Biology subjects

Xu, P.

Publications and source records attributed to Xu, P..

At least 19 recordsLinked to original sources

bioRxiv2026

Central platelet-rich plasma administration promotes weight loss in diet-induced obese mice

Aims: Platelet-rich plasma (PRP) contains diverse bioactive factors, but its potential role in central regulation of energy balance remains unexplored. This study investigated whether central PRP administration alters metabolic homeostasis in diet-induced obese (DIO) mice and examined associated hypothalamic responses under obese and inflammatory conditions. Materials and Methods: PRP was administered intracerebroventricularly in mouse models of DIO and lipopolysaccharide (LPS)-induced systemic inflammation. Body weight, food intake, energy metabolism, and adiposity were assessed alongside hypothalamic microglial morphology and transcriptomic profiling. Results: A single intracerebroventricular PRP injection reduced body weight in DIO mice, with the effect persisting throughout the 7-day observation period. PRP also decreased cumulative food intake, relative gonadal white adipose tissue weight, and fasting glucose, accompanied by increased energy expenditure, oxygen consumption, fat utilization, and physical activity. In both DIO and LPS-treated mice, PRP reduced IBA1-positive area per microglial cell in the arcuate nucleus without altering microglial number. Hypothalamic RNA sequencing revealed changes in immune and metabolic signaling, extracellular matrix organization, and growth-related pathways, suggesting a complex transcriptional response rather than uniform suppression of inflammation. Conclusions: A single central PRP administration induces sustained weight loss and favorable metabolic changes in DIO mice, accompanied by region-specific hypothalamic microglial and transcriptional responses. These findings identify central PRP as a potential modulator of energy balance and provide a basis for investigating its active components and underlying mechanisms.

neuroscience↗

bioRxiv2026

Specific host translation perturbations promote TnSmu1 loss independently of the canonical ImmA regulator

Integrative and conjugative elements (ICEs) are major drivers of horizontal gene transfer and bacterial genome evolution. Although ICE-encoded regulatory circuits have been extensively characterized, the impact of host physiology on the stability of integrated ICEs remains poorly understood. Here, we identify a host-dependent pathway that links specific host translation perturbations to loss of the ICE TnSmu1 in Streptococcus mutans. Analysis of host-gene deletion mutants revealed that disruption of fmt, rnjA, or rnjB--three translation-associated host genes--reproducibly promoted TnSmu1 loss through a mechanism that bypasses the canonical ICE-encoded metalloprotease ImmA but remains dependent on the native attachment site attR. This phenotype was selective, as mutations affecting other essential cellular functions, including protein folding, tRNA modification, cell division, and fatty acid biosynthesis, failed to destabilize TnSmu1 despite undergoing the same experimental evolution and accumulating adaptive genomic changes. Preventing TnSmu1 loss in these translation-associated mutants markedly reduced bacterial growth, whereas loss of the element improved fitness, indicating that ICE elimination alleviates the cost associated with TnSmu1 retention under these conditions. Finally, we show that the relationship between host translation and TnSmu1 stability extends to a genetically distinct S. mutans clinical isolate, although with strain-dependent penetrance. Together, these findings identify host translational state as an important physiological determinant of TnSmu1 stability and reveal that bacterial hosts can influence the maintenance of integrated mobile genetic elements through mechanisms that extend beyond element-encoded regulatory circuits.

evolutionary biology↗

bioRxiv2026

Genome-wide association study and genetic differentiation analysis reveal the genetic basis of Vibrio harveyi resistance in blackspotted croaker (Protonibea diacanthus)

