Biology subjects

Fujisawa, T.

Publications and source records attributed to Fujisawa, T..

15 recordsLinked to original sources

bioRxiv2026

SPATNIC enables H&E-grounded cancer-cell identification across spatial and single-cell transcriptomic modalities in colorectal cancer

Tumors contain diverse non-cancer cells that constitute the tumor microenvironment; thus, identifying which cells are cancerous is a foundational and critical first step in cancer research. However, to our knowledge, no transcriptome-based classifier has been trained using pathologist-defined annotations from matched H&E images at single-cell resolution. Moreover, few tools can be shared across diverse platforms. Here, we present SPATNIC, a variational autoencoder that classifies individual cells as cancer or normal, trained on Xenium cells labeled by a pathologist review of matched H&E images. It matches or exceeds the performance of existing tools in primary colorectal tumors and liver and lung metastases. SPATNIC generalizes zero-shot across Visium, Visium HD, CosMx, Xenium, MERSCOPE, and single-cell RNA sequencing. In public single-cell atlases, it identified a rare POU2F3 tuft-like cancer stem cell population missed by existing annotations and associated with worse survival. In rectal cancer profiled before and after chemoradiotherapy, it separated cancer from normal epithelium in tissue, thus enabling per-patient screening of antibody-drug conjugate (ADC) targets.

cancer biology↗

bioRxiv2026

Catalog of gut microbiota alterations associated with anticancer therapies across multiple cancer types

BackgroundAnticancer therapies can alter the gut microbiota and may affect gut bacteria associated with treatment response. However, most longitudinal studies have focused on specific cancer types or individual treatment regimens, and systematic analyses across diverse cancer therapies remain limited. We analyzed longitudinal fecal microbiota profiles using 16S ribosomal RNA gene amplicon sequencing in the pan-cancer SCRUM-Japan MONSTAR-SCREEN cohort. We included 528 paired pre- and post-treatment fecal samples from 264 patients with advanced solid tumors across 18 cancer types and characterized the gut microbiota alterations associated with 22 anticancer drugs and related clinical factors. ResultsAcross the cohort, Shannon diversity did not significantly change after treatment (mean, 3.81 vs. 3.78; P = 0.58), and pre- and post-treatment samples exhibited no clear separation in ordination space. However, within-patient analysis detected a subtle but significant longitudinal microbiota shift (paired PERMANOVA, P = 0.0001), highlighting the importance of accounting for paired sampling. Clustering of genus-level compositional alterations revealed patient groups with distinct degrees of microbiota alteration, with the largest shifts associated with antibiotic exposure, transition from normal stool to diarrhea, and specific treatment regimens. Multivariable regression analysis of 22 anticancer drugs identified drug-bacteria associations and demonstrated that drugs with similar mechanisms of action, including epidermal growth factor receptor (EGFR) inhibitors and immune checkpoint inhibitors, exhibited similar microbiota change profiles. Targeted analyses highlighted concordant reductions in the Christensenellaceae R-7 group among EGFR inhibitor-exposed patients and depletion of Faecalibacterium among immune checkpoint inhibitor-exposed patients. ConclusionsThis study provides a cross-cancer catalog of microbiota alterations associated with anticancer therapies and highlights therapy-related shifts in the gut ecosystem, including patterns shared by drugs with similar mechanisms of action.

microbiology↗

bioRxiv2026

TogoMCP: Natural Language Querying of Life-Science Knowledge Graphs via Schema-Guided LLMs and the Model Context Protocol

Querying the RDF Portal knowledge graph maintained by DBCLS--which aggregates approximately 60 life-science databases--requires proficiency in both SPARQL and database-specific RDF schemas, placing this resource beyond the reach of most researchers. Large Language Models (LLMs) can, in principle, translate natural-language questions into executable SPARQL, but without schema-level context, they frequently fabricate non-existent predicates or fail to resolve entity names to database-specific identifiers. We present TogoMCP, a system that recasts the LLM as a protocol-driven inference engine orchestrating specialized tools via the Model Context Protocol (MCP). Two mechanisms are essential to its design: (i) the MIE (Metadata-Interoperability-Exchange) file, a concise YAML document that dynamically supplies the LLM with each target databases structural and semantic context at query time; and (ii) a two-stage workflow separating entity resolution via external REST APIs from schema-guided SPARQL generation. On a benchmark of 50 biologically grounded questions spanning five types and 23 databases, TogoMCP achieved a large improvement over an unaided baseline (Cohens d = 1.82, Wilcoxon p < 0.001), with win rates exceeding 80% for question types with precise, verifiable answers. An ablation study shows that all component configurations deliver significant improvements, with MIE schema files providing the largest marginal contribution on mean per-question score ({Delta} = +0.50 relative to a no-MIE condition, two-sided Wilcoxon p = 0.067; 90% bootstrap CI [+0.04, +0.94] excludes zero); a one-line instruction to load the relevant MIE file recovers the same mean improvement as a full procedural protocol, while the protocol additionally reduces downside risk (loss rate 1.6% vs. 4.8%, Fisher p = 0.036). These results suggest a general design principle: concise, dynamically delivered schema context is more valuable than complex orchestration logic for mean-score performance, while procedural guidance plays a complementary role in narrowing variance. Database URL: https://togomcp.rdfportal.org/

