bioRxiv Science⌕ Search

Biology subjects

Ma, S.

Publications and source records attributed to Ma, S..

At least 37 records · Page 2Linked to original sources

Evolutionary Origins of Molecular Programs Underlying Brain Circuitry

The vertebrate pallium harbors independently evolved structures that, nevertheless, support strikingly similar sensory and cognitive circuit architectures. The mechanisms and evolutionary timing driving the emergence of these parallel pallial circuits remain unelucidated. Here, we integrated spatial transcriptomic and single-nucleus RNA-seq datasets from eight representative vertebrate species spanning approximately 500 million years of evolution to reconstruct the evolutionary assembly of the primary sensory-allocortical (Pr-Al) molecular axis, a conserved cortical hierarchical axis defined in our prior work. We uncovered that Pr- and Al-like neuronal identities are deeply conserved across sarcopterygians, encompassing all tetrapods and lobe-finned fishes. Intriguingly, this ancestral neuronal homology is uncoupled from spatially partitioned patterning: only tetrapods further compartmentalized these Pr- and Al- neurons into distinct pallium regions. Functional enrichment of Pr- and Al- gene programs uncovered a conserved tetrapod genetic core suite including MAPK signaling and axon guidance pathways. This core toolkit underwent sequential functional refinement from amphibians through mammals. Notably, mammals and birds convergently evolved association-cortical molecular profiles enriched for synaptic regulatory genes to support advanced cognitive functions. Collectively, this work delineates a stepwise vertebrate pallium evolutionary paradigm that explains how conserved molecular modules shape spatially organized brain circuitry across deep evolutionary time.

evolutionary biology↗

Spatial atlas highlights contribution of C. difficile in early-stage colorectal cancer

Background and AimsIntratumoral heterogeneity with respect to both host and microbial components has emerged as an important contributor to colorectal cancer (CRC) biology. Although the overall taxonomy and abundance of the CRC tumor microbiome have become well characterized, much less is known about bacterial niches within the tumor microenvironment (TME). Escherichia coli and Fusobacterium nucleatum are highly prevalent and abundant species linked to CRC. Less abundant organisms, like Clostridioides difficile and Enterocloster aldenensis, are emerging as potentially important CRC-associated bacteria. Additionally, many studies are confounded by preoperative oral antibiotics or the inclusion of late-stage cancers, both of which may alter the gut microbiome. To better understand the spatial relationships between bacteria and CRC, we generated a molecular atlas based on surgically resected tissue specimens collected from a unique cohort of early-stage CRC patients in whom oral antibiotics were not administered preoperatively. MethodsFrom 20 CRC specimens, we performed matched histopathological analysis, fluorescence in situ hybridization (FISH), whole exome sequencing (WES), 16S rRNA amplicon bacterial DNA sequencing, codetection by indexing (CODEX) multiplex immunofluorescence, and spatial transcriptomics of 756 regions among the specimens. We used 16S rRNA amplicon sequencing data to design and experimentally validate custom bacterial probes that were applied to the spatial transcriptomics. Human colonic organoids were used to validate the relationship between Clostridioides difficile toxin B (TcdB) and an E-twenty-six family transcription factor, ELF3. ResultsWe identified eight intratumoral bacterial niches consisting of a variety of bacterial species, and each niche was associated with unique tumor features. We further defined tumor gene expression patterns correlating with individual bacterial species and show that C. difficile has a disproportionate impact on the tumor transcriptome given its relatively low abundance. Specifically, TcdB induces nuclear localization of ELF3, increased cytosolic {beta}-catenin protein, and upregulated WNT signaling. ConclusionOur study integrates multi-omics to identify bacterial species with biological and spatial relevance in CRC regardless of abundance. These findings will enable further studies to define diagnostic and therapeutic targets for bacteria-associated CRC.

