The error rate of a promising gene-editing tool can be reduced up to 60-fold by introducing mutations that change where the system’s enzyme cuts the DNA strand. The researchers say this ‘next generation’ prime editor could form the basis for a range of advanced tools and applications.
Based on Crispr, prime editing is one of the newest types of genome-editing tools and, in contrast to conventional gene editing, only cuts one strand of DNA, which should mean fewer unexpected deletions or insertions. However, a key remaining challenge with this approach is its high error rate. In this study, researchers from the Massachusetts Institute of Technology (MIT), US, set out to see if they could find a way to engineer the errors out of prime-editing systems.
Vikash Chauhan, a bioengineer at MIT and one of the researchers involved in the study, explains that in prime editing, only one strand of DNA is cut, and this is where the new genetic sequence is attached. ‘On the end of that cut DNA strand, the prime editor adds a new sequence [that] contains the genetic manipulation you want to make,’ he explains. This results in a ‘flap’ of DNA, where the new sequence has been added, but the original strand needs to be kicked out to finish the edit.
‘This is a competitive process, because both the new sequence and the original sequence have very similar DNA sequences, and … in a sense, the cell has a bias for keeping that original sequence,’ says Chauhan. ‘If the new sequence wins the competition, then you get the desired edit. If the original sequence wins suddenly you have this extra bit of DNA sitting around … and that’s what produces the errors.’
DNA scissors learn to relax
To find an effective solution, the team focused on the protein Cas9, the molecular scissors that cut the DNA. Previous work has shown that Cas9 preferentially cuts DNA at specific sites, which can contribute to the incidence of errors. So, the team looked to see if they could ‘relax’ the way Cas9 nicks the DNA and make a version of the protein that would cut the DNA at different points along the strand, which can reduce the stability of the original sequence and allows any edits to take better.
‘There’s a strong connection between mutations that relax the nick positioning and those that led to degradation of the original DNA strand that we wanted to get rid of,’ says Chauhan. ‘We believed that that would help change this competitive process between the new strand and the original strand by simply eliminating the competition.’
Once they had found the desired mutations, they started combining them in a mix and match approach to find a prime editor that had a very low error rate but still reasonable editing efficiency, ending up with a version of Cas9 that had four mutations in total. The last step was to drop the Cas9 into an advanced prime editor architecture.
‘The prime editor has several components strapped together into one large enzyme; the Cas9 component that makes the DNA cuts, reverse transcriptase that writes the new sequence onto the end and then added factors that help boost the efficiency of the process,’ says Chauhan. ‘We took our Cas9 component, dropped it into an architecture that had several of these efficiency boosting components, and what we ended up with was the prime editor we call vPE, and that turns out to be fairly efficient and robust and has very low error rates.’
They calculated that the maximum reduction in the error rate possible with this new prime editor was around a 60-fold reduction. Although Chauhan explains that this kind of improvement will only be achievable in ‘extreme’ cases in systems with a high error rate where everything possible has been done to ensure a good, efficient edit.
However, in more common approaches to prime editing, he says that the ratio of good edits to insertion-and-deletion errors, or indels, was found to improve from around 5 or 6:1 to just over 100:1 with vPE.
Chauhan says the team is currently working on building up the efficiency of vPE even further to make the process more robust, but says that the research community could start using it fairly soon.
Jin-Soo Kim, an expert in engineering biology at Korea Advanced Institute of Science and Technology in Daejeon, South Korea, describes the work as ‘very interesting’. ‘Two years ago, we reported that H840A Cas9, an essential component of prime editors, is not a bona fide nickase, often yielding [double-strand breaks] rather than [single-strand breaks]. As a result, prime editors often produce unwanted indels. Use of bona fide nickase in prime editing can minimise unwanted indels.’
‘I wonder whether the new mutations reported in the paper avoids the formation of [double-strand breaks] in addition to relaxing nick positioning,’ he adds. Kim notes that the approach was ‘straightforward’ and says he would be happy to use vPE in his own research.