Changing pegRNA sequences to reduce self-binding may improve prime editing efficiency

Restriction enzymes (BpiI (#ER1011) and BsmBI (#ER0451)), T4 ligase (#EL0012), Transcript Aid T7 High Yield Transcription Kit (#K0441), Dulbecco’s modified Eagle’s medium (DMEM) (#52100–047), Fetal Bovine Serum (#10 500 064), Turbofect Transfection Reagent (#R0532), Penicillin/Streptomycin (#15140–122), TrypLE Express Enzyme (#12 604 021), Geltrex LDEV-Free Reduced Growth Factor Basement Membrane Matrix (#A1413302) and Qubit dsDNA HS Assay kit (#Q32854↗) were purchased from Thermo Fischer Scientific. mTeSR1 media (#85 850) and Accutase (#07 920) were from STEMCELL Technologies. DNA oligonucleotides and the GenElute HP Plasmid Miniprep kit (# NA0160-1KT) used in plasmid purifications were acquired from Sigma-Aldrich. Q5 High-Fidelity DNA Polymerase (#M0491L), NEB Stable Competent E. coli (# C3040H) were from New England Biolabs Inc. NucleoSpin Gel and PCR Clean-up kit (#740609.250) was purchased from Macherey-Nagel. RNA Clean & Concentrator Kit (#R1018) was acquired from Zymo Research.

The PEAR fluorescent reporter target plasmid with GFP (pDAS12125_PEAR-GFP, Addgene #177 178) and the pegRNA cloning plasmids (pDAS12222-U6-pegRNA-BFP, Addgene #177 181 and pDAS12069-U6-pegRNA-mCherry), were developed by Simon et al. [22]. The mScarlet fluorescent reporter plasmid (pAT9752-BEAR-mScarlet, Addgene #162 991) and the 2^nd nicking sgRNA coding plasmids (pAT9679-sgRNA-BFP (#162 988) and pAT9658-sgRNA-mCherry (#162 987)) were constructed by Tálas et al.(2021) [23].

For the construction of pegRNAs, a two-step cloning procedure was used: first, the spacer coding linkers were cloned into pDAS12222 plasmids between BpiI sites using three units of the BpiI enzyme, two units of T4 DNA ligase, 500 μM ATP, 1 × Green buffer, 50 ng vector, and 0.25 μM of each of the corresponding oligonucleotides. In the second cloning step, linkers containing the RTT-PBS were cloned between the BsmbI sites of the plasmids with the above-described method, but using a different buffer condition (1 mM DTT and 1 × Tango buffer). The sgRNA-expressing plasmids contain either an mCherry or a TagBFP expression cassette in order to monitor transfection efficiency.

For prime editing experiments, either the pCMV-PE2 plasmid (created by Anzalone et al. [1] and acquired from Addgene #132 775) or the same plasmid containing the PEmax modifications (R221K and N394K mutations in the SpCas9 nickase) was used as the PE and PEmax expressing plasmids. For non-prime editing experiments or for control samples, the X330-Flag-dSpCas9 (Addgene #92 113) plasmid was used as the dSpCas9-expressing plasmid, the pX330-Flag-wtSpCas9-H840A (Addgene #80 453) as the nSpCas9-expressing plasmid, and the pX330-Flag-wtSpCas9 (Addgene #92 353) [24] as the WT-SpCas9-expressing plasmid.

The HEK293T (CRL-3216) cells used were obtained from ATCC. The HuES9 human embryonic stem cell line was a gift from Dr. Douglas Melton, Harvard University. (HuES9 NIH Approval number: NIHhESC-09–0022 and Health Care Research Council, Human Reproduction Committee in Hungary, Approval number: 6681/2012-EHR.) The cell lines were authenticated by their respective suppliers, and they tested negative for mycoplasma during our experiments.

The HEK293T cells were cultured in DMEM supplemented with 10% heat-inactivated FBS, 100 units/mL penicillin, and 100 μg/mL streptomycin, at 37°C in a humidified atmosphere of 5% CO₂. HuES9 cells were maintained on Geltrex-coated plates in mTeSR1 media at 37°C and 5% CO₂. HuES cells were passaged every 4–5 days using the Accutase cell dissociation reagent and were placed in mTeSR1 supplemented with 10 µM ROCK inhibitor (Y-27632–2HCl, Selleckchem) for the first 24 hours.