Cancer Biology↗

Novel Synthetic Promoter Armed Oncolytic Herpes Simplex Virus For Treatment of PAX3-FOXO1 Positive Rhabdomyosarcoma

BackgroundDue to the difficulty in targeting oncogenic transcription factors, there are no targeted treatments available for pediatric patients afflicted with PAX3-FOXO1 fusion positive alveolar rhabdomyosarcoma. Additionally, alveolar rhabdomyosarcomas are immunologically cold tumors inherently resistant to immune checkpoint blockade. Viral vectors present a unique opportunity to address this gap in treatment strategies as their presence causes immune infiltration, and a virus can be engineered to rely on tumor specific transcriptional networks for its own replication. MethodologyWe engineered a synthetic promoter which relies on PAX3-FOXO1 to drive expression of a downstream gene. We then placed the HSV-1 RL1 gene encoding ICP34.5 under this promoter within the G47{Delta} oncolytic herpes simplex virus (oHSV), generating a novel oHSV targeting cells harboring the PAX3-FOXO1 fusion gene, which we name oRP3Fus ("Orpheus;" oncolytic recombinant G47{Delta} targeting PAX3-FOXO1 virus). PAX3-FOXO1 specificity was determined by qRT-PCR, protein expression, and chromatin immunoprecipitation. Antitumor efficacy of oncolytic virus and anti-PD1 antibody was determined using immuno-competent mice harboring syngeneic PAX3-FOXO1 expressing rhabdomyosarcoma tumors. Bulk RNA sequencing was performed on infected and mock infected fusion positive and negative cells for differential gene expression analysis. ResultsoRP3Fus kills PAX3-FOXO1 positive human rhabdomyosarcoma lines, while remaining non-toxic to normal human skeletal muscle cells in vitro. In immunocompetent mouse models, both novel and parental viruses improved tumor control when combined with anti-PD1, while oRP3Fus treatment increased T cell infiltration within tumors. RNA and DNA extraction from tumors demonstrate viral expansion within oRP3Fus treated tumors up to three weeks post treatment. Meanwhile oRP3Fus treated mice displayed decreased amounts of necroptosis-associated gene expression within normal organs as compared to groups treated with parental virus. Bulk RNA sequencing of fusion positive and negative human rhabdomyosarcoma lines demonstrated PAX3-FOXO1 increases expression of genes involved in the TNFa-NF{kappa}B signaling axis, as well as innate and adaptive immune signaling. Additionally, ICP34.5 expression from oRP3Fus reduces RIPK1 activation. ConclusionsICP34.5-armed oHSV (oRP3Fus) combined with anti-PD1 provides effective antitumor activity against PAX3-FOXO1-positive alveolar rhabdomyosarcoma in vivo and should be further explored for fusion oncogene-driven tumors. What is already known on this topic?PAX3-FOXO1 fusion-positive and negative rhabdomyosarcoma (RMS) is highly resistant to immune checkpoint blockade. Previously, a PAX3-FOXO1 driven oncolytic adenovirus showed efficacy in an immuno-compromised fusion-positive RMS model. In fusion-negative, immuno-competent RMS models, oncolytic herpes viruses have been shown to improve survival when combined with anti-PD1 therapy, but combinational efficacy was only observed with sex-mismatched models and was never tested on fusion-positive RMS tumors in immunocompetent mice. What this study adds?We improved the specificity of the synthetic PAX3-FOXO1 targeted promoter. We also show, for the first time, efficacy of oncolytic herpes simplex virus-1 in immunocompetent mice harboring orthotopic PAX3-FOXO1-expressing RMS tumors. Our study also demonstrates improvement in survival efficacy in sex-matched tumors, whereas previous reports did not observe complete responses except when using sex-mismatched tumor models. How this study might effect research, practice, or policy?The synthetic promoter generated here can be used for future gene therapy or targeting efforts to study PAX3-FOXO1 or target PAX3-FOXO1 tumors. We observe anti-PD1 therapy benefits survival when paired with an oHSV, demonstrating oncolytic virus can sensitize PAX3-FOXO1 expressing RMS tumors to immunotherapy.

