Biology subjects

He, S.

Publications and source records attributed to He, S..

At least 73 records · Page 4Linked to original sources

bioRxiv2025

Dynamic Retinal Pathology in Glaucoma Progression Revealed by High-Resolution Functional Imaging in Vivo

Glaucoma, a complex and multifactorial neurodegenerative disease, is commonly associated with elevated intraocular pressure and primarily characterized by the progressive loss of retinal ganglion cells (RGCs) and their axons. Despite its prevalence, our understanding and treatment of this disease remain challenging due to the intricate interplay of various pathophysiological factors and the limited capability for in vivo functional study of the disease. In this work, we investigated the dynamic retinal pathology of glaucoma from onset to late stages through longitudinal in vivo high-resolution imaging in a silicone oil-induced ocular hypertension glaucoma mouse model. We developed an optimized adaptive optics two-photon excitation fluorescence microscopy (AO-TPEFM) technique for both morphological and functional assessments of pathological changes in the retina across three distinct glaucoma phenotypes with varying progression rates. Our AO-TPEFM technology visualized the complete process of functional and structural changes in the three most important retinal components related to glaucoma: microvascular vessels, microglia, and RGCs during disease progression. Notably, our functional imaging revealed pathological alterations in microvascular circulation, microglia state, and RGC functionalities at a very early stage of disease development when retinal morphology still appeared normal, providing critical insights into glaucoma pathogenesis. This research also demonstrates that AO-TPEFM is a powerful tool for the in vivo study of general retinal diseases.

bioengineering↗

bioRxiv2025

R27 pilus biogenesis is promiscuous and electrostatic perturbations impair functionality

During conjugation, plasmid DNA is transferred from donor to recipient bacteria via the plasmid-encoded mating pilus, formed as thin helical assemblies of polymerised pilin subunits. In the IncHI1 R27 plasmid-encoded pilus, the TrhA pilin undergoes cyclisation (via a peptide bond between Gly1 and Asp69), essential for conjugation. Gly1 and Asp69 are exposed on the pilus surface and conserved in all TrhA pilins in the Plascad database. Substituting Asp69 with Asn, Ala, Gly, or Arg did not prevent cyclisation or pilus formation, which remained structurally indistinguishable from the wild type. Conjugation efficiency of the Asp69 substitutions across multiple recipient species correlated with side chain size, in the order Asp69Asn > Asp69Ala > Asp69Gly. However, Asp69Arg, as well as Asp69Lys and Gly1Lys substitutions abolished conjugation, likely due to the positively charged pilus surface (opposite to the wild-type negative charge) forming unfavourable electrostatic interactions with the recipient outer membranes inner leaflet, composed solely of zwitterionic phosphatidylethanolamine (PE). Consistently, conjugation was rescued in recipients lacking PE. These findings indicate strong selective pressure to maintain Gly1 and Asp69, as efficient DNA transfer depends on precise electrostatic and steric constraints of the pilus surface.

microbiology↗

bioRxiv2025

Organism-wide spatiotemporal profiling of gene expression utilizing a X-CreERT2/Ai9 tracing system

This study utilizes a novel inducible X-CreERT2/Ai9 system to comprehensively delineate the spatial expression landscape of Fermt2, which encodes a key focal adhesion protein Kindlin-2, across organs in mice. We can confirm all known Kindlin-2-expressing cells and identify any previously unknown Kindlin-2-expressing cells and those that do not express Kindlin-2 in all tissues/organs in mice. We have developed this cost-effective, rapid, and high-fidelity system and implemented its first whole-organism application for comprehensive gene expression profiling. This work not only provides unprecedented insights into roles of Kindlin-2 in diverse physiologies and pathologies, but also establishes the inducible X-CreERT2/Ai9 system as a powerful paradigm-shifting platform for organism-wide pan-organ spatiotemporal expression profiling of any target genes encoding any proteins and micro-RNAs. By enabling high-resolution mapping of drug target expression, this system may lay critical groundwork for elucidating previously unknown adverse and beneficial effects of targeted therapies, offering transformative potential for precision medicine.

biochemistry↗

bioRxiv2025

Genomic Touchstone: Benchmarking Genomic Language Models in the Context of the Central Dogma