In general, transfections of either of the cell types were performed in triplicate, and transfected cells were analyzed by flow cytometry three days post-transfection. Transfection efficiency was considered in both PEAR and genomic experiments. In PEAR experiments, only those mScarlet/GFP-positive cells were considered successfully edited that were also BFP/mCherry-positive (these were used to determine transfection efficiency and are found on pegRNA plasmids). For genomic experiments, samples were always analyzed by flow cytometry first, followed by genomic DNA purification. When evaluating sequencing data, samples were normalized for transfection efficiency.

Flow cytometry analysis was carried out using an Attune NxT Acoustic Focusing Cytometer (Applied Biosystems by Life Technologies). As a rule, signals from a set target minimum of 20 000 viable single cells were acquired by gating based on the side and forward light-scatter parameters. BFP, GFP, mCherry, and mScarlet signals were detected using the 405 (for BFP), 488 (for GFP), and 561 nm (for mCherry and mScarlet) diode lasers for excitation, and the 440/50 (BFP), 530/30 (GFP), 620/15 (mCherry), and 585/16 nm (mScarlet) filters for emission. Attune Cytometric Software v.4.2 was used for data analysis. For the gating of live and GFP/BFP/mScarlet/mCherry positive cells, see Supplementary Fig. S1.

The pegRNAs were synthesized by in vitro transcription. First, the DNA templates were constructed by PCR with the forward primer carrying a T7 promoter sequence. For the PCR, the Q5 High-Fidelity DNA Polymerase was used with the following program: 98°C, 40 s; 35 × (denaturation: 98 °C, 10 s; annealing: see Supplementary Table, 15 s; elongation: 72°C, 15 s; 72°C, 5 min. For the used primer sequences see Supplementary Table). The PCR products were purified using the NucleoSpin Gel and PCR Clean-up Kit. In vitro transcription was performed using the TranscriptAid T7 High Yield Transcription Kit at 37°C for 8 h and followed by the purification of the RNA by the RNA Clean & Concentrator^TM-25 Kit.

The pegRNAs were refolded by heating to 98°C for 2 minutes and a slow cooling down, as described by Zhang et al. [27]

The PEmax protein was overexpressed and purified using the bacterial expression plasmid PE-Max-pet21A (Addgene plasmid # 204 471) developed by the laboratory of Scot Wolfe [16] and by following the protocol of the Wolfe Lab [16, 28] with minor modifications and inclusion of a size exclusion chromatography step to the procedure. Briefly, E. coli Rosetta2(DE3)pLysS cells were transfected with the expression plasmid, and single colonies of transfected cells were picked to grow starter cultures in Luria-Bertani media at 37°C, up to an OD₆₀₀ of about 2. For protein expression, rich growth media (16 g/l tryptone, 10 g/L yeast extract, 5 g/l NaCl, 755.8 mg/L NaH₂PO₄, 5.98 g/l Na₂HPO₄H₂O, pH 7.5) of 3 × 1 L was inoculated by starter cultures to yield starting OD₆₀₀s of 0.01, and the cells were grown at 37°C, 220 rpm, in an incubator-shaker. At the OD₆₀₀ of approximately 0.6, the cultures were briefly chilled on ice (15–20 min), then shifted to 18°C, 220 rpm. At the OD₆₀₀ of approximately 0.8, protein expression was induced by the addition of isopropylthiogalactoside to the media at a 0.7 mM final concentration. The induced cultures were grown for 18 h at 18°C. Cells were harvested by centrifugation (6000 g, 20 min, 4°C) and were resuspended in Ni-NTA buffer (20 mM Tris, 1 M NaCl, 20 mM imidazole, 1 mM TCEP, pH 7.5) supplemented by EDTA-free complete protease inhibitor cocktail tablets (Roche), lysozyme (∼ 0.1 mg/ml) and benzonase nuclease (∼5 units/mL), in order to proceed for cell lysis using repeated 30 s pulse-sonication on ice. The cell lysate was cleared by centrifugation (50 000 g, 50 min, 4°C), and then it was applied to Ni-NTA resin preequilibrated in Ni-NTA buffer. The lysate-loaded resin was incubated on ice for 1 h in 50 mL conical tubes, while gently shaking on a shaker, to capture the His-tagged proteins. The resin with bound proteins was poured onto empty chromatography columns, was washed by 5xCV (column volume) Ni-NTA wash buffer (20 mM Tris, 500 mM NaCl, 20 mM imidazole, 1 mM TCEP, pH 7.5) and the proteins were eluted from the resin in one step by 250 mM imidazole-elution buffer (20 mM Tris, 500 mM NaCl, 250 mM imidazole, 1 mM TCEP, 10% w/v glycerol, pH 7.5). The protein eluate was further purified by cation exchange chromatography using an UNOsphere-S column (Bio-Rad), and a gradient elution with buffers A (20 mM HEPES, 100 mM NaCl, 1 mM TCEP, 10% w/v glycerol, pH 7.5) and B (20 mM HEPES, 1 M NaCl, 1 mM TCEP, 10% w/v glycerol, pH 7.5), while collecting fractions of 1 ml. The fractions containing the PEmax protein confirmed by SDS-PAGE were combined, they were filtered through a 0.22 µm filter, and concentrated by centrifugation using 100 kDa cut-off concentrator tubes (100K MWCO, Pierce). The concentrated protein sample was then subjected to size exclusion chromatography using an ENrich-SEC650 high-resolution column (Bio-Rad) and SEC-buffer (20 mM HEPES, 500 mM NaCl, 1 mM TCEP, 10% w/v glycerol, pH 7.5) and collecting 0.75 mL fractions. The fractions containing the primary PEmax protein peak were combined, filtered through a 0.22 µm filter in a sterile hood, and concentrated by centrifugation (in 100K MWCO concentrator tubes) to at least 50 µM protein. The concentrated protein solution was distributed into small aliquots under sterile conditions and was flash-frozen in liquid nitrogen and placed at −80°C until use.