Vibrio harveyi is a primary causative agent of vibriosis, leading to catastrophic economic losses in the blackspotted croaker (Protonibea diacanthus) aquaculture. However, the genetic architecture underlying its resistance remains poorly understood. In this study, we conducted the first integrated genome-wide association study (GWAS) and genetic differentiation analysis (F_ST) to investigate the genetic basis of V. harveyi resistance. A total of 450 juveniles were challenged, and 300 individuals (comprising 150 resistant and 150 susceptible) were selected for whole-genome resequencing. Genomic heritability (h^2) was estimated at 0.374+/-0.29 for survival time (ST) and 0.237+/-0.185 for survival status (SS), indicating moderate genetic control. Notably, an exceptionally high genetic correlation (r_g = 0.999) was observed between ST and SS, while resistance traits were genetically independent of growth performance (r_g < 0.03). Through the convergence of GWAS and selection sweep signals, a major-effect QTL was identified on chromosome 4. Candidate gene prioritization identified high-confidence targets involved in post-transcriptional regulation and mRNA stability, specifically elavl4 (harboring a top-tier exon variant), rorb, cbpc6, and epn4. Functional enrichment analysis further revealed that resistance is significantly associated with "rRNA processing", "mRNA surveillance pathway", and "nucleocytoplasmic transport", suggesting that efficient RNA metabolic control is a critical determinant of host survival. These findings provide novel insights into the non-canonical immune mechanisms of P. diacanthus and offer robust molecular markers for marker-assisted selection and genomic selection, facilitating the development of disease-resistant strains in the sciaenid industry.

genomics↗

bioRxiv2026

LucaCell: a sequence-centric foundation model for cross-species single-cell analysis

Single-cell foundation models have transformed transcriptomic analysis, yet most rely on fixed gene identifiers that limit transfer across species and data types. Here we present LucaCell, a sequence-centric foundation model that represents genes through pre-trained mRNA sequence embeddings rather than static gene annotations. Gene expression is discretized into bins and modeled with a Transformer encoder, enabling sequence-informed cell representation without a fixed gene-ID vocabulary. Pre-training on 85 million human and mouse single cells, LucaCell is evaluated on human, mouse and lemur gene expression profiles, human chromatin accessibility data, unaligned reads from more than 50 prokaryotic taxa, and five influenza A virus genomes. LucaCell enables manual-mapping-free cross-species cell type annotation and an alignment-free microbial embedding framework that simultaneously distinguishes bacterial species identity and intra-species physiological states. It also improves gene expression reconstruction by incorporating donor-specific exonic SNP information into mRNA sequence embeddings, and predicts cellular viral load across influenza A virus strains while highlighting infection-like transcriptional states in mock-infected cells. These results show that sequence-informed gene representation can improve the generalization of single-cell foundation models across species, data types, and predictive tasks.

bioinformatics↗

bioRxiv2026

RNA-guided transcriptional repression by TIGR-Tas systems

Tandem interspaced guide RNA (TIGR)-TIGR-associated protein (Tas) systems are a widespread family of RNA-guided DNA-targeting proteins whose diversity has remained uncharacterized because their arrays, unlike CRISPR arrays, lack the sequence conservation required by existing annotation tools. We developed TIGRFinder, a motif-based pipeline that identified 6,685 TIGR arrays from genomic and metagenomic data. Phylogenetic and structural analysis of Tas proteins revealed clade-specific insertions in stem-loop binding Tas proteins that co-vary with features of their cognate tigRNAs. Cryo-electron microscopy structures of two stem-loop binding TasR ribonucleoprotein complexes demonstrate how these protein insertions directly accommodate extended tigRNA stems while maintaining DNA binding through catalytically inactive RuvC domains. We also show that these catalytically dead TasR proteins, along with the nuclease-lacking TasA, function as RNA-guided transcriptional repressors. These findings establish TIGR-Tas as a functionally diverse family of RNA-guided effectors, with comprehensive annotations available through TIGRSafari (https://tigr.bio) to support further exploration and engineering.

molecular biology↗

bioRxiv2026

A Multi-Omics Study of Cognitive Resilience to Alzheimer's Disease

Alzheimers disease (AD) is the most severe form of dementia. While significant efforts have been made to identify AD risk factors and develop therapeutics, much less attention has been given to neurobiological factors and molecular signatures associated with cognitive resilience. Substantial evidence suggests that modifiable factors, such as physical and mental activity, may contribute to cognitive reserve and resilience. To this end, we developed a resilient cohort of paired genetics, transcriptomics and proteomics data from the prefrontal cortex of 282 cognitively normal older adults. We conducted extensive preprocessing and rigorous quality control (QC), including QC on sequence metrics and sample matching across multiomics. This cohort can be leveraged for identification of molecular signatures of cognitive resilience to AD and is freely shared with the research community.

