Biology subjects

Zhu, S.

Publications and source records attributed to Zhu, S..

At least 19 recordsLinked to original sources

bioRxiv2026

Bioenergetic and Proteomic Profiling of ARID1A-Deficient Ovarian Clear Cell Carcinoma under Metabolic Challenge

ARID1A loss is associated with increased mitochondrial respiration in ovarian clear cell carcinoma (OCCC), but whether this phenotype is maintained under metabolic challenge remains unclear. Here, we examined mitochondrial function and electron transport chain (ETC) protein abundance in isogenic ARID1A wild-type (WT) and ARID1A-knockout (KO) OCCC cells cultured under physiological glucose, galactose, or the Complex I inhibitor IACS-010759. Across all conditions, KO cells maintained higher basal, ATP-linked, and maximal respiration, greater spare respiratory capacity, and higher coupling efficiency than WT cells, despite lower steady-state ATP levels. In human OCCC tumors, ARID1A-negative cases showed a higher bioenergetic cellular (BEC) index than ARID1A-positive cases, and high BEC was associated with poorer cancer-specific survival. Quantitative proteomics revealed a trend toward higher ETC complex abundance in KO cells, although whole-complex differences did not reach statistical significance. ARID1A loss was associated with selective increases in Complex III and V proteins, whereas metabolic challenge was associated with increases in Complex I and IV proteins. Together, these findings show that the ARID1A-associated mitochondrial phenotype persists across distinct metabolic challenges and is not explained simply by coordinated increases in ETC protein abundance. The association of mitochondrial bioenergetics with clinical outcomes further supports its relevance in human OCCC.

cancer biology↗

bioRxiv2026

Aging Differentially Remodels the Retina and RPE/Choroid Proteome and Metabolism

Purpose: The neural retina and retinal pigment epithelium/choroid (RPE/Cho) are metabolically coupled. Metabolic dysfunction of the RPE/Cho contributes to photoreceptor death in age-related macular degeneration (AMD). This study aimed to determine how aging, a major risk factor for AMD, alters the proteome and metabolic pathways in the retina and RPE/Cho. Methods: We performed quantitative proteomics on neural retina and RPE/Cho from young (8-week) and old (73-week) C57BL/6J mice. The differentially expressed proteins between old and young, as well as between the retina and RPE/Cho, were visualized using volcano plots and heatmaps, and functional analyses were assessed by Gene Ontology enrichment tools. Results: Aging altered the expression of 85 retinal proteins and 272 RPE/Cho, with 16 shared between tissues. In the aged retina, proteins associated with NADP(H) metabolism, neurodegeneration, and immunity were mostly upregulated, while proteins in lipid metabolism, transcription, and epigenetic regulation decreased. In the aged RPE/Cho, all 21 significantly altered lysosomal enzymes and transport proteins increased, together with complement components, and photoreceptor outer segment proteins. These changes accompanied increases in proteins associated with lipid metabolism and protection against lipid peroxidation, while several glutathione metabolism enzymes decreased. Seven shared proteins involved in lipid and NADP(H) metabolism, changed in opposite directions between retina and RPE/Cho. Comparison of the retinal and RPE/Cho proteomes identified 3073 differentially abundant proteins, of which 2452 maintained tissue-specific differences across both age groups, while approximately 17% showed age-dependent shifts in tissue enrichment. These shifts involved proteins associated with metabolism, extracellular matrix, visual function, lysosomal pathways, and innate immunity. Within the metabolic proteome, several glycolytic and fatty acid oxidation proteins showed greater relative enrichment in RPE/Cho with age, whereas glucose transporters and mitochondrial complexes shifted toward the retina. Conclusions: Overall, these findings show that aging differentially remodels the retina and RPE/Cho proteome, accompanied by altered metabolic specialization between the two tissues.

biochemistry↗

bioRxiv2026

Gene dosage effect and reduced core fucosylation in B lymphocytes of FUT8 heterozygous mice