After flow cytometry, genomic DNA was extracted using the Puregene DNA Purification protocol (Gentra Systems Inc). Amplicons for next-generation sequencing were generated from the genomic DNA samples using two rounds of PCR to attach the Illumina handles. (The 1st step PCR primers used to amplify target genomic sequences are listed in the Supplementary Table.) PCR was done in an S1000 Thermal Cycler (Bio-Rad) or PCRmax Alpha AC2 Thermal Cycler using Q5 high-fidelity polymerase supplemented with Q5 buffer, and 150 ng of genomic DNA in a total volume of 25 μL. The thermal cycling profile of the PCR was: 98°C, 30 s; 35 × (denaturation: 98°C, 20 s; annealing: sees; elongation: 72°C, 20 s; 72°C, 5 min. i5 and i7 Illumina adapters were added in a second PCR reaction using Q5 high-fidelity polymerase with supplied Q5 buffer and 1 µL of the first step PCR product in a total volume of 25 µL. The thermal cycling profile of the PCR was: 98°C, 30 s; 35 × (98°C, 20 s; 67°C, 30 s; 72°C, 20 s); 72°C, 2 min. Amplicons were purified by agarose gel electrophoresis. Samples were quantified with the Qubit dsDNA HS Assay kit and pooled. Supplementary Table, 20

Sequencing on an Illumina NextSeq instrument was performed by Delta Bio 2000 Ltd. Reads were aligned to the reference sequence using BBMap. The aligned reads were considered as the total reads for each sample.

Indels at the pegRNA and 2^nd nicking sgRNA target sites were computationally identified from the aligned reads. Indels were searched for at ±2 bp around the nick/cut sites. For each sample, indel frequency was determined as (number of reads with an indel)/(number of total reads). The frequency of single-substitution mutations without indels generated by prime editing was determined as the percentage of (sequencing reads with the intended modification without indels)/(number of total reads). Conversely, the frequency of intended insertions or deletions generated by prime editing was determined as the percentage of (all sequencing reads with the intended modification)/(number of total reads). For these samples, the indel background was calculated from reads containing different types of indels than the intended edit. In both cases, reads with the intended modifications were identified by searching for a sequence stretch containing the desired edit flanked by 5–5 matching nucleotides. When calculating the edit or indel percentage for each sample, the respective background edit or indel percentage derived from untransfected cells was subtracted.

The following programs were used to analyze the NGS data: BBMap 38.08, Samtools 1.8, BioPython 1.71, and PySam 0.13.