The emergence of genomic language models (gLMs) has revolutionized the analysis of genomic sequences, enabling robust capture of biologically meaningful patterns from DNA sequences for an improved understanding of human genome-wide regulatory programs, variant pathogenicity and therapeutic discovery. Given that DNA serves as the foundational blueprint within the central dogma, the ultimate evaluation of a gLM is its ability to generalize across this entire biological cascade. However, existing evaluations lack this holistic and consistent framework, leaving researchers uncertain about which model best translates sequence understanding into downstream biological prediction. Here we present Genomic Touchstone, a comprehensive benchmark designed to evaluate gLMs across 36 diverse tasks and 88 datasets structured along the central dogmas modalities of DNA, RNA, and protein, encompassing 5.34 billion base pairs of genomic sequences. We evaluate 34 representative models encompassing transformers and convolutional neural networks, as well as emerging efficient architectures such as Hyena and Mamba. Our analysis yield four key insights. First, gLMs achieve comparable or superior performance on RNA and protein tasks compared to models pretrained on these molecules. Second, transformer-based models continue to lead in overall performance, yet efficient sequence models show promising task-specific capabilities and deserve further exploration. Third, the scaling behavior of gLMs remains incompletely understood. While longer input sequences and more diverse pretraining data consistently improve performance, increases in model size do not always translate into better results. Fourth, pretraining strategies, including the choice of training objectives and the composition of pretraining corpora, exert substantial influence on downstream generalization across different genomic contexts. Genomic Touchstone establishes a unified evaluation framework tailored to human genomics. By spanning multiple molecular modalities and biological tasks, it offers a valuable foundation for guiding future gLM design and understanding model generalizability in complex biological contexts.

bioinformatics↗

bioRxiv2025

Mitochondrial Genome Variants and Nuclear Mitochondrial DNA Segments in 7331 Individuals from NyuWa and 1KGP

Dysfunctional mitochondria are implicated in various diseases, while little is known about comprehensive characterization of mitochondrial DNA (mtDNA) in Chinese population. Here, we conducted a systematic analysis of mtDNA from 7331 samples, comprising 4129 Chinese samples (NyuWa) and 3202 samples from the 1000 Genomes Project (1KGP). We identified 7216 distinct high-quality mtDNA variants and classified them into 22 macrohaplogroups, and detected 1466 distinct nuclear mitochondrial DNA segments (NUMTs), with 88 mtDNA variants and 642 NUMTs being specific to NyuWa. The genome-wide association analyses revealed that 12 mtDNA variants were significantly correlated with 199 nuclear DNA variants. Our findings revealed that all individuals in both NyuWa and 1KGP harbored common NUMTs, and one-fifth possessed ultra-rare NUMTs, which tended to insert into nuclear gene regions. Compared to 1KGP, significant enrichment of nuclear breakpoints in long interspersed nuclear elements (LINEs) was observed for rare NUMTs in NyuWa. Overall, this study represents the first comprehensive profiling the landscape of Chinese NUMTs and offers the most extensive resource of Chinese mtDNA variants and NUMTs based on high-depth WGS to date, providing valuable reference resources for genetic research on mtDNA-related diseases.

genomics↗

bioRxiv2025

Clinical breakthrough: Biomimetic Scaffolds Based on Soluble Collagen for Regenerating Large Volumes of Highly Functional Tissues

