Biology subjects

Xu, D.

Publications and source records attributed to Xu, D..

At least 73 records · Page 4Linked to original sources

bioRxiv2025

Harnessing the Power of Single-Cell Large Language Models with Parameter Efficient Fine-Tuning using scPEFT

Single-cell large language models (scLLMs) capture essential biological insights from vast single-cell atlases but struggle in out-of-context applications, where zero-shot predictions can be unreliable. To address this, we introduce a single-cell parameter-efficient fine-tuning (scPEFT) framework that integrates learnable, low-dimensional adapters into scLLMs. By freezing the backbone model and updating only the adapter parameters, scPEFT efficiently adapts to specific tasks using limited custom data. This approach mitigates catastrophic forgetting, reduces parameter tuning by over 96%, and decreases GPU memory usage by more than half, significantly enhancing scLLMss accessibility for resource-constrained researchers. Validated across diverse datasets, scPEFT outperformed zero-shot models and traditional fine-tuning in disease-specific, cross-species, and under-characterized cell population tasks. Its attention-mechanism analysis identified COVID-related genes associated with specific cell states and uncovered unique blood cell subpopulations, demonstrating scPEFTs capacity for condition-specific interpretations. These findings position scPEFT as an efficient solution for improving scLLMs utilities in general single-cell analyses.

cell biology↗

bioRxiv2025

The H3K27me3 reader UAD-2 recruits a TAF-12-containing transcription condensates to initiate piRNA expression within heterochromatic clusters

Metazoans utilize the small RNA pathway to regulate gene expression and maintain genome integrity. This pathway directs histone H3 lysine 9 tri-methylation (H3K9me3) or histone H3 lysine 27 tri-methylation (H3K27me3) at target loci to induce transcriptional gene silencing. Interestingly, some small RNAs are generated from genomic loci enriched in H3K9me3 or H3K27me3. However, the transcription mechanism of small RNA precursors from these heterochromatic regions remains unclear. In C. elegans, piRNAs originate from two genomic clusters enriched with H3K27me3 marks, which recruit the H3K27me3 reader UAD-2 and the upstream sequence transcription complex (USTC). Here, we demonstrate that piRNA transcription in C. elegans relies on TAF-12, a subunit of the basal transcription factor IID (TFIID). Depletion of TAF-12 reduces the production of both piRNA precursors and mature piRNAs. TAF-12 interacts with UAD-2 and facilitates piRNA focus formation in germ cell nuclei. We further show that TAF-12 triggers piRNA transcription by recruiting the RNA polymerase II subunit RPB-5, the Mediator complex subunit MDT-8, and the general transcription factors GTF-2F2 and GTF-2H2C. Thus, piRNA transcription within heterochromatic regions depends on the collaboration between histone modification readers, piRNA-specific transcription factors, and core transcription machinery.

genetics↗

bioRxiv2025

Small non-coding RNAs encapsulating mammalian cells fuel innate immunity

Cell surface RNAs, notably glycoRNAs, have been reported, yet the exact surface RNA compositions in different cell types remain unclear. Here, we introduce a comprehensive suite of methodologies for imaging, profiling, quantifying, and exploring the biological functions of specific surface RNAs. Utilizing these techniques, we have identified diverse non-coding RNAs present on mammalian cell surfaces. We confirm the membrane anchorage and quantify the abundance of several representative RNAs on human primary cells. Notably, we discover a significant prevalence of Y RNAs on the surfaces of human monocytes and B cells. We find that these Y RNAs on human monocyte surfaces enrich extracellular histones, regulating interleukin-6 (IL-6) gene expression and subsequent protein secretion upon histone stimulation via NF-{kappa}B and AP-1 activation. Our study not only presents effective approaches for investigating surface RNAs, but also uncovers a previously unrecognized immune activation pathway mediated by surface Y RNAs on monocytes. In briefA comprehensive profiling of surface RNAs across diverse mammalian blood cell types unveil abundant Y RNAs on the surface of human monocytes. Subsequent investigations uncover functional roles of surface Y RNAs on monocytes as "immune sentinel" to enrich extracellular histones, revealing a previously unknown pathway of innate immune activation. HighlightsNovel methodologies for cell surface RNA mapping, validation and functional interrogation. Sequencing of surface RNAs across various mammalian blood cell types offers detailed maps of surface RNAs. Abundant Y RNAs localize on human monocyte surfaces and capture extracellular histones released by adjacent ruptured cells. Surface Y RNAs enrich extracellular histones to facilitate transcription and secretion of IL-6 through NF-{kappa}B and AP-1 activation.

