The progression of prostate cancer (PCa) to metastatic disease remains a critical clinical challenge, making it essential to gain a detailed understanding of the factors that enable certain subclones to invade and spread. Building on our previous study, which examined genomic variations in benign and malignant prostate tissues by inferring copy number alterations (CNA) from spatial transcriptomics data1, this study aims to trace the spatial evolutionary trajectories of PCa subclones from the primary tumor to matched lymph node metastases and to identify alterations in associated microenvironments and early diagnostic markers. To achieve this, we performed spatial transcriptomics on entire prostate axial disks and patient-matched lymph node metastases from 10 individuals. Standard spatial transcriptomics (55 µm) was used for tissue profiling; from the spatial transcriptomics data, copy number alterations (CNAs) were inferred to identify distinct tumor subclones, and a clone-tree was constructed to describe their evolutionary relationships. Additionally, a high-resolution spatial transcriptomics (2 µm) technology was employed to gain a cellular-level view of the immediate tumor-microenvironment (TME) surrounding specific subclones. Analysis of over 1,000,000 barcoded regions identified many distinct tumor subclones, enabling us to trace their evolutionary trajectories spatially. We observed notable subclonal events within the lymph nodes, including polyclonal colonization, indicating multiple dissemination events during the evolution of the primary disease. Exploration of the immediate TME revealed significant cellular heterogeneity and upregulation of genes related to antigen presentation and inflammatory pathways concentrated near ancestral tumor clones, specifically at the tumor border. By focusing on the inferred CNA profiles of the metastasizing clone across patients, we identified several common features. Additionally, we were able to identify potential lethal disease as early as the diagnostic core needle biopsy stage by detecting features of metastasis and pinpointing the metastasizing clone. In summary, our study provides a detailed spatial map of PCa clonal evolution and dissemination, linking primary tumors to nodal metastases and revealing altered cell composition and gene expression around tumor clone borders. Importantly, it demonstrates subclonal events within lymph nodes, polyclonal colonization, and the potential to identify metastasizing clones as early as the diagnostic phase, which holds strong implications for precision oncology and early intervention.

Maintaining postnatal bone growth is crucial for humans to reach their final height. To determine transcriptional networks coordinating this process, we applied spatially resolved transcriptomics to growth plate biopsies obtained from healthy adolescents who underwent epiphysiodesis surgery for idiopathic tall stature. Spatial profiling revealed new markers for each zone of the human growth plate and identified genes associated with poorly understood growth disorders, including the novel hypertrophic zone marker SGMS2. We elaborated on this finding and established that Sgms2 is present in growth plate-derived matrix vesicles, and its activity facilitates mineralization - a process impaired in patients with SGMS2 mutations. By exploring the low transcriptional activity of resting zone chondrocytes, we found that a subset of these cells exists in a functionally quiescent state in vivo, as determined by their predominantly nuclear mRNA, abundant heterochromatin, and ability to exit the G0 phase under specific conditions - features shared with skeletal stem cells in mouse growth plates. Additionally, we identified distinct sub-populations of human resting zone chondrocytes; an exploration of their hierarchy determined that CHRDL2 and/or SFRP5-positive sub-populations were among the least quiescent resting zone cells. In summary, we generated a comprehensive map of gene expression within the human growth plate, revealing novel zone-specific markers, new primary growth disorders, candidate pharmacological targets, and sub-populations of resting zone chondrocytes with features of quiescent stem cells. These results contribute to a better understanding of the cellular and molecular mechanisms governing human height and can facilitate improved diagnosis and treatment strategies for patients with skeletal growth disorders.

While cancer immunotherapies have primarily focused on activation of cytotoxic CD8 killing, CD4 T cell activity is also associated with survival and immunotherapeutic response in numerous cancers. We applied integrated single-cell RNA sequencing and multiplexed protein epitope profiling to breast cancer samples to resolve the complexity of immune cell states within the tumor microenvironment. This approach enhanced phenotypic resolution, identifying three distinct states within the T follicular helper (Tfh) cell cluster. A CXCR4high progenitor state gave rise to two differentiated states: an IGFL2high subset resembling conventional Tfh cells and localised to B cell-rich lymphoid aggregates, and a CD103+ subset, exhibiting features of tissue residency, exhaustion, and cytotoxicity, which co-localised with tumor foci. CD103+ Tfh-like cells were found to interact with CXCL10+ macrophages through production of CCL chemokines and CSF1. A higher CD103+ Tfh to IGFL2high Tfh ratio correlated with improved patient survival and enhanced responses to anti-PD1 checkpoint blockade. These findings integrate Tfh and CD4 with cytotoxic potential in breast cancer, offering new insight into anti-tumor immunity and response to checkpoint blockade.

