Biology subjects

Tang, F.

Publications and source records attributed to Tang, F..

At least 19 recordsLinked to original sources

bioRxiv2026

Rspo1 dependent WNT signaling safeguards ovarian development by suppressing androgen signaling

Embryonic ovarian development requires high levels of WNT/B-catenin signaling yet the molecular basis of this requirement remains unclear. Mutations in the WNT potentiator RSPO1 impair ovarian development in both mouse and human, leading to gonadal masculinization in XX individuals. Here, we show that RSPO1 acts within a restricted developmental window at the onset of ovarian differentiation and that its absence triggers precocious androgen signaling in the developing ovaries. Moreover, androgen receptor inhibition rescues ovarian development in Rspo1 mutant mice. Our findings identify RSPO1/WNT/B-catenin signaling as an essential pathway that suppresses androgen signaling to ensure proper ovarian development in mouse.

developmental biology↗

bioRxiv2026

Non-Invasive Embryo Quality Assessment via Matrix-Optimized Untargeted LC-MS Metabolomics of Spent Embryo Culture Media and Weighted Ensemble Machine Learning

BackgroundNon-invasive embryo quality assessment is a critical unmet need in assisted reproductive technology (ART). Preimplantation genetic testing for aneuploidy (PGT-A) is effective but requires invasive biopsy that may compromise embryo viability. Metabolomics of spent embryo culture media (SECM) offers a non-invasive alternative, yet analytical challenges--limited sample volume, high salt content, and abundant proteins--have hindered standardization and clinical translation. ResultsWe systematically optimized sample preparation for untargeted LC-MS metabolomics of SECM using human serum as a reference. Optimal conditions were highly matrix-dependent: SECM required 7x volume of 50% acetonitrile for extraction and 40% acetonitrile for reconstitution, whereas serum required 10x volume of 100% methanol and 100% water--reflecting that SECM contains more non-polar species than serum. Applying the optimized workflow to 120 clinical SECM samples (72 euploid, 48 aneuploid), we identified 102 differential metabolites between euploid and aneuploid embryos, with prominent enrichment of lipid pathways (fatty acid metabolism, {beta}-oxidation, sphingolipid metabolism) and involvement of amino acid (methionine, tryptophan) and TCA cycle metabolism. A weighted ensemble machine learning model discriminated aneuploid from euploid embryos with an AUC of 0.977, 100.0% specificity, and 89.6% sensitivity. Among 72 euploid embryos stratified by morphological grading (good, fair, poor), metabolic alterations progressed from mitochondrial energy deficiency (good vs. fair) to broader lipid dysregulation (fair vs. poor), with the ensemble model achieving AUCs of 0.944, 0.889, and 0.943, respectively. ConclusionsThis study establishes a rigorously optimized and validated SECM metabolomics workflow that overcomes key analytical barriers in this challenging matrix. Our findings demonstrate that metabolic signatures--particularly in lipid and energy metabolism--are strongly associated with both embryo ploidy and morphological quality, providing biological insights into the metabolic underpinnings of embryo developmental competence. The high predictive performance of the ensemble model supports the feasibility of non-invasive embryo assessment as a complementary tool to existing methods, with potential to reduce reliance on invasive biopsy in ART. External validation in prospective multi-center cohorts is warranted to further assess clinical utility and generalizability. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=122 SRC="FIGDIR/small/741666v1_ufig1.gif" ALT="Figure 1"> View larger version (59K): org.highwire.dtl.DTLVardef@10fbe40org.highwire.dtl.DTLVardef@936273org.highwire.dtl.DTLVardef@246e5dorg.highwire.dtl.DTLVardef@1c9d115_HPS_FORMAT_FIGEXP M_FIG C_FIG

genetics↗

bioRxiv2026

Integrative Multi-Omics Analysis of Gut Microbiota Dysbiosis and Host-Microbiome Interaction Mechanisms in Hypertension