Core fucosylation (alpha1,6-fucosylation) of N-glycans, catalyzed solely by fucosyltransferase 8 (FUT8), plays important roles in development, immunity, and antibody effector functions. Here we report the generation of Fut8 floxed mice using a CRISPR/Cas9-based Extreme Genome Editing (EGE) system, in which loxP sites flank exon 8 of the Fut8 gene and a LoxP-STOP-LoxP-RFP cassette was knocked in downstream of the promoter, rendering the targeted allele functionally null before Cre-mediated recombination. Crossing floxed mice with CMV-Cre transgenic mice yielded complete knockout mice carrying the Cre-recombined allele. A PCR-based genotyping method was established that unambiguously distinguishes the wild-type, heterozygous, and homozygous genotypes (+/+, delta/+, and delta/delta). Intercrossing of heterozygous mice revealed that homozygous knockout mice die during the early postnatal period, consistent with previous reports, whereas heterozygous mice are viable. Flow cytometric analysis of AAL binding to splenic B cells and AAL lectin blotting both demonstrated markedly reduced core fucosylation in heterozygous mice (60.0 +/- 6.8% and approximately 48% of wild-type levels, respectively). These results show that disruption of a single Fut8 allele substantially reduces core fucosylation, and the mouse model described here provides a practical tool for functional studies of core fucosylation.

immunology↗

bioRxiv2026

USP20 Stabilizes FABP5 to Drive Macrophage Foam Cell Formation and Atherosclerosis

Background: Macrophage foam cell formation is a central event in atherosclerosis. Although protein ubiquitination is increasingly recognized as a regulator of macrophage lipid handling, whether deubiquitinating enzymes directly control lipid uptake and metabolic remodeling during atherogenesis remains unclear. Methods and Results: Here, transcriptomic profiling identified ubiquitin-specific protease 20 (USP20) as a previously unrecognized regulator of macrophage foam cell formation. Genetic and pharmacological studies demonstrated that USP20 promoted macrophage lipid accumulation and foam cell formation, whereas macrophage-specific Usp20 deficiency markedly attenuated atherosclerotic lesion formation, lipid deposition, and macrophage accumulation in mice. To elucidate the underlying mechanism, integrative analysis of USP20-interacting proteins and proteomic alterations in Usp20-deficient macrophages identified fatty acid-binding protein 5 (FABP5) as a substrate of USP20. Mechanistically, USP20 interacted with FABP5 and stabilized its protein abundance by preventing ubiquitin-dependent proteasomal degradation. Consequently, USP20-mediated stabilization of FABP5 enhanced PPAR{gamma} occupancy at the Cd36 promoter, thereby increasing Cd36 transcription, ox-LDL uptake, and foam cell formation. Accordingly, Fabp5 knockdown abolished USP20-induced CD36 expression, ox-LDL uptake, and foam cell formation, whereas pharmacological activation of PPAR{gamma} restored CD36 expression in Fabp5-deficient macrophages. Finally, macrophage-targeted delivery of the USP20 inhibitor GSK2643943A markedly reduced atherosclerotic burden without evident systemic toxicity, supporting the therapeutic potential of targeting USP20 in atherosclerosis. Conclusions: USP20 promotes macrophage foam cell formation by stabilizing FABP5 and enhancing PPAR{gamma}-dependent Cd36 transcription. Targeting the USP20-FABP5-PPAR{gamma}-CD36 axis may represent a therapeutic strategy to limit macrophage lipid accumulation, metabolic dysfunction, and atherosclerosis progression.

cell biology↗

bioRxiv2026

Automated acoustic classifiers provide insights into calling patterns of cicadas in tropical Australia