Bicuspid aortic valve, a prevalent congenital malformation, predisposes individuals to severe complications. Although the condition exhibits substantial heritability, known protein-coding and common regulatory mutations explain a minority of cases. To assess the contribution of rare regulatory variants, here we integrate high-resolution three-dimensional genome organization profiling with matched whole-genome sequencing from eight individuals with bicuspid aortic valves and eight with standard tricuspid aortic valves. In bicuspid aortic valve patients, mutation-driven chromatin rewiring affected 1.8-fold more valve development genes than in healthy individuals. Genome-wide in silico analyses show that rare regulatory mutations disrupt the transcriptomes of mesenchymal cell populations necessary for endocardial cushion formation. We identify 198 candidate genes associated with bicuspid aortic valve, revealing pronounced heterogeneity and complex interplay between coding and regulatory mutations. Collectively, our findings establish rare regulatory mutations as contributors to the heritability of bicuspid aortic valve and underscore the need to elucidate their mechanistic roles in disease pathogenesis.

Li et al. introduce CANVAS, an AI framework for translating hematoxylin and eosin images into spatial maps of tumor habitats. By predicting stable, biologically anchored labels with foundation-model image analysis, CANVAS extends habitat mapping to archival pathology samples and suggests a path toward more accessible precision oncology.

Plant leaves originate from the shoot apical meristem (SAM) and undergo a developmental process of highly coordinated gene expression regulation. To date, only a few key regulators have been identified and characterised, so the gene expression cascades responsible for leaf cell specification and differentiation from SAM remain largely elusive. Here, we optimised a spatial transcriptomics protocol using the 10x Genomics Visium system and developed computational pipelines to reconstruct three-dimensional gene expression profiles of the SAM and sequentially developing leaves in maize seedlings. These enabled positional indexing of cells sampled from consecutive developmental stages, revealing dynamic transitions from undifferentiated stem cells in the SAM to functionally differentiated leaf structures. Through spatial-temporal transcriptome analysis, we identified distinct transcriptional programs and key regulatory genes involved in meristem maintenance, leaf primordia initiation, vascular tissue differentiation, and cellular heterogeneity. This approach outperforms the single-cell transcriptome profiling, which lacks temporal and spatial contexts. Our optimised experimental pipeline, which goes from section preparation to data processing, enables the spatial resolution and 3-dimensional mapping of gene expression profiles. The established pipeline is readily applicable to delineating molecular events underlying developmental transitions, cell type specifications, and differentiation in plants.

Localized prostate cancers are heterogeneous and multifocal, with diverse outcomes. Current prognostic methods are epithelium-centric, overlooking the complex cellular landscape within the tumor microenvironment (TME). To further characterize the heterogeneity of the TME, we performed a comprehensive analysis of cancerous and adjacent-benign cores from 24 patients with hormone therapy-naïve localized prostate cancer using single-cell RNA sequencing (scRNA-seq). Integrating copy-number variation and transcriptional signatures enabled epithelial cell classification across a malignant spectrum, revealing widespread molecular perturbation. The analysis revealed patient-unique and shared luminal states and an expansion of club cell phenotypes, suggesting luminal dedifferentiation. Detailed annotation of stromal phenotypes, with a focus on fibroblasts, identified a perineural fibroblast population. Spatial transcriptomics elucidated the precise anatomic distribution of cancer-associated fibroblasts within the prostate cancer TME. Together, this study provides a valuable foundation for advancing the understanding of prostate cancer pathobiology and developing a comprehensive cellular model of the disease. SIGNIFICANCE: Development of a single-cell RNA-sequencing and spatial transcriptomics cellular reference of localized prostate cancer enables identification of a spectrum of malignant epithelial phenotypes and discovery of a perineural class of fibroblast.

Tissue architecture is a product of a multilayered molecular landscape, where even subtle disruptions in the spatial context can initiate or reflect disease processes. Recent advances in high-throughput spatial omics technologies have enabled the investigation of this complexity in stunning detail, providing groundbreaking insights into how spatial molecular organization shapes health and disease. Spatial analysis empowers the discovery of developmental and disease-associated molecular signatures, cell states and multicellular niches, as well as the evaluation of disease heterogeneity within and across organs. This Review examines spatially resolved pathological molecular alterations in a wide range of disease processes, such as developmental disorders, tumorigenesis, fibrosis and injury responses, neurodegeneration, infection and inflammation, through the lens of these universal biological frameworks. We discuss challenges, opportunities and promising examples in advancing these technologies to clinical applications, including the increasing importance of artificial intelligence. Finally, we explore possible avenues for a more comprehensive, multidimensional assessment of tissues.