Functional tissue regeneration in tissue engineering has long been an aspiration of humanity. However, conventional materials such as metal, ceramic, polymer, and decellularized extracellular matrix (dECM) have not yet yielded the desired outcomes. We have developed a collagen biomimetic material based on soluble collagen. This biomimetic material emulates the composition, structure, and function of native tissue. It possesses high strength, high affinity, high nutrient, and low antigenicity. The biomimetic material had been utilized to fabricate some implants (absorbable surgical suture, artificial tendon, hernia patch) and had demonstrated remarkable functional regeneration in animal experiments. In our previous researches, collagen surgical sutures facilitated wound healing, artificial tendons and artificial muscles (hernia patches) successfully repaired extensive defects of tendons and muscles, respectively. These successful tissue engineering endeavors enabled damaged defect to restore its biological functions and structure. Degradation of these biomimetic implants synchronized with the regeneration of the new tissues. Based on these successful functional regenerations, the present study focuses on regeneration of rabbit sciatic nerve. This collagen biomimetic material was employed to construct an artificial nerve tube (length: 2 cm, inner diameter: 2 mm, outer diameter: 3.5 mm) to repair a 2 cm-long defect in the rabbit sciatic nerve. After 36 weeks, the newly formed functional sciatic nerve successfully bridged the damaged ends, while approximately 92% of the artificial nerve tube has degraded. In this study, we provide an overview of the biomimetic material and its medical applications. We believe that these successful regenerations of tendon, muscle, and nerve all stem from the exceptional characteristics of this biomimetic material. We firmly believe that this collagen biomimetic material exhibits remarkable regenerative capabilities in diverse tissues. It can pave the way for the fabrication of increasingly critical artificial implants, which can be utilized to repair defects in human tissues, including artificial esophagi, tracheae, and small blood vessels. In regenerative medicine of induced pluripotent stem cells (iPS), the biomimetic materials can replace temperature-sensitive materials (PIPAAm) to fabricate cell sheets. This approach expedites the process, enhances cellular vitality, and consequently improves regeneration outcomes. This material technology not only provides a revolutionary solution for clinical treatment but also paves the way for a new development direction in regenerative medicine.

bioengineering↗

bioRxiv2025

MYC-induced oncogenesis is dependent on acidic patches within its N-terminal intrinsically disordered domain.

MYC is one of the most enticing therapeutic targets for cancer but clinical-grade inhibitors are still lacking. By site-saturation mutagenesis screening, we identified several evolutionarily conserved acidic patches within the intrinsically-disordered MYC N-terminus that were confirmed to be functionally essential in different cell models and in vivo. Beyond modulating MYCs global transcriptional activity, these negatively charged patches regulate the interaction with chromatin-modifying complexes including those with histone acetyl-transferase activity. One of the key interactions is established with the co-factor TRRAP, a subunit shared between several Histone Acetyl-Transferase complexes. The protein-protein binding between MYC and TRRAP predominantly relies on two of the N-terminal negative clusters that are located outside MYC-Box-II (MBII) and drive oncogenesis. Our work identifies a new multivalent MYC subdomain that presents new therapeutic vulnerabilities providing invaluable insights for the development of new therapeutic approaches.

cancer biology↗

bioRxiv2025

Multitargeted Reduction of Inflammation and Atherosclerosis in Tet2-deficient CHIP via XPO1 Inhibition and Atf3 restoration

TET2 is the second most frequently mutated gene in clonal hematopoiesis of indeterminate potential (CHIP), driving hematopoietic stem cell clonal expansion and increasing the risk of myeloid malignancies. Affected individuals often develop atherosclerotic cardiovascular disease, exacerbated by hyperinflammatory TET2-mutant macrophages. Here, we show that the XPO1 nuclear export inhibitor eltanexor significantly reduces atherosclerotic plaque formation in a mouse model of Tet2-mutant CHIP. In addition, we investigated the mechanisms and gene expression pathways that underlie the proinflammatory phenotype that characterizes Tet2-mutant CHIP. Single-cell CITE-seq identified increased expression of multiple proinflammatory mediators in Tet2-mutant macrophages and in non-hematopoietic cells of the aortic wall, which was reduced by eltanexor treatment. Atf3, which encodes a core transcriptional modulator of inflammation, occupies and regulates the largest enhancer in wild-type macrophages. Tet2 loss diminished ATF3 binding to the regulatory loci of inflammatory mediators, which was restored upon XPO1 inhibition. These results provide new insights into drivers of heightened inflammation in TET2-mutant CHIP and highlight a novel therapeutic strategy for intervention.

immunology↗

bioRxiv2025

Efficient Cell Factory Design by Combining Meta-Heuristic Algorithm with Enzyme Constrained Metabolic Models