molecular biology↗

bioRxiv2025

Phloem-Specific Translational Regulation of Soybean Nodulation: Insights from a Phloem-Targeted TRAP-Seq Approach

Soybean (Glycine max) root nodulation is a symbiotic process that requires complex molecular and cellular coordination. The phloem plays a crucial role not only in nutrient transport but also in long-distance signaling that regulates nodulation. However, the molecular mechanisms underlying phloem-specific regulation during nodulation remain poorly characterized. Here, we developed a phloem-specific Translating Ribosome Affinity Purification sequencing (TRAP-seq) system to investigate the translational dynamics of phloem-associated genes during nodulation. Using a phloem-specific promoter (Glyma.01G040700) combined with the GAL4-UAS amplification system, we successfully captured the translatome of soybean root phloem at early (72 hours post-inoculation, hpi) and late (21 days post-inoculation, dpi) nodulation stages. Differential expression analysis revealed dynamic translational reprogramming, with 2,636 differentially expressed genes (DEGs) at 72 hpi and 8,422 DEGs at 21 dpi. Gene ontology and pathway enrichment analyses showed stage-specific regulatory shifts, including early activation of ethylene and defense pathways and late-stage enhancement of nutrient transport and vascular development. Transcription factor analysis identified GmbHLH121 as a key phloem-specific regulator of nodulation. Functional validation using RNAi knockdown and overexpression experiments demonstrated that GmbHLH121 negatively regulates nodule formation, likely acting downstream of or independently from early nodulation signaling pathways. Additionally, we uncovered dynamic regulation of cell wall-modifying enzymes (PME and PMEI) in the phloem, implicating their role in modulating plasmodesmata permeability and facilitating symplastic connectivity during nodulation. Our findings highlight the critical role of phloem-mediated translational regulation in coordinating root nodulation, emphasizing the phloem as an active regulatory hub for long-distance signaling and symbiotic efficiency.

plant biology↗

bioRxiv2025

SPDEv3.0: A Multidisciplinary Integrated Data Analysis Platform

Plant research encounters significant challenges in integrating multidisciplinary data and ensuring compatibility among diverse analytical methods. The limited automation in many existing tools further complicates data processing, thereby hindering efficiency and interdisciplinary collaboration. In response to these issues, we present SPDEv3.0, a multidisciplinary integrated data analysis platform specifically designed to tackle the complexities of data and the varied analytical requirements in plant sciences. Unlike traditional tools, SPDEv3.0 facilitates seamless cross-disciplinary data integration and automates analysis, featuring over 130 functions spanning plant breeding, molecular biology, genomics, and comparative genomics. The platform enables one-click operations for tasks ranging from data normalization to statistical analysis and visualization, significantly minimizing manual intervention and enhancing research efficiency. With its modular design and user-friendly interface, we believe that SPDEv3.0 will promote interdisciplinary collaboration and provides robust technical support for advancements in plant sciences.

