bioRxiv2025

LabOS: The AI Co-Scientist That Sees and Works With Humans

Modern science advances fastest when thought meets action. LabOS represents the first AI co-scientist that unites computational reasoning with physical experimentation through multimodal perception, self-evolving agents, and XR-enabled, embodied human-AI collaboration. To empower agentic AI with embodied intelligence, we built LabSuperVision, the first vision-language-model (VLM) benchmark spanning biomedical and materials science laboratories, designed to evaluate scientific perception and reasoning. This endowed LabOS with visual reasoning capabilities that surpass those of general-purpose models on laboratory-specific tasks. By connecting multi-model AI agents, smart glasses, and human-AI-robot collaboration, LabOS allows AI to see what scientists see, understand experimental context, and participate in real-time execution through immersive XR interaction and robotic autonomation. Across applications-from cancer immunotherapy target discovery to cell engineering-LabOS shows that AI can move beyond computational design to participation, turning the laboratory into an intelligent, collaborative environment where human and machine discovery evolve together.

bioengineering↗

bioRxiv2025

BioLab: End-to-End Autonomous Life Sciences Research with Multi-Agents System Integrating Biological Foundation Models

Scientific discovery in the life sciences remains hindered by fragmented workflows, narrow-scope computational models, and inefficient links between in silico prediction and wet-lab validation. We present BioLab, a multi-agent system that integrates domain-specialized foundation models to automate end-to-end biological research. BioLab comprises eight collaborating agents, including a Planner, Reasoner, and Critic, orchestrated through a Memory Agent that enables iterative refinement via retrieval-augmented generation and a suite of 219 computational xBio-Tools spanning five biological scales (DNA, RNA, protein, cell, and chemical). These tools are built on the xTrimo Universe, a collection of 104 models derived from six foundation models (xTrimoChem, Protein, RNA, DNA, Cell, and Text), the majority of which achieve state-of-the-art (91-100% SOTA ratios) on domain benchmarks. Across standard reasoning tasks (PubMedQA, MMLU-Pro/Biology, GPQA-diamond), BioLab consistently outperformed leading large language models, including GPT-4o, Gemini-2.5, and DeepSeek-R1. Beyond benchmarks, BioLab autonomously executed a fully computational pipeline for de novo macrophage-targeting antibody design, progressing from target mining to multi-objective antibody optimization, where molecular dynamics simulations revealed structural mechanisms underlying enhanced affinity of optimized variants. Closing the computational-experimental loop, BioLab designed optimized antibodies (Pem-MOO-1, Pem-MOO-2) that achieved IC50 values of 0.01-0.016 nM, markedly surpassing the parental Pembrolizumab (0.027 nM) for PD-1. Functional assays confirmed enhanced pathway blockade and improved multi-parameter performance profiles. Together, these results establish BioLab as a generalizable framework for AI-native scientific discovery, demonstrating how multi-agent systems coupled with foundation models can autonomously generate, execute, and experimentally validate novel biological hypotheses.

bioinformatics↗

bioRxiv2025

STELLA: Self-Evolving LLM Agent for Biomedical Research

The staggering complexity of modern biomedical research has intensified the aspiration for a generalist "Biomedical World Model", yet current AI agents remain constrained by static capabilities and a lack of self-evolution mechanisms. To bridge this gap, we present STELLA, a self-evolving multimodal agent designed to progressively refine its computational reasoning and physical execution through interaction. STELLA operates via a collaborative multi-agent framework (comprising Manager, Developer, Critic, Critic, and Tool Creation agents) that continuously updates reasoning templates and autonomously expands a dynamic "Tool Ocean". We demonstrate STELLAs capabilities on the created Tool Creation Benchmark, where it attains a score of 4.01/5 with 100% task completion, significantly outperforming state-of-the-art models including GPT-5, Claude 4 Opus, and Biomni. Beyond computational metrics, STELLA drives experimentally validated scientific discovery. In oncology, the agent identified Butyrophilin Subfamily 3 Member A1 (BTN3A1) as a novel negative regulator of natural killer (NK) cell function in acute myeloid leukemia (AML), verified via CRISPR knockout studies. In protein engineering, STELLA orchestrated a complete directed evolution workflow for the enzyme strictosidine synthase, identifying variants, notably M276L, exhibiting more than a two-fold improvement in catalytic activity. Finally, the system extends to physical laboratory automation by training Vision-Language-Action (VLA) models through a Decompose-Monitor-Recover mechanism, which increased success rates from 17% to 82%. By integrating autonomous tool evolution, biological discovery, and robotic control, STELLA offers a blueprint for a self-evolving world model in the life sciences.

bioinformatics↗

bioRxiv2025

EpiMII: Integrating Structure and Graph Neural Networks for MHC-II Epitope and Neoantigen Design