genomics↗

bioRxiv2026

Selective GRAB sensors reveal distinct endocannabinoid dynamics in vivo

The endocannabinoid system modulates diverse physiological processes via two endogenous lipid ligands, 2-arachidonoylglycerol (2-AG) and anandamide (AEA); however, their specific spatiotemporal dynamics remain poorly understood owing to the lack of selective probes. Here, we developed GRAB2-AG2.0 and GRABAEA2.0, two genetically encoded fluorescent sensors that selectively detect 2-AG and AEA, respectively. Both sensors exhibited high apparent affinity and molecular specificity for their respective ligands, enabling the real-time detection of 2-AG and AEA release evoked by electrical stimulation in cultured neurons and acute brain slices. In freely behaving mice, these sensors revealed ligand- and context-specific eCB dynamics: aversive stimulation preferentially evoked 2-AG, whereas psychoactive drugs produced distinct 2-AG and AEA responses. Notably, {Delta}9-THC elicited a sustained 2-AG signal in the nucleus accumbens shell, and local deletion of Dagla markedly attenuated both this signal and {Delta}9-THC-induced hypolocomotion. These sensors therefore enable detecting 2-AG and AEA signaling seperately and reveal an endogenous 2-AG component of the behavioral response to {Delta}9-THC.

neuroscience↗

bioRxiv2026

piRNA loss unleashes episodic transposition bursts

In Metazoa, transposon expression is suppressed by the piRNA pathway, and disruption of this pathway leads to rampant transposon expression. However, it remains unclear whether increased transposon expression results in actual transposition, and if so, which and how frequently transposons mobilize upon piRNA loss. Here, we developed a framework to track transposon copy accumulation across generations on a single set of nonrecombining haploid genome in the Drosophila male germline, with or without the piRNA biogenesis factor HP1D/Rhino. Single-fly Nanopore DNA sequencing revealed that multiple transposon families mobilized after 10-45 generations of piRNA loss. Among them, the most prolific across all replicates was copia, producing dozens of new insertions that were distributed across chromosome arms. With genomic DNA collected at every generation, we validated and dated each copia insertion, revealing episodic bursts of transposition that deviated strongly from a Poisson process. Using phylogenetic analysis, we further showed that multiple copia loci--both autonomous and nonautonomous copies--can mobilize. Interestingly, two additional transposon families, mdg3 and invader3, also mobilized episodically, but with distinct timing and magnitude, highlighting the stochastic nature of transposition bursts. Because we did not observe a single transposition event for these elements in controls, our results provide direct evidence that the piRNA pathway tightly suppresses germline transposition. More importantly, our findings argue that disruption of the piRNA pathway does not simply elevate transposition rates, but it instead unleashes punctuated and stochastic bursts of transposon movement that may radically reshape the timing and magnitude of mutational input during evolution.

genetics↗

bioRxiv2026

Complex coacervation reshapes the aggregation landscape of tau

Biomolecular condensates are increasingly implicated in protein aggregation, yet their contribution is often reduced to that of concentrating reactants. Whether the phase state of a protein itself changes how it aggregates remains unclear. Using complex coacervates of a tau repeat-domain construct (K12) with heparin, we show that phase separation does not simply accelerate tau aggregation but redirects it along a distinct kinetic regime. Amyloid nucleation and growth are largely associated with the condensed phase, where intact phase-separated mixtures nucleate with a markedly shortened lag phase, whereas the corresponding dilute phase contributes little to overall amyloid formation. Aggregation kinetics in this regime become largely decoupled from total protein concentration. Because phase equilibrium pins the composition of the dense phase, additional tau partitions largely into the coexisting dilute phase without substantially altering the reacting population. This weak concentration dependence provides a kinetic signature of compartmentalized aggregation, and it recurs across chemically distinct coacervates formed with heparin, RNA, and polyglutamate, pointing to a general feature of coacervate-mediated tau assembly rather than a heparin-specific effect. Aggregation within coacervates also yields fibrils with altered morphology and secondary structure, suggesting access to alternative regions of the assembly landscape, and shows reduced sensitivity to bulk pH perturbations. Together, these results show that condensation changes tau aggregation by defining the local reaction environment: phase equilibrium buffers the dense-phase composition, in turn altering aggregation kinetics and the properties of the amyloid formed.

