This primer by Garam Kim introduces the field of spatial transcriptomics, highlighting its evolution from bulk and single-cell RNA sequencing by preserving tissue context and spatial information. She broadly categorizes current technologies into imaging-based and sequencing-based methods. Imaging-based techniques (e.g., Murfish, SeqFISH, In Situ Sequencing, Expansion Sequencing) offer high subcellular resolution by directly visualizing fluorescently barcoded RNA molecules through repeated imaging cycles. Sequencing-based methods (e.g., Visium, HDST, Slide-seq, Dbit-seq, Stereo-seq, OpenST, Slide-tags) provide untargeted, whole-transcriptome profiling by capturing spatially barcoded RNA from tissue onto arrays or by tagging cells directly, followed by sequencing.
The presentation details the principles, strengths, and challenges of each approach, including considerations like resolution (spatial vs. RNA), sample compatibility (fresh frozen vs. FFPE), technical expertise, and cost. Kim concludes by discussing immediate and future directions for the field, such as improved FFPE compatibility, larger capture areas, faster workflows, enhanced multi-omics integration, and advanced computational methods like 3D reconstruction and machine learning for inferring transcriptomic states from images.
Summarizing the Spatial Transcriptomics Technologies Primer:
0:00 Introduction to Garam Kim: Garam Kim, a postdoctoral fellow at the Broad Institute and Harvard, specializes in developing spatially resolved technologies that combine high-resolution imaging and high-throughput genomics. Her goal is to apply these tools to neurodegenerative diseases and make them accessible to a wider research community.
1:37 Primer Overview: The presentation will introduce spatial transcriptomics, classify technologies, discuss their strengths and challenges, offer practical considerations for choosing a method, and touch on future directions in the field.
Bulk RNA sequencing provides high sensitivity but averages gene expression, losing spatial information (like grinding up all Lego pieces).
Single-cell RNA sequencing allows analysis of individual cell types and states but requires tissue dissociation, thus losing cellular context and spatial arrangement.
Spatial transcriptomics is the advancement that preserves tissue structure, enabling the study of cell-to-cell interactions, microenvironments, and spatial patterning, which offers deeper biological insights into disease and tissue architecture.
Imaging-based methods: Detect RNA through repeated fluorescent imaging cycles using targeted probes, offering high subcellular resolution and morphology details.
Sequencing-based methods: Capture RNA off tissue or via dissociation, then sequence it, enabling non-targeted, whole-transcriptome profiling with resolution depending on the specific technology.
Imaging-based: FISH (e.g., Murfish, SeqFISH) and In Situ Sequencing (e.g., Zenium, StarMap, Expansion Sequencing).
Key Principle: Generate an optical barcode, read signals through sequential imaging and signal removal, then decode to identify transcripts.
Murfish/SeqFISH: Use binary or color codes to identify transcripts across imaging rounds.
In Situ Sequencing: Uses unique nucleotide barcodes and relies on padlock probe hybridization and rolling circle amplification (RCA) to amplify and sequence RNAs directly in tissue.
Expansion Sequencing: Combines tissue expansion (embedding in hydrogel to physically enlarge the sample) with in situ sequencing for higher resolution.
Sequencing-based: Spot-based (barcoded arrays) and Cell-based (tagging cells with spatial info).
Key Principle: Place tissue on a capture array with spatially barcoded DNA, allow RNA to diffuse and bind, then prepare and sequence libraries.
Slide-tags: A cell-based method that tags nuclei in intact tissue with photo-cleavable spatial barcodes, allowing for high-throughput profiling using existing single-cell technologies after dissociation.
33:32 Comparing Spatial Methods: Spatial resolution varies widely (down to 0.25 microns), but single-cell sequencing remains the gold standard for RNA resolution and unbiased transcriptomics. Computational challenges exist in deconvoluting signals from spots and accurately assigning them to individual cells.
Imaging-based strengths: Direct visualization, subcellular resolution, morphological context, co-detection with protein, no tissue dissociation artifacts.
Imaging-based challenges: Specialized equipment, limited to targeted gene panels, molecular crowding, autofluorescence (especially in human tissue), complex/long protocols (photobleaching risk), need for highly specific probes, computationally intensive data analysis.
Sequencing-based strengths: Untargeted whole-transcriptome coverage, high throughput, compatible with existing next-generation sequencing workflows, less specialized equipment.
Sequencing-based challenges: Generally lower true cellular spatial resolution, potential RNA/cell loss during processing, blurring of spatial boundaries due to RNA diffusion, computational complexity in cell assignment, higher cost per sample due to array fabrication.
43:48 Practical Considerations: Researchers should choose a method based on their specific biological hypothesis, gene coverage and resolution requirements (subcellular vs. broad structures), sample type (fresh frozen vs. FFPE tissues), available technical expertise (especially computational), and infrastructure/budget constraints (commercial vs. open-source platforms).
46:52 Future Directions: Key areas for advancement include improving FFPE sample compatibility, reducing cost barriers, streamlining workflows, increasing multi-omics capabilities (combining RNA, protein, etc.), integrating pathology information, and leveraging computational advances like 3D reconstruction and machine learning to infer transcriptomic states from images.
50:22 Conclusion: Spatial transcriptomics has evolved significantly, producing visually stunning and biologically rich data, promising exciting discoveries.