For each CYP gene, gene-specific primers were used, and gene-specific reads were identified based on sequence differences between the two genes. Reads derived from non-gene-specific primer-annealing and from mixed PCR products due to template switching were excluded by exploiting two gene-specific motifs located at different positions of the amplicon.

We analyzed datasets 3, 4, 7, and 8 from the Supplementary Table S2 in the publication of Yu and Kim et al. [17] to compare the average PE efficiencies of pegRNAs grouped according to their PBS length between Library-Small and Library-epegRNA, when used with PE2max and PE4max systems. We included editing data in the analyses only from editing via the conventional ‘NGG’ PAM. In the case of some pegRNAs, we identified more and different editing data for a single pegRNA in the data files; therefore, we excluded these pegRNAs from our analysis.

Statistical significance was assessed by a two-tailed unpaired or paired t-test for comparing two groups. For comparing more than two groups, homogeneity of variances was tested by the Brown–Forsythe test, and normality of residuals was tested by the D’Agostino–Pearson omnibus (K2) test. For data sets with normal distributions, statistical significance was assessed using one-way ANOVA with Tukey’s post-hoc test (when comparing each group to every other group) or with Dunnett’s test (when comparing each group to a control group). In cases where data did not pass normality but fulfilled the assumptions of the Box-Cox transformation, the transformed data were analyzed as above. If not, the Kruskal–Wallis test with Dunn’s test was applied. Statistical tests were performed using GraphPad Prism 9.1.2.

The efficiency of prime editing is decreased with long PBSs [17, 18]. We found that the cleavage activity of pegRNAs with 17-nucleotide-long PBSs is significantly lower or even completely diminished compared to that with their single guide RNA (sgRNA) counterparts. These data also show low prime editing efficiency with 17 nt PBSs when the cleavage activity of the corresponding sgRNAs was also low. (Fig. 1c). The complementarity of the PBS and the spacer, which may contribute to decreased efficiencies with 17 nt PBS, can be reduced by introducing mismatches to the spacer sequences. The mismatch tolerance of the SpCas9 is target- and sgRNA-dependent; it tolerates PAM-distal mismatches well for most targets, but there are a few exceptions where cleavage is inhibited completely [29–31]. We analyzed its mismatch tolerance, aiming at three PEAR plasmid targets integrated into the genome of HEK293T cells [23], the same targets we later examined using the plasmid-based PEAR system. The PEAR system [22], used in either plasmid-based or genome-integrated form, offers easy fluorescence readout to report on PE efficiency by the correction of an inactive splice site of a split mScarlet or EGFP gene separated by an intron sequence. (Supplementary Fig. S2a). To assess its mismatch tolerance on the selected targets, SpCas9 nuclease was used with mismatched sgRNAs, and the results were obtained using new generation sequencing (NGS) (Fig. 1d). In the case of the two mScarlet targets, SpCas9 showed cleavage activities with mismatching spacers above 80% of that of the perfectly matching spacer, while on the GFP target, it was substantially less tolerant for mismatches (Fig. 1d). This latter effect may be related to the use of 21G-sgRNAs, as discussed later. Exploring the observation, reported in the literature [32] and supported by two of the targets above, that SpCas9 generally tolerates one mismatch well at PAM distal positions between the target and the spacer of the sgRNA, we tested if the efficiency of PE can be increased by decreasing PBS-spacer complementarity (Fig. 1e). First, we used two mScarlet targets shifted by 7 nucleotides (Supplementary Fig. S2a) to introduce an AC to GT substitution [22] and designed 18 single-nucleotide and several two- or three-nucleotide mismatches placed primarily into the PAM-distal region of the two spacers (Fig. 1e; Supplementary Fig. S2b, c). We investigated the effect of spacer mismatches on PE3 activity using PBSs of different lengths (10, 13, 17, and 20 nucleotides, the latter discussed later). To define the limits of the effect, we also introduced mismatches at positions 8 and 10 of the spacer, as well as into the portion of the spacer outside the PBS-complementary region.

