Biology subjects

Kawaguchi, K.

Publications and source records attributed to Kawaguchi, K..

At least 19 recordsLinked to original sources

bioRxiv2026

Binary node clustering via contrastive learning for haplotype phasing in de novo genome assembly

Accurate haplotype phasing is essential for high-quality genome assembly, yet de novo phasing of complex genomes without parental data remains challenging. We formulate haplotype phasing as a node clustering problem with overlapping clusters on augmented unitig graphs, where nodes represent contiguous, non-branching DNA sequence fragments and two edge types can encode sequence overlap or Hi-C proximity information. We introduce a contrastive learning framework with a custom objective function and train a graph-transformer-based model, termed grapHiC, to phase paternal, maternal, and homozygous unitig nodes. grapHiC; is the first machine-learning-based method to perform reference-free haplotype phasing and the first approach to directly phase raw unitig graphs without prior simplification. We show that grapHiCaccurately clusters nodes on human-genome-scale graphs and that its predictions can effectively guide phased de novo genome assembly, producing human assemblies with contiguity and phasing quality comparable to the state of the art when integrated with the DipGNNome assembler.

bioinformatics↗

bioRxiv2026

OddCAPS: a simple, low-cost, universal technique for detecting single nucleotide variants

Derived Cleaved Amplified Polymorphic Sequences (dCAPS) assays have been widely performed historically to detect known base substitutions in many model organisms--notably Caenorhabditis elegans, Arabidopsis thaliana, and yeast--where chemical mutagens that induce point mutations are frequently used. With the rise of whole-genome sequencing and genome editing technologies, dCAPS is increasingly applied to detect diverse nucleotide changes across a broader range of species. However, a key limitation of dCAPS is that genomic target sites amenable to primer designs that both preserve PCR amplification and create recognition sites for inexpensive, high-performance restriction enzymes are scarce. Here we report One-step dual-primer dCAPS (OddCAPS), a modification that uses three primers in one reaction to overcome this constraint. Two of these primers, an intermediate primer and a dCAPS primer, sequentially introduce 1-2 base substitutions each into the amplicon, enabling up to four engineered base changes near the nucleotide of interest. By using the intermediate primer at 1/10-1/100 the concentration of the other primers, the desired product is generated directly in a single-tube, one-step PCR. Increasing the number of engineered substitutions improves the chance of using a researchers preferred restriction enzyme. In principle, having eight common restriction enzymes (BamHI, EcoRI, NheI, SalI, BglII, ClaI, HindIII, and MluI) suffices to detect any single-nucleotide variant in any biological or synthetic DNA sequence with this approach.

molecular biology↗

bioRxiv2026

Topological defects and coherent myocardial chirality shape torsional heart contraction

The efficient pumping of the mammalian heart relies on torsional contractile motion generated by its highly ordered three-dimensional (3D) architecture of myocardial fibres. However, the topological principles governing how its complex geometry translates into contractile mechanics remain elusive. Here, we show that the mammalian heart forms a chiral nematic field, a biological analogue to 3D liquid crystals, whose topological organisation underlies its function. Analysis of 3D imaging data revealed disclination lines, continuous assemblies of topological defects characteristic of nematic systems, within the compact myocardium. Finite-element simulations reveal that these defects are not mere structural irregularities but can locally modulate contractile behaviour and reduce mechanical work. In heterotaxy hearts with reversed global anatomy (situs inversus), myocardial fibres retain a predominantly counter-clockwise twist, similar to that of the normal heart, but with a small clockwise component near the base. This decoupling of tissue-level chirality from systemic left-right patterning suggests that cardiac twist is an intrinsic property of the myocardial fibre. Mechanical simulations of situs inversus heart demonstrate that the coherence of transmural chirality, rather than its specific orientation, is critical for contractile efficiency. Together, these findings establish the heart as a topological material and reveal how organised chiral fields generate robust organ-level mechanical function.

biophysics↗

bioRxiv2026

Coding sequence clustering universally predicts fine- and coarse-scale chromatin compartment landscapes

Intra-chromosomal contact maps from many species display a striking plaid pattern, reflecting chromatin compartmentalization at a sub-chromosomal scale. Although widely regarded as a core feature of genome architecture, such patterns are absent in many organisms, and their underlying determinants remain unclear. Here we systematically examine the relationships among sequence features, chromatin compartmentalization, and evolutionary conservation across 247 species spanning five kingdoms. By testing the usual genomic suspects as determinants of compartmentalization within and between species, we identify the coding DNA sequence (CDS) density landscape as the most consistent predictor of compartmentalization, including whether an organism exhibits a fine-scale plaid or coarse-scale non-plaid chromatin contact map. In contrast, correlations with GC content, CpG density, or repeat element composition vary across lineages, contextualizing long-standing observations such as the prominence of chromosome G-banding in amniotes and its absence elsewhere. Notably, compartment organization is conserved across syntenic blocks between species separated by up to 1 billion years of evolution, and this conservation tracks with preservation of CDS density profiles rather than other sequence features. These findings establish the genomic distribution of coding sequences as a universal and deeply conserved organizing principle of nuclear compartment architecture.