Cancer Biology↗

Validation of MHPC512: A Publicly Available Special-Purpose Supercomputer for Molecular Dynamics Simulations of Biomolecular Systems

MHPC512 is a massively parallel, special-purpose supercomputer designed primarily for atomic-level molecular dynamics (MD) simulations of biomolecular systems. It comprises 512 processor units interconnected by a high-speed three-dimensional torus network and employs a custom chip architecture that uses 35-bit fixed-point arithmetic to accelerate computation while controlling precision loss within an acceptable margin. The preprocessor is compatible with GROMACS and AMBER input formats and supports widely used biomolecular force fields (including CHARMM, AMBER, and OPLS/AA), the Neutral Territory method for short-range nonbonded interactions, the k-space Gaussian Split Ewald method for long-range electrostatics, and multiple thermostats, barostats, and integrators. We present a three-tier validation protocol--comparing static energy and virial components, examining ensemble distributions (NVE, NVT, NPT), and evaluating long-time statistical properties--demonstrating that MHPC512 reproduces results consistent with GROMACS and AmberTools. Application examples, including bulk water, dipeptide conformational sampling, folding of fast-folding peptides, membrane-protein systems, lipid self-assembly, and GPCR conformational transitions, further confirm its reliability. MHPC512 has been deployed at multiple supercomputing centers and is publicly accessible, representing a significant advance in high-throughput, large-scale biomolecular MD simulations.

Molecular Biology↗

Modulation of Ferroptosis During Early Mycobacterium tuberculosis Infection Contributes to Beijing Lineage Strain SA161 Virulence

Although spread of internalized Mtb from the initial infected alveolar macrophages (AMs) is a crucial determinant of infection outcomes, the role of cell death in facilitating this spread, and how it is regulated by Mtb remains poorly understood. Ferroptosis, a lipid peroxidation-mediated form of necrotic cell death, contributes to pathology during later stages of infection. However, the role of ferroptosis in early AM cell death remains inadequately defined, and its induction during infection has been primarily studied with the laboratory strain, H37Rv. Using gene set variation analysis of single cell RNAseq data profiling Mtb-infected murine lungs, we found that the hypervirulent Beijing sublineage clinical strain SA161 is associated with an elevated pro-ferroptotic transcriptional response in AMs compared to H37Rv by 17 days post infection. Consistent with these transcriptional profiles, we found that SA161 induced increased lipid peroxidation in comparison to H37Rv during infection in vitro and in vivo. Administration of the lipid peroxidation inhibitor, ferrostatin-1 (Fer-1), reduces this Mtb-induced lipid peroxidation. Notably, we found that administration of Fer-1 to Mtb-infected mice significantly reduced bacterial burden for SA161 at 14 dpi while having no effect on H37Rv. Microscopic analysis of SA161-infected lung lesions at 14 dpi suggests that inhibition of ferroptosis-driving lipid peroxidation results in a greater proportion of AMs amongst infected cells and decreased neutrophil-associated IFN signaling. Collectively, these findings reveal that ferroptosis plays an important role during early infection with virulent clinical strains of Mtb by influencing bacterial spread and signaling of immune cell responders, potentially informing host-directed intervention strategies.

immunology↗

Differences in microbial community structure and soil nutrients between healthy and diseased soils of Zingiber officinale

