Sub-cellular Imaging of the Entire Protein-Coding Human Transcriptome (18933-plex) on FFPE Tissue Using Spatial Molecular Imaging
Rustem Khafizov,Erin Piazza,Yi Cui,Michael Patrick,Evelyn Metzger,Daniel McGuire,Dwayne Dunaway,Patrick Danaher,Margaret L Hoang,Andrew Grootsky,Megan Vandenberg,Shanshan He,Rachel Liu,Michael McKean,Michael Rhodes,Joseph M Beechem
DOI: https://doi.org/10.1101/2024.11.27.625536
2024-12-03
Abstract:Single-cell RNA-seq revolutionized single-cell biology, by providing a complete whole transcriptome view of individual cells. Regrettably, this was accomplished only for individual, tissue-dissociated cells. High-plex spatial biology has begun to recover the x, y, and z-coordinates of single-cells, but typically at the expense of far less than whole transcriptome coverage. To solve this problem, Bruker Spatial Biology has accomplished a commercial-grade panel (CosMx® Spatial Molecular Imager Whole Transcriptome Panel (WTx)), using 37,872 imaging barcodes, capable of sub-cellular imaging of the entire human protein-coding transcriptome. The imaging barcodes are encoded with 156 bits of information (4 on-cycles and 35 dark-cycles per code), at a Hamming Distance of 4 from each other to achieve a very low false-code detection. Key to achieving this high-plex capability was the ability to manufacture imaging barcodes that require no in-tissue amplification (every barcode is manufactured under GMP to contain exactly 30 fluorescent dyes) and uniform, size-exclusion purified, extremely small imaging barcodes (~ 20 nm). A detailed study of six different human FFPE tissue types was performed (Colon, Pancreas, Hippocampus, Skin, Breast, Kidney), yielding over 5.4 billion transcripts from 2.7 million cells. We counted over 1,550 transcripts-per-cell on average and observed 900 unique genes per cell (measured as the median). Single fixed-cells containing well over 10,000 subcellularly imaged transcripts were accomplished. Advancing single-cell imaging to the whole transcriptome level opens a single unified approach to accomplish essentially all single-cell experiments, both imaging and non-imaging. Depending upon the sample type (e.g. fixed-cells, organoids, tissue sections, etc.), the transcripts per cell and genes per cell measured using the whole transcriptome panel often exceeds that obtained by the highest-resolution single-cell RNA-seq, can be performed on a single 5 μm FFPE tissue section, with no dissociation bias (every cell is counted). Pathway analysis within the tumor bed of a colon adenocarcinoma sample found evidence of enrichment in pathways suggestive of an aggressive tumor type, and localized ligand-receptor analysis showed spatially restricted patterns related to adhesion, migration, and proliferation. The high-dimensional whole transcriptome data is streamed directly to a cloud-based Spatial Informatics Platform, allowing for the scalable processing of millions-of-single-cells and billions-of-transcripts per operation. The WTx data are combined with high-resolution antibody-based cell-morphology imaging and data-driven machine-learning cell segmentation algorithms, to generate the most complete view of single cell and sub-cellular spatial biology that has ever been obtained.
Genomics