evolutionary biology↗

bioRxiv2026

CDK12/13 inhibitor, CTX-439, suppresses tumor growth and potentiates BCL-2 family blockade

CDK12 and CDK13 (CDK12/13) regulate transcription by phosphorylating Serine 2 (S2) of the C-terminal domain of RNA polymerase II and have been proposed as therapeutic targets in cancer. Here we report the development of CTX-439, a novel, orally bioavailable, ATP-competitive small-molecule CDK12/13 inhibitor. CTX-439 specifically inhibits S2 phosphorylation and downregulates many genes including those involved in DNA damage repair, thereby exerting a profound anti-cancer effect in vitro and in vivo including breast cancer PDX models. A CRISPR activation screen identified BCL-2 and BCL-xL, anti-apoptotic BCL-2 family members, as genes that when upregulated confer resistance to CTX-439. Simultaneous inhibition of BCL-2/BCL-xL and CDK12/13 rapidly induced apoptosis and significantly suppressed xenograft tumor growth. Mechanistically, CTX-439 downregulates MCL1 protein levels through transcriptional readthrough, shifting cell survival dependency to BCL-2 and BCL-xL. Our study provides novel insights into the anti-tumor effect of CDK12/13 inhibition and proposes a new combination therapy strategy with anti-apoptotic BCL-2 family inhibitors, which may improve therapeutic outcomes in cancer treatment.

cancer biology↗

bioRxiv2025

ADGRG1 blockade drives astrocyte positioning into the extracellular matrix to reduce scar size

After the central nervous system (CNS) injury, inflammatory cells enter the lesion core and a fibrotic extracellular matrix accumulates. Astrocytes subsequently cluster around this matrix, forming a glial scar, but the role of the extracellular matrix in scar formation remains poorly understood. Here, we identify type III collagen, a major fibrotic extracellular matrix component after spinal cord injury, as a key regulator of astrocyte behavior. Using single-nucleus transcriptomics, spatial transcriptomics, and time-lapse cell imaging, we show that type III collagen positions astrocytes at the boundary of the fibrotic matrix by activating the adhesion G protein-coupled receptor G1 (ADGRG1) and its downstream effector RhoA. Blocking ADGRG1 genetically or pharmacologically allows astrocytes to enter the fibrotic matrix, resulting in a smaller scar, improved neuronal regeneration, and better motor recovery. The migration of human-induced pluripotent stem cell-derived astrocytes was inhibited by type III collagen via the ADGRG1-RhoA pathway, supporting their therapeutic potential. Our findings reveal the importance of controlling glial scar positioning via the type III collagen-ADGRG1 axis, providing a novel therapeutic target for CNS injury.

neuroscience↗

bioRxiv2025

DipGNNome: Diploid de novo genome assembly with geometric deep learning and beam-search

De novo genome assembly remains a central challenge in computational biology, particularly for diploid genomes where maternal and paternal haplotypes must be accurately resolved. Existing assemblers achieve impressive results through carefully designed heuristics, yet modern deep learning methods remain largely unexplored in the diploid setting. We present DipGNNome, the first deep learning-based framework for diploid de novo genome assembly. Our approach formulates assembly as an edge classification and traversal problem on haplotype-aware assembly graphs, training graph neural networks (GNNs) to guide contig construction. To enable this, we establish a novel pipeline for generating diploid graphs with ground-truth edge labels, providing the first systematic way to produce training data for machine learning models in this domain. This framework creates a foundation for applying and extending graph-based deep learning to diploid assembly. DipGNNome creates assemblies comparable to state-of-the-art and demonstrates the feasibility of deep learning for diploid assembly and introduces a paradigm that bridges algorithmic genomics with graph representation learning. Our code, dataset and trained model is openly available at https://github.com/lbcb-sci/DipGNNome.

genomics↗

bioRxiv2025

Fc-Free Single Chain Antibody mRNA Therapy for Airway Infection of Multidrug-Resistant Pseudomonas Aeruginosa