To investigate the differences in soil nutrients and microbial community structure in the rhizosphere between healthy and diseased Zingiber officinale plants, soil samples were collected from healthy root-zone soil (ZSH), healthy rhizosphere soil (RSH), diseased root-zone soil (ZSD), and diseased rhizosphere soil (RSD). Diseased soils had significantly higher pH values, whereas RSH showed the strongest acidity. Moreover, pH, AN, and AK contents in diseased soils were significantly higher than those in healthy soils, while SOM and AP were significantly lower. The -diversity of microbial communities in diseased soils was significantly reduced, and the community structure was distinctly differentiated from that of healthy soils. In diseased soils, the abundance of potential pathogenic taxa such as Ralstonia solanacearum increased significantly, while beneficial genera such as Bradyrhizobium decreased. Redundancy analysis and correlation analysis indicated that soil pH, AN, SOM, and AP were the major environmental factors driving changes in microbial community structure. The occurrence of soil-borne diseases in Zingiber officinale is closely associated with soil nutrient imbalance and disruption of microbial community structure. The study identified candidate microbial taxa (e.g., beneficial Sphingomonas, Streptomyces) and key soil properties (pH, available nitrogen) that differentiate healthy from diseased ginger soils. Together, these findings provide a theoretical basis for improving diseased soils through microecological regulation strategies, and also serve as a foundation for generating testable hypotheses in future hypothesis-driven research on ginger soil-borne disease suppression.

microbiology↗

MetworkPy A Python Package for Graph- and Information-theoretic Investigation of Metabolic Networks

SummaryWe present MetworkPy, a python package for investigating in silico genome-scale models of metabolism (GSMM). By using novel graph- and information-theoretic methods to explore the feasible reaction flux space, MetworkPy quantifies network context and simulates metabolic relationships between sets of enzyme-encoding genes without imposing assumptions of optimal growth. To demonstrate utility, we used MetworkPy to identify metabolic features perturbed by the transcription factor ArgR, a known regulator of arginine biosynthesis in Mycobacterium tuberculosis, based on published transcriptome data generated from an argR mutant strain. MetworkPy successfully linked reaction flux shifts in ArgRs transcriptome-constrained GSMM to arginine biosynthesis, which cannot be easily ascertained by conventional constraint-based optimization modeling approaches. MetworkPy offers a flexible toolbox for metabolic contextualization of genes-of-interest in microbial, eukaryotic, and multi-organism systems with potential applications for medicine and bioengineering. Availability and implementationThe MetworkPy package can be retrieved from PyPi (https://pypi.org/project/metworkpy/) and GitHub (https://github.com/Ma-Lab-Seattle-Childrens-CGIDR/metworkpy). Code for analyses performed in this paper can be retrieved from GitHub (https://github.com/Ma-Lab-Seattle-Childrens-CGIDR/metworkpy_application_note) Supplementary InformationSupplementary data are available online at bioRxiv.

bioinformatics↗

Regulation of Small RNAs by Exercise and Their Role in Insulin Sensitivity

We investigated effects of three aerobic exercise interventions, varying in amount and intensity with durations of 8-9-months on small RNA (smRNA) expression and regulatory pathways in skeletal muscle and plasma from 120 participants. Using untargeted smRNA sequencing focused on miRNAs and piRNAs, adjusting for demographics and bodyweight, we identified 124 muscle smRNAs altered by exercise amount and 15 by intensity, and 47 plasma smRNAs altered by intensity and one by amount. These smRNAs were enriched in metabolic, transcriptional, translational, and cell cycle pathways. Exercise-induced changes in several smRNAs-six from muscle and five from plasma-and exercise-induced reduction in body weight, aligned with improvement in insulin sensitivity (p<0.05). These findings demonstrate tissue-specific regulation of smRNAs by exercise and identify potential candidates for exercise mimetics to modulate muscle insulin sensitivity.

physiology↗

Tunable Rigid Spikes on Virus-Like Porous Silica Enable Mechanistically Controlled Nanovaccine Platforms

