š§ Neurobiology
C1q and immunoglobulins mediate activity-dependent synapse loss in the adult brain.
Using chemogenetics, spatial transcriptomics, live cell tracking, and super-resolution microscopy, this study shows that neuronal hyperactivity drives C1q-dependent synapse loss in the adult hippocampus. Antibody-secreting B-lineage cells associate with sites of C1q deposition, and reducing perforant path hyperactivity in an Alzheimerās mouse model lowers local amyloid-Ī², C1q, and synapse loss.
Impact: Reveals an activity- and antibody-driven mechanism for complement-mediated synapse elimination in the adult brain with direct relevance to Alzheimerās disease.
Crowley G et al., https://doi.org/10.1126/science.adv1219
Morphological and functional diversity of spatially resolved vestibular ganglion neuron cell types.
Combining single-cell and spatial transcriptomics, the authors define five transcriptionally distinct vestibular ganglion neuron types with discrete spatial distributions, organ-specific innervation, and characteristic synaptic endings. Genetic labeling of each type reveals a dedicated gravity-sensing population that underlies otolith-dependent vestibulo-ocular reflexes.
Impact: Delivers a cell-typeāresolved spatial atlas and functional framework for vestibular computations and balance control.
Liu R et al., https://doi.org/10.1073/pnas.2530677123
Microglial TDP-43 mediates myelin refinement and represses Tyrobp cryptic exon inclusion in mice.
By deleting TDP-43 specifically in microglia, the study uncovers early-life myelin abnormalities, motor deficits, and an interferon-responsive signature linked to oligodendrocyte dysfunction using spatial transcriptomics and multimodal imaging. Mechanistically, loss of microglial TDP-43 induces cryptic exon inclusion in Tyrobp, truncating DAP12 and disrupting TREM2āDAP12 signaling and myelin clearance.
Impact: Identifies microglial TDP-43 as a key regulator of TREM2āDAP12 signaling and myelin remodeling, with implications for TDP-43 proteinopathies.
Compagnion AC et al., https://doi.org/10.1038/s41593-026-02348-3
š§¬ Cancer Research
AI-predicted spatial transcriptomics unlocks breast cancer biomarkers from pathology.
This study introduces Path2Space, a deep learning model that predicts spatial gene expression patterns directly from routine histopathology slides, trained on large breast cancer spatial transcriptomics datasets. Applying Path2Space to 976 TCGA tumors reveals three spatially distinct tumor microenvironment subgroups with differing survival and improves prediction of chemotherapy and trastuzumab response over bulk sequencing biomarkers.
Impact: Provides a scalable, low-cost route to spatially resolved biomarkers for breast cancer treatment stratification using standard pathology images.
Shulman ED et al., https://doi.org/10.1016/j.cell.2026.04.023
Macrophage-Induced Senescent Cancer-Associated Fibroblasts Promote SASP-Mediated Chemoresistance in Colorectal Cancer.
Using single-cell RNA-seq, spatial transcriptomics, and multiple preclinical CRC models, the authors develop a Cellular Senescence Prediction Model to identify senescent CAFs that strongly associate with poor chemotherapy response. They show macrophage-derived IL1B drives CAF senescence via IL1R1, producing a SASP enriched in IL6 and CXCL12 that promotes chemoresistance, and fibroblast-specific Il1r1 knockout mitigates this effect.
Impact: Pinpoints the IL1BāIL1R1āsCAF axis as a spatially organized driver of stromal-mediated chemoresistance and a targetable vulnerability in colorectal cancer.
Yang S et al., https://doi.org/10.1158/0008-5472.CAN-25-4870
š§ Gastrointestinal & Mucosal Immunology
Primary sclerosing cholangitis displays distinct colonic mucosa topography yet a shared mast cell state with ulcerative colitis.
Integrating single-cell transcriptomics, antigen receptor sequencing, microbiome profiling, and spatial transcriptomics across four colonic regions, this study maps region-specific immune and microbial landscapes in PSC-associated ulcerative colitis versus classic UC. PSC-UC colons harbor distinct right-sided mucosa-adherent microbiota and enriched activated CD8 and Ī³Ī“ T cells even during histologic remission, while sharing a conserved mast cell state with UC.
Impact: Defines a spatially resolved immuneāmicrobiota topography that distinguishes PSC-UC from UC and may explain its unique clinical and cancer risk profile.
Tearle JL et al., https://doi.org/10.1038/s41467-026-75231-1
š§« Developmental & Reproductive Biology
Spatial transcriptomic mapping of postnatal mouse uterine development.
By combining high-resolution in situ transcriptomics with histology, proteomics, genetic models, and functional assays, this work builds a spatially resolved atlas of mouse uterine development from postnatal day 3 to 21. The study shows endometrial glands emerge from luminal epithelium via progressive transcriptional reprogramming and uncovers compartment-specific signaling biases (Wnt, RA, Hedgehog, RTK, Notch, TGFĪ², BMP, Hippo, PI3KāmTOR) that coordinate adenogenesis and epithelial maintenance.
Impact: Offers a comprehensive spatial framework of signaling networks guiding postnatal uterine morphogenesis and gland formation.
Jamaluddin MFB et al., https://doi.org/10.1073/pnas.2600524123
š Transplantation & Xenobiology
Longitudinal multiomics profiling of extracorporeal cross-circulation with pig liver xenografts in human decedents.
In gene-edited porcine liver xenografts cross-circulated with human decedents, the authors integrate longitudinal proteomics, lipidomics, metabolomics, spatial transcriptomics, and histology to dissect hostāxenograft interactions. They observe progressive human immune cell infiltration with loss of porcine resident immune cells, divergent complement activation, platelet adhesion to activated porcine endothelium, and preserved xenograft metabolic support despite anhepatic conditions.
Impact: Delineates cellular and molecular mechanisms underlying thrombocytopenia and compatibility in liver xenograft support, informing safer clinical translation of ELC.
Guo Q et al., https://doi.org/10.1038/s41591-026-04511-6
š§Ŗ Technology & Methods Development
SpatialPEFT: A Parameter-Efficient Fine-Tuning Framework for Spatial Transcriptomics Foundation Models.
SpatialPEFT presents a unified parameter-efficient fine-tuning strategy that combines LoRA, gradient checkpointing, and a spatial-aware adapter to adapt large spatial transcriptomics foundation models (up to 1.4B parameters) on a single 16 GB GPU. The framework cuts peak VRAM usage by more than 87% while boosting downstream spatial annotation accuracy across tasks.
Impact: Democratizes fine-tuning of large spatial omics models by making high-performance adaptation feasible on commodity hardware.
Zou X et al., https://doi.org/10.1093/bioinformatics/btag503
DFCE-KanT: Predicting Spatial Gene Expression from Histology Images via Contrastive Learning.
This work proposes DFCE-KanT, a contrastive learningābased model that leverages spatial context in H&E whole-slide images to predict spatial gene expression as a cost-effective proxy for experimental ST. By better aligning image patches with their corresponding expression patterns, DFCE-KanT addresses limitations of previous deep learning approaches that underutilized spatial structure.
Impact: Advances image-to-transcriptome prediction, enabling broader virtual spatial transcriptomics from routine pathology slides.
Li F et al., https://doi.org/10.1093/bioinformatics/btag502