Biology subjects

Qiu, Y.

Publications and source records attributed to Qiu, Y..

At least 37 records · Page 2Linked to original sources

bioRxiv2026

Diffusion models learn underlying trends in actomyosin networks and predict behavior at unseen filament turnover

Generative diffusion models have demonstrated an ability to produce novel images sampled from the learned underlying data distribution. These models are able to infer system characteristics for parameter combinations that were not seen during training. We investigate the ability of these models to infer trends in biological data from limited samples. Specifically, we consider the response of system scale behaviors such as cortical flow in a simulated actomyosin system as we tune filament turnover rates. We train a diffusion model on coarse grained actin curvature and density heatmap images, and are able to generate images from conditioning variables not seen during training. These images are predictive of nonlinear trends in the system. We also consider characteristics of the system that allows this level of inference, such as the strong linear relationship between average density and filament turnover in the system, and explore minimal underlying dynamics with a motor binding model.

biophysics↗

bioRxiv2026

Inhibition benefits neural system identification

Neural system identification approaches use empirical data to fit the stimulus-response functions of neurons. Augmented by deep neural networks, such models have achieved high predictive performance and allow to perform in silico experiments to test hypotheses. Yet, many of these methods ignore common features in visual systems, such as inhibitory interactions between neurons, which are essential for nonlinear neural computation. Here, we incorporate inhibition as an inductive bias into a deep model for neural prediction and investigate the influence of inhibition on the learned transfer functions. To this end, we employ difference-of-Gaussian (subtraction) and within-channel divisive normalization (division), which have been proposed to relate inhibition to neural processing, in deep networks for predicting visual responses. We observe that incorporating such operations maintains the predictive performance and encourages the learning of biologically plausible kernels reminiscent of neural representation in early vision. Additionally, our in silico experiments demonstrate that implementing either sub-tractive or divisive operation benefits the learning of surround suppression but not cross-orientation inhibition. Interestingly, while division increases the sparsity of activation and reduces the sparsity of weights, subtraction has the reverse effect.

neuroscience↗

bioRxiv2026

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

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

bioinformatics↗

bioRxiv2026

Polarization-engineered aberration-resilient light sheet microscopy

Light sheet fluorescence microscopy enables volumetric imaging with high imaging speed, optical sectioning capability, and reduced photobleaching and phototoxicity, and has become a workhorse in bioimaging. However, widely adopted Gaussian light sheets face an inherent trade-off between axial resolution and field-of-view due to diffraction. State-of-the-art nondiffracting light sheets--including Bessel beam, Airy beam, and lattice light sheet--alleviate this trade-off but suffer from optical aberrations that compromise performance with increasing imaging depth. While the integration of adaptive optics offers a promising solution, such integrated systems are typically complex, expensive, and slow due to the need for serial mapping and correction of spatially varying aberrations across the specimen. Here, we present polarization-engineered aberration-resilient light sheet (PEARLS), a new class of monochromatic nondiffracting light sheet with temporally invariant profile and robustness to optical aberrations. In comparison with existing light sheets, PEARLS showed significantly reduced photobleaching and enhanced aberration-resilience, permitting imaging of three-dimensional subcellular dynamics in optically complex environments. We applied PEARLS for noninvasive observations of biological dynamics in various living systems, revealing phenotypic diversity across spatial and temporal scales--from rapid membrane dynamics and organelle interactions in cultured cells to coordinated mitosis and cell migrations in developing embryos.

