NEW YORK – David Liu and his team at the Broad Institute have developed a new genome editing technique called prime editing, which they say is capable of correcting about 89 percent of all known human pathogenic variants.
Prime editing — which Liu debuted at the Cold Spring Harbor Laboratory Genome Engineering meeting last week and which first author Andrew Anzalone and his colleagues described in a paper published in Nature today — works by directly writing new genetic information into a specified DNA site using a catalytically impaired Cas9 fused to an engineered reverse transcriptase (RT) and programmed with a prime editing guide RNA (pegRNA). The RNA specifies the target site and encodes the desired edit.
"We performed [more than] 175 edits in human cells including targeted insertions, deletions, and all 12 types of point mutations without requiring double-strand breaks or donor DNA templates," the authors wrote. "We applied prime editing in human cells to correct efficiently and with few byproducts the primary genetic causes of sickle cell disease (requiring a transversion in HBB) and Tay-Sachs disease (requiring a deletion in HEXA), to install a protective transversion in PRNP, and to precisely insert various tags and epitopes into target loci."
They further noted that prime editing was supported with varying efficiency in four human cell lines, as well as primary post-mitotic mouse cortical neurons, and that the technique is more efficient with fewer byproducts than homology directed repair (HDR).
"The goal of prime editing was really to develop a new approach to genome editing that could address as many of the different flavors of mutation as possible, meaning to install or correct them in human cells," Liu said in an interview. "Currently, there are more than 75,000 known human DNA changes associated with disease … but this diversity of ways that DNA changes can cause disease really contrasts with the fact that prior to this study there were only two classes of commonly used programmable genome editing agents that work robustly in mammalian cells. The first is the programmable nucleases that includes zinc finger nucleases, talens, CRISPR-Cas9 and other agents. The second class that is widely used in mammalian cell are base editors, which we developed in 2016."
The first class of programmable genome editors have often been likened to scissors because they make double-stranded cuts in DNA. But the most common result of cutting DNA in this way is a mixture of deletions and insertions that cannot be controlled at the cut site, he added. So, while making double-stranded cuts in DNA can be very useful for disrupting genes and removing large segments of DNA, it has proven difficult in most cell types to make precise DNA changes in this way.
In contrast, Liu said, base editors have been likened to pencils — they use the targeting ability of CRISPR, but they work by directly converting one DNA letter to another without making double-stranded breaks. The result is a genome editing technology that works efficiently to correct four of the most common kinds of mutations with few byproducts. But they can't correct the other eight possible types of point mutations and they can't make insertions or deletions.
"The development of prime editing was really designed to be a highly versatile new approach to editing mammalian cells that doesn't make double-stranded breaks," Liu explained. "So, if CRISPR-Cas9 and other programmable nucleases are like scissors and base editors are like pencils, then one can think of prime editors like word processors capable of searching for target DNA sequences and precisely replacing them with edited DNA sequences without major restrictions on either the original or the edited DNA sequences."
The researchers originally hypothesized that an engineered guide RNA could both specify the DNA target and contain new genetic information that would replace target DNA nucleotides. The first version of the prime editor, PE1, contained an RT fused to an RNA-programmable nickase and a pegRNA. It worked by nicking genomic DNA at the target site to expose a 3'-hydroxyl group, to prime the reverse transcription of an edit-encoding extension on the pegRNA directly into the target site.
This resulted in a branched intermediate with a 5' flap that contained the unedited DNA sequence, and a 3' flap that contained the edited sequence copied from the pegRNA. The researchers reasoned that preferential 5' flap excision and 3' flap ligation could drive the incorporation of the edited DNA strand, creating heteroduplex DNA containing one edited strand and one unedited strand. In order to resolve the heteroduplex and permanently install the edit in the target locus, the researchers found that nicking the non-edited DNA strand would bias DNA repair to preferentially replace the non-edited strand.
They then developed a second version of the prime editor, PE2, using an engineered RT to increase editing efficiencies. The third version, PE3, nicks the non-edited DNA strand to induce its replacement and further increase editing efficiency. Typically, in HEK293T cells, the researchers were able to induce editing with 20 percent to 50 percent efficiency with 1 percent to 10 percent indel formation using PE3.
In testing the PE3 version, the researchers targeted five genomic sites in HEK293T cells using sgRNAs that induced nicks 14 to 116 nucleotides away from the site of the pegRNA-induced nick. In four of the five sites, they found that nicking the non-edited strand increased editing efficiency by 1.5- to 4.2-fold compared to PE2.
