Tweet
Spatial transcriptomic technologies offer multiple paths for measuring gene expression within intact tissue, each defined by distinct technical and practical tradeoffs that can impact experimental design. In a recent SEQanswers webinar, Michal Lipinski, Senior Scientist at the Broad Institute’s Spatial Technology Platform, provided a detailed overview of spatial transcriptomics to help researchers decide whether to invest in these methods and how to choose among available technologies.
Why Spatial Context Matters
Single-cell RNA sequencing has become an essential technology for profiling gene expression across thousands of cells, but it lacks information about cellular location within tissue. Traditional imaging approaches, such as immunohistochemistry and in situ hybridization, preserve spatial context but are limited in the number of targets they can measure. Lipinski explained that spatial transcriptomics combines these approaches by pairing high-dimensional gene expression data with spatial information to map transcriptional patterns directly onto tissue architecture.
Sequencing and Imaging Approaches
Spatial transcriptomics methods generally fall into two categories (sequencing or imaging) based on how transcripts are detected. In sequencing-based approaches, tissue sections are placed on substrates containing spatially barcoded capture features, typically with poly-T sequences to bind polyadenylated RNA. After capture, samples undergo reverse transcription and library preparation similar to single-cell RNA-seq, followed by sequencing. The sequencing reads are then mapped back to spatial coordinates using the known barcode locations.
Imaging-based approaches use microscopy to localize predefined sets of transcripts directly within tissue sections. These methods rely on gene-specific probe panels and detect transcripts through in situ sequencing or sequential rounds of hybridization and imaging, ultimately producing images in which individual spots correspond to individual RNA molecules.
Key Considerations in Method Selection
With a rapidly expanding set of spatial transcriptomics options, selecting the most appropriate method has become increasingly challenging. Given the growing complexity of the field, Lipinski advised researchers to consult experts or core facilities early, as hands-on input is often essential for matching a method to a specific application. Along with this guidance, he outlined three main considerations.
Sample Characteristics
The first is sample characteristics, including preservation method, tissue size, and source. Sample preservation is a major limiting factor. Addressing it requires standard molecular biology equipment, high-quality microscopy or slide scanning, access to specialized instruments, and appropriate sectioning tools such as a cryostat or microtome. The two dominant preservation methods are fresh frozen and formalin-fixed paraffin-embedded (FFPE) tissue. Fresh frozen samples are sectioned using a cryostat, while FFPE samples are cut using a microtome. Although FFPE samples dominate clinical archives, many spatial transcriptomics protocols cannot accommodate them. Even among FFPE samples, quality varies widely. RNA degradation during fixation, embedding, or delayed processing can compromise both transcript detection and tissue morphology. Lipinski stressed that not all FFPE samples are equivalent, and poor sample quality can limit the usefulness of spatial assays.
Sample size also matters. Some platforms profile relatively small tissue areas per run, while others can accommodate multiple sections simultaneously. This influences experimental design, throughput, and cost, especially for studies that require large tissue coverage.
Matching Methods to Experimental Goals
The experimental goals are the next significant consideration. This includes the number of genes required, sensitivity, and spatial resolution. Lipinski explained that while protein localization benefits greatly from subcellular resolution, transcript localization is often less informative at that scale because most mRNAs are transcribed in the nucleus and translated in the cytoplasm. Imaging-based platforms can localize individual transcripts at near subcellular resolution, whereas sequencing-based methods depend on the size of their capture features. Feature sizes range from roughly ten microns in some systems down to one or two microns in newer platforms, approaching single-cell resolution. Lipinski notes that one to two micron resolution is often sufficient to resolve individual cells, even if it does not reach classical microscopy resolution.
Ultimately, Lipinski stressed that no single approach is optimal for all applications. Imaging-based methods require researchers to select target genes in advance, similar to antibody-based assays. These predefined panels introduce bias but provide precise spatial localization. Sequencing-based methods offer broader transcriptome coverage, allowing for the discovery of unexpected markers and pathways, though at the cost of lower spatial precision. Selecting a technology, therefore, involves balancing resolution, gene coverage, sample constraints, and cost, with trade-offs inherent to every spatial transcriptomics experiment.
Practical Constraints
The third consideration is practical constraints, including accessibility, reliability, and cost. Cost was a recurring concern throughout the talk. Lipinski described spatial profiling as expensive even in well-funded environments. Imaging-based methods require costly instruments and reagents, with additional expense for customized probe panels. Sequencing-based methods involve expensive consumables and sequencing costs, and their limited tissue coverage can drive up costs for large studies.
Despite these challenges, Lipinski offers strategies to manage expenses. Multiplexing samples can improve cost efficiency, particularly for imaging-based approaches. Selecting the appropriate assay for the biological question is essential. Not all studies require thousands of genes or the highest-resolution platforms. Off-the-shelf panels are substantially less expensive than custom designs and can often meet experimental needs. Profiling more samples at lower cost can increase statistical power and biological confidence.
Collaboration is another cost-saving approach. Sharing resources with other researchers or working through core facilities can reduce barriers to access, as cores often already maintain the necessary instruments and expertise. Lipinski also recommends taking advantage of vendor promotions and running pilot studies before committing to large-scale experiments.
Finally, he suggests combining technologies strategically. Researchers can use sequencing-based spatial methods to identify candidate genes or pathways across tissue regions, then apply imaging-based approaches with targeted panels to localize those signals precisely. This staged strategy can balance discovery and resolution while controlling costs.
Final Advice
Lipinski closed by encouraging researchers to consult experts and core facilities when planning spatial transcriptomics studies. He emphasized that thoughtful experimental design is essential for making effective use of these powerful but resource-intensive technologies. To learn more details about choosing a spatial transcriptomics technology, watch the full webinar here.