Search bioRxiv⌕ Search

Biology subjects

Tu, X.

Publications and source records attributed to Tu, X..

At least 19 recordsLinked to original sources

Programmable Self-Assembly of RNA Nanostructures with >100 Unique Components

Sophisticated biomolecular functions often arise from large, precisely organized architectures, motivating efforts to construct increasingly complex structures through programmed nucleic acid self-assembly. Although RNA offers a richer repertoire of structural motifs and biological functions than DNA, RNA nanostructures constructed to date have remained substantially less complex and less scalable than their DNA counterparts, most notably DNA origami and DNA bricks, which rely on large libraries of synthetic single-stranded (ss) DNA strands. Directly adapting these strategies to RNA, however, faces two major barriers: (i) the high cost of producing large libraries of distinct synthetic ssRNA strands, and (ii) the limited stability of ssRNA under conditions commonly used for assembling multicomponent nucleic acid nanostructures. Here we present the double-stranded RNA (dsRNA) bricks approach, which addresses both barriers by adopting principles inspired by natural RNA systems: (i) many distinct RNA components ("bricks") are encoded within a single precursor transcript and released by enzymatic processing, (ii) each brick's predominantly double-stranded nature enhances stability. The resulting dsRNA bricks self-assemble through programmable branched kissing-loop interactions. Using this approach, we constructed complex two- and three-dimensional RNA nanostructures with more than 100 distinct components, representing, to our knowledge, the largest and most compositionally complex RNA nanoarchitectures reported to date. Overall, the dsRNA bricks approach establishes a scalable and robust framework for constructing complex RNA nanostructures with compositional and architectural sophistication approaching that of DNA-based systems, thereby opening new opportunities for programmable RNA materials and RNA-based devices.

synthetic biology↗

Spatial multi-omics analysis reveals vimentin-high macrophages-endothelial cells niche shapes CAFs heterogeneity in colorectal cancer metastasis

The spatial architecture of the tumor microenvironment (TME) is pivotal in the progression of colorectal cancer (CRC) liver metastasis. By applying high-plex spatial multi-omic mapping and neighborhood analysis to a discovery cohort of colorectal cancer primary tumor (PT) and paired liver metastases (LM), we identified a specialized vimentin-high macrophages-endothelial cells niche that orchestrates cancer-associated fibroblast (CAF) phenotypes. Mechanistically, in primary tumors, vimentin-high macrophages secrete INHBA to activate the ACVR2/TGF-{beta} axis in endothelial cells, driving CAFs toward a myCAF phenotype. Conversely, in liver metastases, these macrophages secrete CXCL9 to trigger the PI3K-Akt/NF-[kcy]B/CXCL12 cascade in endothelial cells, directing CAFs toward an iCAF state. Clinically, high niche activity predicts poor survival. Divergent endothelial signaling in primary versus metastatic lesions exposes site-specific stromal vulnerabilities for therapeutic targeting, with architectural features discernible from routine histopathology. These findings reveal a site-specific regulatory mechanism of the macrophage-endothelial niche, offering a novel and clinically significant biomarker for CRC prognosis.

cancer biology↗

Flexible neuronal participation within a reliable motor sequence

The reliable execution of learned motor sequences poses a challenge: they require reproducible neuron activity patterns for behavioral consistency yet must retain flexibility for modulation depending on internal states and adaptation to external contingencies. The zebra finch, a songbird, provides an ideal system to examine this problem because their vocalizations are highly stereotyped and are associated with precisely time-locked bursts in HVC, a key nucleus involved in song production. However, their songs are not immutable or totally fixed, as they can be modulated during social interactions and can recover after brain lesions. Here, we used calcium imaging of HVC projection neurons during freely produced songs to track activity across repeated renditions of the same song from seconds to weeks. We first confirmed that song-locked calcium-event timing was stable across days. Surprisingly, not every neuron showed a calcium event on every rendition of the song. We referred to the presence of such an event as participation. This participation was a stable, neuron-specific property that varied with the past and future renditions of the songs within an utterance and depended on social context. Song recovery after brain lesion and refinement during learning further revealed that neural participation could change while sequence timing remained stable. These results suggest that stable motor sequences need not rely on rigid repeated activation of the same cellular ensemble but can preserve temporal order while flexibly selecting which neurons contribute to each expression of the behavior. This provides a cellular solution to the reliability-flexibility tradeoff in learned motor control.