Insects are a key group in ecosystems and economics, but are increasingly under threat, making ongoing monitoring essential to preserving them. Traditional monitoring approaches are time consuming, among other limitations. Automated acoustic recognition can be a good solution for monitoring arboreal, soniferous insects, such as cicadas, and the acoustic biology of cicadas in Australia is understudied, especially using automated methods. To evaluate the applicability of automated approaches for detecting cicadas, we used the deep-learning acoustic model BirdNET (v2.4) to perform embedding searches for Cystosoma schmeltzi (lesser bladder), Illyria burkei (eastern rattler), Macrotristria intersecta (corroboree cicada), Thopha sessiliba (northern double-drummer), and Pauropsalta opaca (fairy dust squawker), using audio recorded by Passive Acoustic Monitoring over 15-months at 18 sites in north Queensland. All classifiers achieved high area under the precision-recall curve (AUPRC) ranging from 0.753 to 0.982. We used a precision-prioritised threshold ([≥] 0.85) to constrain the false positive rate while retaining adequate outputs, which were consistent with the documented species ecology and life history. This approach was readily applicable to the Cicadidae, offering a scalable solution for biodiversity monitoring in complex acoustic environments.

ecology↗

bioRxiv2026

Type 2B VWD substitutions trigger enhanced macrophage-mediated clearance that is attenuated by the BT200 aptamer

Type 2B von Willebrand disease (VWD) is characterized by missense variants in exon 28 of the von Willebrand factor (VWF) gene that encodes the VWF-A1 domain. These missense mutations result in single amino acid residue substitutions that cause activation of VWF-A1 and promote spontaneous interaction with platelet GPIba;. Importantly, enhanced VWF clearance has been shown to play a key pathogenic role in >90% patients with Type 2B VWD. Although this VWF clearance occurs via mechanisms that are independent of VWF-platelet complex formation, the biological mechanisms responsible for the increased clearance of type 2B VWD variants remain poorly understood. In this study, we investigated a series of different type 2B amino acid substitutions within the VWF-A1 domain (R1306W, R1308C, W1313C and R1379L). We demonstrate that these variants all exhibit significantly increased macrophage binding. In part, the enhanced macrophage interactions are mediated via increased binding of type 2B VWD variants to LRP1 extracellular cluster II and IV. Our findings further demonstrate that the K1405-K1408 lysine cluster in the VWF-A1 domain plays a key role in enabling enhanced LRP1 interactions for type 2B VWD variants. In addition, we show that type 2B VWF variants demonstrate significantly enhanced interaction with the macrophage MGL receptor. Finally, and importantly from a clinical perspective, we demonstrate that the increase in macrophage binding for type 2B variants is significantly attenuated in the presence of BT200. Collectively, our findings have direct translational relevance with respect to the clinical heterogeneity and treatment of type 2B VWD.

biochemistry↗

bioRxiv2026

Advancing long-read metagenomic binning via single-copy-gene guided contrastive learning

Long-read sequencing advances metagenomics by producing highly contiguous assemblies and more complete metagenome-assembled genomes (MAGs). However, current long-read metagenomic binners fail to incorporate the rich information of long-read assemblies into representation learning and exhibit limited performance on complex datasets. Here, we show that a higher proportion of long-read assembled contigs contain single-copy genes (SCGs) and more SCGs per contig. Therefore, we developed SCGBinner, which leverages SCG-guided contrastive learning to exploit the advantage of long-read data for learning high-quality contig embeddings. SCGBinner consistently outperforms other binning methods across five simulated and seven real-world long-read datasets, especially on real-world high-diversity samples. For a deep agricultural soil metagenome, SCGBinner recovered 71% more high-quality MAGs and 38% more near-complete MAGs than the second-best method. Notably, SCGBinner uniquely recovered 449 novel high-quality species, which shed light on the predicted ecological roles of 65 uncharacterised families and 93 novel genera. Overall, SCGBinner could harness the potential of long-read sequencing to provide unprecedented insights into the microbial dark matter of complex microbial communities.

bioinformatics↗

bioRxiv2026

Active Learning Enables Efficient Directed Evolution of a Far-Red Fluorescent Protein with Minimal Experimental Data