cell biology↗

bioRxiv2026

Detecting and quantifying overparametrization in RNA language models with REDIAL

While RNA language models (LMs) have served as foundation models (FMs) to advanced structural prediction, their evaluation relies heavily on supervised downstream tasks. Such tasks can often mask FM inefficiencies and reflect downstream training set memorization. To address this, here we introduce REDIAL (RNA Embedding perturbation Diagnostics for Language models), a zero-shot, unsupervised framework designed to extract coevolutionary signals directly from the high-dimensional latent spaces of RNA language models. By applying REDIAL, we uncover stark, layer-wise disparities in how popular RNA language models (LMs) internalize structural constraints through a layer-wise dissection and ablation study. Our results showed how such layerwise behavior deviates from protein LMs and is related to design flaws in the architectures. Specifically, we show that current RNA LMs are severely overparameterized relative to the limited sequence diversity of available RNA databases, leading to profound parameter inefficiency and overfitting. Furthermore, we establish that structure-guided pretraining fundamentally improves the signal-to-noise ratio of learned coevolutionary couplings compared to sequence-only baselines. Ultimately, this unsupervised evaluation paradigm exposes critical flaws in current parameter scaling strategies and provides a rigorous diagnostic benchmark to guide the development of more efficient, generalizable foundation models for RNA therapeutics and de novo design.

bioinformatics↗

bioRxiv2026

SNAC-DB: An ML-Ready Database for Antibody and NANOBODY(R) VHH-Antigen Complexes with Expanded Structural Diversity and Real-World Benchmarking

Predicting antibody and NANOBODY(R) VHH-antigen complexes remains a critical challenge for state-of-the-art structure prediction models, limiting their impact in therapeutic discovery pipelines. We introduce SNAC-DB, an ML-ready database and curation pipeline enriched with structural biology expertise, designed to accelerate model accuracy and generalization by providing 31-37% expanded structural diversity over existing resources like SAbDab through comprehensive re-curation that extracts maximum value from available experimental structures. SNAC-DB expands coverage by capturing often-overlooked complexes and accurately identifying complete multi-chain epitopes through improved biological-assembly-based logic. Built for ML practitioners, SNAC-DB provides standardized formats with multi-threshold structure-based clustering to enable principled sample weighting during training. Using a rigorous benchmark of public PDB entries deposited post-May 2024 plus confidential therapeutic structures, we evaluate seven leading models (Protenix-v1, OpenFold-3p2, RosettaFold-3, Boltz-2, Boltz-1x, Chai-1, and AlphaFold2.3-multimer) with evaluation methodology tailored to antibody/NAN-OBODY(R) VHH-antigen complexes to ensure correct handling of multi-chain epitopes, revealing systematic performance gaps: success rates rarely exceed 25%, confidence-based ranking fails to identify best predictions even when accurate structures exist in ensembles, and all models consistently struggle with therapeutically relevant NANOBODY(R) VHHs. Systematic evaluation of sampling strategies demonstrates that while generating 1000 samples per target substantially increases the likelihood of producing accurate structures (oracle selection improves from 11.9% to 50.5%), confidence-based ranking remains nearly flat (between 10.9% and 14.9%), revealing that improved ranking mechanisms represent a more tractable path to performance gains. Finally, fine-tuning GeoDock on SNAC-DB yields higher success rates than training on SAbDab (11.0% vs. 7.1% for antibodies; 7.0% vs. 4.0% for NANOBODY(R) VHHs), suggesting that SNAC-DBs expanded structural diversity translates to improved model generalization. Significance StatementComputational antibody/NANOBODY(R) VHH design shows promise but remains unreliable for therapeutic development. SNAC-DB provides 31-37% expanded structural diversity through comprehensive data curation, immediately accelerating model development. Benchmarking seven leading AI models reveals accuracy rarely exceeds 25% on therapeutic targets, with confidence-based ranking failing to identify correct structures even when they exist in model outputs. Training on SNAC-DB increases prediction accuracy, validating that high-quality, diverse training data is critical for advancing computational methods toward clinical impact.

