Skip to main content
Premium Trial:

Request an Annual Quote

CRISPR/Cas9 Deployed as Cellular Historian


NEW YORK (GenomeWeb) – Scientists are using CRISPR/Cas9 genome editing to record a cell's history in its own DNA, answering questions such as where the cell came from in development, also known as cell lineage tracing, and what biochemical events it has encountered during its lifespan.

To trace cell lineage in early vertebrate development, Jay Shendure of the University of Washington and Alexander Schier of Harvard University led a team that created arrays of synthetic targets in the genome that could pick up different edits as the cells underwent ongoing rounds of division.

They published their technique, called genome editing of synthetic target arrays for lineage tracing (GESTALT), as well as a finding that cells in mature zebrafish derived from very few progenitor cells, in May in Science

Also in May, teams from Timothy Lu's lab at the Massachusetts Institute of Technology and George Church's lab at Harvard posted separate papers to the preprint server BioRxiv on the idea of a self-targeting guide RNA. Lu's team suggested using it as a memory device for cellular events, such as exposure to inflammation or transcription, while Church's team used it as an evolving genomic barcode that could be used in lineage tracing.

Earlier this month, Lu published his team's study in Science.

Lu told GenomeWeb that all the ideas are complementary to each other. "They highlight the utility of getting cells to remember what they've seen and tell you at a later time," he said.

Shendure agreed, adding, "You can imagine tying other sorts of events in a cell to editing as a means of recording when certain things were happening."

For Shendure and Schier, the goal was to address the longstanding challenge of lineage tracing in complex organisms.

"Even though it's been of great interest to the field, dissecting a lineage has been challenging," Shendure said. For simpler model organisms like Caenorhabditis elegans, which has about 1,000 cells, the rounds of division are always the same and the lineage map was figured out decades ago. "For every other organism, generation of comprehensive maps of cell lineage has been an elusive goal."

Shendure pointed to the field of hematology as an example of how difficult it is to determine which cell types are more closely related to others. "Despite decades of study there's still controversy," he said.

Shendure first approached the problem in graduate school, where he studied under George Church at Harvard. It was a similar idea to his recent study, using a genomic array of site-specific recombinases to encode information in a compact sequence. But that was more than a decade ago, pre-CRISPR genome editing, and he didn't spend a lot of time on it.

Fast-forward to now and his lab, working with Schier, developed a way to use CRISPR to make mutations in pre-defined, synthetic sites in the genome. GESTALT introduces an array of target sites into the genome and the cell is injected with a CRISPR system that can target those sites. The editing events can be read through sequencing both DNA and RNA.

If the CRISPR editing is induced in early development, certain editing events and associated mutation records only happen in some cells, creating a new kind of phylogenetic tree.

Instead of being a history of how organisms are related to each other through evolution, it's a map of how cells are related to each other. "Ideally it would edit once per cell cycle, but we would need a much larger array of target sites to pull that off," Shendure said. "And ideally we would be capturing the full span of development rather than early stages."

The fact that it allows scientists to look at development in vivo is a major advantage, Shendure said.

After proving the concept in cell lines, his team applied it to zebrafish zygotes and what they saw surprised them. For many organs and tissues, the cells could trace their lineage to a very small subset of progenitor cells. "For blood, five alleles were in 98 percent of blood," he said. While this was primarily a technology development paper, that finding suggests this method could yield interesting results. Whether there's an inflationary period in development or cell remodeling remains to be seen, but somehow a small subset of cells takes over, Shendure said.

Tim Lu said that he thought about applying his self-targeting guide RNAs to the idea of lineage tracing, but instead focused on advancing his lab's ideas of using DNA as a memory storage device.

"What we did was figure out that if you put a protospacer-adjacent motif domain in the gRNA sequence it should recognize its own DNA locus and target for cleavage," Lu explained. That creates mutations in the gRNA's target sequence that serve as a record of editing events. "The mutations are not always the same, but it allows us to encode memory."

"This is the first prototype," he said. "There are tons of possible improvements we outline, but it shows the concept is a powerful way of having cells record what they encounter and report on it. We can now have studies of cell in a natural environment we want to study, for example, how much of a given inflammatory signal they sense. Anything you can couple into expression or activity of gRNA and Cas9 should be recordable." He said that events linked to transcription should be especially easy to record. "The hard part is to link editing to the level of a protein, not because it's not possible, but sensors for such a thing have not been robustly developed," he said.

Like GESTALT, the self-targeting gRNAs can be read out with sequencing. While there are existing systems like Cre-lox that can be coupled to sense and record events in a cell's genome, they're mostly one-off, binary reporters. Lu's CRISPR system, however, can convey the magnitude of what the cell is seeing and can be easily multiplexed by simply adding more gRNAs. 

And like GESTALT, Lu's technology has major advantages for in vivo studies. Fluorescent reporters can also convey information about a cell, but they're not as useful in vivo.

"There's no sense in using our system in a dish, because then you could use imaging and that's probably much easier to do," Lu said. "But if you really want to understand what a cell is doing, say, in the liver in an animal, it's hard for imaging to do that."

Lu and his co-authors noted in their Science paper that there's lots of room for improvement. The mutations made are random so they don't provide any additional information. "If you can make it so every time you express the gRNA, a defined mutation gets picked up, then you could definitely say how much a particular event gets sensed," Lu said. Another issue is that deletions are more common than insertions, so the target sequence of the gRNA actually gets shorter over time, to the point where it can stop working. Newer versions of the system have target sequences of 30 bases or more to account for shortening.

If paired with single-cell genomics, Lu said his concept could help enable powerful new studies.

"If you could barcode each cell that has a record of its history and [perform] single-cell transcriptome profiling, you could potentially get a genome-wide map of every cell," he said. Moreover, one could link that to what each cell experienced in its lifetime.

He mused that a study could look at two tumor cells in a scenario where each one is exposed to a different amount of the same drug. "You might want to understand how that relates to heterogeneity of cancer response," Lu said. "If you have a memory device that tells you, 'This cell got 2 micrograms versus 20,' then you could do single-cell omics to see what the resulting effects on the transcriptome or metabolome were."

Lu added that his lab primarily uses next-generation sequencing to grab the information from the cells. "The sequence we're reading is actually quite short so we would prefer to have a lot of reads over a short distance, rather than long reads," he said.

Church's paper has not been published in a peer-reviewed journal, but it combined elements of both GESTALT and Lu's self-targeting gRNAs. Dubbed homing guide RNA (hgRNA), Church's team pointed to lineage tracing as an important application. They also said it could be used for "cellular barcoding, molecular recording, dissecting cancer biology, and connectome mapping."