Make-or-break prime editing for genome engineering in Streptococcus pneumoniae

Abstract

CRISPR-Cas9 has revolutionized genome engineering by allowing precise introductions of DNA double-strand breaks (DSBs). However, genome engineering in bacteria is still a complex, multi-step process requiring a donor DNA template for repair of DSBs. Prime editing circumvents this need as the repair template is indirectly provided within the prime editing guide RNA (pegRNA). Here, we developed make-or-break Prime Editing (mbPE) that allows for precise and effective genetic engineering in the opportunistic human pathogen Streptococcus pneumoniae. In contrast to traditional prime editing in which a nicking Cas9 is employed, mbPE harnesses wild type Cas9 in combination with a pegRNA that destroys the seed region or protospacer adjacent motif. Since most bacteria poorly perform template-independent end joining, correctly genome-edited clones are selectively enriched during mbPE. We show that mbPE is RecA-independent and can be used to introduce point mutations, deletions and targeted insertions, including protein tags such as a split luciferase, at selection efficiencies of over 93%. mbPE enables sequential genome editing, is scalable, and can be used to generate pools of mutants in a high-throughput manner. The mbPE system and pegRNA design guidelines described here will ameliorate future bacterial genome editing endeavors.

Fig. 1. Prime editing 2 (PE2) in S. pneumoniae.

A Schematic overview of PE2^S.pn. B Schematic overview of the general cloning strategy of pegRNAs (shown is the tRNA-asp-1 containing vector pVL4393 as backbone). C Editing strategy used to test efficiency of PE2^S.pn in pneumococcal strains expressing either a functional luciferase (VL4200) or a luc gene interrupted by a premature stop codon (VL4297). Single colonies are picked and grown in microtiter plates in the presence of luciferase substrate and bioluminescence is recorded. D Induction of PE2^S.pn by aTc in combination with expression of a pegRNA without 3′ leader, by processing of the tRNA-asp-1, leads to accurate gene editing in ~2% of S. pneumoniae cells. Induction of PE2^S.pn in combination with 3′ extended pegRNAs provides an editing efficiency of ~3%. Each dot represents a randomly picked clone grown in C + Y medium containing luciferin substrate. The maximal bioluminescence (relative light units; RLU) reached, normalized by the optical density (OD), of each clone is shown (AU, arbitrary units). RLU/OD values were normalized against the baseline level of medium luminescence obtained for each experiment. A typical experiment is shown (at least 3 biological replicates, with n = 90). E Schematic overview of pegRNAs containing the S. pneumoniae rpsT terminator (SpTer) or a stretch of 7 Ts. F Prime editing is slightly enhanced in the absence of mismatch repair (ΔhexA) in S. pneumoniae (n = 90). Clones with either an RLU/OD value below 2 × 10² (left), or an RLU/OD value above 7 × 10⁶ (right), were considered to be edited (as also validated by Sanger sequencing, Supplementary Fig. 1).

Fig. 2. Make-or-break prime editing selectively enriches correctly edited clones.

A Schematic overview of mbPE^S.pn. Binding of the spacer sequence to the target region leads to formation of DSBs by the Cas9 domain. The PBS can hybridize to its complementary sequence. The reverse transcriptase (RT) domain of mbPE^S.pn subsequently synthesizes cDNA using the RTT as template. B Schematic depicting the mechanism leading to selective survival of edited clones. The hybridization of the 3′-flap containing the edit is in equilibrium with the unedited strand. Repair with the edited region will prevent repeated DSBs by Cas9. C Induction of mbPE^S.pn by aTc in combination with expression of a structured pegRNA leads to accurate gene editing. ~93% of surviving S. pneumoniae cells after induction are accurately edited (n = 47 each). Cells were grown in C + Y medium containing luciferin substrate. The maximal bioluminescence reached of each clone is shown (RLU: relative light units. OD: optical density). RLU/OD values were normalized against the baseline level of medium luminescence obtained for each experiment. In these experiments, clones with either an RLU/OD value below 2 × 10³ (left), or an RLU/OD value above 2 × 10⁵ (right), were considered to be edited (as also validated by Sanger sequencing). D In the absence of reverse transcriptase (strain VL7598, P_tet-cas9), no induced clones were edited, as indicated by the lack of bioluminescence (n = 44). E mbPE^S.pn in recA-deficient strain shows that RecA-dependent homologous recombination is not required for editing (n = 42).

Fig. 3. mbPE^S.pn is highly scalable and efficient in making base pair substitutions.

A Cloning strategy to produce a library of pegRNAs targeting luc from a ssDNA oligo pool. The oligo pool was amplified by PCR as well as the backbone of pVL4134 that includes the luc spacer sequence and assembled by Golden Gate digestion/ligation followed by transformation to S. pneumoniae containing luc and mbPE2^S.pn. Chromosomal DNA was isolated from colonies grown in absence or presence of aTc (to induce mbPE2^S.pn) and PCR amplified using staggered primers followed by Illumina sequencing (see “Methods”). 2FAST2Q was used to extract and count pegRNAs and luc alleles. B Schematic overview of the pegRNA design encoding for different base pair substitutions. (C, D) Single base pair substitutions and multiple base pair substitutions are installed with high selection efficiency. For mutation frequencies, results of four biological replicates are plotted in box-plots, with first quartile, median, and third quartile indicated. Whiskers indicate the maximum value, at most 1.5x IQR from the first or third quartile. Outliers are plotted in red. For pegRNA frequencies, the mean of four experiments is plotted. C pegRNA abundance of - aTc samples (control, purple) and + aTc samples (orange) are shown. For reference, values of a successful edit (113 bp deletion) and unsuccessful edit (63 bp insertion) are also plotted. D The fraction of luc reads carrying the desired mutation over the total reads in that sample is shown for induced (aTc treated, orange) and non-induced (-aTc; control) samples on the left Y-axis. The average fraction of pegRNA reads for the substitutions is shown as dots on the right Y-axis. Red dots indicate outliers in induced samples. Non-induced samples only have data indicated as lines in double mutant “TCCTGAA” and triple mutant “TGCAGAA”, as only one non-zero datapoint was collected in those samples.

