Biology subjects

Pan, T.

Publications and source records attributed to Pan, T..

At least 19 recordsLinked to original sources

bioRxiv2026

AI-enabled rhodopsin design for blue-light enhanced bacterial growth

We developed an AI-guided design pipeline that generated and validated non-natural microbial rhodopsins with spectral properties not yet known in nature. The pipeline comprised a three-stage in silico design, a genetic algorithm (GA) for sequence generation, a stacked LASSO and XGBoost machine-learning (ML) regressor for spectral prediction and fitness ranking, and a Markov-based sequence plausibility filter to enforce proton pumping like characteristics. Four candidate rhodopsins (APR1, APR2, APR6, and APR7) targeting blue light absorption were designed and AlphaFold3 structural modelling predicted retinal binding pocket architecture consistent with outward proton-pumping function. Experimental characterisation confirmed that all four variants absorbed light at [~]410 nm and significantly promoted the growth of Cupriavidus necator under blue light illumination. This study demonstrates that AI-enabled design can engineer proteins with no natural precedent, generating light-harvesting rhodopsins with novel spectral properties while preserving biological function, marking a significant advance in programmable synthetic biology.

Synthetic Biology↗

bioRxiv2026

DesignMaster: A Multi-Conditional Diffusion Framework for Rational PROTAC Design

MotivationProteolysis-targeting chimeras (PROTACs) enable targeted protein degradation through ternary complex formation with E3 ubiquitin ligase. However, the rational design of PROTACs remains highly challenging due to limited structure-activity relationship data and the vast conformational diversity of linkers. Existing computational approaches can be broadly divided into structure-based ternary modelling methods and fragment-based linker generation models. Although these approaches have advanced PROTAC design, they typically neglect key physicochemical constraints and linker-length control during the generation process, causing the generated PROTACs to lack balanced structural properties required for effective ternary complex formation with drug-like characteristics. ResultsTo address these limitations, we propose DesignMaster, a diffusion-based generative framework that explicitly incorporates linker length and physicochemical properties as controllable conditioning signals. DesignMaster employs an E(3)-equivariant graph Transformer with a gated multi-condition fusion module to inject linker length and physicochemical constraints throughout the diffusion process, enabling fine-grained and constraint-aware molecular generation. Experiments on PROTAC-DB 2.0 and 3.0 demonstrate that DesignMaster outperforms state-of-the-art baselines, with a 3.2% improvement in validity and a 34.4% improvement in recovery. The Case study shows DesignMaster achieves a 51.78% reduction in RMSD when predicting the linker of PROTAC BCPyr targeting 6W7O, highlighting its potential for practical structure-guided PROTAC design. AvailabilityThe source code and datasets are available at https://github.com/ABILiLab/DesignMaster.

bioinformatics↗

bioRxiv2026

Structure-aware geometric graph learning for modeling protease-substrate specificity at scale

Protease-substrate specificity is central to cellular regulation and disease pathogenesis, and accurately modeling its structural determinants remains challenging. Substrate recognition is governed by spatial constraints and higher-order relationships that extend beyond local sequence motifs. Most computational approaches rely predominantly on motif-centric or sequence-based representations, limiting their ability to capture the geometric and relational structure underlying enzymatic specificity. Here, we introduce OmniCleave, a structure-aware geometric graph learning framework for modeling protease-substrate specificity at scale. OmniCleave is trained on 57,278 structure-informed protease-substrate pairs derived from 9,651 substrates spanning over 100 proteases across six distinct families. The framework integrates multi-scale structural graphs with higher-order protease relational topology, explicitly encoding spatial context and inter-protease dependencies within a unified geometric representation. This formulation moves beyond local pattern recognition and enables transferable modelling across six protease families. Across large-scale benchmarks, the framework consistently outperforms existing approaches and reveals interpretable geometric determinants underlying substrate recognition. Experimental validation confirms three novel caspase-3 substrates and 21 cleavage sites predicted by OmniCleave, supporting the biological relevance of the learned representations. Together, OmniCleave provides a scalable geometric framework for modeling protease-substrate specificity, with practical utility for systematic analysis of protease biology.