ObjectiveTo characterize gut microbiota dysbiosis in hypertension and investigate its multilevel interactions with the host immune system. MethodsIntegrated multi-cohort microbiome data were used to evaluate microbial diversity, differential abundance, and co-occurrence network features between individuals with hypertension and healthy controls. The scBPS framework was applied to analyze microbiome-cell associations, enabling the resolution of relationships between key microbial taxa and functional states of immune cells at single-cell resolution. ResultsSeveral potentially protective genera reduced in hypertension and occupied central topological positions in the co-occurrence networks. Single-cell analyses further demonstrated that multiple key genera were closely associated with the functional states of monocytes and T cells (p<0.05). Specifically, Bacteroides and Bifidobacterium were associated with the proliferation and repair of classical monocytes; Butyricimonas showed a negative association with antigen processing and presentation pathways in monocytes; and Oscillospira promoted the transition of dnT cells toward an immunoregulatory state, suggesting its potential role in immune homeostasis. ConclusionsIntegrated multi-omics analyses reveal that hypertension-associated gut microbes may contribute to disease development through immune regulation, providing insights into microbiome-immune interaction mechanisms and potential targets for precision interventions.

microbiology↗

bioRxiv2026

RVQ-Alpha: Bridging Single-Cell Transcriptomics and Large Language Models via Discrete Tokenization and Verifiable Reinforcement Learning

Single-cell RNA sequencing yields continuous expression profiles, whereas large language models operate over discrete autoregressive sequences, leaving no shared computational interface for language-model reasoning over cell states. Existing approaches either keep cellular information outside the LLM vocabulary, consume context per listed gene, or learn reconstruction codes without gene-level grounding. We introduce RVQ-Alpha, which systematically adapts four stages of LLM training (tokenization, supervised fine-tuning, reinforcement learning, and distillation) to single-cell analysis. Multi-codebook Residual Vector Quantization (RVQ) lexicalizes each profile into a compact, hierarchical cellular alphabet in the models native token stream, while a paired decoder reconstructs the corresponding expression profile. Evidence-First supervision grounds these symbols in named genes and expression-linked evidence; Fact-Aware RLVR then penalizes contradictory claims to support auditable reasoning. Task-specific RLVR yields strong experts, but a single Mixed policy underperforms them across all four task families. To recover specialist competence in a unified model, we adopt Multi-Teacher On-Policy Distillation (MOPD), which consolidates these experts into one All-in-One checkpoint without retaining a separate policy for each task. On the CAPSTONE benchmark, a four-task suite with explicit biological shifts and deterministic ontology-aware graders, the unified checkpoint improves over Mixed on all 12 metrics and remains within 0.005 of the task-routed specialist reference on every primary metric. Overall, RVQ-Alpha provides a unified pipeline for grounded, auditable multi-task single-cell analysis.

bioinformatics↗

bioRxiv2026

Unlocking Ultrasensitive Phosphoproteomics of Nanoscale Samples: A Rapid IMAC Enrichment Platform Using Ti-PAN Tip