MHC-II neoantigens play a critical role in immunotherapy, either as direct effectors or through their influence on CD8+ T cells. However, only a small fraction of tumor DNA mutations qualify as functional neoantigens, and current prediction tools often lack accuracy, leading to the low immunogenicity of predicted neoantigens in vivo. Here, we present EpiMII, a Graph Neural Network model for MHC-II epitope design, which learns from the structural features of epitopes to predict their sequences. To train EpiMII, we constructed a reliable, large dataset containing 142,934 MHC-II epitope structures. This approach achieves a 4.2x improvement over ProteinMPNN, with a sequence recovery rate of 78.0% for known MHC-II epitopes in the Protein Data Bank. As a case study, we designed a neoantigen from hepatocellular carcinoma. All five designed epitopes significantly activated CD4+ T cells in vitro and induced secretion of IFN-{gamma} and TNF-. Notably, one epitope treatment significantly reduced tumor volume in mice in vivo. EpiMII offers a novel and efficient approach for identifying MHC-II epitopes/neoantigens, potentially contributing to vaccine development.

immunology↗

bioRxiv2025

KIDINS220 and InsP8 safeguard the stepwise regulation of phosphate exporter XPR1

XPR1 is emerging as the only known inorganic phosphate (Pi) exporter in humans, critical for Pi homeostasis, with its activity stimulated by inositol pyrophosphate InsP8 and regulated by neuronal scaffold protein KIDINS220. Our structural studies reveal InsP8 specifically activates XPR1 in a stepwise manner, involving profound SPX domain movements. Each XPR1 subunit functions with four gating states, in which Pi permeates a constriction site via a "knock-kiss-kick" process. In contrast, KIDINS220 delicately stabilizes XPR1 in a closed conformation through multiple mechanisms, one of which involves trapping the XPR1 1 helix--critical for InsP8 binding--within an interaction hub. InsP8 serves as a key to release KIDINS220s restraint, reinforcing a "key-to-locks" mechanism to safeguard the stepwise activation. Additionally, our study provides direct structural insights into XPR1-associated neuronal disorders and highlights the evolutionary conservation and divergence among XPR1 orthologues, offering a comprehensive understanding of Pi homeostasis across species.

molecular biology↗

bioRxiv2024

RNAGenesis: Foundation Model for Enhanced RNA Sequence Generation and Structural Insights

RNA molecules are central to gene regulation, catalysis, and molecular recognition, and offer broad opportunities for therapeutic applications. However, uncovering their complex sequence, structure, and function relationships, particularly for non-coding RNAs, remains a formidable challenge. Here, we introduce RNAGenesis, a Generalist RNA foundation model that integrates sequence representation, structural prediction, and de novo functional design within a single generative framework. Trained on diverse clustered non-coding RNAs, RNAGenesis leverages a BERT-style encoder, query-based latent compression, and a diffusion-guided decoder enhanced by inference-time alignment with gradient guidance and beam search strategies. Through comprehensive evaluations, RNAGenesis achieves state-of-the-art performance on 11 of 13 tasks in the BEACON benchmark and surpasses structure-aware models in inverse folding, 3D structure prediction, and de novo structure design. We further introduce RNATx-Bench, a dedicated benchmark for RNA therapeutics comprising over 100,000 experimentally validated sequences. RNAGenesis demonstrates strong predictive performance across ASOs, siRNAs, shRNAs, circRNAs, and untranslated region (UTR) variants. Furthermore, RNAGenesis enables functional RNA design, including aptamers targeting IGFBP3 and structurally constrained sgRNA scaffolds. Wet-lab validation confirms aptamer binding with KD values as low as 4.02 nM and up to 2.5-fold improvement in editing efficiency across CRISPR-Cas9, base editing, and prime editing systems. These results position RNAGenesis as a next-generation general-purpose RNA foundation model with broad utility for computational modeling and experimental therapeutic design.

bioinformatics↗

bioRxiv2024

A Multi-Modal Deep Learning Framework with Both Sequence and Structure for Tumor Antigens Prediction

Tumor antigens are key targets in cancer immunotherapies that can be recognized by T cell receptor and induce immune responses. However, precision screening of immunogenic tumor antigens remains a great challenge due to human leukocyte antigen (HLA) restriction and tumor antigen escape. Here, we introduce MultiTAP (Multi-modal Tumor Antigen Predictor), a pioneering multi-modal framework with TCR-peptide-HLA sequence and structure features incorporating an attention mechanism designed to accurately identify tumor antigens with immunogenic properties. By constructing the multi-modal TCR-peptide-HLA Dataset (TPHD) and integrating its sequence and structure, we perform antigen feature enhancement using peptide-HLA (pHLA) structural features at the residue level, achieving interpretable prediction of immunogenicity for tumor antigens. Relative to existing baseline models, the MultiTAP framework has exhibited superior efficacy in predicting the immunogenicity of tumor antigens. Through comprehensive out-of-distribution (OOD) assessments, MultiTAP has maintained predictive robustness across diverse HLA phenotypes and the continuously evolving landscape of epitope distributions. Overall, MultiTAP presents a brand-new and promising approach for cancer immunotherapies that target tumor antigens.