bioinformatics↗

bioRxiv2025

Exploring the mycovirome: novel and diverse mycoviruses in Botrytis cinerea

Botrytis cinerea is a necrotrophic fungal pathogen that causes significant economic losses to many crops, including vegetables, fruits, and ornamental plants. The management of B. cinerea is difficult due to a rise in fungicide resistance. Harnessing mycoviruses that cause reduced virulence (hypovirulence) in B. cinerea is a promising alternative. Over 100 mycoviruses have been identified in Botrytis spp. to date, including several hypovirulence-inducing mycoviruses. This research aimed to further explore, for the first time in Canada, the mycovirome of B. cinerea and identify potential hypovirulence-inducing mycoviruses. Isolates of B. cinerea were collected from fruits and vegetables in the province of Quebec. Fitness and pathogenicity criteria, including sclerotia production, colony morphotype, and lesion size were evaluated. A double-stranded RNA (dsRNA) extraction protocol tailored to the detection of mycoviruses was used to sequence dsRNA from 45 isolates with low fitness/pathogenicity, and an in-house bioinformatics workflow was used to profile the mycovirome. Mycoviruses were identified in 44/45 isolates. Most of these had positive single-stranded RNA or dsRNA genomes, and a small number had negative single-stranded RNA, single-stranded DNA, or reverse transcriptase RNA genomes. Following deep analysis of RNA-dependent RNA polymerase and replication initiation proteins, a total of 62 unique contigs were identified belonging to new strains of mycovirus species. Furthermore, four putative novel mycovirus species belonging to Endornaviridae, Botybirnaviridae, Peribunyaviridae, and Bunyavirales taxa were identified. Several mycovirus species positively and/or negatively co-occurred with B. cinerea isolates collected from strawberry or raspberry. This study revealed a high degree of diversity in the mycovirome of B. cinerea. Species accumulation curve analysis indicated that, with the number of isolates characterized, we were unable to capture the full extent of expected diversity. Nevertheless, we identified potential hypovirulence-inducing mycoviruses, including Botrytis cinerea mitovirus 1, Botrytis cinerea hypovirus 1, and Botrytis porri botybirnavirus 1. Some of these novel mycoviruses belonged to taxa known to produce viral particles, which can be an interesting feature for their use as biocontrol agents (BCA). ImportanceThis study provides the first comprehensive profiling of mycoviruses infecting Botrytis cinerea in Canada, a significant step in understanding how these viruses can naturally limit crop disease. Due to growing resistance against conventional fungicides, new biological methods to control B. cinerea are crucial. By profiling mycoviruses in fungal samples collected in Quebec, we identified several novel viruses that appear to reduce the pathogenicity of B. cinerea. These viruses, known as hypovirulence-inducing mycoviruses, could be used to develop biocontrol agents (BCA), offering a more sustainable disease management alternative. Notably, we found virus families with extracellular potential, which may enable easier application as BCAs in agriculture. This research not only broadens the understanding of fungal virology but also holds promise for innovative, eco-friendly approaches to managing Botrytis cinerea in Canada.

genomics↗

bioRxiv2025

Predicting Kinase-Substrate Phosphorylation Site Using Autoregressive Transformer

Accurately predicting kinase-specific phosphorylation sites remains difficult due to the diversity of kinases and the context-dependent nature of substrate recognition. Importantly, aberrant kinase overactivation is a hallmark of many cancers including colorectal, gastric, liver, and breast tumors where dysregulated kinase signaling promotes malignant transformation, tumor progression, and therapy resistance. This underscores the clinical importance of understanding kinase-substrate relationships and precisely mapping phosphorylation events. In this paper, we introduce two complementary sequence-based architectures that operate directly on full-length substrate and kinase sequences. Stage 1 extends a task-agnostic prediction method, named Prot2Token, to jointly support three tasks: kinase-group classification from substrate sequences alone, kinase-substrate interaction prediction, and kinase-specific phosphorylation-site prediction while incorporating a self-supervised decoder pretraining task that predicts amino-acid positions from encoder embeddings. This pretraining substantially strengthens site prediction. Stage 2 specializes the architecture for phosphorylation-site prediction by replacing causal decoding of Prot2Token with a bidirectional one, yielding further gains. On standard benchmarks, the specialized model consistently outperforms widely used baselines. Beyond in-distribution evaluation, across both in-distribution and zero-shot settings of understudied dark kinases, we show the sign of zero-shot kinase-specific phosphorylation-site prediction capability. Together, these results indicate that jointly modeling substrate and kinase sequences provides a straight-forward, scalable approach to state-of-the-art, zero-shot-capable phosphorylationsite prediction.