Protein phosphorylation is a critical mechanism in cellular signal transduction, playing essential roles in various cellular processes. However, significant challenges of their low abundance, high dynamic range, and complexity are especially prominent for accurate phosphoproteome analysis in nanoscale samples, such as oocytes, where misregulation of phosphorylation can lead to infertility. In this study, we developed a novel phosphopeptide enrichment tip (Ti-PAN) using highly hydrophilic textile PAN fibers, enabling rapid in-situ one-tip IMAC enrichment in just 8 minutes. This approach not only minimizes sample loss but also enhances analytical sensitivity and is easily adaptable to automation for high-throughput applications, demonstrating excellent recovery (>94%) and outstanding selectivity for phosphopeptides with remarkable resistance to interference up to 1:5000. By integrating it with a compatible sample preparation workflow based on n-dodecyl {beta}-D-maltoside (DDM) lysis with the high-sensitivity MS data acquisition method of turboDDA, comprehensive mapping of phosphosites in microsamples was achieved, including those from the ERK interactome, mouse brain slices, and oocytes. Using this approach, we identified 2,139 to 11,296 phosphosites across 645 to 2,168 phosphoproteins from samples ranging from 100 to 100,000 cells. Furthermore, we successfully applied this strategy to analyze the phosphoproteome of mouse oocytes at different developmental stages, using as few as 5 oocytes per sample. This led to the identification of 2,709 phosphosites on 701 phosphoproteins, revealing several key kinases involved in meiotic resumption and oocyte maturation, including STK4 (serine/threonine kinase 4) and DYRK3 (dual specificity tyrosine phosphorylation-regulated kinase 3). To our knowledge, this represents the deepest coverage of phosphoproteomes from such minimal oocyte material reported to date, offering valuable mechanistic insights into developmental controls during oocyte maturation. This Ti-PAN-based method provides a robust, versatile, and automatable platform for large-scale, highly sensitive phosphopeptide enrichment, substantially advancing phosphoproteomic studies in complex biological systems.

biochemistry↗

bioRxiv2026

All-in-one Single-Cell Proteomic Analysis of Protein Alterations in Human Oocytes Undergoing in Vitro Aging

Study questionWhat specific proteomic changes and molecular markers associated with in vitro aging occur in human oocytes? Summary answerSingle-cell proteomic analysis revealed the protein dynamics during the in vitro aging of oocytes. Subsequent functional analysis highlighted key biological processes affected in aged oocytes and identified several genes as candidate key factors for oocyte and early embryo quality. What is known alreadyOvulated oocytes undergo a time-dependent degradation process if fertilization does not occur within a specific window. This aging process leads to morphological, molecular, and epigenetic changes that can compromise oocyte quality and subsequent embryo development. In previous studies, transcriptome sequencing in humans and mice has revealed serial changes in aging oocytes including oxidative stress, mitochondrial dysfunction, DNA damage, and alterations in cell cycle regulation. These changes can lead to apoptosis, chromosomal abnormalities, and epigenetic modifications, all of which negatively affect oocyte quality. However, research regarding protein level changes and regulatory mechanisms in human oocytes during in vitro aging remains unclear. Study design, size, durationUtilizing "All-in-one" single-cell proteomics, we investigated the protein expression levels in human oocytes at the germinal vesicle (GV) stage and metaphase II (M II) stage following 24 hours of in-vitro aging. We analyzed four groups of oocytes: fresh GV (n=8), in vitro-aged GV (n=9), fresh M II (n=11), and in vitro-aged M II (n=11). This approach allowed for a detailed comparison of proteomic changes associated with in vitro aging in oocytes at different developmental stages. Participants/materials, setting, methodsOocytes at GV and MII stage were collected for "All-in-one" single-cell quantitative proteomic respectively. Differentially expressed genes were analyzed using the R package DESeq2 or DEP (filtered with q-value[≤]0.05 and Foldchange[≥]1.5). EGSEA package was used to perform pathway analysis. Main results and the role of chanceThe proteomic analysis of fresh and in vitro-aged human GV and MII oocytes identified 3,268 proteins. In GV oocytes, compared to fresh samples, 73 proteins were upregulated and 90 were downregulated. Gene Ontology (GO) enrichment analyses revealed these DEPs were involved in key biological processes such as lysosome organization, spindle assembly and oxidative stress response. Gene Set Enrichment Analysis (GSEA) revealed a significant down regulation of genes associated with protein translation initiation and methylation pathways, while those in the apoptosis mitochondrial changes were upregulated. In MII oocytes, 63 proteins were upregulated and 69 were downregulated. GO analysis highlighted their involvement in cytoplasmic translation, ribosome biogenesis and oocyte development. GSEA showed that aged MII oocytes exhibited a downregulation of gene sets in protein translation but an upregulation in membrane fusion pathways compared to fresh oocytes. Furthermore, we identified proteins with consistent expression trends across both GV and MII stages, including MRFAP1 and MT2A, as candidate biomarkers for human oocyte aging in vitro. Limitations, reasons for cautionSingle cell proteomic techniques may not fully capture the dynamic range of proteins present within oocytes, leading to potential underrepresentation of low-abundance proteins that could play crucial roles. Furthermore, the heterogeneity within oocytes introduces variability that can obscure the identification of consistent proteomic signatures associated with oocyte quality or aging. Wider implications of the findingsThese findings not only provide an understanding of the protein changes underlying the in vitro aging of human oocytes but also offer potential biomarkers and intervention targets for future research and clinical applications in assisted reproductive technologies. Study funding/Competing interest(s)This project received funding from the National Natural Science Foundation of China (NO.82471693, 22574175); the Natural Science Foundation of Hunan Province (NO. 2024JJ4100), the Hunan Provincial Grant for Innovative Province Construction (2019SK4012), the Hundred Youth Talents Program of Hunan Province (to S.Z.), the Major Scientific Program of CITIC Group (No. 2023ZXKYB34100) and Scientific Research Foundation of Reproductive and Genetic Hospital of CITIC-XIANGYA (YNXM-202313, YNXM-202319, YNXM-202211). The authors have no conflicts of interest to declare.