bioinformatics↗

bioRxiv2025

Reliable repurposing of antibody interactome inside the cell

In biology proximity is paramount and eighty-five percent of the human proteome has at least one documented interacting monoclonal antibody. These molecules penetrate the cytoplasm poorly and are very often non-functional within the cell. Sequence analysis of 106 antibody variable domains alongside the cytoplasmic human proteome shows charge and isoelectric point are characteristics ill adapted to intracellular monodispersity. Characterisation of forty-five single-chain variable fragment (scFv) intrabodies expressed in human cells confirmed charge to have the greatest impact on solubility. We created new interdomain linkers, optimised scFv domain orientation and found variable heavy domain framework sites to be generally positively charged, and promote insolubility, but be amenable to optimisation. This is applied in combination to reduce the search space and refine the products of AI-led inverse folding to create highly soluble, abundant and thermally stable intrabodies that maintain parent antibody epitope recognition. Over six hundred intrabody sequences are described targeting sixty cytoplasmic proteins with linear, conformational, post-translational modification or oligomeric state specificity. Interactions were validated for p53, -synuclein, SOD1, polyQ, FUS/TLS, UCHL1 and GFP. This approach removes obstacles hindering intracellular repurposing of the vast sequenced antibody interactome with applications relevant to many human disease states.

molecular biology↗

bioRxiv2025

Performance and limitations of out-of-distribution detection for insect DNA (meta)barcoding

Successful applications of DNA barcoding/metabarcoding rely on the accurate taxonomic identification of sequence fragments. When biological surveys with DNA (meta)barcoding target underexplored biological communities, sequence-based identification is often conducted using incomplete databases that do not fully cover the regional species pool. Consequently, specimens to be identified may include species not present in reference databases. Such unknown or "out-of-distribution" samples can cause misidentification if left undetected. A similarity cutoff is commonly used to detect out-of-distribution samples before taxonomic assignment, but its effectiveness has not been carefully studied. In this study, we evaluated the performance of out-of-distribution detection for DNA barcoding with genetic distance and deep learning metrics. Using extensively sampled datasets of multiple insect taxa, we measured the performance of identification and out-of-distribution detection under conditions in which genetic variations in species were sufficiently sampled. Although identification with DNA barcoding is a highly accurate process, even with short noisy fragments, out-of-distribution detection was more susceptible to a reduction in performance due to sequence noise and a lack of diagnosable characters. Our results provide guidelines for designing unknown-proof identification procedures by determining factors affecting out-of-distribution detection performance.

bioinformatics↗

bioRxiv2025

MarpolBase: Genome database for Marchantia polymorpha featuring high quality reference genome sequences

The liverwort Marchantia polymorpha is a key model organism for understanding land plant evolution, development, and gene regulation. To support the growing demand for high-quality genomic resources, we present MarpolBase, a comprehensive and integrated genome database that hosts newly assembled, high-accuracy reference genomes for both the male Tak-1 and female Tak-2 accessions, designated as ver. 7.1 reference genomes. These new assemblies, generated using PacBio HiFi long-read sequencing, represent nearly telomere-to-telomere chromosome-level genomes, with improvements in assembly continuity, annotation accuracy, and structural resolution--especially for repeat-rich regions and sex chromosomes. MarpolBase offers not only access to genome sequences and gene annotations but also provides a unified platform for data exploration, comparative analysis, and community-driven gene nomenclature. It includes keyword-searchable gene pages with structural and functional annotations, expression data integration, genome browser visualization, and online analytical and utility tools. By unifying genome assembly, annotation, nomenclature, and analysis tools in a single platform, MarpolBase serves as a central resource for functional genomics and evolutionary studies in M. polymorpha, and a model for future plant genome databases.