The rational design of high-performance microbial cell factories remains a central challenge in sustainable biomanufacturing due to the complexity of metabolic networks and the difficulty of predicting synergistic genetic interventions. Despite recent advances in strain design algorithms, predicting combinatorial targets remains computationally prohibitive due to the combinatorial explosion. Here, we present MetaStrain, a unified computational framework that integrates enzyme-constrained models (ecModels) with meta-heuristic algorithms to identify non-intuitive combinatorial gene targets for improving product yields. MetaStrain first performs pre-screening through a modified enforced objective flux scanning algorithm to reduce the dimensionality of candidate genes and annotate editing strategies. The subsequent search module employs flexible individual encodings compatible with diverse meta-heuristic algorithms and adopts a bottom-up phenotype evaluation strategy enabling efficient exploration of the combinatorial design space. Integrated redundancy analysis tools further identify single and fixed-size combinatorial strategies, facilitating direct experimental implementation. Computational simulation in Saccharomyces cerevisiae reveals significant enhancements in 2-phenylethanol and spermidine biosynthesis, while controlling target count and covering experimentally validated targets. Experimental validation in Escherichia coli further confirmed the algorithms predictive power, achieving up to a 61.25% increase in L-tryptophan titer of the five-target combination strain. Overall, MetaStrain achieves high computational efficiency, stable convergence, and broad adaptability across diverse metabolic targets. This study provides a powerful tool for metabolic engineering, bridging computational prediction and experimental realization, and highlighting the potential of meta-heuristic optimization in synthetic biology.

bioinformatics↗

bioRxiv2025

Spatial multi-omics and deep learning reveal fingerprints of immunotherapy response and resistance in hepatocellular carcinoma

Despite advances in immunotherapy treatment, nonresponse rates remain high, and mechanisms of resistance to checkpoint inhibition remain unclear. To address this gap, we performed spatial transcriptomic and proteomic profiling on human hepatocellular carcinoma tissues collected before and after immunotherapy. We developed an interpretable, multimodal deep learning framework to extract key cellular and molecular signatures from these data. Our graph neural network approach based on spatial proteomic inputs achieved outstanding performance (ROC-AUC > 0.9) in predicting patient treatment response. Key predictive features and associated spatial transcriptomic profiles revealed the multi-omic landscape of immunotherapy response and resistance. One such feature was an interface niche expressing restrictive extracellular matrix factors that physically separates tumor tissue and lymphoid aggregates in nonresponders. We integrate this and other spatially-resolved signatures into SPARC, a multi-omic "fingerprint" comprising scores for immunotherapy response and resistance mechanisms. This study lays groundwork for future patient stratification and treatment strategies in cancer immunotherapy.

cancer biology↗

bioRxiv2025

DEVELOPMENT OF A POTENT MONOCLONAL ANTIBODY FOR TREATMENT OF HUMAN METAPNEUMOVIRUS INFECTIONS

Human metapneumovirus (HMPV) is a major cause of respiratory infections, particularly among vulnerable populations, yet effective therapeutics remain unavailable. Monoclonal antibodies (mAbs) offer a promising approach for both treatment and prevention. Here, we describe the discovery and characterization of 4F11, a highly potent and broadly neutralizing mAb with demonstrated in vitro and in vivo efficacy against HMPV. Using cryo-electron microscopy, we defined a unique mechanism of binding HMPV employed by 4F11, which distinguishes it from previously characterized RSV and HMPV mAbs. 4F11 targets an epitope located at the apex of the prefusion F protein (site O) with a 1:1 stoichiometry, distinct from the 3:1 stoichiometry observed with other HMPV site O antibodies. Unlike other site O antibodies, which penetrate the glycan shield between Asn57 and Asn172, 4F11 binds vertically and directly interacts with the Asn172 glycan, representing a unique glycan-dependent mode of recognition. In vitro, 4F11 displayed high potency and broad neutralization across diverse HMPV strains. It also showed a low propensity for resistance development, with only a single escape mutation (K179E) identified, a mutation not found in any published HMPV sequence to date. Viruses rescued with the K179E escape mutation had significantly decreased fitness in vitro compared to wild-type virus. In a hamster challenge model, 4F11 significantly reduced viral loads in both the lungs and nasal turbinates. These findings highlight 4F11 as a promising candidate for therapeutic development, particularly for immunocompromised individuals and other high-risk groups.