developmental biology↗

bioRxiv2025

Pan-cancer tumor classification by a holistic tumor microenvironment atlas

The tumor microenvironment (TME) heterogeneity presents a major bottleneck to effective cancer immunotherapies. We address this by establishing a pan-cancer tumor classification system based on the holistic TME cellular components. First, using single-cell transcriptomes from 1,271 patients across 26 cancer types, we compile a TME cellular atlas, leading to the identification of recurrent cell states and multicellular modules across cancer types. Among these, we highlight a type I interferon (IFN-I)-related multicellular module, including IFIT1+ tumor-associated macrophages, as a critical contributor to immune-activation TMEs. From a holistic perspective, pan-cancer TMEs exhibit trichotomous patterns predominantly driven by T, myeloid, and stromal compartments. Leveraging fine-grained TME cellular composition, we further stratify all tumors into 10 stable groups. These groups display varying responses to immune checkpoint blockade (ICB), and each rationally selected therapeutic strategy specifically perturbs the corresponding expression signature. Our pan-cancer tumor classification scheme reveals the underlying patterns of TME heterogeneity and provides a framework for stratifying patients, guiding treatment selection, and developing novel therapeutic options.

cancer biology↗

bioRxiv2025

Decoding the Oral Microbiome: Metagenomic Insights into Host Metabolic Health

The oral microbiome may capture system-specific information about host metabolic health, yet large-scale, multi-system evidence remains scarce. We analyzed 9,431 participants in the Human Phenotype Project (HPP), integrating buccal-swab oral whole-metagenome profiles with 44 metabolic measures spanning liver ultrasound, continuous glucose monitoring (CGM), and dual-energy X-ray absorptiometry (DXA). Using a microbiome-wide association study (MWAS) framework, we constructed a multilayer map across strains, gene families and pathways, revealing widespread associations: 213 strains, 124,603 gene families and 299 pathways were significantly associated with metabolic measures. Prioritizing the strongest and cross-phenotype signals, we identified multiple oral features with most significant associations to metabolic health. For example, acyl carrier protein (ACP) was associated with lower liver inflammation and reduced adiposity, whereas polyamine biosynthesis and ceramide -oxidation tracked higher glucose variability and adverse liver and adiposity phenotypes. Leveraging these MWAS-derived signals, we trained disease classification models using phenotype-selected oral features, which outperformed full-feature oral models across six metabolic diseases. These association signals were also robust in oral-health sensitivity analyses in HPP, and key BMI and waist-circumference associations directionally replicated at the genus level in an independent cohort (n = 20,293). Together, these findings provide a population-scale oral-metabolic association map and highlight the potential of oral microbial markers as non-invasive tools for metabolic risk stratification.