bioinformatics↗

bioRxiv2026

The functional organization of retinal input to the mouse superior colliculus

The superior colliculus integrates retinal input to drive rapid, adaptive visual behavior, yet how the functional diversity of retinal ganglion cell types is represented in superior colliculus remains poorly understood. Using chronic two-photon calcium imaging of retinal ganglion cell axonal boutons in awake mice, we recorded over 200,000 boutons across superficial superior colliculus layers -- a scale that enabled systematic comparison with large-scale ex vivo retinal datasets. This revealed that the superior colliculus receives a near-complete sampling of retinal ganglion cell functional diversity. Functionally distinct response types were organized in systematic laminar gradients: not only response properties such as direction selectivity and contrast suppression, but retinal response types themselves varied systematically with depth. To probe how this organized input encodes natural scenes, we trained a "digital twin" deep network model on natural movie responses and validated its generalization to parametric stimuli, including cell type identification. Leveraging this model to generate predicted responses to looming stimuli, we identify a discrete subset of retinal response types tuned for collision detection at low angular thresholds -- a specialization embedded within a broader, non-specialized retinal population. The digital twin is made publicly available as a community resource. Together, these findings provide a comprehensive functional map of retinal drive to the superior colliculus and an in silico platform for linking retinal cell types to behaviorally relevant superior colliculus computations.

neuroscience↗

bioRxiv2026

π-MSNet: A billion-scale, AI-ready living proteomics data portal

Artificial intelligence (AI) is reshaping proteomics workflows, delivering remarkable gains in both peptide identification sensitivity and quantitative performance. However, the potential of deep learning models in proteomics has not been fully exploited due to the scarcity of large-scale, high-quality and consistently labeled datasets. Here, we present {pi}-MSNet, a billion-scale, AI-ready living mass spectrometry (MS) data portal. Using a uniform identification and quality control workflow, it comprises over 1.66 billion MS/MS spectra, 501 million peptide-spectrum matches (PSMs), and 9 million precursors from 36,356 LC-MS/MS runs across ten instrument types and 55 diverse species. Through community collaboration, the data are shared via international, interactive, and living web resources. Enabled by the built-in MSNetLoader Python API for seamless and scalable data access--with native support for PyTorch and TensorFlow--{pi}-MSNet provides an AI-ready data framework for efficient training and systematic benchmarking of multiple models across three representative tasks (e.g., MS/MS spectrum prediction, retention time prediction, and de novo peptide sequencing). In particular, by retraining multiple models on {pi}-MSNet, we achieved consistent performance improvements over their original versions. These improved models were subsequently integrated into the {pi}-MSNet agent to enable interactive, deployment-free use. Through SDRF (Sample and Data Relationship Format) metadata, an open-source cloud analysis workflow, and a community-driven interactive data portal that supports continuous data submission, {pi}-MSNet serves as a living, AI-ready resource for reproducible benchmarking, robust model training, and accelerated AI innovation in proteomics.

bioinformatics↗

bioRxiv2026

Exploration of Orally Disintegrating Tablet for Sublingual Vaccination against Mucosal Bacterial Infection

Oral mucosal bacterial infections impose a substantial global disease burden, yet current clinical management typically reduces microbial load only transiently and rarely establishes durable protection at the oral surface. Analysis of 200 patients with periodontitis revealed that elevated levels of pathogen-specific salivary secretory Immunoglobulin A (sIgA) were strongly associated with reduced bacterial burden and improved clinical periodontal outcomes, identifying sIgA as a key determinant of effective oral protection. Guided by this observation, we developed a sublingual, orally disintegrating tablet vaccine (Capot) that incorporates bacterial extracellular vesicles providing a comprehensive repertoire of native antigens and multiple pathogen-associated molecular patterns, encapsulated within a calcium phosphate nanoshell to enable safe transmucosal delivery to submandibular lymph nodes. The rapidly disintegrating tablet format minimizes inadvertent swallowing and enhances local mucosal bioavailability. In mice and non-human primates, Capot induced robust and long-lasting salivary sIgA responses without overt oral mucosal or gastrointestinal inflammation and conferred strong protection against primary, recurrent, and antibiotic-resistant periodontitis. Together, these findings establish sublingual tablet vaccination as a practical strategy for selectively engaging oral mucosal immunity and preventing chronic bacterial diseases at oral mucosal surfaces.