Fluorescent proteins are fundamental tools for cellular imaging. Most fluorescent proteins in routine use, including GFP, are derived from the jellyfish Aequorea victoria and emit blue-green light, which is strongly absorbed and scattered by tissue, limiting imaging depth. Far-red and near-infrared fluorescent proteins, engineered from bacteriophytochromes, address this limitation because far-red light penetrates tissue considerably further. However, these proteins are typically much dimmer than their A. victoria -derived counterparts. Improving brightness by conventional directed evolution requires screening large random mutant libraries, a process that is slow, labor-intensive, and often impractical outside specialized laboratories. We utilized an active-learning-guided directed evolution workflow that identified improved variants from substantially less data than conventional screening. Each round coupled automated, miniaturized cell-free protein expression directly from a DNA template without cloning or cell culture, with a machine-learning model retrained on cumulative sequence-function data to nominate the most informative variants for the next round. Applied to miRFP670nano3, this workflow screened 120 variants across successive rounds and identified twelve with improved brightness, the best four-fold brighter in bacterial systems. However, these gains did not translate when the variants were evaluated in mammalian cells, indicating that performance can be strongly dependent on cellular context. Retrospective simulation across benchmark datasets from ProteinGym showed that performing more experimental batches with fewer samples per batch consistently accelerated convergence to high-fitness sequences. Incorporating protein-language-model derived zero-shot fitness priors also accelerated convergence, but only in proportion to how well each prior score correlated with the true fitness landscape. Together, these findings established generalizable design rules, favoring smaller acquisition batches and confidence-weighted priors, for engineering proteins from minimal experimental data. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=55 SRC="FIGDIR/small/744534v1_ufig1.gif" ALT="Figure 1"> View larger version (11K): org.highwire.dtl.DTLVardef@14cd238org.highwire.dtl.DTLVardef@7d8334org.highwire.dtl.DTLVardef@30e9f2org.highwire.dtl.DTLVardef@14f3a0c_HPS_FORMAT_FIGEXP M_FIG C_FIG

synthetic biology↗

bioRxiv2026

T50 O-GlcNAc PKM2 promotes aerobic glycolysis and PDAC progression via ARNT-CDC27-AKT pathway

Pancreatic ductal adenocarcinoma (PDAC) is one of the most lethal cancers and more evidence suggests that glucose metabolism plays a significant role in the development and progression with glycosylation at multiple sites potentially being a key characteristic. However, the underlying mechanisms remain insufficiently studied. Here, we first confirmed the presence of O-GlcNAc glycosylation modifications at Thr50 on PKM, which is highly expressed in PDAC and strongly correlated with poor prognosis. Further, we found that PKM expression was significantly positively correlated with CDC27 levels, and mutation of the Thr50 O-GlcNAc site in PKM abolished the upregulation of CDC27. We confirmed that O-GlcNAc-modified PKM enhances nuclear translocation of ARNT, which binds to the CDC27 promoter to upregulate its expression. Finally, we demonstrated that reduced expression of CDC27, as a key component of the APC/C complex, leads to downregulation of ubiquitination at the K11 site of PPP2CA, resulting in upregulation of PPP2CA expression, in turn, reduces AKT phosphorylation, ultimately driving PDAC regression by inhibiting aerobic glycolysis. Thus, we delineate a novel O-GlcNAcylation-dependent pathway where PKM drives PDAC progression through ARNT-mediated CDC27 transcriptional activation and AKT-mediated glycolysis.

cancer biology↗

bioRxiv2026

Bacterial surface display enables lysis-independent joint host-pathogen single-cell profiling

Heterogeneity in host-pathogen interactions arises from variation in both host cell state and pathogen state, yet most single-cell methods capture these features separately. Joint profiling is limited by fundamental technical mismatches between host- and pathogen-derived material, particularly differences in cell wall structure, lysis requirements, and molecular abundance. Here we introduce a lysis-independent strategy that enables unified measurement of intracellular bacterial presence and state alongside host single-cell profiles. We repurpose bacterial surface display to encode promoter activity and bacterial identity as antibody-detectable signals, rendering bacterial features compatible with existing antibody-based single-cell assays. This approach is modular across promoters, epitope tags, display scaffolds, and bacterial species, including Escherichia coli and Mycobacterium tuberculosis. Surface-displayed reporters are detectable during intracellular infection and can be read out by flow cytometry and droplet-based single-cell RNA sequencing without pathogen-specific lysis optimization or pre-sorting on pathogen signal. Applying this method to infected macrophages, we link heterogeneous bacterial uptake to heterogeneous expression of phagocytosis-associated host programs. This strategy enables scalable, joint host-pathogen single-cell measurements and expands the range of pathogens and states accessible to high-throughput single-cell analysis.

