🧠 Neuro-oncology & Brain Disease
A blueprint for local and distal invasion programs in glioblastoma.
Using single-cell RNA-seq, spatial transcriptomics, and multiplexed imaging in orthotopic glioblastoma xenograft models, this study disentangles local versus long-range (distal) invasion programs in brain tissue. Distally invading tumors are enriched for oligodendrocyte progenitor–like cells following peri-axonal routes, while locally invading tumors show mesenchymal-like states associated with peri-vascular invasion, each with distinct transcriptional programs and spatial niches.
Impact: Provides a multiscale “invasion blueprint” that links glioblastoma cell states, invasion routes, and spatial microenvironments to inform targeted therapies.
Chanoch-Myers R et al., https://doi.org/10.1038/s41467-026-70470-8
Decoding neurodegeneration one cell at a time.
This review synthesizes how single-cell transcriptomics, epigenomics, and spatial transcriptomics are reshaping our understanding of selective neuronal vulnerability across major neurodegenerative diseases. By mapping vulnerable cell types, disease-associated molecular states, and their local tissue context, the authors outline how these technologies are moving the field from descriptive atlases toward mechanistic and causal insights.
Impact: Establishes a roadmap for using single-cell and spatial genomics to uncover the molecular basis of neurodegeneration and guide therapeutic development.
Gautier O et al., https://doi.org/10.1172/JCI199841
Spatial transcriptomics uncovers vasculature-centered cellular interactions driving Japanese encephalitis progression in a mouse model.
Leveraging Stereo-seq to capture host and viral transcriptomes in situ during Japanese encephalitis virus infection, this work builds a spatiotemporal atlas of brain pathology in mice. The study identifies vasculature-centered immune cell niches, including Ly6c2⁺ inflammatory populations, and dissects how their spatial organization drives neuropathology and disease progression.
Impact: Delivers a high-resolution spatial framework for understanding and targeting neuroinflammatory circuits in viral encephalitis.
Ou Z et al., https://doi.org/10.1038/s41467-026-70047-5
🧬 Cancer Research & Tumor Microenvironment
FineST: contrastive learning integrates histology and spatial transcriptomics for nuclei-resolved ligand-receptor analysis.
FineST is a deep contrastive learning framework that fuses histology images with spatial transcriptomics to achieve nuclei-level segmentation, high-resolution RNA imputation, and detailed ligand–receptor interaction mapping. Applied to VisiumHD and Xenium datasets across several tumor types, FineST outperforms existing tools in cell type prediction and uncovers tumor–immune communication patterns at invasive fronts, tertiary lymphoid structures, and therapy resistance niches.
Impact: Introduces a powerful AI-driven pipeline to resolve cell–cell communication and immune interactions in cancer from routine histology plus spatial data.
Li L et al., https://doi.org/10.1038/s41467-026-70528-7
Spatial heterogeneity of MDSCs mediated by ANXA1-FPRs signaling drives immune suppression in OSCC progression.
Combining single-cell and spatial transcriptomics, this study maps the heterogeneous distribution and states of myeloid-derived suppressor cells (MDSCs) in oral squamous cell carcinoma. It reveals that ANXA1–FPR signaling underpins spatially organized MDSC-driven immune suppression, particularly impacting CD8⁺ T-cell function in specific tumor niches.
Impact: Identifies spatially restricted ANXA1–FPR–dependent MDSC programs as actionable targets to enhance immunotherapy in OSCC.
Li F et al., https://doi.org/10.1038/s41467-026-70861-x
Multimodal imaging reveals a lysosomal drug reservoir that drives heterogeneous distribution of PARP inhibitors.
Using a patient-derived explant pipeline integrating multimodal imaging and spatial transcriptomics, the authors show that PARP inhibitor rucaparib accumulates heterogeneously at the single-cell level in ovarian tumors. Lysosomal sequestration forms an intracellular drug reservoir, enriching lysosomal and apoptotic signatures in high-drug regions and differentially influencing response across weak base PARP inhibitors.