Virus-like particles represent an emerging and promising vaccine platform. However, these particles are inherently mechanically soft and have limited control over particle surface architecture, thereby constraining their immunological control. Herein, we report the rational design of bioinspired virus-like porous silica (VLPSi) nanoparticles (NPs) with tunable and mechanically rigid spike architectures that function dually as antigen delivery carriers and immune adjuvants. Using ovalbumin (OVA) as a model antigen, we systematically elucidate the spiky structure-function relationship in antigen delivery and immune response. VLPSi NPs exhibit good biocompatibility, sustained antigen release, and markedly enhanced cellular uptake and endosomal escape compared with soft spike and spherical counterparts. Mechanistic investigations combining molecular dynamics simulations and proteomic analyses reveal that rigid spike architectures reduce the energetic barrier for cellular internalization and concurrently activate dual pathways involving endosomal Toll like receptors and calcium signaling. Consequently, VLPSi with long spikes elicit significantly enhanced humoral and cellular immune responses, outperforming the particles with shorter spikes, spherical shape as well as clinically used alum adjuvant. To demonstrate translational potential, bioinspired antibacterial vaccines were produced by loading Staphylococcus aureus surface protein rEsxB. The VLPSi-based vaccine elicited robust protective immunity to achieve complete (100%) survival following lethal challenge without detectable adverse effects, whereas traditional Alum-adjuvanted formulation conferred only minimal protection, with a survival rate of 10%. Collectively, this work establishes VLPSi with tunable spikes as a mechanistically controlled platform for next generation vaccines. Graphic Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=145 SRC="FIGDIR/small/720861v1_ufig1.gif" ALT="Figure 1"> View larger version (47K): org.highwire.dtl.DTLVardef@1f6c13eorg.highwire.dtl.DTLVardef@1090d07org.highwire.dtl.DTLVardef@1364926org.highwire.dtl.DTLVardef@fc68ab_HPS_FORMAT_FIGEXP M_FIG C_FIG

bioengineering↗

OneGenomeRice (OGR): A Genomic Foundation Model for Rice

The transition of genomics to a predictive intelligence discipline is driven by the advent of genomic foundation models. While substantial progress has been observed in all-life and human-centric models, plant species, particularly for the staple crops, remains hindered by a lack of models. Here we introduce OneGenome-Rice (OGR), a rice (Oryza sativa) genomic foundation model pre-trained on a genomic dataset comprising 422 high-quality genomes of cultivated and wild rice. OGR is engineered upon a Mixture of Experts (MoE) transformer architecture with 1.25-billion parameters and supports an ultra-long context window of up to 1 million base (Mb) pairs at single-nucleotide resolution. A comprehensive benchmark demonstrated that OGR significantly outperforms existing state-of-the-art multi-plant or all-life genome models in 11 categories (e.g. motif identification, sweep detection, etc). We further demonstrated the utility of OGR in several downstream applications, such as indica-japonica subspecies introgression analysis, identification of agronomy trait-associated functional loci and prediction of gene expression from DNA sequences. These results establish OGR as a promising foundational computational infrastructure for rice functional genomics and precision breeding. The OGR and its fine-tuned models, including pretrained weights, training code and the rice genomic benchmark suit, have been fully opened.

genomics↗

Decoupling Detection and Classification to Improve Morphological Phenotype Analysis of Sickle Red Blood Cells in Full-Scope Microscopy

Microscopy-based analysis of red blood cell (RBC) morphology is widely used to study phenotypes in sickle cell disease (SCD). Although AI models have been developed to automate classification, most are trained on pre-cropped single-cell images and thus struggle with full-scope microscopic images containing densely packed cells and diverse morphologies, which require both accurate detection and fine-grained classification. We propose an end-to-end computational framework to identify individual RBCs in full-scope microscopy images and classify them into five morphological categories: discocytes (DO), echinocytes (E), elongated and sickle-shaped cells (ES), granular cells (G), and reticulocytes (R). We first evaluate advanced detection-classification models, including You Only Look Once (YOLO) and Detection Transformers (DETR), and demonstrate that while these models effectively detect cells, their classification performance falls short of specialized classifiers trained on single-cell images, particularly for minority phenotypes. To address this limitation, we introduce a two-step framework in which a YOLO-based detector localizes and crops individual cells from full-scope images, followed by a fine-tuned DenseNet121 ensemble classifier that assigns each cell to one of the five morphological categories. The proposed framework achieves a detection-level F1-score of 0.9661 and a weighted-average classification F1-score of 0.9708, with an overall classification accuracy of 97.06%. Compared with the single-step YOLO26n baseline, the two-step pipeline yields a macro-average F1-score improvement of +0.1675, with particularly substantial gains for minority classes (E: +0.1623; G: +0.2774; R: +0.2603). Overall, this hybrid framework demonstrates a practical strategy for adapting fast, general-purpose detection models to domain-specific biomedical tasks by combining them with specialized classifiers, delivering both efficiency and high accuracy for scientific and clinical image analysis.