immunology↗

bioRxiv2025

Breakthrough in Functional Muscle Regeneration: Biomimetic Collagen Scaffold Restores Abdominal Wall Defects in a VML-rabbit Model

Functional regeneration of volumetric muscle loss (VML) remains a significant challenge in the field of regenerative medicine. We have developed a novel biomimetic scaffold (0.4 mm thick hernia patch) derived from soluble collagen. This biomimetic scaffold exhibits exceptional strength, low immunogenicity, and structurally mimics the natural muscle fiber arrangement. The scaffold is utilized to repair a VML rabbit model (30 x 30 mm defect in abdominal wall) using nylon sutures for connection. It was observed that the VML site was gradually covered by regenerated tissue, which consisted of 2-3 mm thick functional muscle. By 32 weeks post-surgery, the newly formed muscle tissue had covered the majority of the VML area, as evidenced by morphological observations and histological evaluations. Approximately 24 weeks after surgery, the scaffold underwent complete degradation. This degradation exhibited a strong correlation with muscle regeneration. Furthermore, it was observed that the nylon sutures gradually migrated towards the VML center as muscle regeneration progressed, and nylon sutures exhibited an adverse impact on muscle regeneration.. The study did not employ exogenous cells or growth factors. However, the collagen scaffold effectively stimulated endogenous muscle regeneration. In contrast to the previous limited partial recovery or fibrotic scar healing, the collagen scaffold successfully induced genuine structural and functional muscle regeneration. The 0.4 mm thick scaffold (hernia patch) exhibits a different structure from the 2-3 mm thick natural abdominal muscle wall. This suggests that a simple scaffold can regenerate functional tissue with a complex structure. This research represents the first successful functional regeneration of VML in clinically relevant animal models through the utilization of artificial materials. This technology possesses transformative potential in the treatment of muscle defects arising from trauma, tumor resection, hernia, or congenital anomalies. Furthermore, it holds substantial medical potentials in addressing neuromuscular diseases, such as Duchenne muscular dystrophy (DMD), amyotrophic lateral sclerosis (ALS), and spinal muscular atrophy (SMA), by collaborating with advanced technologies including CRISPR-Cas9, adeno-associated virus (AAV) gene therapy, and stem cell technology.

bioengineering↗

bioRxiv2025

Contrastive Learning-Enhanced Drug Metabolite Prediction Drives the Discovery of an Orally Bioavailable RSK4 Inhibitor for Esophageal Cancer

Drug metabolism liabilities are a major bottleneck in drug development, necessitating accurate metabolite prediction tools. While AI-driven methods have advanced, challenges remain in achieving high-precision, robustness, broad reaction coverage and seamless integration into the drug development workflow. To address this, we first constructed the most comprehensive human-specific drug metabolism database to date, encompassing 11,665 human metabolic reactions and 2,497 high-quality rules. Leveraging this resource, we developed ConMeter (Contrastive learning and feature interaction based Metabolic Reaction prediction), a novel model integrating chemical feature interaction and contrastive learning. ConMeter demonstrates significant improvements, achieving [~]80% accuracy in identifying at least one metabolite within the top 5 predictions and showing an average 25% relative enhancement in ranking true positives compared to state-of-the-art methods. The comprehensive database also expands the range of predictable metabolic reactions. To demonstrate the utility of ConMeter, we integrated it into a metabolite prediction-guided drug design strategy aimed at addressing the low oral bioavailability of a metabolically vulnerable RSK4 lead compound -- a pervasive issue among known RSK inhibitors. This strategy successfully yielded R636, a novel RSK4 inhibitor exhibiting remarkably improved bioavailability (a 63-fold increase, reaching 63% absolute bioavailability) while maintaining its potent activity. Notably, R636 showed favorable safety profile and significant anti-tumor efficacy against esophageal squamous cell carcinoma (ESCC) in vitro and in two distinct patient-derived xenograft (PDX) mouse models. This work introduces ConMeter as a powerful and practical tool for high-accuracy metabolite prediction, demonstrating its potential to accelerate lead optimization, exemplified by the discovery of R636 as a promising orally bioavailable candidate for ESCC.