bioengineering↗

bioRxiv2026

The brain-meningeal interface functions as a reservoir and entry site for brain parenchymal macrophages

Microglia are regarded as self-maintaining brain parenchymal macrophages without contribution from adult hematopoiesis. Nevertheless, peripheral macrophage engraftment into the brain has been reported, but the biological variables governing central nervous system (CNS) macrophage niche access remain unclear. We show that CNS macrophage engraftment is determined by the balance between tissue-resident macrophage (TRM) self-renewal and the temporal alignment of niche opening with the availability of engraftment-competent cells and proximity to the vacated niche, rather than prolonged niche vacancy. Fate mapping revealed that monocytes entering the subdural space undergo border-associated macrophage (BAM)-like differentiation, clonal expansion, and transpial migration into the parenchyma, whereas mature BAMs directly repopulate selectively vacated parenchymal niches. We further identify a parenchymal macrophage population with a peripheral BAM-like transcriptional program in aged and neurodegenerative human brains, challenging the dogma of MG-exclusivity. Together, these findings establish a predictive framework for interpreting macrophage maintenance and replacement and for developing macrophage-based therapies.

neuroscience↗

bioRxiv2026

Nuclear Cx43 restrains microglial neurotoxicity during brain development

Microglia are essential for sculpting the developing brain, yet the molecular mechanisms that select beneficial overreactive phagocytosis remain incompletely understood. Connexin 43 (Cx43, encoded by GJA1 in humans) is best known as a gap junction and hemichannel protein, although its non-canonical, channel-independent functions are increasingly recognized. We found that Cx43 is highly expressed in microglia during the perinatal period in human and mouse, whereas proportion of full-length multimeric Cx43 unexpectedly localizes to the nucleoplasm. Deletion of microglial Cx43 in mice during development instigates a transient neurotoxic state with microgliosis, upregulated phagocytic and complement pathways, excessive neuronal apoptosis, translating into depressive-like and cognitive deficits in the adulthood. Notably, neither microglia-specific deletion of Cx43 in adulthood nor hemichannel blockade recapitulate these changes, indicating a channel-independent, developmental stage-specific neuroprotective mechanism. Nucleus-targeted Cx43 overexpression suppresses neurotoxic markers and neural apoptosis. Nuclear Cx43 interacts with transcriptional regulators to restrain proinflammatory gene programs, nuclear import of Cx43 is driven by neurogenic niche-derived bFGF, which triggers AKT-mediated phosphorylation of a C-terminal nucleus localization signal (NLS), 14-3-3 binding, and importin-dependent nucleus translocation. These findings reveal a developmentally restricted nuclear Cx43 function that restrains microglial neurotoxicity while promoting microglial physiological functions thus expanding connexin biology to transcriptional co-regulation and pointing to a potential avenue for therapeutic intervention.

neuroscience↗

bioRxiv2026

Microscopy-informed structural connectivity mapping in the in vivohuman brain via domain adaptation

