Spatially resolved transcriptomics, which Nature Methods named as its Method of the Year for 2020, allows researchers to look at transcriptional activity while preserving the spatial relationship of different transcripts.
In its editorial announcing the decision, the Nature Methods editors wrote that "this maintenance of spatial context is crucial for understanding key aspects of cell biology, developmental biology, neurobiology, tumor biology and more, as specialized cell types and their specific organization are crucially tied to biological activity and remain poorly explored on the scale of whole tissues and organisms."
At the subcellular level, too, spatial information is critical. In a neuron, for example, whether a given protein is active at a synapse or in the cell body can matter a great deal.
There are several different methods for tagging and sequencing transcripts in situ. In the January 29, 2021, issue of Science, investigators at the Massachusetts Institute of Technology's McGovern Institute and Harvard University's Wyss Institute for Biologically Inspired Engineering reported an addition to the technology arsenal with expansion sequencing.
Expansion sequencing combines two methods developed in the laboratories of corresponding authors Edward Boyden and George Church, respectively.
"Our core contribution is to make the process of spatial transcriptomics as precise as possible," Boyden, who is an investigator at the McGovern Institute and the Howard Hughes Medical Institute, told BioWorld.
The method's key difference, compared to other methods that have been developed for in situ transcriptomics, is that the tissues being studied are magnified physically, not just via imaging.
Magnifying glass... of water
In expansion microscopy, one of expansion sequencing's precursor technologies, tissues are treated with water-absorbent polymers. In previous work, Boyden and his team demonstrated that these polymers could swell tissues, and that swelling occurred evenly in all directions, preserving the 3 dimensional organization of tissues. Boyden and his colleagues expanded tissues by up to 100-fold, a magnification that enabled light microscopes to see structures that would otherwise be invisible to them.
In the work now published in Science, the authors combined up to 30-fold expansion with fluorescent in situ sequencing, developed in Church's lab.
Using neurons and tumor biopsies as examples, they showed that expansion sequencing could be used to look at both prespecified sets of transcripts, and untargeted transcripts.
The team was able to gain new insights into the biology of both tissue types.
In neurons, they identified transcription factors outside of the nucleus in the dendrites, which make synaptic connections with other neurons.
"A neuron has one nucleus, but thousands of connections," Boyden said, and the discovery of transcription factors in some of those connecting structures suggests that they are able to operate more independently of the nucleus than has previously been realized.
They also compared different cell types and found similar distributions of transcripts, suggesting there might be general rules governing the transport of RNA transcripts.
In tumor biopsies of metastatic breast cancer, they discovered that "immune cells change their gene expression based on how close or far away from other cell types they are," Boyden said.
B cells, for example, expressed up to four times as much higher levels of s100A8, which regulates inflammation and immune responses, when they were near tumor cells that expressed epidermal growth factor receptor (EGFR).
Boyden said that the team is currently working to make the technology accessible, and useful, to the largest possible number of researchers.
To that end, the team is in the process of writing a more detailed paper describing the expansion sequencing protocols, and "still trying to figure out how to help different end users use it," "we are part of lots of different groups," he said, including Cancer Research UK and the Chan Zuckerberg Initiative Human Cell Atlas pilot program, to understand what researchers want spatial transcriptomics for. "I really believe in technologies that everybody can use."
Boyden said that another important way to expand the method's uses is to develop good machine learning and AI techniques to work with the terabyte-size datasets generated by spatial transcriptomics.
Boyden said that currently, machine learning approaches are used to "pre-process datasets -- 'here are these seven features I want you to pay attention to.' So if you see a pattern, that was us putting forth hypotheses that we can test."
Ultimately, a more powerful approach would be to develop algorithms that, like the sequencing itself, could be either targeted or agnostic. "That's where the wild frontier is," Boyden said.