With the growing threat of antimicrobial resistance (AMR), alternatives to antibiotics are urgently needed. In this study, we developed mRNA-based therapeutics encoding single-chain variable fragment (scFv) antibodies that target the type III secretion system of Pseudomonas aeruginosa, a key AMR pathogen. Delivered intravenously via lipid nanoparticles, the scFv mRNA effectively protected mice from airway infections by mitigating lung inflammation, reducing bacterial load, and improving survival. Notably, this approach proved effective in clinically relevant models involving immunocompromised mice infected with multidrug-resistant, exoU-positive (highly cytotoxic) clinical isolates. The study also explores the potential of Fc-free scFv formulations, leveraging mRNA technology to overcome their short half-life through sustained protein production. Importantly, Fc-free scFv antibodies migrated more efficiently from the bloodstream to the airway epithelium--the primary site of infection--than their Fc-conjugated counterparts, resulting in efficient therapeutic outcomes. Overall, mRNA-encoded Fc-free scFv antibodies represent a promising strategy for combating AMR infections.

immunology↗

bioRxiv2025

Sequence grammar and dynamics of subcellular translation revealed by APEX-Ribo-Seq

Local translation at specific subcellular regions is proposed to define appropriate protein destinations and functions. However, our understanding of local translation still remains far from complete due to a lack of versatile analytical tools. Here, we developed a new method, termed APEX-Ribo-Seq, that integrates ribosome profiling (Ribo-Seq) with APEX2-based proximity labeling. We mapped local translation across 14 distinct organelles and identified over 3,000 genes undergoing localized translation, encoding components of complexes that function at these specific sites. RNA language-model predictors captured compartment-associated information beyond protein sequence, and model-derived attribution nominated sequence-sensitive transcript regions for mechanistic follow-up. Gene-set-level RBP enrichment and knockdown experiments separately identified RBPs associated with compartment-specific translation. The utility of APEX-Ribo-Seq was demonstrated in primary neurons, revealing compartment-specific translatomes in dendrites, axons, and presynapses. Extending this approach in vivo, we applied APEX-Ribo-Seq to germ granules in Drosophila embryos, profiling the temporal transition from translationally silent mRNA storage to localized translation during early embryogenesis. Our results provide a comprehensive view of the spatiotemporal translatome and its regulation.

molecular biology↗

bioRxiv2025

Linear ubiquitination triggers Amph-mediated T-tubule biogenesis

T-tubules are specialized invaginations of the plasma membrane essential for muscle contraction. While their physiological importance is well established, the mechanisms underlying T-tubule formation remain elusive. Here, we identify LUBEL/RNF31, a ubiquitin E3 ligase responsible for linear (M1-linked) ubiquitination, as a key regulator of T-tubule biogenesis through proximity proteomics and RNAi screening in Drosophila. Loss of LUBEL leads to the formation of Amphiphysin (Amph)-positive membrane sheets instead of tubular networks in muscle cells. Mechanistically, the ubiquitin ligase activity of LUBEL, and direct interaction with Amph, a BAR domain protein involved in membrane tubule extension, are crucial for proper T-tubule morphology. LUBEL and M1-linked ubiquitin chains assemble into condensates on membranes, facilitating Amph-mediated membrane tubulation. Notably, the Amph-LUBEL/RNF31 interaction is evolutionarily conserved across a broad range of species, underscoring a fundamental role for linear ubiquitination in membrane remodeling. Our findings uncover a previously unrecognized role for linear ubiquitination in membrane deformation driven by BAR domain proteins.

cell biology↗

bioRxiv2025

Distribution of myogenic stem cell activator, hepatocyte growth factor, in skeletal muscle extracellular matrix and effect of short-term disuse and reloading

Hepatocyte growth factor (HGF) is a key myogenic stem cells (satellite cells) activator, that resides in the extracellular matrix (ECM). However, HGF distribution in the ECM varies depending on the muscle fiber type. Furthermore, aging impedes the binding of HGF to its receptors due to nitration by peroxynitrite (ONOO-). Though oxidative stress increases rapidly during muscle disuse atrophy, satellite cells are rapidly activated upon reloading. In this study, we investigated the distribution of HGF in the ECM in various muscle fiber types, and examined nitration of HGF in disuse and reloading models. Immunofluorescence staining was performed on the soleus (Sol), plantaris (Pla), and gastrocnemius (Gas) muscles of 10-week-old mice. Three mice were used to assess HGF distribution, while 12 mice, divided into control, disuse, and reloading groups were used for qualitative evaluation of nitrated HGF (nitroHGF). In Sol muscle, type IIa and IIx muscle fibers exhibited higher HGF distribution in the ECM (70.7{+/-}1.9% and 63.9{+/-}2.6%, respectively) than type I fibers (28.0{+/-}1.7%; p<0.001). In Pla and Gas muscle, type IIa (65.4{+/-}2.3% and 70.3{+/-}2.2%, respectively) and type IIx fibers (55.3{+/-}1.8% and 61.4{+/-}3.2%, respectively) had significantly higher HGF distribution in the ECM than type IIb fibers (13.1{+/-}1.8% and 5.9{+/-}1.6%; p<0.001, respectively). The amount of nitroHGF increased in the disuse group compared to that in the control group but decreased in the reloading group compared to that in the disuse group. This preferential HGF distribution around type IIa and IIx muscle fibers indicates a distinct mechanism for satellite cell activation, differing from the satellite cell-rich environment associated with type I fibers and the lower HGF association with type IIb fibers. Disuse-induced HGF nitration may inhibit satellite cell activation. Reloading likely triggers mechanisms that counteract nitration, enabling satellite cell reactivation in young muscle.