Characterising human brain connectivity remains a major challenge in neuroscience. Multimodal datasets combining diffusion MRI with high-resolution microscopy in the same brain offer a unique link between macroscopic imaging and microstructural detail, but we lack tools to leverage these data to improve connectivity estimates for in vivo human imaging. We present a deep learning model that predicts high-resolution microscopy-informed fibre orientations from diffusion MRI. The model uses microscopy-derived three-dimensional fibre orientation maps as biologically grounded training targets. It is trained on a bespoke macaque dataset integrating in vivo MRI, postmortem MRI, and whole-brain microscopy, and then translated to in vivo human imaging. We use domain adaptation to predict fibre orientations from diverse MRI datasets: first to bridge differences in tissue state in the macaque (postmortem to in vivo), and then to generalise across species (macaque to human). Our method derives microscale-informed fibre architecture from diffusion MRI without requiring microscopy at inference. It leverages data that can easily be acquired only in animal models whilst generalising to in vivo human diffusion MRI with minimal acquisition requirements. The microscopy-informed fibre orientation distributions support biologically meaningful tractography, enhancing superficial white matter and cortical-subcortical pathway delineation for in vivo human data. More broadly, this work establishes a general framework for transferring microstructural information from microscopy to non-invasive imaging, enabling biologically informed mapping of brain connectivity.

neuroscience↗

bioRxiv2026

A modular neural circuit for computing the motion of objects

Computing the direction of a moving object requires integrating motion signals from its component edges into coherent patterns. The representation of pattern motion in visual cortex has been extensively studied, yet its underlying neural circuitry remains unknown. Using high-density recordings in macaque area MT, we show that selectivity to pattern motion emerges from a functionally and anatomically distinct cortical circuit. We demonstrate that neurons specialized for encoding component and pattern motion comprise distinct cell types arranged in a hierarchical circuit in which pattern neurons integrate inputs from a range of component neurons. Furthermore, we show that component and pattern neurons are spatially segregated into modules arranged systematically across cortical columns encoding direction of motion. This architecture and circuit align with a classic solution to the problem of computing the motion of objects.

neuroscience↗

bioRxiv2026

Spatially organized cancer-associated fibroblast subtypes partition cutaneous carcinomas into immune-active and contracted, immune-repressed niches

Basal cell carcinoma (BCC) and cutaneous squamous cell carcinoma (SCC) are the most common keratinocyte-derived malignancies, yet they differ markedly in invasiveness, metastatic potential, and immune contexture. Although cancer-associated fibroblasts (CAFs) are increasingly recognized as key regulators of tumor architecture and tumor immunity, the spatial organization of distinct CAF subtypes in cutaneous carcinomas and their functional relationship with immune cells remains incompletely understood. Using a 33-plex imaging mass cytometry (IMC) panel, we profiled 28 regions of interest (ROIs) from 17 human BCC and SCC specimens, encompassing more than 739,000 single cells, and integrated these data with RNA fluorescence in situ hybridization (RNA-FISH), immunohistochemistry (IHC), multiplex immunofluorescence, and in vitro functional assays. We identified four fibroblast populations, including immunomodulatory CAFs (iCAFs), matrix CAFs (mCAFs), myofibroblast-like CAFs (myoCAFs), and reticular fibroblasts (retFIBs), and found that aggressive tumor subtypes were characterized by increased stromal area, extracellular matrix deposition, and altered CAF composition. CAF composition differed most prominently across BCC subtypes, with nodular BCC enriched for mCAFs and infiltrative BCC showing increased myoCAF density, consistent with a shift toward a contractile stromal program. Spatial analyses revealed distinct CAF-immune niches: iCAFs localized to immune-cell-rich, inflamed niches enriched for activated and/or exhaustion-associated immune-cell marker programs, whereas myoCAFs occupied fibroblast-dense, immune-poor niches with globally reduced immune activation. mCAFs were preferentially associated with immune cell accumulation in the stroma and spatial immune compartmentalization, with limited immune cell presence within tumor nests. At the invasive front, CAF-immune coupling was highly subset-dependent, with iCAFs linked to antigen-experienced T-cell states and myoCAFs linked to immune exclusion. In vitro, patient-derived CAF cultures from myoCAF-rich biopsies showed enhanced collagen-gel contraction, with cultures enriched for MCAM+ CAFs displaying increased contractile capacity. Aggressive tumor variants displayed increased stromal nuclear YAP/TAZ, while complementary single-cell pathway analysis supported a mechanically remodeled stromal microenvironment in which mCAFs contribute ECM/matrix-remodeling programs and RGS5/myoCAF-like populations show enhanced mechanotransduction-associated signaling, rather than a uniform CAF-wide increase in canonical YAP/TAZ transcriptional output. Together, these findings define spatially organized CAF programs in cutaneous carcinomas and identify myoCAF-rich stromal niches as a recurrent feature of aggressive, immune-repressed tumor architecture. These results nominate CAF composition as a biomarker of immune architecture and a potential determinant of therapeutic response.