neuroscience↗

BiomniBench: Process-level Evaluation of LLM Agents for Real-world Biomedical Research

LLM agents now perform real biomedical research, but evaluating them rigorously is hard. Outcome-only benchmarks fail in two ways. First, a correct final answer can come from memorization, reward hacking, or wrong reasoning that produces the right number by chance. Second, valid alternative analyses are marked wrong simply because they differ from the reference. We introduce BiomniBench, a process-level evaluation framework that scores the full agent trajectory against expert-designed, task-specific rubrics. Our first release, BiomniBench-DA, contains 100 data-analysis tasks across 17 task types, 5 disease areas, and a general-biology category, each based on a paper from journals such as Nature, Cell, and Science and co-developed with an original author or a domain expert. Benchmarking frontier and open-weight models across four agent harnesses reveals three findings. Frontier and open-weight bases cluster within a few points of each other, with substantial headroom for all models. The agent harness shifts scores by more than the gap between successive model generations. Agents reliably ground claims in real sources yet consistently fall short on method selection, biological interpretation, and scientific reasoning. BiomniBench is the first process-level benchmark for LLM agents in biomedical research, providing the dimension-level diagnostics that outcome scoring cannot. Datasethuggingface.co/datasets/phylobio/BiomniBench-DA

bioinformatics↗

Brainwide dopamine dynamics across sleep-wake transitions

Dopaminergic signaling plays a critical role in regulating arousal and transitions between sleep states, yet brainwide dynamics underlying these transitions remain incompletely characterized. Here, we employed multi-site fiber photometry with the dopamine sensor GRAB-DA2m, combined with electroencephalogram (EEG)/electromyogram (EMG) recordings, to systematically assess regional dopamine (DA) dynamics across sleep-wake states in the medial prefrontal cortex (mPFC), striatal subregions, central amygdala (CeA), and midbrain nuclei in mice. We found that DA levels prominently increased in the mPFC, dorsolateral striatum (DLS), ventral tegmental area (VTA), substantia nigra pars compacta (SNc), and dorsal raphe nucleus (DRN) during transitions from non-rapid eye movement (NREM) sleep to wakefulness (WAKE), supporting their role in arousal. Conversely, DA decreased in the CeA and nucleus accumbens lateral parts (NAc-L) during these transitions. During transitions from NREM to rapid eye movement sleep (REM), DA elevations were observed in the CeA and middle accumbens subregion (NAc-M), rather than NAc-L, while other regions exhibited decreases. Cross-regional DA correlations revealed synchronized network activity across sleep-wake transitions. Optogenetic activation of VTA and DRN dopamine neurons induced robust DA release in cortical and subcortical regions, and chemogenetic activation promoted wakefulness selectively via VTA and DRN, but not SNc DA neurons. These results elucidate distinct brainwide DA dynamics across state transitions and highlight differential roles for DA signaling in modulating sleep-wake states.

neuroscience↗

A modality gap in personal-genome prediction by sequence-to-function models

Sequence-to-function (S2F) models trained on reference genomes have achieved strong performance on regulatory prediction and variant-effect benchmarks, yet they still struggle to predict inter-individual variation in gene expression from personal genomes. We evaluated AlphaGenome on personal genome prediction in two molecular modalities--gene expression and chromatin accessibility--and observed a striking dichotomy: AlphaGenome approaches the heritability ceiling for chromatin accessibility variation, but remains far below baseline for gene-expression variation, despite improving over Borzoi. Context truncation and fine-mapped QTL analyses indicate that accessibility is governed by local regulatory grammar captured by current architectures, whereas gene-expression variation requires long-range regulatory integration that remains challenging.

bioinformatics↗

Integrating physiology and transcriptomics reveals the mechanisms underlying the differences in cotton (Gossypium hirsutum L.) growth and development under different photoperiods