Most single mismatches in the PAM-distal (beyond position 10) PBS-complementary region increased PE efficiency with up to 7.2-fold (Fig 1e). Out of these 46 mismatching pegRNAs, one decreased the editing efficiency, while 24 significantly increased it. As expected, beyond the PBS-complementary region, no mismatches increased the efficiency of prime editing, while 2 mismatches out of the 7 significantly decreased PE efficiency. At the PAM-proximal positions 8 and 10, however, 7 out of the 12 mismatches significantly decreased PE efficiency, suggesting that PAM-proximal mismatches between the target DNA and the spacer caused a greater inhibition to the process than how much the reduced complementarity between the spacer and the PBS benefited it. (Fig. 1e, Supplementary Fig. S2b, c)

These results confirm that the efficiency of prime editing can be increased by introducing single mismatches to the PBS-complementary region of the spacer of the pegRNAs. Apparently, weakening the spacer-PBS interactions at PAM-distal positions can compensate for, or even outweigh, the negative effect of disrupting spacer–target pairing in the case of most mismatches placed within the PBS-complementary region of the spacer in the targets tested above.

We also examined the third PEAR target on which SpCas9 showed little mismatch tolerance (Fig 1d). As expected, none of the mismatches increased editing efficiency, in accordance with the above interpretation of the impact of the mismatch tolerances (Supplementary Fig. S3a, b)

We hypothesize that SpCas9 likely opens the two DNA strands at the target sequence by a few nucleotides longer than the 20-nucleotide-long segment complementary to the spacer. Thus, extending the length of the PBS to more than 17 nucleotides may increase the length of the PBS:non-target DNA strand duplex, which could potentially increase the efficiency of the reverse transcription initiation at certain targets, without elongating the inhibitory spacer-PBS complementarity. Increasing the PBS length to 20 nucleotides enhanced PE efficiency for one of the two targets tested (Supplementary Fig. S2d). However, the corresponding mismatches did not increase editing efficiency more with PBS20 than with PBS17. (Fig. 1e, Supplementary Fig. S2b, c)

We also tested the effect of introducing multiple mismatches simultaneously into the spacer. When all mismatches were in the PBS-complementary region, 10 out of 17 pegRNAs resulted in a significant increase in editing efficiency, while in the case of 3 pegRNAs, a significant decrease was observed (Fig. 1e, Supplementary Fig. S2b, c). This indicates that although multiple mismatches may have a higher beneficial impact on PE efficiency, they may cause a decrease in efficiency more often than single mismatches do.

To confirm that disrupting intramolecular self-complementarity can enhance editing efficiencies not only in the plasmid-based PEAR system but also in a genomic context, we tested the effect of spacer mismatches using 26 mismatching pegRNAs on mScarlet target 1 integrated into the genome (Fig. 1f, Supplementary Fig. S2e). Consistent with the plasmid-based assays, none of the spacer mismatches were detrimental to editing efficiency, with improvements of up to two-fold, even with the shorter (10 and 13 nt) PBS lengths.

Evidently, reducing the complementarity between the spacer and the PBS can increase the efficiency of PE. However, we hypothesized that introducing mismatches into the PBS instead of the spacer might be more beneficial for certain targets where SpCas9 has a low tolerance for spacer mismatches.

We considered three effects of mismatches regarding their impact on editing efficiency. (i) Mismatches can reduce the stability and the formation of the PBS:non-target strand heteroduplex, which can negatively influence the initiation of reverse transcription, and thus, the efficiency of PE. (ii) Reverse transcriptase (RT) forms non-sequence-specific interactions with the phosphate groups and sugar backbone of the heteroduplex, specifically with the 3′ nucleotides of the DNA primer and the 5′ nucleotides of the RNA. Thus, mismatches in this region can reduce the binding of RT to the heteroduplex and ultimately decrease the efficiency of PE. (iii) Mismatches reduce the complementarity between the spacer and PBS, which can lead to a reduction of inhibitory effects that, in turn, may increase the efficiency of prime editing. This effect is likely stronger if the mismatch occurs in the middle of the PBS rather than at its ends [33].

We examined whether, as a result of the aforementioned three effects, we would observe an increase in PE3 efficiency using both shorter (10 and 13 nt) and longer (20 nt) PBSs and the three formerly selected targets. For the two mScarlet targets, with the shorter PBSs (PBS10 and PBS13), mismatches rarely increased PE efficiency significantly (in 1 out of the 23 mismatching pegRNAs tested), most of them significantly reduced it (Fig. 2a, Supplementary Fig. S4), and with longer PBSs (PBS20), mismatches increased editing by up to 6.7-fold, with only 4 of the 14 mismatched pegRNAs failing to reach at least a 1.5-fold improvement. The negative effects appeared to be more pronounced with shorter PBSs, in which the inhibitory complementary interaction is inherently weaker. In contrast, with longer PBSs, the inhibitory effect is stronger and the RNA:DNA heteroduplex remains more stable even in the presence of mismatches. Interestingly, in the case of the GFP target with the longer PBS, mismatches at the same positions abolished editing efficiency (Supplementary Fig. S3b).