cancer biology↗

bioRxiv2026

Genos-m: a foundation model for human-associated microbial genomes

Human-associated microbial genomes encode extensive strain-level diversity and niche-specific gene repertoires that are critical to host health. However, these complex sequence features remain difficult to capture using general-purpose DNA foundation models, highlighting the need for dedicated representation learning tailored to the human microbiome. Here, we introduce Genos-m, an open-source foundation model for human-associated microbial genome representation. Genos-m was pretrained on approximately 1.2 trillion nucleotide tokens from a curated microbial genome corpus, including human-associated prokaryotic isolates, high-quality metagenome-assembled genomes (MAGs) and bacteriophages, supplemented with GTDB species-level representative genomes to broaden prokaryotic taxonomic breadth. The model uses a sparsely activated Mixture-of-Experts (MoE) Transformer architecture, with 4.7 billion total parameters, approximately 330 million activated parameters per forward pass and a maximum context length of one million base pairs. We evaluated frozen Genos-m representations across short-sequence and gene-level tasks, biosynthetic gene cluster (BGC)-based regional sequence tasks, whole-genome strain phenotype prediction, and zero-shot transfer on prokaryote-related RNAfitness assays. Across these benchmarks, Genos-m consistently ranked among the leading comparison models, with the best performance in five of eight gene-fitness regression tasks and in BGC type classification. Using sparse autoencoders, we identified sparse features in Genos-m hidden activations that aligned with annotated ORFs, intergenic regions, and tRNA and rRNA loci. In downstream applications, Genos-m-derived genome-informed species representations in-corporated into a human microbiome self-supervised learning model improved colorectal cancer (CRC)-control classification over conventional species-abundance random forest models. Genos-m also generated stable sample-level embeddings from as few as 10,000 metagenomic reads, retaining gut microbial community structure that distinguished geographic origin and aligned with enterotypes defined from full-depth taxonomic profiles. Together, these results support Genos-m as a reusable representation model for microbial genomes and metagenomes, with conclusions bounded by the reported datasets, task definitions and evaluation protocols. Genos-m model weights, inference code, and usage documentation are publicly available on GitHub (https://github.com/BGI-HangzhouAI/Genos-m) and Hugging-Face (https://huggingface.co/BGI-HangzhouAI/Genos-m).

bioinformatics↗

bioRxiv2026

Integrin-independent Tie2 activation using de novo designed proteins

The Angiopoietin-Tie2 pathway is a key regulator of vascular stability, but therapeutic exploitation has been limited by the poor developability of Angiopoietin-1 (Ang1), and there are key unresolved mechanistic questions. Ang1 binds both Tie2 and 5{beta}1 integrin, and the role of the 5{beta}1 interaction in signaling has been unclear. We used RFdiffusion to design a stable, high-affinity Tie2 minibinder that has no affinity for 5{beta}1. The binder is a selective antagonist in monomeric form, and a potent agonist when assembled into an octavalent architecture (H8mb) which drives Tie2 clustering. H8mb signals as potently as Ang1, indicating that integrin engagement is not required for Tie2 activation. However, the duration of signaling is reduced, and internalization of H8mb-Tie2 complexes is more rapid, suggesting that integrin may function as a co-receptor for Ang1 that prolongs signaling by extending the lifetime of the receptor ligand complex on the cell surface. In a mouse model of acute respiratory distress syndrome (ARDS), H8mb markedly improved survival. These results demonstrate that de novo designed receptor binders can enable dissection of co-receptor control of signaling dynamics, and the more stable and manufacturable H8mb provides routes to more developable therapeutic candidates.

biochemistry↗

bioRxiv2026

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↗