Limits of bacterial prime editing

Prime editing is a technology for precisely modifying genes without creating DNA double-strand breaks. The researchers developed PE-STAR in Escherichia coli by combining DNA repair modulation with selection of edited cells, reporting editing efficiencies of about 80–90% under specific editing conditions. These figures represent the proportion carrying the desired edit in the population after selection.

In bacteria, enzymes that process editing intermediates and DNA repair pathways affect editing outcomes. Earlier approaches had low efficiency or restrictions on the length of sequences that could be modified. The team built PE-STAR using strain ES95, derived from E. coli MG1655 by deleting three exonuclease genes: sbcB, exoX, and xseA.

Reinforcing repair pathways and SOS-based selection

Removing the three enzymes can reduce degradation of the edited DNA strand produced by reverse transcription. The researchers also extended the culture period after transformation and overexpressed the RecJ enzyme. The authors proposed a model in which RecJ enhances processing of the competing, unedited strand in the 5′ direction, creating conditions that favor incorporation of the edited strand into the genome.

To increase the proportion of edited cells, the team also introduced an SOS circuit that responds to DNA damage. They linked PE3, which creates a single-strand break in each of the two DNA strands, to LexA-regulated guide RNA expression and designed the MAD7 nuclease to eliminate a counterselection plasmid carrying the gene for the toxin CcdB. Subsequent induction of toxin expression reduces unedited cells that retain the plasmid. The editing fraction increased by this process must not be interpreted as the survival probability of edited cells.

Efficiency varied by edit type, length, and position. A single-base insertion in xylB reached 82.41% under specific conditions, while 5 bp and 10 bp deletions reached about 90%. By contrast, efficiency for 10 bp insertions increased from 0.44% to 6.78%, and for 30 bp insertions from 11.76% to 27.40%. The system supported insertions, deletions, and replacements of up to 46 bp, but these did not all show the same efficiency. At galK and lacZ, improvements of up to 16-fold over the existing PE2 system were observed. Efficiency at xylB was assessed by sequencing the target region in colonies collected after selection, while efficiency at galK and lacZ was evaluated using colony color phenotypes.

Integrating large genetic pathways and remaining challenges

A separate recombination step was used to insert long DNA cassettes. First, PE-STAR installed a 46 bp attB recognition site in the chromosome. Bxb1 integrase was then used to integrate a 3.2 kb green fluorescent protein (GFP) cassette and an 8.0 kb riboflavin biosynthesis pathway in separate integrations. Correct integration was confirmed in all colonies examined after selection. This 100% figure does not represent the success rate among all input cells or across the entire two-step process.

In the strain carrying the riboflavin pathway, the riboflavin concentration in the culture supernatant reached a maximum of 3.23 mg/L during 96 hours of cultivation. However, optical density at 600 nm (OD600), a measure of cell density, was lower in the strain with the integrated pathway, at 2.72 compared with 3.27 in the control, indicating a growth burden. These results provide laboratory validation linking precise editing with installation of a metabolic pathway in E. coli. Performance in other bacteria and productivity at industrial scale remain to be validated.

The methods and results are available in the open full text of the Nucleic Acids Research paper. DOI: 10.1093/nar/gkag285.