bioinformatics↗

bioRxiv2025

Using Autoregressive-Transformer Model for Protein-Ligand Binding Site Prediction

AO_SCPLOWBSTRACTC_SCPLOWAccurate prediction of protein-ligand binding sites is critical for understanding molecular interactions and advancing drug discovery. Existing computational approaches often suffer from limited generality, restricting their applicability to a small subset of ligands, while data scarcity further impairs performance, particularly for underrepresented ligand types. To address these challenges, we introduce a unified model that integrates a protein language model with an autoregressive transformer for protein-ligand binding site prediction. By framing the task as a language modeling problem and incorporating task-specific tokens, our method achieves broad ligand coverage while relying solely on protein sequence input. We systematically analyze ligand-specific task token embeddings, demonstrating that they capture meaningful biochemical properties through clustering and correlation analyses. Furthermore, our multi-task learning strategy enables effective knowledge transfer across ligands, significantly improving predictions for those with limited training data. Experimental evaluations on 41 ligands highlight the models superior generalization and applicability compared to existing methods. This work establishes a scalable generative AI framework for binding site prediction, laying the foundation for future extensions incorporating structural information and richer ligand representations. The code, model, and datasets are available at this link.

bioinformatics↗

bioRxiv2025

Extending Prot2Token: Aligning Protein Language Models for Unified and Diverse Protein Prediction Tasks

AO_SCPLOWBSTRACTC_SCPLOWComprehensive protein function and property prediction remains a major challenge due to the vast diversity of sequences, structural variations, and limited labeled data. Existing models are often specialized to be task-specific, requiring independent training, which limits scalability. To address this, we extend Prot2Token, a unified autoregressive framework that focuses on the post-training alignment of pre-trained protein language models (PLMs), to new applications. Our approach enables next-token prediction across new applications of proteinprediction tasks, including protein-protein structure similarity, 3D structure prediction, mutation stability, post-translational modifications (PTMs), substratekinase phosphorylation sites, protein-protein affinity, and protein-ion binding sites. We introduce a self-supervised pre-training stage for the decoder, enhancing model initialization and improving downstream predictions. By integrating a causal autoregressive transformer with a pre-trained ESM-2 encoder, our model effectively aligns diverse protein tasks within a single framework. Additionally, we discuss the opportunities and limitations of this approach, providing insights for future research in optimizing PLMs as a general tool for broader biological applications. Code is available on GitHub Repository.

bioinformatics↗

bioRxiv2025

Lamin A/C safeguards replication initiation by orchestrating chromatin accessibility and PCNA recruitment

Lamin A/C, a nuclear lamina protein, is essential for maintaining nuclear architecture, organizing chromatin and preserving genomic stability. However, its role in regulating DNA replication remains unclear. This study investigates how Lamin A/C regulates proper activation of replication origins by orchestrating the chromatin structure and recruitment of proliferating cell nuclear antigen (PCNA). We demonstrate that Lamin A/C stabilizes replication domains (RDs) by restricting chromatin mobility, preserving spatial organization, and maintaining accessibility. Furthermore, Lamin A/C interacts with PCNA and sequesters a pool of PCNA, to regulate its recruitment to replication machinery. The loss of Lamin A/C leads to an increase in chromatin dynamics, RD accessibility, and PCNA availability at RDs, which together trigger excessive activation of replication origins, leading to replication stress and DNA damage. These disruptions prolong the S phase and compromise genome stability, highlighting Lamin A/C as a critical gatekeeper of balanced replication initiation. Our findings reveal Lamin A/Cs dual role in chromatin organization and replication machinery regulation, offering valuable insights into its involvement in replication-associated diseases and potential therapeutic opportunities through targeting replication dynamics.