bioinformatics↗

bioRxiv2026

ACENet: a graph neural network for predicting enzyme pHmin using structure and sequence insights

The acid activity of enzymes, characterized by the minimum pH at which enzymes remain active (pHmin), is crucial for industrial applications in acidic environments. However, the rational design of acid-active enzymes remains challenging due to limited understanding of sequence-structure- activity relationships under acidic conditions. Here, we propose ACENet, a graph neural network that predicts enzyme pHmin by integrating surface features of protein structures with evolutionary representations derived from the large-scale protein language model ESM-2. ACENet achieved a Pearson correlation coefficient of 0.85 on the test dataset, significantly outperforming other deep learning baseline models and maintains stable pHmin predictions under various conditions. Even on a subset of the dataset with less than 20% homology, the PCC remains above 0.5, with an RMSE (Root mean square error) less than 1.4. ACENet also present excellent performance in the annotation of pHmin for homologous proteins and the predictive screening of minimal active pH in protein mutants. Remarkably, ACENet could identify the catalytic region as key determinants of acid activity through residue-level interpretability analysis. Overall, ACENet accelerates the development of highly efficient biocatalysts for diverse applications where acidic conditions predominate.

bioinformatics↗

bioRxiv2026

MolX: A Geometric Foundation Model for Protein-Ligand Modelling

Understanding how small molecules interact with protein binding pockets is central to structure-based drug discovery. Accurately modelling these interactions requires capturing the 3D geometry and physicochemical complementarity of binding interfaces, yet existing computational approaches encode proteins and ligands separately or rely on simplified structural representations that do not explicitly model cross-entity spatial relationships. Such decoupled representations restrict their capacity to capture interface-level geometric constraints that arise from protein-ligand coorganisation. Here we present MolX, a Graph Transformer foundation model that jointly learns geometric and chemical representations of protein pockets and ligands from large-scale 3D structural data. Integrating over 3 million protein pockets and 5 million molecules, MolX represents both entities as E(3)-equivariant graphs to preserve spatial geometry and chemical context. The architecture employs dual E(3)-equivariant graph Transformer encoders to model pocket and ligand embeddings, ensuring representations remain invariant to rotation, translation, and reflection. MolX is pretrained using a hybrid learning paradigm that combines supervised biochemical objectives, logP and energy-gap regression, with self-supervised geometric objectives, coordinate re-construction, and atom-type prediction, fostering generalisable molecular understanding. Across eight downstream benchmarks, including antibody-drug conjugates (ADC), proteolysis-targeting chimeras (PROTAC), molecular glue, and PCBA activity prediction, as well as binding affinity and physicochemical property regression, MolX achieves consistent state-of-the-art performance and strong cross-domain generalisation. Furthermore, MolX incorporates a sparse autoencoder module to decompose latent representations into interpretable biological components, thereby revealing the pocket-ligand interactions that drive prediction outcomes. Together, MolX establishes a scalable and interpretable foundation model for molecular representation learning, providing a unified framework for predicting and interpreting complex small-molecule-protein interactions.

bioinformatics↗

bioRxiv2026

Polymer-lipid hybrid nanoparticle enhances mRNA delivery and T cell-mediated immunity

mRNA vaccines have transformed prophylactic immunization against infectious diseases as well as therapeutic interventions for cancer. However, their effectiveness against emerging viral variants and a range of malignancies continues to be hindered by suboptimal induction of T cell-mediated immunity. To overcome this limitation, here we developed a polymer-lipid hybrid nanoparticle (PLNP) platform engineered to improve mRNA delivery to antigen-presenting cells (APCs) and to potentiate T cell responses. Relative to conventional mRNA lipid nanoparticle (LNP) vaccines, mRNA PLNP vaccines demonstrated markedly improved lymph node targeting, APC activation, Th1-biased pro-inflammatory cytokine response, and antigen-specific T cell expansion while retaining robust humoral immunity. Remarkably, mRNA-PLNP vaccines generated approximately 50% more antigen-specific CD8+ T cells than mRNA-LNP vaccines across multiple antigens, including SARS-CoV-2 spike, influenza hemagglutinin, and ovalbumin. In prophylactic applications, mRNA PLNP vaccine provided complete protection against SARS-CoV-2 variants. As a therapeutic approach in a melanoma model, mRNA PLNP vaccination resulted in enhanced tumor control and significantly prolonged survival compared to LNP-based formulations. Collectively, these results establish PLNP as a versatile and broadly applicable platform for augmenting mRNA vaccine efficacy through improved mRNA delivery and T cell priming, offering promising implications for infection prevention and cancer immunotherapy.