Impact: Uncovers lysosomes as a key spatially variable determinant of PARP inhibitor distribution and efficacy, pointing to strategies to overcome resistance.
R Moncayo C et al., https://doi.org/10.1038/s41467-026-70558-1
Hot Zones for Liver Cancer: Metabolic Zonation, Ferroptosis, and the Origins of HCC.
This commentary highlights a study that uses spatially restricted genetic engineering and spatial transcriptomics to trace how hepatocellular carcinoma arises within the liver’s zonated metabolic landscape. Despite greater expansion of periportal zone 1 clones, tumors preferentially originate from pericentral zone 3 hepatocytes, driven by glutathione S-transferases that protect against ferroptosis and can reprogram zone 1 cells into tumor-initiating states.
Impact: Defines “tumorigenic zonation” and exposes ferroptosis-related vulnerabilities as spatially encoded targets for HCC prevention and therapy.
Barrows KM et al., https://doi.org/10.1158/0008-5472.CAN-25-5839
🛠️ Technology & Methods Development
FineST: contrastive learning integrates histology and spatial transcriptomics for nuclei-resolved ligand-receptor analysis.
FineST integrates a histology foundation model with spatial transcriptomics through contrastive learning to boost resolution from spot-level to nuclei-level analyses. By improving RNA imputation, nuclei segmentation, and ligand–receptor inference, it recovers subtle communication patterns in complex tumor microenvironments that are missed by current pipelines.
Impact: Sets a new standard for histology-aware spatial transcriptomics analysis, enabling finer-grained cellular and communication mapping.
Li L et al., https://doi.org/10.1038/s41467-026-70528-7
Celcomen: spatial causal disentanglement for single-cell and tissue perturbation modeling.
Celcomen is a generative graph neural network grounded in causal inference that disentangles intra- and intercellular gene regulatory programs from spatial transcriptomics data. Validated on simulated and real datasets from glioblastoma, fetal spleen, and lung cancer, it can generate counterfactual “virtual tissue” responses to perturbations, mimicking experimentally inaccessible conditions.
Impact: Provides a causal modeling framework to virtually test perturbations in spatially resolved tissues, accelerating in silico hypothesis generation and therapy design.
Megas S et al., https://doi.org/10.1038/s41467-026-69856-5
Metabolic RNA Labeling-Enabled Time-Resolved Single-Cell RNA Sequencing.
This Account reviews the evolution of metabolic RNA labeling–based time-resolved scRNA-seq, detailing methods such as Well-TEMP-seq, scDUAL-seq, and Dyna-vivo-seq that capture RNA synthesis and degradation dynamics at single-cell resolution. The authors also describe emerging efforts to merge these temporal measurements with spatial transcriptomics to chart spatiotemporal regulation of gene expression and cell fate in tissues.
Impact: Outlines a powerful toolkit for integrating time and space into single-cell transcriptomics, opening new windows on dynamic gene regulation in vivo.
Yin K et al., https://doi.org/10.1021/acs.accounts.6c00010
🧫 Immunology & Inflammatory Disease
Single-cell and spatial profiling reveal cDC2A-CXCL13(+)CD8(+) T-epithelial cell crosstalk and cytotoxicity through TNFRSF9 in cutaneous and mucosal lichen planus.
By combining single-cell RNA-seq, spatial transcriptomics, and proteomics from patients with cutaneous and mucosal lichen planus, this study maps a highly cytotoxic immune microenvironment. It identifies cDC2A dendritic cells, CXCL13⁺ CD8⁺ T cells, and epithelial cells engaged in TNFRSF9-dependent crosstalk that drives T-cell infiltration and keratinocyte apoptosis.
Impact: Pinpoints spatially organized immune–epithelial circuits in lichen planus, highlighting TNFRSF9 signaling as a potential therapeutic axis.
Jiang R et al., https://doi.org/10.1038/s41467-026-70506-z