Spatial transcriptomics is a molecular biology technique that maps the expression levels of genes across the physical space of tissue samples, preserving information about the spatial location and organization of cells. Unlike traditional transcriptomics methods that homogenize tissue into a single sample, spatial transcriptomics maintains spatial context, enabling researchers to understand how gene expression varies across different regions of a tissue and how it relates to cellular organization and tissue architecture.1)
Spatial transcriptomics combines high-throughput gene expression profiling with spatial localization, allowing researchers to simultaneously measure which genes are active in specific locations within intact tissue sections (([https://www.nature.com/articles/s41576-021-00325-y|Rao et al. - Spatial transcriptomics and imaging-based spatial omics|Nature Reviews Genetics]]]). The technique preserves the two-dimensional or three-dimensional structure of tissue while capturing transcriptomic information at near-single-cell or subcellular resolution. This spatial dimension is critical for understanding tissue organization, as gene expression patterns often vary significantly across different anatomical regions and cellular microenvironments.
The foundational principle involves physically arraying samples on surfaces containing either probes or capture sequences that can detect RNA molecules and encode their spatial positions. After tissue sectioning and placement on these spatially-indexed surfaces, hybridization or enzymatic reactions enable capture of RNA transcripts, with the capture location automatically recording spatial coordinates. Subsequent sequencing or fluorescence imaging then quantifies expression levels for each detected transcript at its specific spatial location (([https://science.sciencemag.org/content/353/6294/78|Ståhl et al. - Visualization and analysis of gene expression in tissue sections by spatial transcriptomics|Science]]]). Different platforms employ variations in this basic approach, including in-situ hybridization arrays, bead-based capture, and high-resolution imaging techniques.
Spatial transcriptomics has emerged as a particularly powerful approach for understanding tumor biology and heterogeneity. In cancer research, the technique reveals how gene expression varies across tumor regions, including differences between tumor centers and edges, and between malignant cells and surrounding stromal and immune cell populations. This spatial perspective illuminates the microenvironmental context in which cancer cells operate, including proximity to blood vessels, immune cell infiltration patterns, and interactions with fibroblasts and other supporting cells.
The technique enables identification of distinct cellular phenotypes and their spatial relationships within tumors, supporting investigation of mechanisms driving treatment resistance, metastatic potential, and progression. Researchers can correlate gene expression signatures with tumor architecture, revealing how spatial organization influences cancer cell behavior. This capability extends to identifying novel therapeutic targets by exposing spatial dependencies and context-specific vulnerabilities (([https://www.nature.com/articles/s41556-023-01197-5|Wu et al. - Understanding spatial organization of tumors using spatial transcriptomics|Nature Cell Biology]]]).
Commercial and research implementations now incorporate spatial transcriptomics as part of comprehensive multimodal characterization of tumors, capturing diverse data types including spatial gene expression, protein abundance, morphological features, and genomic variants. These integrated datasets provide increasingly rich molecular portraits of tumor samples, supporting development of improved diagnostic and prognostic tools.
Multiple technological platforms for spatial transcriptomics have been developed, each with distinct tradeoffs between resolution, throughput, and ease of implementation. Array-based methods use predefined capture arrays with known spatial positions, enabling detection of hundreds to thousands of genes with spatial coordinates. Imaging-based approaches use fluorescent in-situ hybridization (FISH) techniques to directly visualize individual RNA molecules, achieving single-molecule sensitivity but typically covering fewer genes simultaneously. Sequential rounds of imaging can be performed to increase gene coverage through multiplexing strategies.
Common platform architectures include those using DNA-barcoded oligonucleotide spots on glass slides for capture-based detection, light-directed synthesis of oligonucleotide arrays for high-throughput in-situ capture, and microtechnology approaches using microfluidics or other precision positioning systems. Computational processing of spatial transcriptomics data involves registering captured transcripts to spatial coordinates, normalizing expression measurements across spatial regions, and integrating with morphological or immunological data (([https://genomebiology.biomedcentral.com/articles/10.1186/s13059-021-02407-x|Asp et al. - Spatial transcriptomics enables discovery of spatial cell interactions|Genome Biology]]]).
Advanced analytical methods employ graph-based approaches to define spatial neighborhoods, statistical models to identify significantly localized expression patterns, and machine learning techniques to predict cell types or functional states based on spatial gene expression signatures.
Despite significant advances, spatial transcriptomics faces several technical and analytical challenges. Spatial resolution remains a limiting factor for many applications, with most methods capturing expression data at cellular or subcellular resolution rather than achieving single-molecule localization across large tissue areas. Scale and throughput considerations mean that comprehensive analysis of large tissue samples requires either high computational overhead or acceptance of reduced resolution in some contexts.
RNA capture efficiency and detection sensitivity vary across different spatial locations and sample types, potentially introducing systematic biases in spatial expression maps. Integration of spatial transcriptomics data with other omics modalities (genomics, proteomics, metabolomics) remains technically challenging and requires sophisticated alignment and computational approaches. Interpretation of spatial expression patterns is complicated by cellular heterogeneity within spatial regions and the presence of multiple cell types within individual analysis units.
Ongoing research addresses these limitations through development of higher-resolution imaging and capture strategies, improved computational methods for cell type deconvolution within spatial regions, and integration frameworks for multimodal spatial omics data. Enhanced protocols are being developed to increase detection sensitivity, expand the number of simultaneously measurable genes, and scale analyses to larger tissue areas while maintaining spatial information fidelity.