Photoperiod is a crucial environmental factor that regulates plant growth and flowering. While the photoperiodic flowering pathway has been extensively studied, the mechanisms underlying photoperiodic regulation of cotton (Gossypium hirsutum L.) growth and development remain unclear. In this study, we selected different genotypes of cotton varieties, XJ12 and XJ21-11, to investigate the effects of long-day (LD) and short-day (SD) conditions on cotton photosynthesis, dry matter accumulation, flowering time, and related gene expression, utilizing physiological index measurements, transcriptomics, and WGCNA analysis. The results indicated that LD treatment significantly enhanced photosynthetic parameters and dry matter weight in both cotton varieties, with the net photosynthetic rate (Pn) being 154.79% and 111.86% higher under LD compared to SD treatment. Additionally, the flowering time of the two varieties was advanced by 12 and 13 days, respectively, under LD treatment relative to SD treatment. LD conditions promoted gene expression in sucrose and starch metabolism, facilitating cotton growth. Furthermore, LD inhibited the expression of genes associated with flavonoid biosynthesis, thereby altering the secondary metabolism of cotton, and activated transcription factors through signaling pathways (including phytohormone signaling and MAPK signaling pathways) and biological clock rhythms, which induced earlier flowering in cotton. This study provides new insights into the molecular mechanism of the effect of photoperiod on cotton growth and development.

physiology↗

Atypical developmental remodeling of dopamine neurons involves AKT-GSK3β signaling and glia-mediated axon degeneration

Neuronal remodeling is essential for sculpting neural circuits, and its disruption has been implicated in neurodevelopmental and neuropsychiatric disorders. Yet the molecular and cellular diversity of remodeling across neuron types remains incompletely understood. Here, we uncover a distinct remodeling mode in a subtype of Drosophila dopamine neurons (DANs) critical for learning, memory, sleep, and locomotion. Unlike the stereotypical pruning-then-regrowth paradigm, these DANs undergo a transient axon overgrowth followed by selective pruning during metamorphosis. Remarkably, DAN axon pruning proceeds independently of canonical ecdysone signaling and instead involves neuron-intrinsic AKT-GSK3{beta} signaling and extrinsic glial activity. Disruption of AKT-GSK3{beta} signaling alters microtubule stability and impairs glial recruitment and clearance of axonal debris. Notably, the role of AKT-GSK3{beta} is cell-type specific, underscoring mechanistic diversity in remodeling programs. These findings reveal an unexpected overgrowth-then-pruning developmental trajectory, establishing DANs as a powerful model to uncover the mechanisms underlying neuronal remodeling, circuit maturation, and neurodegeneration.

neuroscience↗

Arabidopsis Group I Pumilio RNA-binding factors are vital for embryo development and balancing between growth and stress resistance

Pumilio (PUM) RNA binding proteins are crucial for regulating gene expression by binding to a conserved motif in the 3'-untranslated region (3'-UTR). Despite their importance, the role of PUM in plants is largely unexplored. Here, we investigated the functions of Arabidopsis group I PUMs (APUM1-APUM6), which are ubiquitously expressed and localized in the cytosol. While single apum mutants exhibit no visible phenotypes, CRISPR/Cas9-generated apum1 apum2 apum3 triple mutants (apum1/2/3) display reduced growth in both vegetative and reproductive organs, alongside hypersensitivity to various stresses. Remarkably, apum1/2/3/4 quadruple mutants are embryonically lethal, highlighting their essential role in embryo development. Transcriptomic profiling revealed that differentially expressed genes (DEGs) upregulated in apum1/2/3 are enriched in pathways related to photosynthesis, stress responses and anthocyanin biosynthesis, while downregulated DEGs are associated with biotic stress defense and hydrogen peroxide metabolism. This suggests that APUM1/2/3 act as molecular hubs balancing plant growth and stress adaptation. Biochemical assays showed that recombinant APUM homologous domains bind to the 5-UGUGUAUA-3 core motif in the 3-UTR of the transcription factor Production of Anthocyanin Pigment1 (PAP1), crucial for anthocyanin biosynthesis. Notably, single nucleotide substitutions, except for the third U, do not affect binding, while multiple mutations disrupt interaction. Consistently, apum1/2/3 mutants accumulate significantly more anthocyanin than wild-type plants. Furthermore, we predicted 7053 potential target genes for APUM1/2/3, with 1609 present among the upregulated DEGs in apum1/2/3. Taken together, our findings demonstrate that group I APUMs are vital posttranscriptional regulators, providing a new perspective on the trade-offs between growth and stress resilience in plants.

