In October 2018 we published a blog (CRISPR Base Editors: An Upgrade for Treatment of Genetic Disease) describing some of the concerns surrounding the use of the CRISPR technology, and highlighting a relatively new version called CRISPR Base Editor(BE) that seeks to avoid some of those pitfalls. While development of a slew of CRISPR-derived technologies continues unabated, the recent October 2019 Nature publication by Anzalone and team discussing a new version of CRISPR which potentially avoids even more unwanted byproducts while also offering a much wider toolkit for DNA editing, caught our eye.
Factors affecting the current CRISPR technology
- CRISPR versions that use fully functional Cas9 create double stranded breaks (DSBs) in DNA. These versions often rely on endogenous DSB repair systems using homology directed repair (HDR) to incorporate changes in target DNA which very often leads to mixtures of unwanted insertions and deletions (indels). The action of endogenous non-homologous end joining (NHEJ) based repair also contributes to indel formation in these cases.
- Current versions of CRISPR base editors do not require DSBs and are able to efficiently install the four transition mutations (C->T, T->C, G->A, and A->G). However, the current base editors are not capable of the eight transversion mutations (C->G, C->A, G->C, G->T, A->C, A->T, T->A, and A->T).
- No DSB-free CRISPR-based system had previously been described that allows for targeted insertion or deletion of single or multiple bases.
Development of Prime Editors
Anzalone and team (2019) have leveraged several ingenious molecular strategies to attempt to create a novel version of CRISPR that avoids the aforementioned pitfalls. These changes involved modifications to the Cas9 molecule to add, remove, and optimize specific functionalities. It also involved changes to the guide RNA (gRNA) that targets Cas9 to specific DNA sequences. Multiple versions of the new Prime Editor (PE) CRISPR-derived system were created iteratively to enhance its capabilities and each version was tested before further improvements were attempted. Firstly, the Cas9 capability was augmented by adding the ability to reverse transcribe additional RNA sequence from the gRNA into the target sequence. Secondly, the Cas9 fusion protein was recoded to enhance its biochemical activities. Thirdly, additional gRNAs were added to increase editing efficiency and specificity.
Image credit: Anzalone et al., (2019)
Prime Editor 1
The initial version called Prime Editor 1(PE1) contains wild-type Moloney murine leukemia virus reverse transcriptase (M-MLV RT) fused to the C-terminus of Cas9 H840A nickase (a Cas9 where one of two nuclease activities is non-functional). This is paired with an engineered gRNA called a primed editing guide RNA (pegRNA). The pegRNA has the established gRNA sequence to direct where CRISPR binds and is extended at its 3’ end to include a primer binding site (PBS) for the RT to initiate from, and an RT template stretch with desired edits encoded for the changes required in the target DNA. When PE1 functions as designed, the complex binds target DNA and nicks the PAM (protospacer-adjacent motif)-containing strand. The 3’ extension of the pegRNA then binds the PBS in the nicked strand and allows RT to add the additional DNA sequence containing the desired edits. Those edits can include any of the 12 single-base changes, small to moderate insertions or deletions and even combinations of all these types of changes. Following PE1 nicking, RT action, and dissociation of the complex we are left with a branched DNA intermediate with a 3’ flap containing the edited/reverse transcribed sequence and a 5’ flap with the original sequence. The flaps are redundant and thermodynamics would favor hybridization of the 5’ flap. However, 5’ flaps are preferred substrates for certain endonucleases involved in lagging-strand DNA synthesis. Equilibration between the flaps and action of endonuclease can result in the flap with the edited sequence now in hybrid. Once this remaining flap is ligated, the edits can become permanent following endogenous DNA repair.
Image credit: Anzalone et al., (2019)
Prime Editor 2
In an effort to improve the efficiency of PE1, the authors screened a series of known mutants in M-MLV RT that can increase thermostability, processivity, DNA:RNA substrate binding, or inactivate RNase H activity. A total of 19 PE1 variants containing these mutations in M-MLV RT singly or in groups were tested. A version containing 5 individual mutations combined into one molecule was selected and termed PE2. PE2 is up to 5-fold improved over PE1 in editing point mutations and is significantly more efficient in directing insertions and deletions as well.
Prime Editor 3 and 3b
The resolution of mismatched heteroduplex after editing by repair determines whether the edits become permanent in the target DNA. To enhance the editing efficiency Anzalone and team (2019) added a second, standard sgRNA that would result in direct nicking of the non-edited strand when paired with Cas9 H840A nickase. Incorporation of this second sgRNA into PE2 was designated PE3. PE3 generally was more efficient in editing by several fold though depending on the location of the second sgRNA higher levels of indel formation could be triggered presumably by DSB induction of NHEJ repair. The authors cleverly got around this by designing the second sgRNA to only recognize the edited strand but not the original allele. This ensures that the second nicking event takes place after editing of the original target and this version was designated PE3b.
Comparisons and conclusions
Prime editing was tested in various mouse and human cell lines and was found capable of generating correct edits in all cases though editing efficiency varied. PE was compared with several BE versions – which was the focus of our original blog post – and was found to be somewhat less efficient in conversion of single target bases within the base editing windows of BE. However, PE led to less bystander editing of non-target bases. Off-target editing by PE was also investigated at known sites of off-target action by Cas9 located in multiple different genes. Incidences of unwanted off-target editing were found to generally be multiple-fold lower compared to standard Cas9 at known off-target sites, and on-par with or better than BE at those same sites.
In summary, the new PE system shows great promise in improving the future of CRISPR directed gene-editing. It will offer capabilities complementary to the BE systems, likely functions better in many circumstances, and significantly improves efficiency and specificity compared to cases of Cas9 with exogenous templates to direct HDR. Although these are very promising developments, testing a broader range of cell lines, as well as in vivo testing will be necessary before these advantages will have been fully established. Since the Cas9 H840A-M-MLV RT fusion product is relatively large some further consideration has to be given with regards to delivery via viral vectors due to size limitations (Yang et al., 2019).
Anzalone et al. Search-and-replace genome editing without double-strand breaks or donor DNA. (2019) Nature, Dec;576(7785):149-157.
Yang et al. One Prime for All Editing. (2019) Cell, Dec 12;179(7):1448-1450.