cancer biology↗

bioRxiv2025

Breaking the Tendon Repair Barrier: A Biomimetic Soluble Collagen Scaffold Enables Full Functional Regeneration of Achilles Tendon in Rabbits

This groundbreaking research employs biomimetic artificial tendons for the first time globally to successfully regenerate a 2 cm defect in the rabbit Achilles tendon. The artificial tendon, woven by collagen threads derived from soluble collagen, is a biomimetic tendon with a composition, structure and function akin to natural tendons. The biomimetic tendon exhibits exceptional tensile strength of (43.4 MPa), surpassing that of the natural Achilles tendon (39.1MPa). The remarkable mechanical properties, coupled with its favorable biocompatibility and low immunogenicity, underscore the substantial clinical application potential of this biomimetic material. A biomimetic tendon was implanted to repair a rabbit model of a 2 cm tendon defect without employing exogenous cells or growth factors. Twenty weeks after transplantation, the mechanical properties of the regenerated tendon were restored to approximately 80.1% of those of the natural Achilles tendon. This biomimetic tendon not only provides mechanical support during the initial phase of cell recruitment but also sustains strength to facilitate functional tendon reconstruction until material degradation and complete tendon regeneration. The regenerated tendon tissue exhibited regularly arranged dense collagen fibers along the longitudinal axis of the natural tendons, demonstrating high mechanical strength and seamless integration. Histological analysis further revealed a progressive enhancement in collagen fiber diameter, density, and structural integrity within regenerated tendons over time, closely resembling the characteristics of natural Achilles tendons eventually. Notably, there were no apparent immune reactions or inflammatory responses during the experimental cycle. These findings unequivocally demonstrate the remarkable efficacy of biomimetic tendons in facilitating functional tissue regeneration, underscoring their significant medical and scientific value. This study not only presents a novel treatment approach for tendon injury repair but also establishes the foundation for future clinical tendon and ligament regeneration and repair, with extensive clinical application prospects.

bioengineering↗

bioRxiv2025

Bona fide hematopoietic stem cells in zebrafish originate from the supra-intestinal artery

Hematopoietic stem cells (HSCs) in vertebrates are widely believed to originate from hemogenic endothelial cells (HECs) in the dorsal aorta (DA) through endothelial-to-hematopoietic transition. However, conclusive in vivo DA-specific lineage-tracing studies to confirm this origin are lacking. Here, leveraging infrared light-mediated temporospatial-resolution lineage-tracing technology, we demonstrate that in zebrafish, HSCs do not originate from the DA. Instead, following the DA hematopoiesis, which produces hematopoietic precursors with limited renewal potential, lifelong renewable HSCs emerge from the supra-intestinal artery (SIA). Remarkably, while the DA produces hematopoietic cells from unipotent HECs, the SIA generates hematopoietic cells from both unipotent HECs and bipotent vessel-resident hemangioblasts (VRHs). These VRHs undergo multiple rounds of division and budding, generating both hematopoietic cells and SIA-resident endothelial cells. The distinct characteristics of DA and SIA cells are further reinforced by single-cell transcriptomic profiling. Our findings challenge the current paradigm of HSC development, underscoring the complexity of vertebrate hematopoiesis.

developmental biology↗

bioRxiv2025

Assessment of nucleic acid structure prediction in CASP16

Consistently accurate 3D nucleic acid structure prediction would facilitate studies of the diverse RNA and DNA molecules underlying life. In CASP16, blind predictions for 42 targets canvassing a full array of nucleic acid functions, from dopamine binding by DNA to formation of elaborate RNA nanocages, were submitted by 65 groups from 46 different labs worldwide. In contrast to concurrent protein structure predictions, performance on nucleic acids was generally poor, with no predictions of previously unseen natural RNA structures achieving TM-scores above 0.8. Even though automated server performance has improved, all top-performing groups were human expert predictors: Vfold, GuangzhouRNA-human, and KiharaLab. Good performance on one template-free modeling target (OLE RNA) and accurate global secondary structure prediction suggested that structural information can be extracted from multiple sequence alignments. However, 3D accuracy generally appeared to depend on the availability of closely related 3D structure templates, and predictions still did not achieve consistent recovery of pseudoknots, singlet Watson-Crick-Franklin pairs, non-canonical pairs, or tertiary motifs like A-minor interactions. For the first time, blind predictions of nucleic acid interactions with small molecules, proteins, and other nucleic acids could be assessed in CASP16. As with nucleic acid monomers, prediction accuracy for nucleic acid complexes was generally poor unless 3D templates were available. Accounting for template availability, there has not been a notable increase in nucleic acid modeling accuracy between previous blind challenges and CASP16.