plant biology↗

GAME: Genomic API for Model Evaluation

The rapid expansion of genomics datasets and the application of machine learning has produced sequence-to-activity genomics models with ever-expanding capabilities. However, benchmarking these models on practical applications has been challenging because individual projects evaluate their models in ad hoc ways, and there is substantial heterogeneity of both model architectures and benchmarking tasks. To address this challenge, we have created GAME, a system for large-scale, community-led standardized model benchmarking on user-defined evaluation tasks. We borrow concepts from the Application Programming Interface (API) paradigm to allow for seamless communication between pre-trained models and benchmarking tasks, ensuring consistent evaluation protocols. Because all models and benchmarks are inherently compatible in this framework, the continual addition of new models and new benchmarks is easy. We also developed a Matcher module powered by a large language model (LLM) to automate ambiguous task alignment between benchmarks and models. Containerization of these modules enhances reproducibility and facilitates the deployment of models and benchmarks across computing platforms. By focusing on predicting underlying biochemical phenomena (e.g. gene expression, open chromatin, DNA binding), we ensure that tasks remain technology-independent. We provide examples of benchmarks and models implementing this framework, and anticipate that the community will contribute their own, leading to an ever-expanding and evolving set of models and evaluation tasks. This resource will accelerate genomics research by illuminating the best models for a given task, motivating novel functional genomic benchmarks, and providing a more nuanced understanding of model abilities.

bioinformatics↗

Investigating Data Size, Sequence Diversity, and Model Complexity in MPRA-based Sequence-to-Function Prediction

We created the MPRA Dataset Collection (MDC), a curated resource of MPRA data from 12 studies comprising over 150 million labeled DNA subsequences. These datasets include both random and natural genomic sequences paired with diverse functional outputs such as gene expression and splicing efficiency. Using this collection, we build sequence-to-function (S2F) predictive models of regulatory elements and analyze these models to uncover insights into the relationship between training data requirements, experimental design, and model generalizability. By offering a high-quality, machine learning-ready repository, MDC accelerates the development of robust computational tools for deciphering the mechanisms of gene regulation.

bioinformatics↗

A scalable approach to investigating sequence-to-expression prediction from personal genomes

Sequence-to-function (S2F) models hold the promise of evaluating arbitrary DNA sequences, providing a powerful framework for linking genotype to phenotype. Yet, despite strong performance across genomic loci, these models often struggle to capture inter-individual variation in gene expression. To address this, we propose personal genome training--training models to make genotype-specific predictions at a single locus. We introduce SAGE-net, a scalable framework and software package for training and evaluating S2F models using personal genomes. Using SAGE-net, we systematically explore model architectures and training regimes, showing that personal genome training improves gene expression prediction accuracy for held-out individuals. However, performance gains arise primarily from identifying predictive variants, rather than learning a cis-regulatory grammar that generalizes across loci. This lack of generalization persists across a wide range of hyperparameters. In contrast, when applied to DNA methylation (DNAm), personal genome training enables improved generalization to unseen individuals in unseen genomic regions. This suggests that S2F models may more readily capture the sequence-level determinants of inter-individual variation in epigenomic traits. These findings highlight the need for further exploration to unlock the full potential of S2F models in decoding the regulatory grammar of personal genomes. Scalable software and infrastructure development will be critical to this progress.

genomics↗

Comparative spatiotemporal single cell transcriptomes reveal rewiring of pre-existing regulations during emergence of Kranz anatomy in C4 grasses

