Factors affecting prime editing efficiencies

Like Comment
Read the paper

Background: Prime editors

 First described less than a year ago, prime editor (PE) is one of the newest gene editing technologies. PE, which consists of a fusion between a reverse transcriptase domain (RT) and a Cas9-nickase domain (Cas9-H840A), forms a complex with a prime editing guide RNA (pegRNA) to edit at a desired site. The pegRNA contains a spacer that searches for a specific target, a guide RNA scaffold sequence that binds to the Cas9 domain, and an additional 3’-extension sequence that includes a reverse transcription template (RT template) with the desired edit and a primer binding site (PBS). When the pegRNA-PE complex binds to the target site, a cDNA complementary to the RT template, and incorporating the desired edit, is synthesized, the edited strand is switched with the original target sequence, and the unedited strand of DNA is repaired to contain the desired edit.

High-throughput experiment

 The prime editing efficiency can vary a great deal depending on the design of the pegRNA. As the initial PE study showed, the length of the PBS and RT template typically influence efficiency. The authors of this original study recommended general rules for designing pegRNAs on the basis of their data derived from 11 target sequences. However, we believed that a more accurate understanding of factors affecting prime editing efficiencies would require a much larger scale analysis involving many more pegRNAs and target sequences. Our research team undertook a high-throughput (HT) experiment using a library of 54,836 pairs of pegRNAs and target sequences. Prime editing efficiencies at these targets were examined in HEK293T cells.

 These HT data allowed us to improve our understanding of the factors that influence prime editing efficiency. Generally, pegRNAs having a 11- to 13-nt PBS and a 10- to 12-nt RT template showed high prime editing efficiencies, which is in line with recommendations from the initial study. In addition, editing the protospacer adjacent motif (PAM) shows higher efficiency. However, new information also emerged from our study. For example, we compared the average editing efficiencies of pegRNAs with each combination of PBS and RT template lengths when the most efficient pegRNA at each target was selected. Surprisingly, the average editing efficiencies under these optimal combinations of PBS and RT template lengths were the highest when the lengths of the PBS and RT template were short (for example, a 7-nt PBS and a 10- to 12-nt RT template) and decreased as the PBS and RT template lengths increased (Fig. 2d). In addition, the paper that first reported prime editors stated that having a G as the last nucleotide of the RT template would reduce the prime editing efficiency, but our HT experiment showed no such tendency. Rather, certain RT template-PBS combinations showed higher editing efficiencies when the last templated nucleotide was a G. More details about many other factors that influence prime editing can be found in our Nature Biotechnology paper.

DeepPE, PE_type, and PE_position: Computational models for predicting prime editing efficiencies

Using the dataset from our HT experiment, we developed a prime editing efficiency prediction model for pegRNAs named ‘DeepPE’. DeepPE searches an input sequence for a position that contains an appropriate PAM, and then calculates and displays a prime editing score for 24 PBS-RT template combinations for each pegRNA. Furthermore, we developed two more computational models that predict the efficiencies of one version of PE, PE2, for various other types of editing, such as insertion and deletion (PE type), and positions (PE position) than were evaluated above.

 Through DeepPE, we can predict pegRNA activity without doing a wet experiment. Our research team is currently working on additional experiments using this exciting genome editing tool. We hope that our study will contribute to a better understanding of prime editing, ultimately making this technology more efficient.

Article: https://doi.org/10.1038/s41587-020-0677-y.

DeepPE webpage: http://deepcrispr.info/DeepPE/


  1. Anzalone, A. V. et al. Search-and-replace genome editing without double-strand breaks or donor DNA. Nature 576, 149–157 (2019).

Goosang Yu

PhD course student, Yonsei Univ. College of Medicine