When a chef develops a new recipe, they methodically add and remove individual ingredients to see how each of them alters the final dish. When scientists try to understand the role of genes in the body, they employ a similar tactic using genome editing. Currently, the most popular tool in their toolbox is CRISPR, with applications ranging from cancer therapeutics to treatments for genetic diseases like sickle cell anemia and β-thalassemia.1 However, this genome editing staple still has its limitations.
“It’s really hard to pack the genes encoding these proteins into the viruses that are used for their delivery into the cells,” said Tautvydas Karvelis, a genome biologist at Vilnius University. Even when CRISPR nucleases are directly delivered into cells, their large protein sizes present limitations. For example, the commonly used Cas9 is about 1,400 amino acid residues long.
In a recent study published in Nature Methods, Gerald Schwank, a genome biologist at the University of Zurich, and his team described a tiny, but efficient, nuclease that works as well as some of the current Cas proteins but is less than half their size.2
“It’s like a new class of tools that can be used for genome editing, not just as a principle,” said Karvelis, who was not involved in the study.
In 2021, Karvelis discovered a compact RNA-guided protein capable of cutting DNA: TnpB.3 Compared to other CRISPR nucleases, TnpB is much smaller with approximately 400 amino acid residues. However, TnpB has a lower editing efficiency and limited target range.
So, Schwank set out to improve TnpB’s performance. He and his team optimized TnpB for mammalian gene editing and engineered variants of the proteins to improve the target range. They also built a machine learning model to predict how well a guide RNA will perform for a set of target sequences, saving future users the headache of multiple trials.
In its unaltered state, the editing efficiency of TnpB is between zero and 20 percent, which is lower than that of the smallest CRISPR-Cas9 ortholog, CjCas9. “We asked if we can really make TnpB efficient enough to use it,” Schwank said. So, the team did two things. They optimized TnpB’s codon sequence for mammalian cells and attached a small tag to the protein that guided it to the nucleus. The team observed a 4.4-fold increase in the editing efficiency of this modified nuclease, surpassing most of the commonly used RNA-guided endonucleases, including CjCas9. They called this variant TnpBmax.
When tested on a large library of target sites inserted into mammalian cells, TnpBmax performed remarkably, causing insertions and deletions of DNA sequences at a rate of approximately 70 percent. But any change to an essential five base region upstream of the DNA target caused the efficiency to drop dramatically. This short region with the sequence 5’ TTGAT 3’, called the transposon adjacent motif (TAM), occurs infrequently and constrains where TnpBmax can work its magic.
To relax these limits, Schwank and the team tested the interactions between different versions of TnpBmax and TAM variants. Replacing lysine with alanine at the 76th position allowed TnpBmax to recognize TAMs that had cytosine or thymine at the second position and guanine or thymine at the third position. This sequence occurs four more times in the genome than the original TAM. The changes to TnpBmax or TAM did not affect editing efficiency.
Schwank wanted to include an additional feature to the tool. He and his team built a model that can predict whether a guide RNA sequence can edit a given DNA sequence with an efficiency of at least 70 percent. “Before, it was more or less a gamble. Having a TAM site meant that, in principle, the site should be targetable, but one didn’t really know if there will be substantial efficiency,” Schwank said. “Now, with this model, it makes it a lot easier to figure out whether you can use the tool or not.”
“Having this model so that people can design their own guides will help with the usability of the system,” said Omar Abudayyeh, a genome engineer at Harvard Medical School who was not involved in the study.
Finally, to put the tool to the ultimate test, the authors targeted genes in the brain and liver of mice. They observed editing efficiencies of 65 percent and 75 percent, respectively. In the liver, the team edited a gene involved in cholesterol metabolism and observed a consequent decrease in the levels of blood cholesterol in the mice.
“At the time of their discovery, there was a big question on how effective TnpBs would be as genome editing enzymes,” said Jonathan Gootenberg, a genome engineer at Harvard Medical School who was not involved in the study. “With this engineering, they really are competitive.”