microbiology↗

bioRxiv2025

A fully-open structure-guided RNA foundation model for robust structural and functional inference

RNA language models have achieved strong performances across diverse downstream tasks by leveraging large-scale sequence data. However, RNA function is fundamentally shaped by its hierarchical structure, making the integration of structural information into pre-training essential. Existing methods often depend on noisy structural annotations or introduce task-specific biases, limiting model generalizability. Here, we propose structRFM, a structure-guided RNA foundation model that is pre-trained on millions of RNA sequences and secondary structures data by integrating base pairing interactions into masked language modeling through a novel pair matching operation. We further introduce MUSES (multi-source ensemble of secondary structures) to mitigate model bias, and a dynamic masking ratio to balance the structure-guided mask and nucleotide-level mask. structRFM learns joint knowledge of sequential and structural data, producing versatile representations, including classification-level, sequence-level, and pairwise matrix features, that support a broad spectrum of downstream adaptations. structRFM ranks among the top models in zero-shot homology classification across seventeen biological language models, and sets new benchmarks for secondary structure prediction. structRFM further derives Zfold, which enables robust and reliable tertiary structure prediction, with consistent improvements in estimating 3D structures and their accordingly extracted 2D structures, achieving a pronounced about 20% performance gain compared with baselines and comparable performances with AlphaFold3 on CASP15-natural, CASP16, and RNA-Puzzles datasets. In functional tasks such as internal ribosome entry site identification, structRFM achieves a whopping 48% performance gain in F1 score. Furthermore, state-of-the-art performances in extensive experiments across novel RNA families and long non-coding RNAs indicate the robustness and generalizability of structRFM. These results demonstrate the effectiveness of structure-guided pre-training and highlight a promising direction for developing multi-modal RNA language models in computational biology. To support the broader scientific community, we have made the 21-million sequence-structure dataset and the pre-trained structRFM model fully open-source, facilitating the development of multimodal foundation models in biology.

bioinformatics↗

bioRxiv2025

Immunomodulation of the Innate Host Response by Mesenchymal-Derived Versican during Influenza A Virus Infection

Viral and bacterial lung infections place a significant burden on public health. Versican, an extracellular matrix (ECM) chondroitin sulfate proteoglycan, coordinates the innate immune response in multiple experimental models. Versicans potential as an immunomodulatory molecule makes it a promising therapeutic target for controlling the hosts immune response to lung infection. However, versicans contribution to lung inflammation, injury, and immune cell activity during influenza A virus (IAV) infection represents a critical knowledge gap. To address our central hypothesis that mesenchymal-derived versican is pro-inflammatory and enhances the innate immune response to IAV infection, we generated a tamoxifen-inducible mouse deficient in mesenchymal-derived versican (B6. Col1a2-CreERT+/-/Vcantm1.1Cwf, Col1a2/Vcan-/-). We report that mesenchymal-derived versican plays a critical role in neutrophil, monocyte, and dendritic cell migration into the lungs and airways early in IAV infection. Intriguingly, mesenchymal-derived versican deficiency had the most substantial negative impact on neutrophil emigration into the lungs. We found that neutrophils were less adhesive to the ECM of Col1a2/Vcan-/- mouse lung fibroblasts (mLFs), which had a significant decrease in versican compared to wild-type mLFs. Additionally, Col1a2/Vcan-/- mLFs treated with poly(I:C) in vitro have reduced cell-associated hyaluronan. These findings suggest that fibroblast-derived versican is necessary for adhesion to lung fibroblasts by neutrophils as they transit into the lung interstitium and airways from the pulmonary vasculature. Our findings demonstrate that mesenchymal-derived versican is a key regulator of the early host immune responses to IAV. NEW & NOTEWORTHYWe report the novel finding that mesenchymal-derived versican is critical for neutrophil, monocyte, and dendritic cell migration into the lungs and airways early in influenza A virus infection. Additionally, a differentiated neutrophil-cell line is less adherent to versican-deficient fibroblasts, and versican-deficient fibroblasts have significantly reduced cell-associated hyaluronan (HA) content in vitro. These findings suggest that mesenchymal-derived versican and cell-associated HA are necessary for the adhesion of neutrophils and monocytes to lung fibroblasts.