These results strongly support the idea that the intramolecular complementarity of the PBS to the spacer can have an inhibitory effect on the efficiency of prime editing, reinforcing conclusions drawn from previous studies using different experimental approaches [16, 27].

Next, we examined how creating deletions instead of mismatches in the long PBSs (PBS20) would alter the effect of spacer-PBS complementarity relative to the DNA:PBS heteroduplex. Consecutive deletions ranging from 1 to 4 nucleotides in length were introduced starting from position 6 to 15 of the PBS sequence (Supplementary Fig. S5, S6a). For comparison, we also present results from experiments using PBSs with lengths that matched the intact PBS segments upstream of the introduced deletions. Similarly to mismatches, deletions also increased editing efficiency at several positions in comparison to the no-deletion-containing long PBS pegRNA, either approximating or even exceeding the efficiency achieved with the optimal PBS (Fig. 2b, Supplementary Fig. S6b). Deletions starting at the 13th position yielded some of the highest efficiencies for both targets, and therefore, we focused on deletions at this position for further investigation.

We also investigated whether the editing efficiency-enhancing effects of spacer mismatches and PBS deletions are cumulative when combined (Fig. 3a). Our results showed that all tested mismatches significantly increased the editing efficiency of the deletion-only pegRNAs for mScarlet target 2, while none of the added mismatches caused a significant decrease in prime editing activity compared to the deletion-only pegRNAs for the two examined targets. The same outcomes were observed when these pegRNAs were compared to those of optimal PBS lengths (Supplementary Fig. S7a). Therefore, based on the PEAR experiments, the most effective strategy seems to be introducing a mismatch mutation into the spacer and a deletion mutation into the PBS.

We also explored the potential to enhance editing by disrupting the intramolecular self-complementary interaction between the spacer and PBS on genomic targets in HEK293T and HuES cells.

We repeated editing experiments on 10 genomic targets previously characterized in the literature [1], using both the pegRNAs employed in the original study and newly designed pegRNAs, modified to reduce the spacer-PBS complementarity in HEK293T cells. These new pegRNAs contained a deletion at position 13 in a 20-nucleotide-long PBS and an additional transition or transversion mismatch in the spacer at either position 13, 15, or 18 (Supplementary Fig. S7b).

Two out of the 10 edits showed significantly higher editing with pegRNAs harbouring both modified spacer and PBS than the pegRNA with the literature-derived optimal PBSs (Fig. 3b). This, together with the PEAR experiments in Supplementary Fig. S7a, indicates that by combining spacer and PBS modifications, it is possible to achieve higher editing efficiency than with the traditional pegRNAs with optimal PBS. Although it apparently requires further extensive optimization.

Comparing these pegRNAs modified in both the spacer and PBS regions to the corresponding deletion-only pegRNAs, however, showed that the increased efficiency of the altered pegRNAs seen in Fig. 3b did not necessarily require the additional mismatches. Only two pegRNAs had a significant increase in editing efficiency (PRNP target) compared to the corresponding pegRNAs without the additional mismatch (Supplementary Fig. S7b).

The pegRNAs with only the PBS deletion achieved editing efficiencies similar to the literature-tested pegRNAs in all cases (Fig. 3c). This suggests that using a 20-nucleotide-long PBS with a deletion may offer an efficient alternative, approximating the performance of the optimal PBS, and it also reduces the burden of optimizing the combination of modifications. We termed this approach SPELL (Streamlined Prime Editing with fixed-Length PBS Leverage).