biochemistry↗

bioRxiv2024

Transcriptional Dynamics Uncover the Role of BNIP3 in Mitophagy during Muscle Remodeling in Drosophila

Differentiated muscle cells contain myofibrils and well-organized organelles, enabling powerful contractions. Muscle cell reorganization occurs in response to various physiological stimuli; however, the mechanisms behind this remodeling remain enigmatic due to the lack of a genetically trackable system. Previously, we reported that a subset of larval muscle cells is remodeled into adult abdominal muscle through an autophagy-dependent mechanism in Drosophila. To unveil the underlying mechanisms of this remodeling, we performed a comparative time-course RNA-seq analysis of isolated muscle cells with or without autophagy. It revealed both transcriptional dynamics independent of autophagy and highlighted the significance of BNIP3-mediated mitophagy in muscle remodeling. Mechanistically, we found that BNIP3 recruits autophagic machinery to mitochondria through its LC3-interacting (LIR) motif and minimal essential region (MER), which interact with Atg8a and Atg18a, respectively. Loss of BNIP3 leads to a substantial accumulation of larval mitochondria, ultimately impairing muscle remodeling. In summary, this study demonstrates that BNIP3-dependent mitophagy is critical for orchestrating the dynamic process of muscle remodeling. Impact statementA time-course RNA-seq analysis of muscle remodeling reveals transcriptional dynamics independent of autophagy and highlights the significance of BNIP3 in mitochondrial degradation in vivo.

cell biology↗

bioRxiv2024

Engineered acetylation patterns drive large-scale chromatin organization in vitro

Chromatin organization plays a crucial role in gene regulation [1, 2, 3], but disentangling the contributions of various epigenetic components to gene-scale chromatin structure remains challenging. While in vitro chromatin reconstitution enables controlled studies on the effect of bio-chemical factors on the structure, current methods are either limited to short arrays or lack control over histone modification patterns. Here we directly test how histone modification affects higher-order chromatin architecture by characterizing gene-scale reconstituted chromatin using single-molecule microscopy and in vitro Hi-C. We reconstitute 20-kilobase chromatin arrays with histone modification patterns controlled at 12-nucleosome resolution, achieving complete assembly of 96 nucleosomes in the designed order as confirmed by atomic force microscopy and longread sequencing. Observing end-to-end fluctuations of the reconstituted arrays, we find that increasing the density of acetylated nucleosomes leads to larger structural fluctuations with longer relaxation times, consistent with the predictions of a polymer model with hydrodynamic interactions. We demonstrate through in vitro Hi-C how acetylation reduces contact frequency between nucleosomes and induces open conformations. In heterogeneously modified arrays, differential contact probabilities between acetylated and unmodified regions lead to distinct structural domains. The results establish the physical principles by which histone modifications directly modulate chromatin architecture through altered nucleosome-nucleosome interactions, providing a quantitative framework for understanding and engineering genome organization.

biophysics↗

bioRxiv2024

Integer topological defects provide a new way to quantify and classify cell sheets

Sheets of confluent cells are often considered as active nematics, with accumulation at [Formula] topological defects and escape from [Formula] defects being widely recognized. However, collective dynamics surrounding integer-charge defects remain poorly understood, despite its biological importance. By using microfabricated patterns, we induce diverse +1 topological defects (aster, spirals, and target) within monolayers of neural progenitor cells. Remarkably, cells are consistently attracted to the core of +1 defects regardless of their type, challenging existing theories and the conventional extensile/contractile dichotomy. We trace back the origin of this accumulation behavior to previously overlooked nonlinear active forces using a combination of experiments and a continuous theory derived from a cell-level model. Our findings demonstrate that +1 topological defects can reveal key features of active nematic systems and offer a new way to characterize and classify cell layers.