molecular biology↗

bioRxiv2025

Unveiling Gene Modules at Atlas Scale through Hierarchical Clustering of Single-Cell Data

A major challenge in scRNAseq analysis is how to recover the biologically meaningful cell ontology tree and conserved gene modules across datasets. Data integration and batch-effect correction have been the key to effectively analyze multiple datasets, but often fail to disentangle cell states in heterogeneous samples, such as in cancer and the immune system. Here we present super single cell clustering (SuperSCC), a novel algorithm that utilizes machine-learning models to discover cell identities and gene modules from multiple datasets without the need of data integration. Of note, SuperSCC can be implemented both in cell lineage and cell state level, thereby building the hierarchy of cell programs with specific cell identity and gene modules. Such information has the great potential to identify the shared rare populations across datasets regardless of batch effect and benefits label transfer for mapping cell labels from reference to query. We used SuperSCC to perform atlas level data analysis on more than 90 datasets and build a cell state map of complex tissue in healthy and diseased stages, such as human lung. We show that SuperSCC outperforms existing approaches in identifying cellular context, has better annotation accuracy, and outlines gene modules that indicate conserved immune cell status in lung microenvironments.

bioinformatics↗

bioRxiv2025

Quantitative phosphoproteomic analysis of testes from Iqcn-deficient mice highlights the significance of calmodulin signaling in spermiogenesis

Calmodulin (CaM) plays a crucial role in sperm function. Studies have reported that proteins containing IQ motif interact with CaM, subsequently engaging with downstream target proteins known as calmodulin-binding proteins (CaMBPs). It is reported that the loss of interaction between IQCN and CaM is mainly manifested as decreased motility, leading to fertilization failure and male infertility. However, no relevant reports have been published detailing which CaMBPs exist and the mechanisms by which they are regulated. In this study, we conducted a comprehensive quantitative proteomic and phosphoproteomic analysis of mouse testes from wild-type (WT) and Iqcn knockout (Iqcn-/-) mice. The results indicated that Iqcn deficiency substantially rewires the downstream phosphorylation signaling pathway, while not causing equivalent changes in protein levels. Among the total 577 differentially regulated phosphorylated sites in our results, most of them (494/577) belong to calmodulin-binding proteins (CaMBPs). Gene ontology analysis of these differentially phosphorylated CaMBPs showed enrichment in male gamete generation, actin cytoskeleton organization and microtubule cytoskeleton organization process, demonstrating IQCN regulates sperm function by interacting with CaM, which in turn affects the phosphorylation level of CaMBPs. Further kinase substrate network analysis revealed that most kinases with substrates phosphorylation sites up-regulated were tyrosine kinases, and the inhibition assay showed that FGFR4 and SYK tyrosine kinases are important for sperm motility and progressive motility. In summary, this study reveals the interaction between IQ motif-containing protein IQCN and CaM, which regulates the phosphorylation of downstream CaMBPs and is involved in the related processes of spermiogenesis and sperm function.

developmental biology↗

bioRxiv2025

Reprogrammable 4D Tissue Engineering Hydrogel Scaffold via Reversible Ion Printing