immunology↗

bioRxiv2026

Myeloid CD209 impairs T-cell activation by inducing ICAM-2-ERM-dependent cortical stiffening

Resistance to PD-1-based immunotherapy remains a major challenge in cancer treatment, especially in immunologically "cold" tumors characterized by T-cell exclusion and impaired effector function. Tumor-associated myeloid cells are important mediators of this resistance, but whether they directly suppress T-cell immunity through mechanical regulation remains unclear. Here, we identify DC-SIGN (CD209), a myeloid-expressed C-type lectin receptor, as a mechanosuppressive regulator of antitumor T-cell immunity. DC-SIGN is selectively enriched in tumor-infiltrating M2-like macrophages and monocytes, and its expression is associated with poor clinical response to PD-1 blockade and elevated TIDE scores across multiple cancer types. Mechanistically, DC-SIGN directly engages ICAM-2 on T cells and activates ROCK1-ERM signaling. This interaction increases T-cell cortical stiffness, destabilizes T cell-antigen-presenting cell interactions, and attenuates T-cell receptor signaling. Blocking the DC-SIGN-ICAM-2-ERM pathway restores T-cell activation and promotes CD8+ T-cell infiltration and effector function in mouse, humanized, and ex vivo human tumor systems. Importantly, DC-SIGN blockade overcomes resistance to PD-1-based immunotherapy in refractory and immunologically "cold" tumor models, and further enhances the response to PD-1 inhibition. Together, our findings define a myeloid-derived mechanical immune checkpoint and reveal T-cell mechanosuppression as a targetable mechanism of tumor immune evasion and immunotherapy resistance.

immunology↗

bioRxiv2026

Fusion gene discovery in single cells from high throughput long read single cell transcriptomes

High throughput long-read single cell RNA sequencing allows for fusion gene discovery in thousands of cells in parallel, which is however constrained by the lack of robust computational tools. Here, we develop LongFUSE that employs the XOR logic operation and stringent filtering criteria for the accurate detection of cell-specific fusion genes and their splicing isoforms. LongFUSE outperforms existing tools in both simulation and real data, demonstrating high accuracy and robustness.

bioinformatics↗

bioRxiv2026

Hypatia: a set of quantitative methods for profiling isoform across cell populations

High-throughput long-read single-cell RNA-sequencing enables isoform-level study across single cells, yet methods for systematically assessing cell-to-cell variations remain limited. Here, we develop Hypatia, a comprehensive platform for dissecting isoform complexities across cell populations, devising Tsallis entropy and Cramers V to facilitate robust comparative profiling. Hypatia revealed prominent isoform species variations and usage shifts across cell-types in glioblastoma, renal cell carcinoma, and heart, highlighting clinically relevant applications for studying isoform-derived, cell-specific functions.

bioinformatics↗

bioRxiv2025

Behavioral Opportunism and Altered Dopamine Dynamics in Mice Exposed to Early Life Adversity

Early life adversity (ELA) confers risk for reward-related psychopathologies. These risks may stem from adaptations optimizing reward pursuit in anticipation of unreliable, resource poor environments. One rational adaptation to poor, unreliable environments is Behavioral Opportunism: updating expectations more slowly and acting vigorously only when reward is immediately available. To systematically test the impact of ELA on behavioral strategies and underlying reward processing mechanisms, we exposed mice to resource restriction (limited bedding and nesting materials for 7 days) to manipulate the reliability and quality of early life care. Subsequently, we tested adults reward learning and decision making in a two-arm bandit task and recorded dopamine signaling using dLight1.2 fiber photometry in the nucleus accumbens core. Exposure to ELA led to poorer choice discrimination, impaired learning, and decreased adaptation to changes in reward availability. Furthermore, ELA mice were slower to choose between levers but were faster to retrieve immediately available rewards when delivered, consistent with a strategy of behavioral opportunism. Dopamine signaling predicted behavior in both rearing conditions, and its fluctuations were strongly predictive of faster retrieval in ELA mice and an increased likelihood of choice repetition, implying that aberrant dopamine signals underlie slowed learning and vigorous action for immediately available rewards. To understand key features of maternal interactions driving these effects, we used home cage video monitoring to quantify maternal behaviors, continuously, across early life. We found that specific experiential outcomes, such as maternal kicking, intensified behavioral opportunism in adults, predicting poorer bandit task performance beyond the group effect of ELA. Behavioral opportunism provides an explanatory framework for interpreting altered reward processing and reward pursuit in adulthood for individuals exposed to ELA.