Fig. 4. mbPE for DNA insertions and genome deletions.

A, C Schematic overview of the approach used to test insertion and deletion efficiency in luc using mbPE. For pegRNAs encoding insertions, additional nucleotides were added between the RTT and PBS. For deletions, the RTT portion of the pegRNA was progressively moved within the luc gene, increasing deletion size, whilst the RTT remained constant. B, D Possibility of making insertions (B), and deletions (D) using mbPE. For each modification size, the fraction of luc reads carrying that mutation over the total reads in that sample, is displayed on the left axis with bars, whilst the fraction of pegRNA reads from uninduced samples carrying the mutation is displayed with dots mapping to the right axis. Resulting means of four biological replicates are plotted. E Verification of the ability to make deletions in luc, through targeted deletion of 10, 63, 127, or 184 bp (n = 45, n = 28, n = 28). Clones with an RLU/OD value below 10³ were considered to be edited (as also validated by Sanger sequencing). The percentages indicate the fraction of clones that were phenotypically observed to be edited. Note that for the 184 bp deletion, no clone showed correct editing by Sanger sequencing (*).

Fig. 5. pegRNA design rules for mbPE^S.pn.

A Schematic overview of the approach used to test mbPE^S.pn efficiency at inserting a single adenine in luc. B Combinations of RTT and PBS with statistically significant differences in pegRNA abundance between uninduced vs induced are outlined. Overrepresented combinations in induced samples show higher log2 fold changes (log₂FC) and are colored bright yellow. Most overrepresented pegRNAs of the induced library compared to uninduced suggest that a 16 or 17 nts for the PBS and 14 or 15 nts for the RTT yield the highest editing efficiency. Statistical significance was defined as an |log₂FC | > 1, and an adjusted p-value < 0.05. Statistical testing was performed using the Wald test provided by DESeq2, including correction for multiple testing. Results based on four biological replicates.

Fig. 6. mbPE can introduce edits at sites distal from the PAM.

A Schematic overview of the RTT design strategy used to edit bases distal from the PAM site. Every pegRNA introduces a single PAM mutation (G to C) and a second, purine to pyrimidine (or vice versa) transversion introduced at varying distances from the PAM sequence. B Fraction of correctly edited luc alleles and pegRNA read counts over the total reads in that sample of DNA extracted from cells grown in the absence (control) or presence of aTc. The average fraction of pegRNA reads for the substitutions is shown as dots on the right Y-axis. Edits are observed up to 91 bp from the PAM site. Resulting means of four biological replicates are plotted. C Validation of the maximum editing distance through individually cloned pegRNAs. The mutations conferred by the pegRNAs all encode the G to C PAM mutation and a second mutation at varying distances. The 33, and 45 bps substitutions from the PAM site, confer a premature stop codon, whilst the pegRNAs encoding mutation either 72 or 91 bps from the PAM site resulted in a glycine to cysteine (72 bps) or isoleucine to asparagine (91 bps) substitution (n = 21 each). The percentages indicate the fraction of clones that were phenotypically observed to be edited. Clones with an RLU/OD value below 3 × 10³ were considered to be edited. The mutation at 91 nts (*) could only be confirmed by sequencing individual clones, as the amino acid change does not affect luciferase function. This showed an editing efficiency of 4.8 % (1 out of 21 tested colonies) with the correct genome edit.

Fig. 7. PBP1a is in a complex with MpgA and RodZ and sequential genome editing using mbPE.

A Schematic overview of the principle of the split luc system. B Schematic overview of the pegRNA used to fuse a SmBit tag to pbp1A using mbPE. C PBP1a interacts with MpgA and RodZ. Each dot represents the average of 15 measurements of a technical replicate, with the size of the dot representing the SEM. Control strains only expressing pbp1A-SmBit (labeled as D39V) or pbp1A-SmBit together with the abundant DNA-binding protein HU-LgBit did not demonstrate bioluminescence. D PBP1a and MpgA co-localize in space and time. Live cell fluorescence microscopy of cells of strain VL7454 (mScarlet-PBP1a, GFP-MpgA) grown at 37 °C in C + Y medium. Scale bar: 2 mm. Demograph of exponentially growing cells of strains VL5532 (GFP-PBP1a) and VL5816 (GFP-MpgA). Representative data are shown from one biological replicate, data of at least 1,000 cells are plotted. E, F Sequential gene editing in S. pneumoniae using mbPE. Schematic overview of cloning regime with the two alternative pegRNA vector backbones (pVL4133, spec; pVL7227, cat) to generate a triple genome edited S. pneumoniae strain (Δ63-luc, Δcps2A-stop, ΔlytA-stop) in a sequential manner.