To further test SPELL, we compared the efficiency of seven previously employed pegRNAs [34] targeting the CYP1A1, CYP1A2, CYP2B6, RYR2, and KRT12 genes to new pegRNAs designed to achieve the same edit but with a 20-nucleotide-long PBS containing a deletion at position 13. In 4 out of 7 cases, the pegRNAs with the long, non-optimized PBS performed similarly to their original counterparts, while three showed significantly lower editing efficiency (Supplementary Fig. S8a). We further examined an additional set of nine pegRNAs, five of which (CDKL1, CTNNB1, CXCR4, IL2RB, and HBG) had been characterized previously [14, 35], and compared their activity to that of the corresponding SPELL pegRNAs. In this set, the majority of SPELL pegRNAs exhibited editing efficiencies comparable to pegRNAs with the short PBS (Fig. 4a).

Altogether, across all 25 pegRNAs analyzed in HEK293T cells, no significant difference in overall activity was observed between pegRNAs with a short PBS and those with a 20-nt PBS containing the deletion at position 13 (SPELL pegRNAs) (Fig. 4b).

Additionally, we tested two targets in the LRRK2 and GBA1 [36] genes in HuES cells. For the GBA1 target, the pegRNAs with the long, non-optimized, single-deletion-containing PBS achieved a 1.5-fold improvement in editing efficiency, compared to literature-derived PBS, while in the case of the LRRK2 gene, there was no significant difference (Fig. 4c).

Examining the indel background of the pegRNAs analyzed above revealed no indication that SPELL pegRNAs would exhibit increased indel formation (Supplementary Figs S7b and S8b–d). Furthermore, neither off-target editing (Fig. 4d) nor off-target indel formation (Fig. 4e) appeared elevated when using SPELL pegRNAs, as compared to the two pegRNAs known from the literature [1, 37].

We also tested the SPELL approach using RNP delivery. An additional consideration arose from the study of Ponnienselvan et al., as they reported that when purified protein or mRNA-encoded prime editor was complexed with synthetic, end-protected pegRNAs, the optimal PBS length differed from that observed with plasmid-delivered pegRNAs, which lack end protection, for both targets they examined. However, importantly, when describing RNP or RNA delivery, their work consistently used synthetic, end-protected pegRNAs; therefore, the effects of delivery method and end protection were not examined independently.

To investigate how the delivery method affects editing efficiency using various pegRNAs, we performed RNP-based editing experiments on five targets using 3′-unprotected in vitro–transcribed pegRNAs with short PBSs (7 and 8 nt), pegRNAs with PBS12, and SPELL pegRNAs. In two of the five cases, the tested pegRNAs showed similar performance patterns between plasmid and RNP delivery (Supplementary Fig. S8e; CDKL5 and COL7A1 T2). For one target, however, the 8 nt PBS pegRNA yielded higher editing efficiency than PBS12 when delivered as an RNP (Supplementary Fig. S8e; CACNA1C T2). In contrast, for two targets, RNP delivery reduced the activity of pegRNAs with short, 7 or 8-nucleotide PBSs (Supplementary Fig. S8f).

These observations indicate that, in some cases, the delivery method may influence which PBS length performs best for a given target. A more comprehensive investigation, particularly regarding the performance of SPELL pegRNAs in the context of RNP delivery, will require further systematic study.

To investigate how the weakened spacer:PBS annealing and the PBS region downstream of the deletion site contribute to the activity of SPELL pegRNAs, we performed several additional experiments.

In Fig. 2b, we directly compared pegRNAs carrying the non-altered, intact PBS region to their corresponding PBS-deletion pegRNAs. In this sense, a pegRNA with intact PBS corresponds to a SPELL pegRNA where the PBS segment upstream of the deletion is the same; for example, a deletion at position 13 in a PBS20 construct corresponds to an intact 12 nt PBS. In 6 out of the 13 comparisons, the PBS-deletion pegRNAs yielded significantly higher editing efficiencies than their intact-PBS counterparts. This indicates that, in several cases, the 3′ end region of the PBS (i.e. the nucleotides downstream of the SPELL deletion site) contributes to activity. Although the most intuitive interpretation is that this increase in activity results from the annealing of the PBS end to the non-target strand, the mechanism underlying this contribution is not clear.