animal behavior and cognition↗

bioRxiv2025

Sex-biased duplicates are rapidly generated during Drosophila tRNA repertoire evolution

Eukaryotic genomes encode hundreds of transfer RNA genes ostensibly to ensure efficient translation. How these seemingly redundant tRNA copies arise and are naturally selected in metazoan genomes and populations remains largely unexplored, owing to challenges in accurate sequencing of tRNA loci and their transcripts. We leveraged long-read genome assemblies of 24 Drosophilid species to infer the origination times of the entire Drosophila melanogaster tRNA gene repertoire and found that continuous gain and loss of tRNA duplicates throughout 60 million years of divergence has resulted in rapid taxonomic restriction of isodecoding and isoaccepting copies--even producing one isodecoder specific to D. melanogaster. Moreover, we identified patterns of global codon usage, especially in lineage-specific genes, incongruous with translational efficiency hypotheses. Through generation of tRNA sequencing in D. melanogaster we observed that recently duplicated, taxonomically restricted tRNA copies had sexually dimorphic patterns of expression, fragmentation, and nucleoside modification. Our work implicates the emergence of taxonomically restricted tRNA genes as sources of regulatory diversity and reveals that sexual and natural selection affect the evolutionary dynamics of the tRNA gene family in metazoan species.

evolutionary biology↗

bioRxiv2025

A high-resolution diel survey of surface ocean metagenomes, metatranscriptomes, and transfer RNA transcripts

The roles of marine microbes in ecosystem processes are inherently linked to their ability to sense, respond, and ultimately adapt to environmental change. Capturing the nuances of this perpetual dialogue and its long-term implications requires insight into the subtle drivers of microbial responses to environmental change that are most accessible at the shortest scales of time. Here, we present a multi-omics dataset comprising surface ocean metagenomes, metatranscriptomes, tRNA transcripts, and biogeochemical measurements, collected every 1.5 hours for 48 hours at two stations within coastal and adjacent offshore waters of the tropical Pacific Ocean. We expect that this integrated dataset of multiple sequence types and environmental parameters will facilitate novel insights into microbial ecology, microbial physiology, and ocean biogeochemistry and help investigate the different mechanisms of adaptation that drive microbial responses to environmental change.

microbiology↗

bioRxiv2025

Brain biomechanics governs mitotic fidelity of embryonic neural progenitors

Accurate chromosome segregation is essential to maintain genetic stability and prevent the onset of diseases such as developmental disorders, infertility, and cancer. While many intrinsic factors and processes involved in mitosis have been extensively characterized, less is known about how extrinsic factors and tissue properties contribute to mitotic fidelity. In this study, using both in vivo and ex vivo systems, in combination with pharmacological perturbations and high-resolution microscopy, we investigated mitosis in apical radial glial (aRG) cells--the primary neural progenitors in the developing mammalian brain. We found that the high cell density typical of early neurogenic stages enhances microtubule polymerization rates from the spindle poles, thereby influencing chromosome segregation. Mechanistically, our results indicate that cortical actin and elevated cortical tension function as sensors of biomechanical stress during mitosis. These findings identify biomechanics as a threat to mitotic fidelity in the embryonic brain.

cell biology↗

bioRxiv2025

Predicting Disease-Specific Histone Modifications and Functional Effects of Non-coding Variants by Leveraging DNA Language Models