biophysics↗

bioRxiv2025

Tumor hypoxia adaptation depends on FLAD1 mediated mitochondrial metabolic reprogramming

Hypoxia, a hallmark of solid tumors, promotes the malignant progression and is challenging to target. Metabolic reprogramming and the resulting metabolic vulnerabilities provide a promising strategy to target tumor hypoxia. Here we systematically compared the metabolic network differences between hypoxic and non-hypoxic cells, and developed a deep learning model, "DepFormer", to predict the dependent metabolic genes in hypoxic tumor cells. The performance of DepFormer was validated using CRISPR screening dataset. Oxidative phosphorylation was identified as the most significantly hypoxia-dependent metabolic pathway, and FLAD1 was predicted to be one of the key hypoxia-dependent metabolic genes. FLAD1 locus is amplified, and FLAD1 expression is upregulated in various tumor types, especially in hypoxic tumors. FLAD1 depletion compromises tumors adaptation to hypoxia by disrupting mitochondrial complex II activity, leading to an imbalance between succinate and fumarate, and consequent failure to adapt to hypoxia. Subsequently, we identified a drug-like inhibitor of FLAD1, which selectively inhibits the growth of tumor cells under hypoxia. Our findings reveal FLAD1 as an innovative therapeutic target for hypoxic tumors.

cancer biology↗

bioRxiv2025

China Autism Brain Imaging Consortium: Charting Brain Growth in Chinese Children with Autism

Autism Spectrum Disorder (ASD) is a lifelong neurodevelopmental condition characterized by atypical brain growth. While advances in neuroimaging and openly sharing large-sample datasets such as the Autism Brain Imaging Data Exchange (ABIDE) have improved understanding of ASD, most studies focus on adolescents and adults, with early brain development-critical for diagnosis and intervention-remaining underexplored. Existing research predominantly involves Western samples, offering limited insight and generalizability into non-Caucasian populations. We introduce the China Autism Brain Imaging Consortium (CABIC) (https://php.bdnilab.com/resources/), a grassroots effort by researchers across the country to aggregate previously collected multi-site structural MRI datasets and phenotypic information from 1,451 autistic children and 1,119 typically developing children, covering an age range from early childhood to school age (1.0 - 12.92 years). Here, we present this resource and depict brain growth charts to push forward a more comprehensive understanding of the brain development in Chinese autism children. We constructed brain growth charts that reveal a developmental shift in autistic children, transitioning from early overgrowth to delayed maturation. Regional analyses identified distinct atypical trajectories across specific brain regions. Individual deviation scores quantified inter-subject variability, characterizing the heterogeneity of brain development in ASD. Comparative analyses between CABIC and ABIDE highlighted differences potentially attributable to ethnicity and culture, advancing our understanding of cross-population neurodevelopmental diversity. CABIC MRI datasets will be shared publicly to foster investigation of the potential neural mechanisms underlying ASD in non-Western populations and support efforts toward precision medicine for autistic individuals across diverse backgrounds.

neuroscience↗

bioRxiv2025

Persistent fatty acid catabolism during plant oil synthesis

Plant lipids are an essential energy source for diets and are a sustainable alternative to petroleum-based fuels and feed stocks. Fatty acid breakdown during seed germination is crucial for seedling establishment but unexpected during seed filling. Here, we demonstrate that the simultaneous biosynthesis and degradation of fatty acids begins early and continues across all phases of oil filling and throughout the photoperiod. Tests in camelina, rapeseed and engineered high-oil tobacco line confirmed that concomitant synthesis and breakdown in oil-producing tissues over development is the rule rather than the exception. We show that transgenics, designed for significantly higher fatty acid biosynthesis, fail to achieve a proportional increase in storage lipid levels due to increasing degradation, potentially explaining the underperformance of engineered lines compared to expectations.