bioengineering↗

bioRxiv2026

A novel subset of hepatocytes is simultaneously gluconeogenic and de novo lipogenic in the fed state and is naturally insulin resistant

It is generally accepted that hepatic gluconeogenesis, the synthesis of glucose from non-carbohydrate substrates is active in the fasted state and inactive in the fed state. In contrast, de novo lipogenesis is active in the fed state and is inactive in the fasted state. Here, we used targeted single cell RNA-seq, HCR RNA-FISH, and PrimeFlow in normal physiological mouse liver, and identified a subpopulation of periportal hepatocytes that simultaneously co-express both gluconeogenic and lipogenic genes in the fed state. Euglycemic-hyperinsulinemic clamps further demonstrated that this novel hepatocyte subpopulation is naturally insulin resistant. Spatial metabolic imaging coupled with stable isotope tracing analyses revealed individual hepatocytes that simultaneously undergo both gluconeogenesis and de novo lipogenesis. These dual-positive hepatocytes were also present in human hepatocytes from humanized mouse livers. Moreover, the number of dual-positive hepatocytes increased in high-fat diet-fed mice, suggesting a paradigm shift in our understanding of how the liver becomes insulin resistant.

molecular biology↗

bioRxiv2026

Multi-Omic Analyses of Dietary Fatty Acid-Microbe-Host Interactions Reveal Metaorganismal Lipid Metabolic Crosstalk Impacting Cardiometabolic Disease

Following a meal, our gut microbiome and human cells collaborate via metaorganismal metabolic circuits to produce diverse nutrient metabolites that systemically circulate to influence health and disease. Although there are now several examples of bacterial fiber-, amino acid-, and micronutrient-derived metabolites impacting cardiometabolic disease, very little is known in regards to how diet-microbe-host interactions impact lipid homeostasis. Here we address this by defining dietary fatty acid substrate availability in germ-free versus conventionally-raised mice coupled to deep multi-omic metabolic phenotyping. Our data demonstrate that the effects of dietary saturated (SFA), monounsaturated (MUFA), and polyunsaturated fatty acids (PUFA) on the host lipidome, transcriptome, proteome and metabolome are uniquely impacted by resident microbiota. Also, the hepatic levels of both pro-inflammatory and pro-resolving lipid mediators are strongly influenced by dietary fatty acid-microbe interactions. This study presents a unique resource to the nutrition and metabolism research community to advance our understanding of metaorganismal lipid metabolism. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=104 SRC="FIGDIR/small/705588v1_ufig1.gif" ALT="Figure 1"> View larger version (50K): org.highwire.dtl.DTLVardef@4fc7e3org.highwire.dtl.DTLVardef@1cc1dc8org.highwire.dtl.DTLVardef@1b7648dorg.highwire.dtl.DTLVardef@12a8088_HPS_FORMAT_FIGEXP M_FIG O_FLOATNOGRAPHICAL ABSTRACTC_FLOATNO C_FIG

biochemistry↗

bioRxiv2026

Fast-ripples are emergent properties of neuronal networks

Fast-ripples (250-500 Hz) have been proposed as a promising biomarker in epilepsy, but their specificity remains unclear. In particular, it is uncertain whether they reflect chance coincident neural activity or distinctly generated pathological entities. We combined in silico simulations, neuronal cultures, the intra-hippocampal kainate rat model of hippocampal epilepsy, and human microwire recordings to investigate whether fast-ripples occur more frequently than expected by the chance aggregation of action potentials. Our simulations showed that chance aggregation can generate fast-ripples and that their incidence changes depending on brain state, an observation that we confirmed in our rodent data. The likelihood of exceeding chance depended on biological system complexity and vigilance state: fast-ripples in neuronal cultures did not surpass chance levels, whereas those in awake - but not sleeping - rodents did. Similarly, the incidence of fast-ripples in awake human recordings was ~30% greater than expected by chance. As such, our findings suggest that most fast-ripples reflect stochastic network activity rather than distinctly generated pathological entities. This work does not rule out FRs as potential indicators of epileptogenic tissue, but it does challenge prevailing assumptions about their generation and specificity. Their higher prevalence in epileptogenic tissue is likely primarily due to increased excitation and/or neural synchronization, rather than peculiar abnormalities in network behavior.