BackgroundEpigenetic modifications play a vital role in the pathogenesis of human diseases, particularly neurodegenerative disorders such as Alzheimers disease (AD), where dysregulated histone modifications are strongly implicated in disease mechanisms. While recent advances underscore the importance of accurately identifying these modifications to elucidate their contribution to AD pathology, existing computational methods remain limited by their generic approaches that overlook disease-specific epigenetic signatures. ResultsTo bridge this gap, we developed a novel large language model (LLM)-based deep learning framework tailored for disease-contextual prediction of histone modifications and variant effects. Focusing on AD as a case study, we integrated epigenomic data from multiple patient samples to construct a comprehensive, disease-specific histone modification dataset, enabling our model to learn AD-associated molecular signatures. A key innovation of our approach is the incorporation of a Mixture of Experts (MoE) architecture, which effectively distinguishes between disease and healthy epigenetic states, allowing for precise identification of AD-relevant epigenetic modification patterns. Our model demonstrates robust performance in disease-specific histone modification prediction, achieving mean area under receiver-operating curves (AUROCs) ranging from 0.7863 to 0.9142, significantly outperforming existing state-of-the-art methods that lack disease context. Beyond accurate modification site prediction, our framework provides important biological insights by successfully prioritizing AD-associated genetic variants, which show significant enrichment in disease-relevant pathways, supporting their potential functional role in AD pathogenesis. These findings suggest that differential modification loci identified by our model may represent key regulatory elements in AD. ConclusionsOur framework establishes a powerful new paradigm for epigenetic research that can be extended to other complex diseases, offering both a valuable tool for variant effect interpretation and a promising strategy for uncovering novel disease mechanisms through epigenetic profiling.

bioinformatics↗

bioRxiv2025

Delayed forebrain excitatory and inhibitory neurogenesis inSTRADA-related megalencephaly via mTOR hyperactivity

Biallelic pathogenic variants in STRADA, an upstream regulator of the mechanistic target of rapamycin (mTOR) pathway, result in megalencephaly, drug-resistant epilepsy, and severe intellectual disability. This study explores how mTOR pathway hyperactivity alters cell fate specification in dorsal and ventral forebrain development using STRADA knock-out human stem cell derived brain organoids. In both dorsal and ventral forebrain STRADA knock-out organoids, neurogenesis is delayed, with a predilection for progenitor renewal and proliferation and an increase in outer radial glia. Ventrally, interneuron subtypes shift to an increase in neuropeptide-Y expressing cells. Inhibition of the mTOR pathway with rapamycin results in rescue for most phenotypes. When mTOR pathway variants are present in all cells of the developing brain, overproduction of interneurons and altered interneuron cell fate may underlie mechanisms of megalencephaly, epilepsy, and cognitive impairment. Our findings suggest mTOR inhibition during fetal brain development as a potential therapeutic strategy in STRADA deficiency.

neuroscience↗

bioRxiv2025

Balanced DNA-to-Cytoplasm Ratio at the 2-Cell Stage Is Critical for Mouse Preimplantation Development

Vertebrate development typically begins with a diploid fertilized egg but can also be initiated parthenogenetically with a haploid maternal genome, though such development fails to reach completion. Among vertebrates, mammalian haploid embryos are particularly prone to arrest during early preimplantation stages for reasons that remain unclear. Here, using mouse embryos with varying cell sizes and genome ploidy, we show that developmental arrest is primarily caused not by haploidy itself but an imbalance between genome ploidy and cytoplasmic volume. Haploid embryos exhibit cytokinesis failure, frequently accompanied by chromosome segregation failure due to malformed spindles beginning at the second mitosis, with about half of the affected blastomeres subsequently arresting before the morula stage. Strikingly, restoring the DNA-to- cytoplasm (D/C) ratio by halving the cytoplasmic volume alleviates these defects. Similarly, diploid embryos with doubled cytoplasmic volume show mitotic and developmental abnormalities resembling those of haploid embryos, whereas tetraploid embryos with the same volume develop normally. These findings demonstrate that a proper D/C ratio, rather than genome ploidy, is critical for mouse embryos to reach the morula stage. Moreover, halving the D/C ratio at later developmental stages, such as the late G2 phase of the 1-cell stage or during the 2-cell stage, progressively improves developmental outcomes. When the D/C ratio is reduced at the end of the 2-cell stage, development proceeds normally. Our findings identify a critical temporal window during which a balanced D/C ratio is essential for mitotic fidelity and successful preimplantation development. Graphic Abstract O_FIG O_LINKSMALLFIG WIDTH=198 HEIGHT=200 SRC="FIGDIR/small/651589v1_ufig1.gif" ALT="Figure 1"> View larger version (49K): org.highwire.dtl.DTLVardef@72a664org.highwire.dtl.DTLVardef@10a22c1org.highwire.dtl.DTLVardef@6fbedforg.highwire.dtl.DTLVardef@5c5b1d_HPS_FORMAT_FIGEXP M_FIG C_FIG O_LIHalved D/C ratio impairs preimplantation development in mouse embryos C_LIO_LIHalved D/C ratio causes cytokinesis failure from the second mitosis C_LIO_LIDelaying D/C ratio halving to late 2-cell stage alleviates developmental defects C_LI