genomics↗

bioRxiv2024

Priming Epigenetic Landscape at Gene Promoters through Transcriptional Activation in Mammalian Germ Cells

In mammalian germ cells, a cycle of erasing and resetting epigenetic information is crucial for sex-specific gametogenesis and early embryogenesis. Specific chromatin states at promoters allow genes to be expressed at the appropriate time during germ cell development, however, the mechanism that establishes such characteristic chromatin states remains unclear. Here, we show that, in mouse male embryonic germ cells, promoters of developmental and housekeeping genes are selectively protected from de novo DNA methylation, coinciding with their transient upregulation and genome-wide erasure of H3K27me3. Importantly, a specific level of H3K4me3 density at the promoters serves as a potent probe that distinguishes between hypo- and hyper-DNA methylation states. Subsequent restoration of H3K27me3 contributes to establishing bivalent chromatin states that prime genes for activation at later developmental stages. These findings reveal a molecular framework shaping the epigenetic landscape at promoters with long-term effects beyond germ cell development.

genomics↗

bioRxiv2024

Giant extrachromosomal element "Inocle" potentially expands the adaptive capacity of the human oral microbiome

Survival strategy of bacteria is expanded by extrachromosomal elements (ECEs). However, their genetic diversity and functional roles for adaptability are largely unknown. Here, we discovered a novel family of intracellular ECEs using 56 saliva samples by developing an efficient microbial DNA extraction method coupled with long-read metagenomics assembly. Even though this ECE family was not hitherto unidentified, our global prevalence analysis using 476 salivary metagenomic datasets elucidated that these ECEs reside in 74% of the population. These ECEs, which we named, "Inocles", are giant plasmid-like circular genomic elements of 395 kb in length, having Streptococcus as a host bacterium. Inocles encode a series of genes that contribute to intracellular stress tolerance, such as oxidative stress and DNA damage, and cell wall biosynthesis and modification involved in the interactions with oral epithelial cells. Moreover, Inocles exhibited significant positive correlations with immune cells and proteins responding to microbial infection in peripheral blood. Intriguingly, we examined and found their marked reductions among 68 patients of head and neck cancers and colorectal cancers, suggesting its potential usage for a novel biomarker of gastrointestinal cancers. Our results suggest that Inocles potentially boost the adaptive capacity of host bacteria against various stressors in the oral environment.

genomics↗

bioRxiv2024

DFAST_QC: Quality Assessment and Taxonomic Identification Tool for Prokaryotic Genomes

MotivationAccurate taxonomic assignments of genomic data are crucial across various biological databases. With a rapid increase in submitted genomes in recent years, ensuring precise classification is important to maintain database integrity. Mislabeled genomes can confuse researchers, hinder analyses, and produce false results. Therefore, there is a critical need for computationally efficient tools that ensure accurate taxonomic classification for data to be deposited into genomic databases. ResultsHere we introduce DFAST_QC, a quality control and taxonomic classification tool of prokaryotic genomes based on NCBI and GTDB taxonomies. We benchmarked DFAST_QCs performance against NCBI taxonomy assignments, showing high consistency with them. Our results demonstrate that DFAST_QC achieves high consistency to NCBI taxonomy classification. Availability and implementationDFAST_QC is implemented in Python and is available both as a web service (https://dfast.ddbj.nig.ac.jp/dqc) and as a stand-alone command line tool. The source code is available under the GPLv3 license at: https://github.com/nigyta/dfast_qc, and the conda package is also available from Bioconda. The data and scripts used for the benchmarking process are publicly available on GitHub (https://github.com/Mohamed-Elmanzalawi/DFAST_QC_Benchmark). Contactyt@nig.ac.jp Supplementary informationSupplementary data are available at Bioinformatics online.

genomics↗

bioRxiv2023

Zinc controls histone acetyltransferase KAT7 activity to maintain cellular zinc homeostasis