biophysics↗

bioRxiv2026

Dystrophic changes of nigrostriatal axons harboring a Synj1 Parkinson mutation suggest catastrophic failure of endocytic mechanisms

Synaptojanin 1 is a brain enriched phosphoinositide phosphatase implicated in endocytosis at the synapse. A mutation (R258Q) that selectively impairs its Sac1 phosphatase domain causes early onset familial Parkinsonism. Neurons of mice with this mutation display synaptic vesicle traffic defects across the brain, but selective dystrophic changes in a subset of dopaminergic axons in the dorsolateral striatum. Using correlative light microscopy-FIB-SEM of mutant mouse striata to visualize in 3D these abnormal structures we show that they represent clusters of focal axonal dilations harboring massive, onion-like DAT enriched plasma membrane infoldings, generally localized next to cell bodies of neighboring cells, often engulfing evaginations of such cells. This dysmorphia was associated with a deficit in dopamine release in the same striatal region. Given the involvement of Synj1 in endocytic mechanisms, these structures may reflect an imbalance between exocytosis and endocytosis. Their occurrence only in a subset of axons suggest a vulnerability threshold of these axons beyond which the expansion of the plasma membrane is not counteracted by compensatory mechanisms.

neuroscience↗

bioRxiv2026

Orthogonal Acoustic Control of Gene Expression Using a Synthetic T7RNAP Rapa-Inducible Dimerization System

Precise spatiotemporal regulation of engineered microbes remains a critical bottleneck in synthetic biology. While ultrasound is extensively utilized for imaging and drug delivery, its translation into bacterial chassis is hindered by the lack of stringent biochemical triggers. Here, we present a rationally designed, ultrasound-responsive hybrid molecular switch based on a strict acoustic-biochemical "AND-gate". We engineered a highly sensitive split-T7 RNA polymerase system, which the dimerization and subsequent gene transcription can only be triggered in the presence of both ultrasound and a PEG-modified rapamycin. By systematically optimizing the acoustic parameters, we deployed this spatiotemporal switch to dynamically regulate a microbial consortium. With lysisE suicide protein as the output module, we achieved precise and programmable tuning of bacterial population in a co-culture system. This acoustic gating strategy may provide a robust and versatile toolkit for complex microbiome engineering and dynamic biomanufacturing. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=158 SRC="FIGDIR/small/733485v1_ufig1.gif" ALT="Figure 1"> View larger version (41K): org.highwire.dtl.DTLVardef@6d229corg.highwire.dtl.DTLVardef@fb5d1eorg.highwire.dtl.DTLVardef@9e905dorg.highwire.dtl.DTLVardef@17e3a1_HPS_FORMAT_FIGEXP M_FIG Graphic Abstract C_FIG