plant biology↗

bioRxiv2025

SAMCell: Generalized Label-Free Biological Cell Segmentation with Segment Anything

BackgroundWhen analyzing cells in culture, assessing cell morphology (shape), confluency (density), and growth patterns are necessary for understanding cell health. These parameters are generally obtained by a skilled biologist inspecting light microscope images, but this can become very laborious for high throughput applications. One way to speed up this process is by automating cell segmentation. Cell segmentation is the task of drawing a separate boundary around each individual cell in a microscope image. This task is made difficult by vague cell boundaries and the transparent nature of cells. Many techniques for automatic cell segmentation exist, but these methods often require annotated datasets, model retraining, and associated technical expertise. ResultsWe present SAMCell, a modified version of Metas Segment Anything Model (SAM) trained on an existing large-scale dataset of microscopy images containing varying cell types and confluency. We find that our approach works on a wide range of microscopy images, including cell types not seen in training and on images taken by a different microscope. We also present a user-friendly UI that reduces the technical expertise needed to use this automated microscopy technique. ConclusionsUsing SAMCell, biologists can quickly and automatically obtain cell segmentation results of higher quality than previous methods. Further, these results can be obtained through our custom GUI without expertise in Machine Learning, thus decreasing the human labor required in cell culturing.

bioinformatics↗

bioRxiv2025

scBSP: A fast and accurate tool for identifying spatially variable features from high-resolution spatial omics data

Emerging spatial omics technologies empower comprehensive exploration of biological systems from multi-omics perspectives in their native tissue location in two and three- dimensional space. However, sparse sequencing capacity and growing spatial resolution in spatial omics present significant computational challenges in identifying biologically meaningful molecules that exhibit variable spatial distributions across different omics. We introduce scBSP, an open-source, versatile, and user-friendly package for identifying spatially variable features in high-resolution spatial omics data. scBSP leverages sparse matrix operation to significantly increase computational efficiency in both computational time and memory usage. In diverse spatial sequencing data and simulations, scBSP consistently and rapidly identifies spatially variable genes and spatially variable peaks across various sequencing techniques and spatial resolutions, handling two- and three-dimensional data with up to millions of cells. It can process high-definition spatial transcriptomics data for 19,950 genes across 181,367 spots within 10 seconds on a typical desktop computer, making it the fastest tool available for handling such high-resolution, sparse spatial omics data while maintaining high accuracy. In a case study of kidney disease using 10x Xenium data, scBSP identified spatially variable genes representative of critical pathological mechanisms associated with histology.

bioinformatics↗

bioRxiv2025

Isolation of a novel plant growth-promoting Dyella sp. from a Danish natural soil

Environmental soils are natural reservoirs of unexplored microbes, including potentially beneficial microbes to improve plant performance. Here, we isolated 75 bacterial strains from surface-sterilized roots of Arabidopsis thaliana (Arabidopsis) grown in a natural soil derived from an alder swamp. Culture-dependent isolation of individual strains from the roots followed by monoassociation-based screening identified seven bacteria that promoted Arabidopsis seedling weight. Of those, we identified a new species from the Dyella genus which increased biomass of tomato and Arabidopsis seedlings in agar as well as the shoot biomass of Arabidopsis grown in both the alder swamp soil and potting soil. Dyella sp. A4 specifically promoted the elongation of lateral roots without affecting lateral root number and primary root elongation. The new Dyella sp. A4 expands the toolbox of biostimulants for plant growth promotion via modulating root architecture.