Many of the worlds most productive food and bioenergy crops use C4 photosynthesis, which have high photosynthetic efficiency due to a Kranz anatomy-based CO2 concentrating mechanism. Here, we took a comparative transcriptomics approach using single cell spatial transcriptomes of leaf primordia for maize (Zea mays) and single cell RNA-seq (scRNA-seq) atlases of corresponding leaf tissues for three C4 species (Zea mays, Sorghum bicolor, Setaria viridis) and one C3 rice (Oryza sativa) to study regulatory networks involved in the development and evolution of Kranz anatomy. We show that the formation of Kranz anatomy involves extensive recruitment and modification of pre-existing regulatory modules, especially the SHR-SCR module and the auxin signaling pathway. Members of the INDETERMINATE DOMAIN (IDD) family transcription factors, including IDD7 and IDDP1, contribute to the modification of the involved modules. These extensive recruitment and modification of existing genetic regulatory programs underlie the repeated emergence of C4 photosynthesis during evolution.

plant biology↗

Pan-cistrome analysis of the leaf accessible chromatin regions of 214 maize inbred lines

Characterizing the noncoding regions of the genome, particularly the cis-regulatory elements (CREs) located in accessible chromatin regions (ACRs) in gene promoters, is crucial for understanding how gene expression is regulated and how genotype contributes to phenotypic diversity in plants.In this study, we used ATAC-seq to resequence and map the ACRs of 214 maize inbred lines. We identified 82,174 ACRs and reported that 39.55% of them exhibited significant variation across the population. Next, we used the accessibility of ACRs as a quantitative feature and performed a chromatin associability GWAS (caGWAS), resulting in 27,004 caQTLs. Among them, 2,463 were predicted to disrupt the cis-regulatory elements that TF binds in these ACRs, suggesting that TF binding affects chromatin accessibility. Specifically, we identified two caACRs that regulate the expression of fad7, which encodes a fatty acid desaturase and affects the linolenic acid content.Our findings highlight the dynamic nature of maize ACRs and demonstrate that chromatin accessibility information could be used for GWASs, offering key insights into the genetic and regulatory mechanisms of chromatin accessibility and its impact on complex traits in plants.

genomics↗

Evolution of cell-type-specific accessible chromatin regions and the cis-regulatory elements that drive lineage-specific innovation

Cis-regulatory elements (CREs) are critical in regulating gene expression, and yet understanding of CRE evolution remains challenging. Here, we constructed a comprehensive single-cell atlas of chromatin accessibility in Oryza sativa, integrating data from 103,911 nuclei representing 126 discrete cell states across nine distinct organs. We used comparative genomics to compare cell-type resolved chromatin accessibility between O. sativa and 57,552 nuclei from four additional grass species (Zea mays, Sorghum bicolor, Panicum miliaceum, and Urochloa fusca). Accessible chromatin regions (ACRs) had different levels of conservation depending on the degree of cell-type specificity. We found a complex relationship between ACRs with conserved noncoding sequences, cell-type specificity, conservation, and tissue-specific switching. Additionally, we found that epidermal ACRs were less conserved compared to other cell types, potentially indicating that more rapid regulatory evolution has occurred in the L1-derived epidermal layer of these species. Finally, we identified and characterized a conserved subset of ACRs that overlapped the repressive histone modification H3K27me3, implicating them as potentially silencer-like CREs maintained by evolution. Collectively, this comparative genomics approach highlights the dynamics of plant cell-type-specific CRE evolution.

plant biology↗

A Supervised Contrastive Framework for Learning Disentangled Representations of Cell Perturbation Data