Synthetic Biology↗

bioRxiv2026

Cardiomyocytes Undergo a Mesenchymal-Like Fate Transition in Myocardial Fibrosis

BACKGROUNDMyocardial fibrosis, a pathological hallmark of adverse cardiac remodeling and heart failure, has been conventionally attributed to the activation of resident fibroblasts. Although recent studies suggest contributions from non-fibroblast lineages, direct in vivo genetic evidence that cardiomyocytes can undergo a mesenchymal-like fate transition during myocardial fibrosis remains absent. This study aims to investigate whether such a transition occurs and to elucidate the underlying regulatory mechanisms. METHODSHuman myocardial infarction (MI) tissues were analyzed by immunohistochemistry and integrated with public single-nucleus RNA sequencing (snRNA-seq) data to detect mesenchymal-like signatures in cardiomyocytes. Genetic lineage-tracing was performed in MI mice, and in cardiomyocyte-specific Hgs (hepatocyte growth factor-regulated tyrosine kinase substrate) gene knockout mice, to map the fate of cardiomyocyte-derived cells. Mechanistic insights were obtained through proteomic and snRNA-seq analysis of Hgs knockout hearts and validated through gain- and loss-of-function experiments targeting Aldh1a2 (aldehyde dehydrogenase 1 family member A2). RESULTSIn human MI samples, a subset of cardiomyocytes showed reduced expression of cardiomyocyte markers concurrent with acquisition of mesenchymal-associated markers. Genetic lineage tracing demonstrated that adult cardiomyocytes can adopt a mesenchymal-like cell fate during post-MI remodeling. We identify HGS as a factor constraining this transition. Hgs knockout in adult cardiomyocytes upregulated Aldh1a2, triggered the mesenchymal-like fate transition, and gave rise to cells expressing markers of activated fibroblasts or osteoblasts, accompanied by pronounced myocardial fibrosis and calcification. Forced Aldh1a2 overexpression in cardiomyocytes drove the mesenchymal-like fate transition in vitro and in vivo, whereas Aldh1a2 deletion in cardiomyocytes mitigated MI-induced myocardial fibrosis. CONCLUSIONSThis study provides in vivo genetic evidence that adult cardiomyocytes possess the capacity to undergo a mesenchymal-like fate transition under pathological conditions. Our data suggest that HGS and ALDH1A2 serve as regulators of the transition, offering a new basis for understanding cellular and molecular mechanisms of myocardial fibrosis. Novelty and SignificanceO_ST_ABSWhat Is Known?C_ST_ABSO_LIMyocardial fibrosis is primarily driven by resident fibroblast activation, with additional contributions from cardiac CD34+ cells, pericytes, and macrophages. C_LIO_LIAdult cardiomyocytes exhibit phenotypic plasticity and transdifferentiate into epicardial-like or pacemaker cells under specific conditions. C_LI What New Information Does This Article Contribute?O_LIA subset of cardiomyocytes adopts a mesenchymal-like cell fate during myocardial fibrosis, marked by downregulation of cardiomyocyte identity markers and loss of aligned cell-cell contacts. C_LIO_LIThese cells acquire mesenchymal morphology and markers, ECM components, migratory gene signatures, and proliferative capacity. C_LIO_LIHGS and ALDH1A2 act as regulators of this mesenchymal-like fate transition. C_LI Myocardial fibrosis drives heart failure progression, yet the cellular sources of pathological fibroblasts remain incompletely defined. Here, we demonstrate that a subset of cardiomyocytes adopts a mesenchymal-like cell fate during myocardial fibrosis by using an integrated approach combining human MI samples, murine genetic lineage tracing, and snRNA-seq. Mechanistically, we identify HGS and ALDH1A2 as regulators of this transition. Cardiomyocyte-specific Hgs deletion upregulates Aldh1a2, triggering the mesenchymal-like fate transition. Furthermore, Aldh1a2 overexpression drives this transition, while its deletion attenuates MI-induced fibrosis. These findings reveal a previously unrecognized plasticity of adult cardiomyocytes and identify potential therapeutic targets for fibrotic heart disease.

genetics↗

bioRxiv2026

Self-supervised Internal Learning Enhances Isotropic Resolution for Three-dimensional Fluorescence Microscopy

