Microfluidic channels provide a means to deliver barcodes encoding spatial information to a tissue, which allows co-profiling of gene expression and proteins of interest in a spatially resolved manner.
The plethora of sequencing tools have broadened our understanding of how cells function and develop. Yet the majority of sequencing tools leave out the spatial context where cells reside. Such spatial context can be essential to understanding how cells organize within a three-dimensional environment and how cells interact with each other.
Back in 2013, Rong Fan from Yale University was intrigued by a conversation with colleague Kathryn Miller-Jensen: they noticed that trypsinizing cancer cells off the substrate could perturb the measurements of the signaling network. Since then, Fan has been thinking about how to fix and measure cellular states on a substrate or in a tissue without cell dissociation. Barcoding strategies were introduced in massively parallel single-cell RNA sequencing and have substantially advanced the single-cell field. Yet, Fan says, “I was never satisfied with the random barcoding approach.” He hoped to have a method for ‘deterministic barcoding’ of a tissue — delivering barcodes to a given cell in a specific location.
In a sense, the advent of spatially resolved transcriptomics, our Method of the Year 2020, solved Fan’s problem. Sequencing-based Visium from 10x Genomics, Slide-seq, and high-definition spatial transcriptomics (HDST), among others, enable mapping the transcriptomic landscape of a tissue section. Visium, Slide-seq and HDST share the same principle of retaining RNAs’ spatial information by capturing them on barcoded solid-phase substrates such as microarrays or beads.
Instead of capturing RNAs, Rong Fan’s group report a microfluidics platform that delivers DNA barcodes into a tissue at a specific location. The researchers can then sequence the spatially barcoded mRNA and profile proteins of interest at the same time. They named this technology “deterministic barcoding in tissue for spatial omics sequencing” (DBiT-seq).
In DBiT-seq, the parallel channels in microfluidic chip confine DNA barcodes to deliver them in a strip across a fixed tissue slide. The DNA barcodes comprise a spatial barcode, a ligation linker, and an oligo(dT) sequence for annealing to poly(A)-tailed mRNAs; this step is followed by in situ reverse transcription. The strip-like confinement, however, only provides one-dimensional information. To achieve a two-dimensional array of pixels, Fan and his colleagues applied a crossflow scheme, in which they removed the first microfluidic chip and clamped on a second chip to deliver a separate set of DNA barcodes in the perpendicular direction. The two sets of barcodes are joined via the ligation linker and store the spatial information of the respective mRNA. After two rounds of microfluidic flow, the tissue retains its morphology and allows optical or fluorescence imaging to associate individual pixels with gene expression patterns.
In addition to mapping mRNA, DBiT-seq also profiles proteins of interest by applying a panel of antibody-derived DNA tags to the tissue before clamping the first microfluidic chip. The antibody DNA tags contains a unique barcode and a poly(A) tail that can be annealed to the oligo(dT) sequence in the first set of barcodes.
The researchers applied DBiT-seq to an embryonic day (E) 10 mouse embryo to validate its performance at a resolution of 50 µm since comprehensive spatial datasets are available for mouse embryos. Clustering of spatial transcriptomes revealed the major tissue types during early organogenesis. They then increased the resolution to 25 µm pixel size and co-mapped the differentially expressed genes and proteins in the brain region. They further zoomed into the subregion of the developing eye in E10 and E11 mouse embryos at a resolution of 10 µm and determined the changes in gene expression patterns. Furthermore, the integration of DBiT-seq and single-cell transcriptomics data allowed the annotation of cell types and the visualization of cell distributions.
In addition to mouse embryos, Fan notes, “As of today, we have performed DBiT-seq on not only mouse embryos but also adult heart, vessel, tonsil, lymph node, kidney, pancreas and skin.” Moreover, DBiT-seq is compatible with immunostained tissue slides and allows the study of transcriptome and protein expression and of cell morphology. Fan says, “We are thinking about how to better integrate tissue histology images and machine learning to deconvolve single cell-transcriptomes in DBiT-seq pixels.”
Looking forward, Fan and his team hope to improve the throughput and mapping window of DBiT-seq so that tens of tissue sections can be profiled at one experiment. The incorporation of multiplex immunofluorescence and multiplex single-molecule fluorescence in situ hybridization (smFISH) may further improve the resolution of DBiT-seq. Since DBiT-seq is not limited to barcoding mRNAs and proteins, the incorporation of epigenomics sequencing will make the updated DBiT-seq even more powerful.
Liu, Y. et al. High-spatial-resolution multi-omics sequencing via deterministic barcoding in tissue. Cell https://doi.org/10.1016/j.cell.2020.10.026 (2020).
About this article
Cite this article
Tang, L. Multiomics sequencing goes spatial. Nat Methods 18, 31 (2021). https://doi.org/10.1038/s41592-020-01043-w