neuroscience↗

bioRxiv2026

Ab initio prediction of RNA structure ensembles with RNAnneal

RNA utilizes three-dimensional structure in addition to sequence to carry out diverse functions in gene expression and disease. Much like well-folded proteins, RNAs adopt specific three-dimensional structures to carry out their function. Yet comparatively few RNA structures have been solved by atomic resolution structural techniques, in part because unlike structured proteins, RNAs fold into heterogeneous ensembles of interconverting structures that pose a challenge for high-resolution structure probing methods. In this work, we introduce RNAnneal as a method for RNA structural ensemble prediction that seamlessly integrates generative deep learning with statistical physics and molecular dynamics modeling. Given the primary sequence, RNAnneal uses ab inito (i.e., first principles) modeling to sample an ensemble of 3D structures, which, in turn, are used to train an ensemble of unsupervised deep learning models. The RNAnneal score, representing the consensus of the deep learning models, is then used to evaluate the 3D structures. We evaluated RNAnneal structures against 16 experimentally-resolved conformations (ERCs) of riboswitch RNAs and found that pseudoknot-free (PK-free) ERCs were well-reproduced by RNAnneal, with clear avenues for improving performance even on PK-comprising structures. Furthermore, we found that the RNAnneal score outperforms the Rosetta score and a state-of-the-art RNA forcefield on the task of classifying ERCs from decoys. We then introduce the interaction entropy as a measure of conformational heterogeneity within an ensemble and use it to assess our predictions. RNAnneal thus provides a generalizable framework for predicting RNA structural ensembles that will accelerate RNA-targeted drug discovery and the design of functional RNA molecules.

biophysics↗

bioRxiv2026

AADAT-Driven Metabolic Control of Malate and CoQ10 Shapes Immune Evasion in Triple-Negative Breast Cancer

Compared to other subtypes of breast cancer, triple-negative breast cancers (TNBC) have fewer treatment options and exhibit a worse prognosis. Through integrated transcriptomic, metabolomic, immunohistochemical, spatial, and clinical analyses, we identify the mitochondrial enzyme, -aminoadipate aminotransferase (AADAT) as a previously unrecognized metabolic immune checkpoint in TNBC. AADAT mRNA and protein were significantly upregulated in human TNBC, and high AADAT expression was associated with reduced intra-tumoral CD8 T-cell density and inferior survival. Genetic silencing of AADAT in orthotopic murine TNBC models curtailed primary tumor growth and distant metastasis in a CD8 T-cell-dependent manner, enhanced effector T-cell activation, and sensitized tumors to dual PD-1/CTLA-4 blockade. Mechanistically, unbiased metabolomics showed increased malate levels after AADAT knockdown. Additionally, 4-hydroxyphenylpyruvate, an essential precursor for coenzyme Q10(CoQ10) biosynthesis, decreased following AADAT knockdown, suggesting an impaired mitochondrial electron transport chain. CoQ10 supplementation restored metabolic balance and reversed malate accumulation caused by AADAT knockdown, indicating that AADAT helps maintain CoQ10-supported redox homeostasis, thereby preventing malate buildup and export. Notably, malate addition directly boosted CD8 T-cell oxidative metabolism, increased the NAD/NADH ratio and reactive oxygen species, and augmented TNF- and IFN-{gamma} production. In vivo, malate supplementation in drinking water phenocopied AADAT knockdown, restored the response to paclitaxel plus anti-PD-1 therapy in multiple independent syngeneic TNBC models with de novo or acquired resistance to immunotherapy, reduced tumor burden, and prolonged survival. In patient cohorts, higher spatially clustered intra-tumoral malate is associated with co-localization of functional CD8 T cells, decreased exhausted T-cell neighborhoods, and superior post-chemotherapy outcomes. These data position AADAT as a central metabolic orchestrator of immune escape in TNBC and nominate oral malate as a readily translatable adjuvant to reverse chemo-immunotherapy resistance in TNBC. Statement of SignificanceAADAT defines a metabolic-immune axis driving immune evasion and therapy resistance in triple-negative breast cancer. Blocking AADAT or administering oral malate reactivates CD8 T-cell immunity and sensitizes chemo-immunotherapy-resistant tumors to these agents. These findings uncover a readily translatable metabolic vulnerability with potential to improve outcomes for patients with aggressive breast cancer subtypes.