Three-dimensional fluorescence microscopy often exhibits anisotropic resolution because axial information is poorly sampled and more blurred than lateral information, which complicates quantitative interpretation of fine 3D structures. Although optical remedies and computational restoration have been explored, many approaches require demanding system calibration or rely on accurate PSF models and assumptions that are difficult to satisfy across all samples and modalities. Here we present DeepIso, a self-supervised isotropy restoration framework that couples supervised pretraining with an internal-learning inference stage to estimate degradation directly from the measured volume. Without explicit PSF specification or enforced lateral-axial structural equivalence, DeepIso recovers axial frequency content and improves the continuity of elongated structures while retaining fine features, with superior performance over existing computational approaches in terms of both visual inspection and quantitative metrics. The method is validated on synthetic benchmarks and experimental datasets, demonstrating isotropy enhancement across confocal, light-sheet, and 3D structured illumination microscopy, thereby supporting downstream volumetric analysis including segmentation and tracking.

bioengineering↗

bioRxiv2026

Discovery of ARQ-501 as a potent PafA inhibitor that elevates KatG levels to potentiate isoniazid in mycobacteria

The prokaryotic ubiquitin-like protein (Pup) conjugation system (PPS), which is essential for Mycobacterium tuberculosis (Mtb) virulence but absent in humans, presents an attractive drug target. Here, we report the discovery of ARQ-501, a quinone-based, covalent, substrate-competitive inhibitor of the Pup ligase PafA. ARQ-501 exhibited potent anti-mycobacterial activity against Mtb under host-mimicking stress conditions and within macrophages. We further identified the catalase-peroxidase KatG, essential for activation of the frontline prodrug isoniazid (INH), as a pupylation substrate. ARQ-501 inhibits KatG pupylation, causing its accumulation and creating a selective synergy with INH. This quantity over quality mechanism successfully rescued INH activation by the clinically prevalent KatG S315T mutant in enzymatic assays and enhanced INH efficacy against clinical S315T isolates to variable degrees. This work identifies a novel class of PafA inhibitors and a previously unrecognized role of pupylation in regulating KatG, offering a potential therapeutic avenue to combat drug-resistant tuberculosis.

microbiology↗

bioRxiv2026

An 8 Gene Bevacizumab Resistance Signature Predicts Prognosis and Reveals Immunosuppressive Microenvironment in Colorectal Cancer

BackgroundBevacizumab resistance severely limits long-term efficacy in metastatic colorectal cancer (CRC). This study aimed to develop and validate a bevacizumab resistance-associated gene signature for prognosis prediction and immune microenvironment characterization in CRC. MethodsTwo GEO datasets (GSE19862, GSE86582) with bevacizumab response data and TCGA-COAD/READ RNA-seq data were analyzed. Overlapping differentially expressed genes (DEGs) linked to both CRC progression and bevacizumab resistance were identified. An 8-gene signature (AXIN2, PSORS1C1, KRT74, SLC2A3, STIL, IL33, GALNT6, HSD11B2) was constructed via univariate Cox and LASSO-Cox regression. ResultsIn the TCGA cohort, high-risk patients had shorter overall survival (OS; log-rank P < 0.0001). Time-dependent ROC yielded 1-year AUC = 0.638, 3-year AUC = 0.657, and 5-year AUC = 0.757. Multivariate Cox regression confirmed the risk score as an independent prognostic factor. External validation in GSE39582 (optimal cutoff = -1.49) replicated these findings: high-risk patients had inferior OS (P = 0.0016) with acceptable 1/3/5-year AUCs and retained independent prognostic value (HR = 1.634, P = 0.00415). CIBERSORT and ESTIMATE analyses showed that the high-risk group was characterized by increased M2 macrophages and neutrophils, higher immune and stromal scores, and reduced activated memory CD4+ T cells, monocytes, and activated dendritic cells (all P < 0.05). GSEA highlighted enrichment of TNF-/NF-{kappa}B, IL-6/JAK/STAT3, and immune checkpoint pathways in the high-risk group. AXIN2 (HR = 0.829, P = 0.032) was an independent protective factor, while PSORS1C1 (HR = 1.356, P = 0.048) was an independent risk factor. ConclusionThe 8-gene bevacizumab resistance signature robustly predicts prognosis and reflects an immunosuppressive microenvironment closely linked to bevacizumab failure in CRC. These findings provide novel insights into immune-mediated resistance and support clinical risk stratification.