Zinc is an indispensable micronutrient for optimal physiological functions, and zinc deficiency has been implicated in the pathogenesis of various human diseases. One potential mechanism underlying such pathogenic effects is the alteration of gene expression caused by zinc deficiency; however, the details of this process remain largely unexplored. Here, we show that during zinc deficiency, the histone acetyltransferase KAT7 loses its enzymatic activity, leading to the attenuated acetylation of histone H3 at Lys14 (H3K14ac) at enhancer regions. Physiologically, the decrease in H3K14ac leads to the upregulation of the expression of ZIP10, a plasma membrane-localized zinc transporter, thereby facilitating the import of extracellular zinc to maintain cellular zinc homeostasis. Moreover, prolonged zinc deficiency in mice induced by a zinc-deficient diet or high-fat diet, accompanied by decreased H3K14ac levels in the liver, upregulated the expression of genes associated with intracellular lipid droplet formation, leading to the accumulation of lipids within liver tissue. Our findings demonstrate that cells respond to zinc deficiency by converting it into an epigenetic signal that drives physiological or pathophysiological biological processes.

cell biology↗

bioRxiv2023

An ancient split of germline and somatic stem cell lineages in Hydra

In many animals, germ cell segregation occurs during early embryogenesis to protect the genome, but its origin in basal metazoans is controversial. Here, we show in the freshwater polyp Hydra by clonal analysis and transgenic animals that interstitial stem cells comprise two separate stem cell populations, i.e., germline and multipotent somatic stem cells. We isolated genetically labelled stem cells for a global transcriptome study and discovered a broad set of germline-specific/enriched genes including Prdm9, Pax5, Dmrt1. In an alternative splicing analysis, we identified many genes with germline-specific isoforms; among them, male-specific isoforms of Dmrt1 and Snf5. The somatic interstitial stem cell lineage was characterized by numerous neuronal control genes like Neurog. But all stem cells in Hydra also share a core of stemness genes that has its roots in unicellular eukaryotes. This suggests an evolutionary scenario in which, at the emergence of animal multicellularity, there was an early split into a stable germline and different somatic stem cell lineages.

developmental biology↗

bioRxiv2022

Horizontal transmission of penicillin binding protein 1A caused a nationwide spread of β-lactam resistance in pneumococci

The emergence and spread of drug-resistant bacteria continue to be a global crisis. The mechanism of the resistance spread via mobile genetic elements such as plasmids is well known, however, the impact of the natural transformation on the spread remains unclear. Streptococcus pneumoniae is well known to be a transformable pathogen by natural competence and they become {beta}-lactam resistance by the acquisition of chromosomal genetic elements including mutated PBPs by natural transformation. To trace the transmission of pneumococcal PBPs among nationwide pediatric population and analyze the impact of transformed PBPs to {beta}-lactam resistance, we collected and analyzed more than 1300 isolates of S. pneumoniae through nationwide surveillance study for pediatric pneumococcal diseases between 2012-2017 in Japan. We discovered a high prevalence of a specific PBP1A type (pbp1a-13) in {beta}-lactam resistant pneumococci that had a 370SSMK substitution in their {beta}-lactam binding SXXK motif, suggesting that this pbp1a-13 transferred horizontally between different clones resulting in emergence and spread of {beta}-lactam resistant pneumococcal clones. Divergence dating analysis suggested that pbp1a-13 was inserted into major resistant lineages in the early 1990s through the 2000s, before introduction of pneumococcal conjugate vaccines in Japan. Our additional analysis for pneumococcal isolates that were recovered in the 90s in Japan suggested that pbp1a-13 in GPSC1 (serotype 19F-CC236) and GPSC14 (serotype 23F-CC242) isolates were the origin of the currently spread pbp1a-13. We provide evidence of pbp1a horizontal transmission at a nationwide scale and highlight the importance of PBP profile monitoring for identifying the emergence and spread of resistant pneumococci lineages.

microbiology↗

bioRxiv2021

Image-based taxonomic classification of bulk biodiversity samples using deep learning and domain adaptation