bioengineering↗

OmicGlaze: Spatial Multi-Omic Mapping of Traumatic Brain Injury

Traumatic brain injury (TBI) is a major cause of mortality and long-term disability worldwide, giving rise to complex neurological complications that impact millions of individuals each year. Cellular stress and neuronal injury vary dramatically across cortical layers, vascular niches, and between the ipsilateral (injured) or contralateral (uninjured) hemispheres. There is a critical need for quantitative measures that capture the spatial distribution of injury-induced cellular changes, as well as the gene regulatory elements that drive them. Here, we developed OmicGlaze, an experimental and computational workflow for systematically profiling the spatial transcriptome and epigenome of mouse brains following mild traumatic brain injury. We established a spatial scoring system, and identified region-specific biological processes post injury, including changes in neuronal activities, cellular stress, immune response, and gliosis. Spatial assay for transposase-accessible chromatin with sequencing (Spatial ATAC-seq) generated the first epigenetic map of traumatic brain injury near single-cell resolution. Notably, we identified the Activator Protein-1 family transcription factor Atf3 as a key gene regulator of injury-induced cellular stress. Together, these spatial multi-omics analyses revealed gene regulatory network in TBI and provided a broadly applicable framework for dissecting cellular and molecular mechanisms underlying complex neurological disorders.

systems biology↗

STAT4-dependent regulation of neuroinflammation in atherosclerosis

Atherosclerosis is linked to an increased risk of cognitive decline, with chronic inflammation being a common feature of both pathologies. IL-12 activates STAT4 to regulate myeloid cell functions, and blockade of this pathway alleviates cognitive impairment in Alzheimers models. However, the mechanisms connecting vascular pathology to neuroinflammation remain unclear. Here, we examine whether STAT4 functions as a common mediator of neuroinflammation in atherosclerosis. We demonstrate that LysMCre-specific STAT4 deficiency ameliorates deficits in long-term memory in low-density lipoprotein-deficient (Ldlr-/-) mice fed a high-fat diet (HFD-C). STAT4 deficiency moderately reduces Ser199-phosphorylated Tau burden. Atherosclerosis alters brain immune composition, characterized by increased numbers of CD45+ leukocytes, activated microglia, and activated T and B cells, whereas STAT4 deficiency attenuates these effects. Nanostring gene-expression pathway analysis further highlights the importance of STAT4 in regulating multiple neuroinflammatory pathways and the Rhodopsin-like receptor signaling, which is associated with synaptic plasticity. LysMCre-specific STAT4 deficiency supports microglial efferocytosis in atherosclerotic Ldlr-/- mice and increases the number of efferocytotic macrophages. Accordingly, STAT4 deficiency also reduced neuronal death. Overall, our data reveal an important role for myeloid-driven STAT4 expression in the pathogenesis of cognitive decline associated with atherosclerosis, mediated through impaired efferocytosis and enhanced leukocyte activation, leading to increased brain neuroinflammation.

immunology↗

X-Cell: Scaling Causal Perturbation Prediction Across Diverse Cellular Contexts via Diffusion Language Models

