NEW YORK – The Broad Institute researchers who invented Slide-seq have found a new way to get spatial transcriptomics data, one that should work with any single-nucleus sequencing (sNuc-seq) method.
Fei Chen and Evan Macosko's new method is called Slide-tags and differs from its predecessor in one very important way.
"It uses the same barcoded arrays, but the process is almost backwards," Chen said. Instead of capturing RNA onto a barcoded array, it releases a barcode into the nucleus of cells in a tissue slide. "It's a little bit like GPS," he said, where spatial location is triangulated by analyzing the number of barcodes found in each nucleus.
Slide-tags thus produces whole-transcriptome data that is "indistinguishable" from sNuc-seq methods but also has location data, Chen said. Moreover, it addresses key issues with sequencing- and imaging-based spatial transcriptomics platforms currently on the market.
"Some of the imaging-based approaches have oversold the extent to which individual cells can be segmented," Macosko said. "It's just a hard problem. We think we've skirted around that," given that each nucleus is already separable from every other one with the barcodes.
Another benefit is that it works in any tissue for which there have been optimized sNuc-seq protocols. In a paper published Wednesday in Nature, Chen and Macosko's team demonstrated the use of Slide-tags in fresh-frozen human brain, lung, kidney, gut, and tonsil, as well as mouse hippocampus.
"This paper is interesting in many ways," said Bony De Kumar, director of operations for the Yale School of Medicine's Center for Genome Analysis. "For one, it's actually tech agnostic." While the authors mostly used 10x Genomics' single-cell sequencing for the paper, they've also used Becton Dickinson's Rhapsody as well as Fluent BioSciences' tube-based method. De Kumar suggested that split-pool barcoding and even single-cell DNA sequencing methods could also be used.
Cell segmentation was the other main advantage. "Tonsil tissue is very dense, and they could effectively do segmentation on that," he said. "That makes it really interesting."
Slide-tags has, in a way, come all the way back to its roots in single-cell sequencing. At its heart are the barcoded beads that label transcripts in a cell nucleus. These are what enabled Slide-seq to capture transcripts for sequencing, in turn taken from Drop-seq, an early single-cell isolation method.
For Macosko, it represents a "unique opportunity" to unite spatial and single-cell methods. Many spatial transcriptomics methods have sought to put the breadth of data obtained from single-cell RNA-seq into the context of tissues. Approaches such as Rong Fan's deterministic barcoding in tissue for spatial omics sequencing (DBiT-seq), 10x's Visium, and NanoString's GeoMx Digital Spatial Profiler have taken this approach, at the expense of spatial resolution. More recently, higher-resolution methods have taken a targeted approach, often multiomic, such as Akoya Biosciences' PhenoCycler-Fusion, 10x's Xenium, and NanoString's CosMx Spatial Molecular Imager.
With Slide-tags, Chen and Macosko feel as if they've captured the best of both worlds. Using 10-micron beads, proof-of-concept experiments in mouse brain showed that they were able to assign spatial locations to about half of the profiled nuclei with no effect on cell typing, molecules per cell, and gene expression profiling. The spatial placement rate has since been raised to about 60 percent, Chen said. "Thus, Slide-tags generated data that are almost indistinguishable from sNucRNA-seq with a theoretical 3 micrometer spatial localization accuracy," they wrote.
But by tagging the nuclei, Chen suggested they've skirted the issue of resolution. "Part of the reason to have higher resolution in sequencing-based methods is to deal with the issue of cell mixing," he said. "Now that we have cells that are unmixed, we just want to know their neighbors and the arrangement of those cells."
And the data obtained are richer, Chen said, by an order of magnitude on a per-cell basis. "The lack of mixing between cells is really important for finding molecular subtypes in any tissue," he said. Slide-tags also enables multiomic analysis, such as ATAC-seq, T-cell receptor sequencing, and clonality of copy number variation in tumor cells.
In addition to technical advantages, the method could make spatial data more accessible, according to Anoja Perera, director of sequencing and discovery genomics at the Stowers Institute for Medical Research. Her lab has used Slide-seq and is working with tech from Curio Bioscience, a startup cofounded by Chen and Macosko to commercialize the technology, which "is a great option for labs without funding for capital equipment with costly assays," she said.
While obtaining the arrays for Slide-tags from a commercial source might increase costs, "given the type of data it can generate, this method would be completely worth it," she said. "Perhaps in the future, one of the novel tube-based approaches [to single-cell sequencing] can be used, making the entire process capital equipment-free."
Chen and Macosko have applied for a patent on Slide-tags. "We believe [Curio is] interested in offering Slide-tag arrays," Macosko added.
Most of the cost of doing Slide-tags comes from single-cell sample preparation and sequencing, which can still be thousands of dollars per sample. "The only added cost is the array," Macosko said. "In our labs we can fabricate those cheaply." He estimated that this adds about 10 percent to the cost of doing sNuc-seq with 10x's Chromium — one of the priciest methods of single-cell sequencing — although pricing for that assay can vary.
With single-nucleus isolation protocols in place for a variety of tissues and plenty of options for sequencing transcriptomes, Slide-tags could enable new types of studies. "I think early on, a lot of the direct applications will be in generating hypothesis from clinical research specimens," Chen said.
"You can now think about doing case-control studies for spatial," Macosko said. "Before it was too hard to optimize tissue; different samples can have different qualities, and it's hard to be systematic."
Perera noted that nuclei isolation is "nontrivial" and would take time for a given user to optimize that step for a given tissue. But she said she was "very excited" about the paper.
Chen and Macosko are clearly also excited about the future of this method. "We really think it's the future of single-cell [sequencing]," Macosko said. "It's the most exciting spatial tech we've been a part of developing."