Shape changeable hydrogel scaffolds recapitulating morphological dynamism of native tissues have emerged as an elegant tool for future tissue engineering (TE) applications, due to their capability to create morphodynamical tissues with complex architectures. Hydrogel scaffolds capable of preprogrammable, reprogrammable and/or reversible shape transformations would widely expand the scope of possible temporal shape changes. Current morphable hydrogels are mostly based on multimaterial, multilayered structures, which involve complicated and time-consuming fabrication protocols, and are often limited to single unidirectional deformation. This work reports on the development of a transformable hydrogel system using a fast, simple, and robust fabrication approach for manipulating the shapes of soft tissues at defined maturation states. Simply by using an ion-transfer printing (ITP) technology (i.e., transferring Ca2+ from an ion reservoir with filter paper and subsequent covering on a preformed alginate-derived hydrogel), a tunable Ca2+ crosslinking density gradient across the hydrogel thickness has been incorporated, which enables preprogrammable deformations upon further swelling in cell culture media. Combining with a surface patterning technology, cell-laden constructs (bioconstructs) capable of morphing in multiple directions are deformed into sophisticated configurations. Not only can the deformed bioconstructs recover their original shapes by chemical treatment, but at user-defined times they can also be incorporated with new, different spatially controlled gradient crosslinking via the ITP process, conferring 3D bioconstruct shape reprogrammability. In this manner, unique "3D-to-3D" shape conversions have been realized. Finally, we demonstrated effective shape manipulation in engineered cartilage-like tissue constructs using this strategy. These morphable scaffolds may advance 4D TE by enabling more sophisticated spatiotemporal control over construct shape evolution.

bioengineering↗

bioRxiv2024

Deep generalizable prediction of RNA secondary structure via base pair motif energy

Deep learning methods have demonstrated great performance for RNA secondary structure prediction. However, generalizability is a common unsolved issue on unseen out-of-distribution RNA families, which hinders further improvement of the accuracy and robustness of deep learning methods. Here we construct a base pair motif library that enumerates the complete space of the locally adjacent three-neighbor base pair and records the thermodynamic energy of corresponding base pair motifs through de novo modeling of tertiary structures, and we further develop a deep learning approach for RNA secondary structure prediction, named BPfold, which learns relationship between RNA sequence and the energy map of base pair motif. Experiments on sequence-wise and family-wise datasets have demonstrated the great superiority of BPfold compared to other state-of-the-art approaches in accuracy and generalizability. We hope this work contributes to integrating physical priors and deep learning methods for the further discovery of RNA structures and functionalities.

bioinformatics↗

bioRxiv2024

Building a literature knowledge base towards transparent biomedical AI

As artificial intelligence (AI) continues to advance and scale up in biomedical research, concerns about AIs trustworthiness and transparency have grown. There is a critical need to systematically bring accurate and relevant biomedical knowledge into AI applications for transparency and provenance. Knowledge graphs have emerged as a powerful tool that integrates heterogeneous knowledge by explicitly describing biomedical knowledge as entities and relationships between entities. However, PubMed, the largest and most comprehensive repository of biomedical knowledge, exists primarily as unstructured text and is under utilized for advanced machine learning tasks. To address the challenge, we developed LiteralGraph, a computational framework to extract biomedical terms and relationships from PubMed literature into a unified knowledge graph. Using this framework, we established the Genomic Literature Knowledge Base (GLKB), which consolidates 14,634,427 biomedical relationships between 3,276,336 biomedical terms from over 33 million PubMed abstracts and nine well-established biomedical repositories. The database is coupled with RESTful APIs and a user-friendly web interface that makes it accessible to researchers for various usages. We demonstrated the broad utility of GLKB towards transparent AI in three distinct application scenarios. In the LLM grounding scenario, we developed a Retrieval Augmented Generation (RAG) agent to reduce LLM hallucination in biomedical question answering. In the hypothesis generation scenario, we elucidated the potential functions of RFX6 in type 2 diabetes (T2D) using the vast evidence from PubMed articles. In the machine learning scenario, we utilized GLKB to provide semantic knowledge in predictive tasks and scientific fact-checking.