Causal models of cellular systems hold the promise to empower broad biological discovery, including the systematic identification of novel targets for drug discovery. Predicting how genetic and pathway perturbations reshape gene expression across diverse cellular contexts is a prerequisite for building generalizable cellular foundation models. However, current methods typically fail to extrapolate beyond their training distributions because they rely predominantly on observational expression atlases rather than interventional perturbation data. We present X-Atlas/Pisces, the largest genome-wide CRISPRi Perturb-seq compendium to date, comprising 25.6 million perturbed single-cell transcriptomes across 16 biologically diverse contexts, including widely used cell lines, induced pluripotent stem cells (iPSCs), resting and CD3/CD28 activated Jurkat T lymphoma cells, and multi-lineage differentiating iPSCs. Leveraging this resource, we develop X-Cell, a diffusion language model that predicts perturbation responses by iteratively refining control-to-perturbed state transitions through cross-attention to multi-modal biological priors derived from natural language, protein language models, interaction networks, genetic dependency maps, and morphological profiles. X-Cell outperforms existing state-of-the-art models by up to five-fold on key metrics such as Pearson{Delta} (correlation between predicted and observed perturbation-induced log-fold changes), and demonstrates zero-shot prediction of T cell inactivating perturbations in stimulated Jurkat cells. We scale X-Cell to 4.9 billion parameters (X-Cell-Ultra), the largest causal perturbation model to date. We demonstrate for the first time that perturbation prediction follows power-law scaling with an exponent matching large language models. X-Cell-Ultra demonstrates zero-shot generalization to novel biological contexts, including unseen iPSC-derived melanocyte progenitors and primary human CD4+ T cells from multiple donors, and outperforms all baselines after self-supervised test-time adaptation. These results demonstrate that coordinated scaling of causal perturbation data and model capacity yields foundation models capable of generalizable perturbation prediction across cellular contexts, with potential applications for improving computational target identification, validation, and context-specific therapeutic prioritization.

systems biology↗

BioReason-Pro: Advancing Protein Function Prediction with Multimodal Biological Reasoning

Protein function annotation is fundamental to understanding biological mechanisms, designing therapeutics, and advancing biomedical research. Current computational methods either rely on shallow sequence similarity or treat function prediction as isolated classification tasks, failing to capture the integrative reasoning across sequence, structure, domains, and interactions that expert biologists perform to infer function. We introduce BioReason-Pro, the first multimodal reasoning large language model (LLM) for protein function prediction that integrates protein embeddings with biological context to generate structured reasoning traces. A key input into BioReason-Pro is the set of GO term predictions made by GO-GPT, our autoregressive transformer that captures hierarchical and cross-aspect dependencies of GO terms. BioReason-Pro is trained via supervised fine-tuning on synthetic reasoning traces generated by GPT-5 for over 130K proteins and further optimized through reinforcement learning. It achieves 73.6% Fmax on GO term prediction and an LLM judge score of 8/10 on functional summaries, substantially outperforming previous methods. Evaluations with human protein experts show that BioReason-Pro annotations are preferred over ground truth UniProt annotations in 79% of cases. Remarkably, BioReason-Pro predicted a novel interaction partner for the renal cancer biomarker RCDG1, which we confirmed in the lab by co-immunoprecipitation. In other binding-partner predictions, its per-residue attention localized to the exact contact residues resolved in cryo-EM structures. Together, GO-GPT and BioReason-Pro establish a framework for protein function prediction that combines precise ontology modeling with interpretable biological reasoning.

molecular biology↗

Deciphering context-dependent epigenetic program by network-based prediction of clustered open regulatory elements from single-cell chromatin accessibility

Large cis-regulatory domains, spanning tens to hundreds of kilobases, are pivotal in orchestrating cell-state-specific transcriptional programs that define cellular identity. However, existing single-cell analytical frameworks lack the capacity to identify these higher-order structures, thereby obscuring the coordinated, domain-level epigenetic regulation essential for complex biological processes. To address this, we introduce enCORE, a computational framework that leverages enhancer-enhancer interaction networks to determine Clustered Open Regulatory Elements (COREs) solely from single-cell ATAC-sequencing data. Our approach faithfully recapitulates established hematopoietic hierarchies and resolves lineage-specific regulatory programs by recovering canonical master transcription factors, frequent chromatin interactions, and enrichment of fine-mapped immune-related disease-associated genome-wide association study (GWAS) variants. In colorectal cancer, enCORE captures tumor-associated H3K27ac landscapes and prioritizes USP7 as a potential therapeutic candidate, supported by in silico perturbation. Collectively, our framework provides a powerful and scalable platform for deciphering the complex epigenetic architectures underlying human development and disease.