biophysics↗

bioRxiv2024

Structure, dynamics and evolution of the Candida albicans multi-drug resistance ABC transporter CDR1

The pleiotropic drug resistance transporter Cdr1 from Candida albicans plays a crucial role in antifungal resistance. Here, we present high-resolution cryo-electron microscopy structures of C. albicans Cdr1 in multiple functional states: nucleotide-bound, substrate-bound, and apo forms. The 3.5 [A] resolution structure of Cdr1 in complex with ATP and ADP reveals the molecular details of its asymmetric nucleotide-binding sites (NBS), with ATP bound to the deviant NBS1 and ADP to the canonical NBS2. Structures of Cdr1 bound to rhodamine 6G (3.4 [A]) and Oregon Green 488 (3.5 [A]) in complex with these nucleotides elucidate the pleiotropic substrate-binding pocket and highlight how nucleotide exchange drives conformational change required for transport. Additionally, we determined a 3.7 [A] resolution structure of Cdr1 in the detergent LMNG without nucleotides, as well as a 3.5 [A] resolution structure with nucleotides but no substrate, representing an apo state. We complemented these structural insights with molecular dynamics simulations to understand substrate binding dynamics, ancestral sequence reconstruction to trace the evolution of key functional motifs, and analysis of clinical isolates from sequence databases to identify potential resistance-associated variations. Comparison of these structures provides new insights into the conformational changes associated with the transport cycle of this asymmetric ABC transporter, revealing how ATP binding at the deviant NBS1 allosterically regulates the canonical NBS2, driving ATPase activity. This work significantly advances our understanding of the molecular mechanisms underlying multidrug resistance in pathogenic fungi and provides a structural and evolutionary framework for the rational design of Cdr1 inhibitors to combat antifungal resistance.

microbiology↗

bioRxiv2024

FoldMark: Protecting Protein Generative Models with Watermarking

Proteins are the principal architects of life, fueling advances in bioengineering, drug discovery, and synthetic biology. The integration of generative AI with computational protein science has revolutionized protein design while also posing dual-use risks, such as enabling the creation of pandemic-capable proteins that require strong biosecurity safeguards. Here, we introduce FoldMark, a first-of-its-kind watermarking strategy leveraging distributional and evolutionary principles tailored for protein generative models, balancing watermark capacity and structural quality. FoldMark achieves over 95% watermark bit accuracy at 32 bits with minimal impact on structural quality (>0.9 scTM scores) for leading models including AlphaFold3, ESMFold, RFDiffusion, and RFDiffusionAA. For user tracing, FoldMark can successfully trace up to 1 million users. To validate FoldMark in wet lab, we applied it to structure-based design of EGFP and CRISPR-Cas13, showing wildtype-level function (98% fluorescence, 95% editing efficiency) and >90% watermark detection, demonstrating its practical utility for safeguarding AI-driven protein research.

bioinformatics↗

bioRxiv2024

IGF2 mediates Hippo signaling to control liver size

The Hippo pathway is a central mechanism in organ size control, but the mediator for its function remains elusive. Here, we show that the expression of insulin-like growth factor II (IGF2) is directly induced by YAP/TAZ transcription cofactors of the Hippo pathway in a developmental stage- and cell type-specific manner. In mouse livers, Igf2 expression is sustained by YAP/TAZ in hepatoblasts and immature hepatocytes at fetal and neonatal stages, coupling with rapid cell proliferation and liver size expansion, whereas turned off in matured hepatocytes following YAP/TAZ inactivation to prevent liver overgrowth. In contrast, YAP/TAZ fails to regulate Igf2 expression in cholangiocytes and diverse liver mesenchymal cells where epigenetic barriers, including DNA and histone methylation, are implanted near Igf2 promoters. Furthermore, IGF2 activates IGF1R signaling and promotes normal and neoplastic liver growth, while the inactivation of IGF2 or IGF1R effectively blocks YAP/TAZ-induced hepatomegaly and hepatoblastoma. In conclusion, our findings reveal that IGF2 is a bona fide mediator of the Hippo pathway, indicating a fundamental role of the Hippo-IGF2-IGF1R signaling axis in organ size regulation and malignancies.

cell biology↗