plant biology↗

bioRxiv2025

Super-omics and PCAS reveal rheumatoid arthritis as a tumor-like disease

Rheumatoid arthritis (RA) is a debilitating systemic autoimmune disorder that significantly impairs quality of life. To elucidate the molecular alterations in RA progression, we performed comprehensive super-omics analyses on synovial tissues from patients with joint trauma, arthritis, and RA. These analyses included transcriptomics, proteomics, metabolomics, microbiomics, and fourteen PTMs (phosphorylation, acetylation, lactylation, O-GlcNAc glycosylation, arginine monomethylation, lysine monomethylation, lysine dimethylation, lysine trimethylation, succinylation, malonylation, glutarylation, tyrosine nitration, N-glycosylation, and O-glycosylation). Additionally, we developed a protein-centered association study (PCAS) method to integrate these complex datasets. Using this approach, we identified key proteins, such as STK17B, and its interactors, which may be crucial in RA pathogenesis. Tumor-like features, including aberrant angiogenesis and epithelial-mesenchymal transition, were observed in RA, alongside the potential involvement of oncogenes and tumor suppressor genes. Finally, we constructed a molecular interaction blueprint of RA, providing a comprehensive framework for advancing RA pathogenesis, diagnosis, and therapy.

immunology↗

bioRxiv2024

Single-cell RNA sequencing analysis reveals new insights for intersex gonad initiation and early sex differentiation of goats

Polled intersex syndrome in goats severely impacts productivity and economic benefits, serving as a potential model for human reproductive defects. However, its specific molecular and cellular regulatory mechanisms remain unclear. This study combined single-cell RNA-sequencing (scRNA-seq) with tissue morphology and immunofluorescence detection to investigate the cellular heterogeneity, pseudo-time differentiation trajectory, differentially expressed genes, and regulatory networks of female, intersex, and male gonads during the 55-65 days embryonic stage of goats. The results indicated significant heterogeneity and cell fate transition in the somatic and germ cells of female, intersex, and male gonads at different developmental stages. Stromal cells may initiate female and intersexual gonad development, and epithelial cells may be the source of gonad cells in male fetuses. The abnormal expression of FOXL2, SOX9, AMH, DMRT1, RSPO1, INSL3, TCF21 and CSF1R may cause female-to-male sex reversal. The new candidate genes like PHLDB2, ZNRF3 and TEAD1 may contribute to the gender differentiation. Additionally, TREM2+ macrophages analogous to those in humans were identified in male gonads, while genes such as CSF1R and TYROBP potentially regulate macrophage immune privilege mechanisms. Early gonadal development in goats may also be regulated by circadian rhythm mechanisms similar to humans, with genes such as NR1D1, LGR4, and ATG5 involved in granulosa cell development, while genes such as ATG5, NRIP1, and FBXW11 may affect intersex trait development. In general, this study provided an important reference for exploring molecular and cellular mechanisms of intersex initiation during early embryonic development in goats and also for modelling human reproductive defects. Author summaryIn this study, we investigated the cell fate trajectory and key gene expression profile of goat fetus gonads using scRNA-seq. The analysis revealed cellular heterogeneity among normal male, normal female, and intersex gonadal somatic and germ cells, alongside their dynamic fate transitions across various developmental states. Pseudo-time trajectory analysis suggested that stromal cells might act as the initiation signal for female and intersexual gonad development, while epithelial cells could serve as the primary source of gonadal cells in male fetal goats. GO functional enrichment and KEGG analyses confirmed that abnormal expression of FOXL2, SOX9, AMH, DMRT1 and RSPO1 was linked to the emergence of intersexual gonads. Newly identified candidate genes including LRP2, PITX2, ARX, LHX9, PHLDB2, ZNRF3 and TEAD1 may contribute to sex-specific differences, while INSL3 and TCF21 could be linked to intersexual gonads. We identified TREM2+ macrophages analogous to those in humans, which could regulate immune privilege mechanisms. Circadian rhythmic changes were observed during early goat gonadal development.

developmental biology↗