biophysics↗

bioRxiv2024

Homeotic and non-homeotic patterns in the tetrapod vertebral formula

Vertebrae can be differentiated into five categories along the body axis in tetrapods, with its numerical distribution known as the vertebral formula. The vertebral formula is a principal tool for connecting development and phylogeny [1]. This is largely due to its robust relationship with the conserved clusters of Hox genes [2], which exhibit expression boundaries coincident with vertebral divisions [3-11]. One avenue for variations in the vertebral formula is thus through Hox-mediated homeotic transformations, which manifest as a relatively fixed sum of adjacent vertebral counts. This expectation is borne out in the mammalian thoracolumbar count [12], but to date, no similar vertebral patterns have been found. Here we conduct a systematic search by generating a large dataset of complete vertebral formulae in a diverse range of tetrapod species and probing the variance of linear combinations of vertebrae. We uncover additional mammalian homeotic patterns, but also unexpected balances between distal vertebrae not comprehensible with Hox-mediated regionalization. One distal pattern appears during the progression from theropods to birds, demonstrating its phylogenetic importance. We further show that several vertebral counts correlate with posterior intergenic distances in the HoxB gene cluster. By creating a vertebral formula database and mathematically defining patterns, our work establishes a quantitative approach for comparative genomics in morphology.

evolutionary biology↗

bioRxiv2024

Transcription of damage-induced RNA in Arabidopsis was frequently initiated from DSB loci within the genic regions

DNA double-strand breaks (DSBs) are the most severe DNA lesions and need to be removed immediately to prevent loss of genomic information. Recently, it has been revealed that DSBs induce de novo transcription from the cleavage sites in various species, resulting in RNAs being referred to as damage-induced RNAs (diRNAs). While diRNA synthesis is an early event in the DNA damage response and plays an essential role in DSB repair activation, the location where diRNAs are newly generated in plant remains unclear, as does their transcriptional mechanism. Here, we performed RNA sequencing of diRNAs that emerged around all DSB loci in Arabidopsis thaliana under the expression of the exogenous restriction enzyme Sbf I and observed 88 diRNAs in 360 DSB loci. Most of the detected diRNAs originated within the active genes and were transcribed from DSBs in a bidirectional manner. Furthermore, we found that diRNA elongation tends to terminate at the boundary of an endogenous gene located near DSB loci. Our results provide reliable evidence for understanding the importance of de novo transcription at DSBs and show that diRNA is a crucial factor for successful DSB repair.

molecular biology↗

bioRxiv2024

Geometric deep learning framework for de novo genome assembly

The critical stage of every de novo genome assembler is identifying paths in assembly graphs that correspond to the reconstructed genomic sequences. The existing algorithmic methods struggle with this, primarily due to repetitive regions causing complex graph tangles, leading to fragmented assemblies. Here, we introduce GNNome, a framework for path identification based on geometric deep learning that enables training models on assembly graphs without relying on existing assembly strategies. By leveraging symmetries inherent to the problem, GNNome reconstructs assemblies with similar or superior contiguity compared to the state-of-the-art tools across several species, sequenced with PacBio HiFi or Oxford Nanopore. With every new genome assembled telomere-to-telomere, the amount of reliable training data at our disposal increases. Combining the straightforward generation of abundant simulated data for diverse genomic structures with the AI approach makes the proposed framework a plausible cornerstone for future work on reconstructing complex genomes with different ploidy and aneuploidy degrees. To facilitate such developments, we make the framework and the best-performing model publicly available, provided as a tool that can directly be used to assemble new haploid genomes.

bioinformatics↗

bioRxiv2023

Low albumin status accompanies multi-layered immunosuppressive phenotypes in metastatic breast cancer patients

Low albumin status is prevalent in advanced cancer patients, but the pathophysiology associated with this anomaly remains largely unexplored. To address this, we aim to search correlations of albumin levels with the transcriptome against peripheral blood mononuclear cells and the plasma metabolome within the same patients having metastatic breast cancers. We confirm that metastatic breast cancer patients exhibit low albumin levels in varying degrees without prominent systemic inflammation. Our data demonstrate that low albumin levels correlate with transcriptome signatures indicative of "neutrophil activation and T-cell down-regulation," an immunosuppressive phenotype. We also find that immunoregulatory metabolites, such as arginine, are reduced in plasma in an albumin-correlated manner, further corroborating systemic immunosuppression. These results are verified using a mouse model of breast cancer. We conclude that low albumin status in metastatic breast cancer patients accompanies immunosuppressive phenotypes, which is likely unfavorable for anti-cancer immunotherapy and thus can be a cause of unsuccessful treatment outcomes.

cancer biology↗