CRISPR technology, combined with single-cell RNA-Seq, has opened the way to large scale pooled perturbation screens, allowing more systematic interrogations of gene functions in cells at scale. However, such Perturb-seq data poses many analysis challenges, due to its high-dimensionality, high level of technical noise, and variable Cas9 efficiency. The single-cell nature of the data also poses its own challenges, as we observe the heterogeneity of phenotypes in the unperturbed cells, along with the effect of the perturbations. All in all, these characteristics make it difficult to discern subtler effects. Existing tools, like mixscape and ContrastiveVI, provide partial solutions, but may oversimplify biological dynamics, or have low power to characterize perturbations with a smaller effect size. Here, we address these limitations by introducing the Supervised Contrastive Variational Autoencoder (SC-VAE). SC-VAE integrates guide RNA identity with gene expression data, ensuring a more discriminative analysis, and adopts the Hilbert-Schmidt Independence Criterion as a way to achieve disentangled representations, separating the heterogeneity in the control population from the effect of the perturbations. Evaluation on large-scale data sets highlights SC-VAEs superior sensitivity in identifying perturbation effects compared to ContrastiveVI, scVI and PCA. The perturbation embeddings better reflect known protein complexes (evaluated on CORUM), while its classifier offers promise in identifying assignment errors and cells escaping the perturbation phenotype. SC-VAE is readily applicable across diverse perturbation data sets.

bioinformatics↗

Investigating the cis-Regulatory Basis of C3 and C4 Photosynthesis in Grasses at Single-Cell Resolution

While considerable knowledge exists about the enzymes pivotal for C4 photosynthesis, much less is known about the cis-regulation important for specifying their expression in distinct cell types. Here, we use single-cell-indexed ATAC-seq to identify cell-type-specific accessible chromatin regions (ACRs) associated with C4 enzymes for five different grass species. This study spans four C4 species, covering three distinct photosynthetic subtypes: Zea mays and Sorghum bicolor (NADP-ME), Panicum miliaceum (NAD-ME), Urochloa fusca (PEPCK), along with the C3 outgroup Oryza sativa. We studied the cis-regulatory landscape of enzymes essential across all C4 species and those unique to C4 subtypes, measuring cell-type-specific biases for C4 enzymes using chromatin accessibility data. Integrating these data with phylogenetics revealed diverse co-option of gene family members between species, showcasing the various paths of C4 evolution. Besides promoter proximal ACRs, we found that, on average, C4 genes have two to three distal cell-type-specific ACRs, highlighting the complexity and divergent nature of C4 evolution. Examining the evolutionary history of these cell-type-specific ACRs revealed a spectrum of conserved and novel ACRs, even among closely related species, indicating ongoing evolution of cis-regulation at these C4 loci. This study illuminates the dynamic and complex nature of CRE evolution in C4 photosynthesis, particularly highlighting the intricate cis-regulatory evolution of key loci. Our findings offer a valuable resource for future investigations, potentially aiding in the optimization of C3 crop performance under changing climatic conditions.

plant biology↗

GetPrimers: a generalized PCR-based genetic targeting primer designer enabling easy and standardized targeted gene modification across multiple systems

Genetic targeting (e.g., gene knockout and tagging) based on polymerase chain reaction (PCR) is a simple yet powerful approach for studying gene functions. Although originally developed in classic budding and fission yeast models, the same principle applies to other eukaryotic systems with efficient homologous recombination. One-step-PCR-based genetic targeting is conventionally used but the sizes of the homologous arms that it generates for recombination-mediated genetic targeting are usually limited. Alternatively, gene targeting can also be performed via fusion PCR, which can create homologous arms that are orders of magnitude larger, therefore substantially increases the efficiency of recombination-mediated genetic targeting. Here we present GetPrimers (https://www.evomicslab.org/app/getprimers/), a generalized computational framework and web tool to assist automatic targeting and verification primer design for both one-step-PCR-based and fusion-PCR-based genetic targeting experiments. Moreover, GetPrimers by design runs for any given genetic background of any species with full genome scalability. Therefore, GetPrimers is capable of empowering high-throughput functional genomics assays at multi-population and multi-species levels. Comprehensive experimental validations have been performed for targeting and verification primers designed by GetPrimers across multiple organism systems and experimental setups. We anticipate GetPrimers to become a highly useful and popular tool to facilitate easy and standardized gene modification across multiple systems. Take AwaysDeveloped GetPrimers for generalized primer design of PCR-based genetic targeting. GetPrimers shines with its high versatility and genome-wide scalability. GetPrimers was validated across multiple organisms and experimental setups.

genetics↗