Vertebrate embryogenesis is a remarkably dynamic process during which numerous cell types of different lineages generate, change, or disappear within a short period. A major challenge in understanding this process is the lack of topographical transcriptomic information that can help correlate microenvironmental cues within the hierarchy of cell fate decisions. Here, we employed Stereo-seq, a high-definition spatially resolved transcriptomic technology, to dissect the spatiotemporal dynamics of gene expression and regulatory networks in the developing zebrafish embryos. We profiled 91 embryo sections covering six critical time points during the first 24 hours of development, obtaining a total of 152,977 spots at a resolution of 10x10x15 µm3 (close to cellular size) with spatial coordinates. Meanwhile, we identified spatial modules and co-varying genes for specific tissue organizations. By performing the integrated analysis of the Stereo-seq and scRNA-seq data from each time point, we reconstructed the spatially resolved developmental trajectories of cell fate transitions and molecular changes during zebrafish embryogenesis. Our study constitutes a fundamental reference for further studies aiming to understand vertebrate development.

Cancer cells interact with a wide variety of other cell types, but our understanding of microenvironmental heterogeneity and how it influences tumor phenotypes is limited. While single-cell RNA-seq (scRNA-seq) has helped define these TME cell types, it provides limited information on the mechanisms that define how individual tumor cells interact with TME. Here, we integrate spatial transcriptomics with scRNA-seq to define the architecture and nature of nascent tumor and surrounding microenvironment cells as they come into contact through the process of invasion. Using a well-defined transgenic zebrafish model of BRAFV600E-driven melanoma, we identify a transcriptionally unique “interface” cluster localized at the boundary between tumor cells and surrounding tissues. Using an unbiased, data-driven approach, we identify spatially-patterned gene modules specific to the interface and show that the interface is a distinct transcriptional entity that histologically resembles the microenvironment but transcriptionally resembles the tumor. By complementing ST with scRNA-seq, we demonstrate that the interface is composed of specialized tumor and microenvironment cells. Both cell types in the interface upregulate a common set of cilia genes, and we find enrichment of cilia proteins only where the tumor meets the TME. Cilia gene expression is regulated by ETS-family transcription factors, which normally act to suppress their expression outside of this region. This unique ETS-driven interface transcriptional state is conserved across ten different human patient samples, suggesting this is a conserved feature of human melanoma. Taken together, our results demonstrate the power of spatial and single-cell transcriptomics techniques in uncovering novel biological mechanisms that drive tumor invasion into new tissues. 

We performed spatially resolved transcriptomics with sub-single-cell resolution in zebrafish embryos at the one-cell stage, which allowed us to identify a class of mRNAs that is specifically localized at an extraembryonic position in the yolk sac, the vegetal pole. The 3’ UTRs of these localized genes are enriched in specific sequence motifs. Comparison to two frog species revealed relatively low conservation of localized genes, but high conservation of sequence motifs. In vivo RNA labeling followed by scRNA-seq revealed that a large number of the localized transcripts are specifically transported to the primordial germ cells.

Spatial transcriptomics by RNA-tomography (Tomo-seq) on whole or sub-dissected embryos (processed along the anterior-to-posterior axis) and on transversal thick slices (processed along the dorsal-to-ventral axis). 2019-03-15