biophysics↗

bioRxiv2025

Dimethyl fumarate reprograms cervical cancer cells to enhance antitumor immunity by activating mtDNA-cGAS-STING pathway

BackgroundCervical cancer (CC) remains a significant global health challenge for women, especially in advanced stages where effective treatments are limited. Current immunotherapies, including PD-1/PD-L1 blockades and adoptive T cell therapies, show limited response rates and durability. Dimethyl fumarate (DMF), an FDA-approved drug for autoimmune diseases, has demonstrated that it might exhibit direct antitumor activity in several cancers. However, its influence on anti-tumor immunity and its function in CC remain poorly understood. This study aims to investigate the therapeutic potential of DMF in CC models and elucidate its underlying mechanisms of action. MethodsCC cell lines and mouse models were treated with DMF. Transcriptional level changes in cervical cancer cells following DMF treatment were analyzed by RNA-seq. Mitochondrial DNA (mtDNA) release, and cGAS-STING activation were assessed via qPCR, immunofluorescence, immunoblotting and ELISA. CD8+ T cell recruitment was analyzed by flow cytometry. Combinatorial therapies (DMF + anti-PD-1/TILs) were tested in syngeneic or patient-derived xenografts (PDX) models. ResultsDMF treatment induces mitochondrial dysfunction in tumor cells, resulting in the release of mtDNA into the cytosol. The cytosolic mtDNA activates the cGAS-STING-TBK1 pathway and type I interferon response, leading to the secretion of CCL5 and CXCL10, thereby enhancing CD8 T cell infiltration. Additionally, DMF exhibits synergistic effect with PD-1 blockade in murine CC model, and can enhance the therapeutic efficacy of adoptively transferred T cells toward CC patient-derived xenografts model. ConclusionThis work elucidated that DMF reprograms CC cells to activate the mtDNA-cGAS-STING pathway, fostering a chemokine-rich microenvironment that recruits CD8+ T cells. Its synergy with PD-1 blockade or TIL therapy underscores its potential as an immunomodulatory adjuvant. These findings suggest that DMF holds promise as a novel immunotherapeutic strategy for improving clinical outcomes in CC.

immunology↗

bioRxiv2025

Scaling Large Language Models for Next-Generation Single-Cell Analysis

AO_SCPLOWBSTRACTC_SCPLOWSingle-cell RNA sequencing has transformed our understanding of cellular diversity, yet current single-cell foundation models (scFMs) remain limited in their scalability, flexibility across diverse tasks, and ability to natively integrate textual information. In this work, we build upon the Cell2Sentence (C2S) framework, which represents scRNA-seq profiles as textual "cell sentences," to train Large Language Models (LLMs) on a corpus comprising over one billion tokens of transcriptomic data, biological text, and metadata. Scaling the model to 27 billion parameters yields consistent improvements in predictive and generative capabilities and supports advanced downstream tasks that require synthesis of information across multi-cellular contexts. Targeted fine-tuning with modern reinforcement learning techniques produces strong performance in perturbation response prediction, natural language interpretation, and complex biological reasoning. This predictive strength enabled a dual-context virtual screen that nominated the kinase inhibitor silmitasertib (CX-4945) as a candidate for context-selective upregulation of antigen presentation. Experimental assessment in human cell models unseen during training supported this prediction, demonstrating that C2S-Scale can effectively guide the discovery of context-conditioned biology. C2S-Scale unifies transcriptomic and textual data at unprecedented scales, surpassing both specialized single-cell models and general-purpose LLMs to provide a platform for next-generation single-cell analysis and the development of "virtual cells."

bioinformatics↗