cancer biology↗

bioRxiv2026

Tunable Bessel beam two-photon fluorescence microscopy for high-speed volumetric imaging of brain dynamics

High-speed volumetric imaging of the brain is essential for linking diverse cellular events to tissue-level functions. However, the brains structural and dynamic heterogeneity--spanning microns to millimeters and milliseconds to hours--requires imaging techniques with tunable spatiotemporal resolution, flexible 3D sampling, and compatibility with targeted perturbations. Here, we present tunable Bessel beam two-photon fluorescence microscopy (tBessel-TPFM), a compact, low-cost, and versatile platform for intravital brain imaging across millimeter scale with subcellular resolution. tBessel-TPFM transforms slow 3D volume scans into fast 2D frame scans via an axially elongated Bessel focus, achieving acquisition rates [~]100-fold faster and reduced motion artifacts compared with conventional TPFM. Exploiting its full tunability of the Bessel focus, we applied tBessel-TPFM for quantitative mapping of cerebral blood flow and neurovascular coupling in normal and ischemic stroke mice. Unlike existing Bessel focus generation methods, the axial center of tBessel-TPFM remains fixed at the objective focal plane during profile tuning. Leveraging this advantage, we integrated tBessel-TPFM with simultaneous 3D targeted optogenetic stimulation for volumetric neuronal connectivity mapping. We also tracked microglial process dynamics following single-cell laser ablation, revealing diverse neuroimmune responses across spatial and temporal scales. By combining high speed, deep penetration, tunable sampling, and multimodal perturbation, tBessel-TPFM empowers a broad spectrum of neurobiological investigations--from vascular physiology and functional connectivity to neuroimmune interactions.

bioengineering↗

bioRxiv2026

Proteolytic processing of Argonautes in multiphase condensates promotes transgenerational epigenetic inheritance

Epigenetic information can be memorized in perinuclear condenates to transmit across generations to modulate gene expression in descenent. Here, we identify a conserved N terminal proteolytic mechanism in Caenorhabditis elegans that is essential for small RNA-mediated transgenerational epigenetic inheritance (TEI) and for maintaining germline immortality. Two conserved, perinuclear germ granule subcompartments-enriched N-terminal processing peptidases, APP-1 and DPF-3, act cooperatively to trim N-terminal intrinsically disordered regions of the Argonaute proteins WAGO-1, WAGO-3, and WAGO-4. This processing shields these Argonautes from proteasome-mediated protein degradation, prevents the loss of their associated small RNAs, and promotes their accumulation in germ granules subcompartments. In contrast, unprocessed WAGOs are recognized and destabilized by the conserved E3 ubiquitin ligases RNF-1 and UBR-7. We propose that this cooperative proteolytic mechanism adds another layer of sspatiotemporal regulation over small RNA-directed epigenetic inheritance and may represents a broadly conserved strategy for regulating protein homeostasis and gene expression programs.

cell biology↗

bioRxiv2025

A Comprehensive Treatment-Induced Resistance Atlas of Glioblastoma Reveals a Fibrotic Niche Shielding the Tumor from Immunotherapy