In order to minimize the formation of indels and double-strand breaks, they designed sgRNAs with spacers that match the edited strand, but not the original allele. Using this version of the editor, which they termed PE3b, the researchers found that mismatches between the spacer and the unedited allele disfavored gRNA nicking until after editing of the PAM strand. This resulted in thirteenfold lower average indels compared to PE3, without any evident decrease in editing efficiency.
"To demonstrate the targeting scope and versatility of prime editing with PE3, we performed all 24 possible single-nucleotide substitutions," the authors wrote. "These 24 edits collectively cover all 12 possible transition and transversion mutations, and proceeded with editing efficiencies (containing no indels) averaging 33±7.9 percent, with 7.5±1.8 percent average indels."
Importantly, they added, long-distance RT templates were also capable of efficient prime editing. Using PE3 with a 34-nucleotide RT template, they installed point mutations at positions +12, +14, +17, +20, +23, +24, +26, +30, and +33 in the HEK3 locus. Other RT templates that were 30 nucleotides or more in length also supported prime editing, suggesting that prime editing is not substantially constrained by the availability of a nearby PAM sequence.
In his presentation at the CSHL conference, Liu also emphasized the high specificity of prime editing compared to other editing techniques. Prime editing requires three separate DNA hybridization events to work — first, the target site guides RNA binding, then the primer binding site in the pegRNA hybridizes with the nicked target DNA strand, and then they require a third DNA hybridization event between the three prime flap containing the edit and the target DNA. If any of those three base-pairing events occur, then prime editing won't proceed to completion. Because of this, the researchers hypothesized that prime editing would create a lower number of off-target events, and they indeed found that editing with PE2 and PE3 was more specific than editing with Cas9.
"Since prime editing is a new method, there is no genome-wide, unbiased prime editing off-target detection method," Liu told GenomeWeb. "So, instead, we looked at known Cas9 off-target sites, and what we found is that prime editors seem to induce far fewer off-target edits than Cas9 at known Cas9 off-targets sites. We think this higher parent DNA specificity arises from the fact that prime editing requires three separate DNA base pairing steps in order to make a productive edit, whereas traditional CRISPR and other genome editing methods require only one such event, namely the DNA base-pairing between the target site and the guide RNA."
Collectively, the researchers used prime editing to perform 19 insertions up to 44 base pairs in length, 23 deletions up to 80 bp, 119 point mutations including 83 transversions, and 18 combination edits at 12 endogenous loci in the human and mouse genomes at locations ranging from 3 bp upstream to 29 bp downstream of a PAM without making explicit double-stranded breaks.
During a press briefing before the release of the study, Liu noted that it is likely prime editors are capable of insertions and deletions that are larger than the ones demonstrated in the paper. However, he added, the researchers suspect that attempting super long insertions and deletions would require a larger pegRNA, which may be too big to fold correctly. They are currently researching this question.
"These results establish prime editing as a remarkably versatile genome editing method," the authors concluded. "Because 85 percent to 99 percent of insertions, deletions, indels, and duplications in ClinVar are 30 bp [or less], in principle prime editing can correct up to about 89 percent of the 75,122 pathogenic human genetic variants in ClinVar."
Importantly, Liu said in the interview, prime editing offers a complement to base editing and nuclease-based editing. Each technique has strengths and weaknesses because they all operate by different mechanisms. In a direct comparison between prime editors and best editors, the researchers found that current base editors often edit with higher efficiency and fewer byproducts than prime editors while prime editors can target more flexibly without significant PAM restriction, and they also offer greater precision.
"You can pick out exactly which base or bases you want to edit [with prime editors], even if that base is buried in a sea of similar bases," Liu added.
Prime editing may also have uses beyond pathogenic variant correction. Because of its ability to add significant stretches of DNA at targeted sites in human or other mammalian cells, the technology could have potential as a base for diagnostics and could be useful in agriculture and as a research tool.
"There really isn't a way to efficiently and without an excess of byproducts make most of the kinds of precise changes one might want to make," Liu said. "So, simply just to test the effects of a mutation or a specific gene, prime editing offers some new and potentially quite useful capabilities compared to the previously available methods."
The researchers are continuing to advance the properties of prime editors themselves, including doing work on developing a fourth version of the technology, PE4. They are already in discussions with other researchers who are looking to use PE4 in screens, Liu said. He also noted that he and his colleagues are working to apply prime editors in animals and plants in order to test some of the technology's most promising applications, as well as experimenting with different ways to deliver prime editors into cells.
The Broad said it is making prime editing tools available for non-exclusive licensing for research and manufacturing by companies, and for the commercial development of tools and reagents. For human therapeutic use, the institute has licensed the technology to Prime Medicine, and Prime has awarded a sublicense to Beam Therapeutics for the use of prime editing in certain fields and for certain applications. Liu is a consultant and co-founder of both companies.