bioinformatics↗

bioRxiv2024

NR2F2 is required in the embryonic testis for Fetal Leydig Cell development

Male genital development in XY mammalian fetuses is triggered by the action of hormones, including testosterone, secreted by the developing testes. Defects in this process are a cause for Differences in Sex Development (DSD), one of the most common congenital abnormalities in humans. Fetal Leydig Cells (FLC) play a central role for the synthesis of masculinizing hormones in the developing testes. Yet, the genetic cascade controlling their differentiation is poorly understood. Here we investigate the role of the orphan nuclear receptor NR2F2 (COUP-TFII) in FLC development. We report that NR2F2 is expressed in interstitial progenitor cells of the mouse embryonic testes and is downregulated upon their differentiation into FLC. By using two mouse models for conditional mutation of Nr2f2 in the developing testes, we demonstrate that NR2F2 is required for testis morphogenesis and FLC development. NR2F2 acts in interstitial progenitors to regulate the initiation and progression of FLC differentiation. These results establish NR2F2 as an essential regulator of FLC development and steroid hormone synthesis in the mouse fetal testis and provide an entry point to understand the etiology of 46, XY DSD associated with pathogenic NR2F2 variants.

developmental biology↗

bioRxiv2024

Rice OsPLATZ3 controls ROS homeostasis by inhibiting ROS-Scavenging Activity during Tapetum Degeneration

OsPLATZ3, a transcription factor of rice (Oryza sativa) belonging to the PLATZ family, is involved in the repression of transcriptional activity. PLATZ3 exhibits preferential expression in the tapetal cells during anthers development, regulating tapetum degeneration by controlling the dynamics of reactive oxygen species (ROS) in rice male reproductive processes. The Osplatz3 exhibits enlarged tapetum with delayed degeneration, significantly lower ROS levels from S5 to S12 in anthers compared to the wild-type, resulting in complete male sterility. In the anthers of Osplatz3, there are significant changes in the regulators of programmed cell death (PCD)-induced tapetal degradation, such as TDR, EAT1, and their effectors OsCP1, OsC6, OsAP25, and OsAP37. These findings indicate that OsPLATZ3 is a new addition to the tapetum PCD regulatory network. In the late stages of anthers development, Osplatz3 exhibits phenotypes associated with oxidative stress. Transcriptome analysis revealed significant differential expression in 229 genes associated with maintaining ROS homeostasis. Furthermore, Electrophoretic Mobility Shift Assay (EMSA) demonstrate that OsPLATZ3 can bind to the promoter region of OsAPX9, a gene encoding a plant ascorbate peroxidase domain protein with putative ROS scavenging capabilities. These findings suggest that OsPLATZ3 serves as a crucial regulatory factor in tapetum PCD by suppressing ROS scavenging activity.

plant biology↗

bioRxiv2023

SKA2 regulated hyperactive secretory autophagy drives neuroinflammation-induced neurodegeneration

High levels of proinflammatory cytokines induce neurotoxicity and catalyze inflammation-driven neurodegeneration, but the specific release mechanisms from microglia remain elusive. We demonstrate that secretory autophagy (SA), a non-lytic modality of autophagy for secretion of vesicular cargo, regulates neuroinflammation-mediated neurodegeneration via SKA2 and FKBP5 signaling. SKA2 inhibits SA-dependent IL-1{beta} release by counteracting FKBP5 function. Hippocampal Ska2 knockdown in mice hyperactivates SA resulting in neuroinflammation, subsequent neurodegeneration and complete hippocampal atrophy within six weeks. The hyperactivation of SA increases IL-1{beta} release, initiating an inflammatory feed-forward vicious cycle including NLRP3-inflammasome activation and Gasdermin D (GSDMD)-mediated neurotoxicity, which ultimately drives neurodegeneration. Results from protein expression and co-immunoprecipitation analyses of postmortem brains demonstrate that SA is hyperactivated in Alzheimers disease. Overall, our findings suggest that SKA2-regulated, hyperactive SA facilitates neuroinflammation and is linked to Alzheimers disease, providing new mechanistic insight into the biology of neuroinflammation.

neuroscience↗