bioinformatics↗

bioRxiv2026

Spatiotemporally dynamic noradrenergic regulation of cortical networks

Brain activity and cognition exhibit state-dependent fluctuations that may reflect the influence of neuromodulatory systems, including norepinephrine (NE). Although the activity of noradrenergic neurons is strongly coupled to sleep-wake cycles and pupil dynamics, suggesting a global arousal signal, recent evidence suggests potentially modular partitioning of these cells. In addition, it is unclear whether different neuromodulatory signals exhibit distinct spatiotemporal patterns. We performed simultaneous, dual color mesoscopic imaging of either NE and calcium or NE and acetylcholine (ACh) signals across the neocortex, along with high-density electrophysiology, to investigate the relationship between NE release and neural activity. We find that the pattern of cortical NE signaling varies with behavioral state and is associated with enhanced functional connectivity. Simultaneous imaging of NE and ACh reveals spatiotemporally dynamic coupling between signals. Finally, NE signaling and functional connectivity are disrupted by sleep deprivation. Overall, our findings demonstrate that NE provides a multimodal signal that links transitions in behavioral state to cortical network interactions.

neuroscience↗

bioRxiv2026

Guidance for high-quality functional gene embeddings from large language models

Large language models (LLMs) are increasingly used to generate gene embeddings, yet systematic benchmarks of prompting strategies and practical guidance for obtaining biologically meaningful representations remain limited. Here we present GEbench, an evaluation framework for assessing LLM-derived gene embeddings across different tasks, prompting strategies, and LLM architectures. GEbench revealed that embedding quality depends primarily on whether the input text contains explicit functional information, rather than on sparse gene identifiers or model size. Identifier-based embeddings showed weak biological organization, whereas embeddings derived from functional descriptions consistently achieved stronger functional separation and predictive performance. Notably, Self-Des, which extracts embeddings from model-generated gene function descriptions, enabled locally deployable LLMs to generate high-fidelity representations that approach the quality of expert-curated databases. Genome-scale analyses further supported these findings, indicating that explicit functional descriptions are an effective design principle for generating high-quality gene embeddings from LLMs.

bioinformatics↗

bioRxiv2026

A genome-wide in vivo screen reveals fitness pathways required for streptococcal infective endocarditis

Infective endocarditis (IE) is a life-threatening disease most often caused by blood-borne bacteria that infect previously damaged cardiac tissue. Despite the importance of this disease, the genetic basis for IE virulence remains poorly defined. Here, we present the first genome-wide in vivo analysis of bacterial fitness in a vertebrate model of IE. We identified 146 genes in Streptococcus sanguinis required for IE fitness, the majority of which had not previously been linked to endocarditis. These determinants cluster into conserved metabolic, cell envelope, transport, and regulatory pathways, representing a vast reservoir of potential targets for novel antimicrobial intervention. A subset of these genes was examined in Streptococcus mutans; all were found to be essential for IE fitness in this distantly related oral species as well, suggesting broad conservation. Using experimental evolution, we further show that disruption of key fitness pathways triggers reproducible compensatory "bypass" mechanisms that reveal the inherent physiological constraints of the IE fitness landscape and identify vulnerable nodes for multi-target drug strategies. Together, these findings redefine streptococcal infective endocarditis as a disease shaped by conserved bacterial fitness networks that may be exploited for therapeutic development. HighlightsO_LIA genome-wide in vivo screen identified 146 Streptococcus sanguinis genes required for infective endocarditis fitness. C_LIO_LI94% of these genes represent previously unrecognized determinants of endocarditis. C_LIO_LIMultiple pathways--including CoA biosynthesis, the shikimate pathway, and rhamnan synthesis--are required for cardiac colonization. C_LIO_LIA subset of infective endocarditis fitness factors are conserved between S. sanguinis and S. mutans, with species-specific adaptations. C_LIO_LIExperimental evolution revealed compensatory metabolic networks that buffer IE fitness defects. C_LI

genetics↗