bioinformatics↗

Multi-component functionalized Bifidobacterium longum hydrogel for multi-target integrated therapy of colitis-associated anxiety and depression

Inflammatory bowel diseases (IBDs) are frequently accompanied by anxiety and depression, largely driven by perturbed gut-brain axis signaling. However, current oral therapies remain constrained by the spatial and functional separation between intestinal inflammation and central nervous system dysfunction. Here, we present a comprehensive gut-brain dual region integrated therapeutic strategy based on functionalized Bifidobacterium longum hydrogel (INPs@BL@Gel), in which baicalin and tyrosine are coordinated with Fe(III) to form infinite coordination polymers (ICPs), coated with inulin, assembled onto Bifidobacterium longum (BL), and subsequently encapsulated within a pH- and matrix metalloproteinase-responsive silk fibroin-gelatin hydrogel. INPs@BL@Gel exhibits high drug-loading, effective gastric protection, inflammation-triggered release, and long-term intestinal colonization. Within the inflamed intestine, BL and components synergistically suppress inflammatory responses, restore gut microbiota homeostasis, and promote intestinal barrier repair through multi-target integrated therapy. Importantly, BL combined with components markedly enhances the production of beneficial neuroactive metabolites such as homovanillic acid and short-chain fatty acids, which integrated regulate neuroinflammation, preserve synaptic function, and facilitate blood-brain barrier repair via the gut-brain axis. In vivo studies demonstrate that INPs@BL@Gel not only exert potent therapeutic efficacy against colitis and effectively alleviate associated depression, but also reshape the gut microbiota and restore barrier integrity, achieving an remarkable comprehensive therapeutic effect. O_FIG O_LINKSMALLFIG WIDTH=158 HEIGHT=200 SRC="FIGDIR/small/710940v1_fig1a.gif" ALT="Figure 11"> View larger version (59K): org.highwire.dtl.DTLVardef@1a831f3org.highwire.dtl.DTLVardef@1676617org.highwire.dtl.DTLVardef@15f99c1org.highwire.dtl.DTLVardef@1677cc4_HPS_FORMAT_FIGEXP M_FIG O_FLOATNOScheme 1.C_FLOATNO (a) Schematic diagram of the design and preparation of functionalized Bifidobacterium longum hydrogel. (b) Exploration of the mechanism of INPs@BL@Gel in treating colitis-associated anxiety and depression through a dual-site multi-target synergistic strategy. C_FIG

pathology↗

Periaqueductal gray neurotensin neurons drive simultaneousthreat response and reinforcement

The periaqueductal gray (PAG) is a midbrain structure known to influence responses to both threat and reward. The PAG sends projections to the ventral tegmental area (VTA), a region critical for regulating motivated behavior via dopamine release. We previously identified a population of VTA-projecting PAG neurons that express the peptide neurotensin (Nts), a potent dopamine neuron activator. Here we find that PAG-Nts neurons co-release glutamate and Nts in the VTA to drive dopamine neuron activation. These neurons are activated by threats and threat-predictive cues and are inhibited by entry into a shelter and during reward consumption. Optogenetic stimulation elicits a robust threat response, including freezing and tail rattle, but remarkably can also drive intracranial self-stimulation. This operant reinforcement behavior is dopamine dependent while the threat response is not. Together, these results identify a dual-output circuit that engages the dopamine system, likely to increase the salience of environmental stimuli, while simultaneously driving specific threat response behaviors.

neuroscience↗