Complex bulk samples of invertebrates from biodiversity surveys present a great challenge for taxonomic identification, especially if obtained from unexplored ecosystems. High-throughput imaging combined with machine learning for rapid classification could overcome this bottleneck. Developing such procedures requires that taxonomic labels from an existing source data set are used for model training and prediction of an unknown target sample. Yet the feasibility of transfer learning for the classification of unknown samples remains to be tested. Here, we assess the efficiency of deep learning and domain transfer algorithms for family-level classification of below-ground bulk samples of Coleoptera from understudied forests of Cyprus. We trained neural network models with images from local surveys versus global databases of above-ground samples from tropical forests and evaluated how prediction accuracy was affected by: (a) the quality and resolution of images, (b) the size and complexity of the training set and (c) the transferability of identifications across very disparate source-target pairs that do not share any species or genera. Within-dataset classification accuracy reached 98% and depended on the number and quality of training images and on dataset complexity. The accuracy of between-datasets predictions was reduced to a maximum of 82% and depended greatly on the standardisation of the imaging procedure. When the source and target images were of similar quality and resolution, albeit from different faunas, the reduction of accuracy was minimal. Application of algorithms for domain adaptation significantly improved the prediction performance of models trained by non-standardised, low-quality images. Our findings demonstrate that existing databases can be used to train models and successfully classify images from unexplored biota, when the imaging conditions and classification algorithms are carefully considered. Also, our results provide guidelines for data acquisition and algorithmic development for high-throughput image-based biodiversity surveys.

zoology↗

bioRxiv2021

SPART, a versatile and standardized data exchange format for species partition information

A wide range of data types can be used to delimit species and various computer-based tools dedicated to this task are now available. Although these formalized approaches have significantly contributed to increase the objectivity of SD under different assumptions, they are not routinely used by alpha-taxonomists. One obvious shortcoming is the lack of interoperability among the various independently developed SD programs. Given the frequent incongruences between species partitions inferred by different SD approaches, researchers applying these methods often seek to compare these alternative species partitions to evaluate the robustness of the species boundaries. This procedure is excessively time consuming at present, and the lack of a standard format for species partitions is a major obstacle. Here we propose a standardized format, SPART, to enable compatibility between different SD tools exporting or importing partitions. This format reports the partitions and describes, for each of them, the assignment of individuals to the "inferred species". The syntax also allows to optionally report support values, as well as original trees and the full command lines used in the respective SD analyses. Two variants of this format are proposed, overall using the same terminology but presenting the data either optimized for human readability (matricial SPART) or in a format in which each partition forms a separate block (SPART.XML). ABGD, DELINEATE, GMYC, PTP and TR2 have already been adapted to output SPART files and a new version of LIMES has been developed to import, export, merge and split them.

bioinformatics↗

bioRxiv2020

Structural basis of the protochromic green/red photocycle of the chromatic acclimation sensor RcaE

Cyanobacteriochromes (CBCRs) are bilin-binding photosensors of the phytochrome superfamily that show remarkable spectral diversity. The green/red CBCR subfamily is important for regulating chromatic acclimation of photosynthetic antenna in cyanobacteria and is applied for optogenetic control of gene expression in synthetic biology. They are suggested to combine the bilin C15-Z/C15-E photoisomerization with a change in the bilin protonation state to drive their absorption changes. However, structural information and direct evidence of the bilin protonation state are lacking. Here we report a high-resolution (1.63[A]) crystal structure of the bilin-binding domain of the chromatic acclimation sensor RcaE in the red-absorbing photoproduct state. The bilin is buried within a "pan" consisting of hydrophobic residues, where the bilin configuration/conformation is C5-Z,syn/C10-Z,syn/C15-E,syn with the A-C rings co-planar and the D-ring tilted. Three pyrrole nitrogens of the A-C rings are covered in the -face with a hydrophobic lid of Leu249 influencing the bilin pKa, whereas they are directly hydrogen-bonded in the {beta}-face with the carboxyl group of Glu217. Glu217 is further connected to a cluster of waters forming a hole in the pan, which are in exchange with solvent waters in molecular dynamics simulation. We propose that the "holey pan" structure functions as a proton-exit/influx pathway upon photoconversion. NMR analysis demonstrated that the four pyrrole nitrogen atoms are indeed fully protonated in the red-absorbing state, but one of them, most likely the B-ring nitrogen, is deprotonated in the green-absorbing state. These findings deepen our understanding of the diverse spectral tuning mechanisms present in CBCRs. Significance StatementGreen/red CBCRs are one of the most important CBCR subfamilies owing to their physiological roles in cyanobacteria phylum and optogenetic applications. They are known to utilize a change in the bilin protonation state to drive the marked change in green/red absorption, but the structural basis of the protochromic green/red photocycle are not well understood. Here, we have determined the crystal structure of the chromatic acclimation sensor RcaE of this subfamily in the photoproduct state, demonstrating a unique conformation of the bilin and its interacting residues. In addition, we provide direct evidence of the protonation state of the bilin via NMR analysis. These findings bring insight to our understanding of the molecular mechanisms underlying the spectral diversity of CBCRs.

biochemistry↗