We further examined the role of the 3′ end region of the PBS by scrambling this segment in three pegRNAs (). For one pegRNA, scrambling the PBS end reduced editing efficiency below that of the pegRNA with the intact PBS. For the second pegRNA, scrambling increased editing efficiency even further, whereas for the third one, scrambling had no significant effect. These results indicate that the PBS end region in SPELL pegRNAs may either enhance or reduce activity through effects that are independent of PBS:DNA annealing (e.g. pegRNA stability or protein-RNA interactions). It is also likely that the PBS end region mediates multiple effects simultaneously, with the relative contribution of each varying across pegRNAs. Supplementary Fig. S8g

We also investigated whether the deletion at position 13 influences SpCas9 nuclease activity (Fig. 4a). Interestingly, we did not observe reduced nuclease activity with the 20 nt PBS compared to shorter PBSs, as would be expected from increased PBS:spacer inhibition, nor did the SPELL deletion increase nuclease activity relative to PBS20, as would be expected from reduced PBS:spacer inhibition, for any of the pegRNAs tested.

At first sight, this may appear unexpected, given that in 5 out of the 9 targets PBS20 pegRNAs showed reduced prime editing activity, and in 4 of these cases the SPELL deletion increased editing efficiency (Fig. 4a). Our interpretation is that the increased inhibition of prime editing by stronger spacer:PBS interactions does not necessarily manifest at the level of SpCas9 nicking. This conclusion is reminiscent of our earlier findings [38], where RTT-scaffold interactions strongly inhibited prime editing efficiency while having no observable effect on nuclease activity.

These observations suggest that, for certain targets, likely including several examined here, the inhibitory effect of the increasing PBS length, and thereby the strengthening of spacer:PBS interactions, does not further impair SpCas9 nicking. Instead, the inhibition appears to act at a subsequent step of the prime editing process, most plausibly by interfering with efficient RT priming.

In conclusion, our results strongly support the presence of an inhibitory effect exerted by complementary spacer-PBS interactions in prime editing. We demonstrate that reducing these interactions can substantially improve editing efficiency, and that such modified pegRNAs can outperform pegRNAs containing the otherwise optimal PBS. However, achieving this improvement requires extensive optimization.

Our findings partially overlap with those of Fei et al. [39], who primarily focused on spacer:protospacer interactions. In their study, the edits introduced by prime editing typically altered the target sequence (positions 1–3 and 5–6), and mismatches added to the spacer further reduced re-nicking of the target site. They also speculated that mismatches at certain positions may affect the stability of the spacer:protospacer duplex, and that duplex flexibility, as predicted by AlphaFold, may correlate with prime editing outcomes. Consistent with their observations, we also found that spacer-mismatched pegRNAs can reduce indel formation at several targets, although indels may increase at others (Supplementary Fig. S7b).

However, Fei et al. did not consider the effects of PBS length or PBS melting temperature—factors that determine the effective position and the impact of the introduced mismatches—nor did they explicitly report on these parameters. In contrast to their emphasis on spacer–protospacer interactions, our work focuses on inhibition arising from spacer:PBS complementarity. Our results also caution against drawing firm conclusions from a limited number of targets or pegRNAs and demonstrate that altering the sequence of either the spacer or the PBS can have multiple consequences beyond simply weakening intramolecular spacer:PBS interactions.

SpCas9 is generally tolerant of mismatches, particularly those located at PAM-distal positions. Thus, it was somewhat unexpected that the GFP target did not tolerate spacer mismatches without a reduction in cleavage activity. The sgRNAs and pegRNAs targeting this locus contained an appended 5′ G nucleotide to satisfy the U6 promoter requirement [40], resulting in a 21-nt spacer. Fu et al. demonstrated that a 5′ GG extension increases nuclease fidelity by weakening off-target cleavage [40], and we subsequently showed that appending a single 5′ G can similarly enhance fidelity [41]. The GFP target may therefore display increased sensitivity to mismatches because of its 21-nt spacer length. Although the effect of a 5′ G extension can depend on the intrinsic cleavability ranking of the target [42], pegRNAs with 21-nt spacers may be less amenable to our approach.

We also observed that pegRNAs with 17-nt PBSs exhibited reduced nuclease activity (Fig. 1c). Surprisingly, however, altering the strength of the spacer:PBS interaction—despite its clear influence on prime editing efficiency—did not seem to affect cleavage activity among pegRNAs sharing the same spacer sequence (Fig. 4a). This suggests that spacer:PBS complementarity also influences steps other than the initial SpCas9 nicking.

Finally, our study demonstrates that high editing efficiency can be achieved for many targets without extensive optimization by using SPELL—a simplified strategy based on a single-nucleotide PBS deletion.