Therapeutic resistance in IDH-wildtype glioblastoma (GBM) is driven by an intricate interplay between cellular plasticity and protective microenvironmental niches. By constructing the GRIT-Atlas--a massive transcriptomic compendium encompassing nearly one million high-quality single cells across 17 cohorts--and cross-analyzing it with 48 independent Visium spatial transcriptomics sections, we discover a recurrent "Spatial Resistance Triad" comprising mesenchymal-like (MES-like) malignant cells, myeloid-derived suppressor cells (MDSCs), and collagen-secreting cancer-associated fibroblasts (CAFs). This large-scale spatial integration reveals that the triad physically fortifies microvascular proliferation (MVP) and pseudopalisading necrosis (PAN) niches. To substantiate these findings at true single-cell resolution, we deploy high-plex spatial molecular imaging (CosMx SMI) to map over 400,000 single cells across a clinical validation cohort. This high-resolution architecture firmly validates the triads geography and, critically, provides structural gradient evidence unmasking a spatiotemporal continuum wherein PAN emerges as a direct functional consequence of MVP advancement and subsequent vascular collapse. Mechanistically, stromal CAF-derived COL6A1 engages CD44 receptors on MES-like cells to accelerate a potent neural stemness cascade within these protective domains. Utilizing a blood-brain barrier (BBB)-penetrant multi-library compound screen, we identify Lacidipine as a multi-modal stromal disruptor that successfully silences CAF activation and halts matrix secretion. Incorporating Lacidipine into the standard chemo-immunotherapy backbone (TMZ + anti-CSF1R) completely dismantles the protective desmoplastic matrix niche, forcing a profound collapse of intracranial tumor burden and dramatically extending overall survival in orthotopic models. Collectively, our study leverages unprecedented single-cell and spatial scale to provide a definitive blueprint of GBM therapeutic evasion, establishing matrix-targeted intervention as a mandatory prerequisite for successful glioblastoma eradication.

cancer biology↗

bioRxiv2025

CK2 inhibitor, CX-4945, enhances BH3 priming and promotes apoptosis of venetoclax-resistant AML by targeting antiapoptotic proteins

Acute myeloid leukemia (AML), the most common hematologic malignancy, generally has a poor prognosis. Despite initial favorable responses to the BCL2 inhibitor venetoclax (VEN), remission is transient, and AML is eventually fatal. Resistance to VEN is primarily due to the overexpression of anti-apoptotic proteins, including MCL-1, BCL2L1 (BCL-XL), and BCL2A1. Casein kinase II (CK2) is a serine-threonine kinase and a known suppressor of apoptosis. We and others have reported that protein kinase CK2 activity is high in leukemic stem cells (LSCs) and associated with resistance to chemotherapy. We have shown that the selective CK2 inhibitor, CX-4945, suppresses BCL-XL and has a significant anti-tumor effect in AML preclinical models. CK2 expression and activity are high in venetoclax-resistant AML (VR-AML) cell lines. Genetic and pharmacological inhibition of CK2 significantly altered VR-AML gene signature, decreased MCL-1 protein level, increased BH3 priming and sensitized VR-AML cells to apoptosis. More importantly, CX-4945 selectively targeted LSCs (CD34+CD38-) and chemoresistant (CD123+CD47+) subpopulation in VR-AML. CX-4945 combined with VEN decreased leukemia burden and prolonged the survival of VR-AML cell line-derived and patient-derived xenografts compared to either drug alone. The combinatorial treatment was well tolerated in mice without additional myelosuppression or organ toxicity. CX-4945 (silmitasertib) is being tested in several early-phase clinical trials against adult and pediatric cancers. These preclinical results support the use of CX-4945 in combination with VEN to overcome resistance to apoptosis and re-sensitize VR-AML to chemotherapy. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=135 SRC="FIGDIR/small/696284v1_ufig1.gif" ALT="Figure 1"> View larger version (43K): org.highwire.dtl.DTLVardef@6a902forg.highwire.dtl.DTLVardef@2028f5org.highwire.dtl.DTLVardef@160fd4dorg.highwire.dtl.DTLVardef@95da53_HPS_FORMAT_FIGEXP M_FIG C_FIG

pharmacology and toxicology↗