cell biology↗

bioRxiv2025

KcatNet: advancing genome-wide enzyme turnover number prediction through structural enzymatic characterization

Enzyme turnover numbers (Kcat) are fundamental kinetic constants that quantify enzymatic efficiency. Systematic studies of Kcat are essential for characterizing the mechanisms underlying proteomic composition and cellular metabolism. However, experimental measurements of Kcat remain limited and prone to noise. To address this, we present KcatNet, a geometric deep learning model designed for high-throughput prediction of Kcat in metabolic enzymes across all organisms, leveraging paired enzyme sequence and substrate representations. KcatNet consistently outperforms existing predictors, particularly for enzymes with high catalytic efficiency, and demonstrates strong generalization to enzymes that are dissimilar to those in the training set. Furthermore, KcatNet uncovers structural mechanisms and interaction patterns within enzyme-substrate complexes, providing insights into architectural principles that are inaccessible with existing methods by harnessing the representational power of large-scale protein language models. We applied KcatNet to genome-scale Kcat prediction across diverse yeast species, improving proteome allocation predictions by integrating its outputs into metabolic models. Experimental validation further confirmed the models ability to identify enzyme mutants with enhanced activity. By bridging the gap between sequence, structure, and function, KcatNet establishes a robust foundation for advancing understanding of molecular-level mechanisms and accelerating enzyme engineering efforts.

bioinformatics↗

bioRxiv2025

Two-Stage CD8+ CAR T-Cell Differentiation in Patients with Large B-Cell Lymphoma

Chimeric antigen receptor (CAR) T-cell therapy has expanded therapeutic options for patients with diffuse large B-cell lymphoma (DLBCL). However, progress in improving clinical outcomes has been limited by an incomplete understanding of CAR T-cell differentiation in patients. To comprehensively investigate CAR T-cell differentiation in vivo, we performed single-cell, multi-modal, and longitudinal analyses of CD28-costimulated CAR T cells from infusion product and peripheral blood (day 8-28) of patients with DLBCL who were successfully treated with axicabtagene ciloleucel. Here, we show that CD8+ CAR T cells undergo two distinct waves of clonal expansion. The first wave is dominated by CAR T cells with an exhausted-like effector memory phenotype during the peak expansion period (day 8-14). The second wave is dominated by CAR T cells with a terminal effector phenotype during the post-peak persistence period (day 21-28). Importantly, the two waves have distinct ontogeny and are biologically uncoupled. Furthermore, lineage tracing analysis via each CAR T cells endogenous TCR clonotype demonstrates that the two waves originate from different effector precursors in the infusion product. Precursors of the first wave exhibit more effector-like signatures, whereas precursors of the second wave exhibit more stem-like signatures. These findings suggest that pre-infusion heterogeneity mediates the two waves of in vivo clonal expansion. Our findings provide evidence against the intuitive idea that the post-peak contraction in CAR abundance is solely apoptosis or extravasation of short-lived CAR T cells from peak expansion. Rather, our findings demonstrate that CAR T-cell expansion and persistence are mediated by clonally, phenotypically, and ontogenically distinct CAR T-cell populations that serve complementary clinical purposes.

immunology↗