Increasing intracellular dNTP levels improves prime editing efficiency

Abstract

In primary cell types, intracellular deoxynucleotide triphosphate (dNTP) levels are tightly regulated in a cell cycle-dependent manner. We report that prime editing efficiency is increased by mutations that improve the enzymatic properties of Moloney murine leukemia virus reverse transcriptase and treatments that increase intracellular dNTP levels. In combination, these modifications produce substantial increases in precise editing rates.

Extended Data Fig. 1:. PEmax** outperforms other prime editor variants for precise editing by PE2 and Twin-PE systems in HEK293T cells, U2OS cells and Fibroblasts

(a) PEmax mediated PE2 editing at HEK4 (20-bp PE-tag insertion) in HEK293T and U2OS cells with increasing amounts of PEmax and pegRNA expression plasmid delivered by transfection to define subsaturating editing conditions. Insertion of 20-bp PE-tag (blue) sequence (b) and 53-bp PE-tag+Flag-tag (blue and lowercase red) sequence (c) in HEK293T and U2OS cells with different prime editor variants (PE2 system) under unsaturated conditions (transfection with 1000ng prime editor + 330ng pegRNA). (d) PEmax mediated prime editing at FANCF (+5 G to T) in fibroblasts, EbvB cells and HBE cells with different amounts of PEmax mRNA/synthetic FANCF pegRNA delivered by electroporation to define subsaturating editing conditions. (e) Prime editing at FANCF (+5 G to T) in fibroblasts with various prime editor variants (PE2 system) under unsaturated conditions (electroporation with 1000ng prime editor mRNA + 50 pmol pegRNA). (f) Evaluation various prime editor variants for the installation of the 38-bp Bxb1 attB site at HEK3 site in HEK293T cells by Twin-PE. For all datasets, values and error bars represent the mean ± SD of n = 3 independent replicates from targeted amplification deep-sequencing. Two-way ANOVA was used to compare the intended edit and unintended edit across all the groups for each graph, where PEmax was used as a control column for multiple comparisons (Supplementary table); * indicates P ≤0.05, ** indicates P ≤0.01, *** indicates P≤ 0.001, **** indicates P ≤0.0001 and ns indicates not significant.

Extended Data Fig. 2:. PEmax protein expression constructs and their purification.

(a) Architecture of the PEmax, PEmax* and PEmax** mammalian expression plasmid, mRNA and protein constructs. PEmax variants encoded by mRNA and protein constructs contain four nuclear localization signal (NLS) sequences, whereas the mammalian expression plasmid encoded PEmax variants contain three NLSs. (b) SDS-PAGE analysis of purified C-terminally His-tagged PEmax and PEmax** protein. Black arrow indicates the desired protein product (n=1). (c) Protein yield per liter of bacterial culture for PEmax and PEmax** (n=1).

Extended Data Fig. 3.. Emax** improves editing rates in zebrafish embryos and adult mice.

P Comparison of PEmax and PEmax** RNP editing efficiency at various concentrations for introducing (a) L907F at the tek locus in zebrafish embryos (target codon indicated in green, n=3 independent replicates). One-way ANOVA was used to compare the intended edit across all the groups for each graph; * indicates P < 0.05, **** indicates P ≤0.0001 (also see Supplementary table). Comparison of zebrafish embryo survival rate 24 hours post RNP injection for different treatment groups for (b) R841W and (c) L907F at the tek locus (each n=3). One-way ANOVA was used to compare the survival across all the groups for each graph (Supplementary table); ns indicates P > 0.05, * indicates P < 0.05. (d) Comparison of in vivo PE3 editing rates for the introduction of Q155H at Pcsk9 mediated by PEmax and PEmax** mRNA LNP formulations delivered to mice through retro-orbital injection (rates determined from 3 liver punches from each of 5 mice in each cohort, n = 15). Two-way ANOVA was used to compare the intended edit and unintended edit across all the groups for each graph (Supplementary table), PEmax was used as a control column for multiple comparisons; **** indicates P ≤0.0001.

Extended Data Fig. 4:. Comparison of 20-bp PE-tag sequence insertion at HEK4 site mediated by various prime editor variants under different conditions in HEK293T and U2OS cells

(a) Cell viability assay for determining the optimal dN dose in growth media for use with HEK293T cells. Cell viability was monitored every 24 hours following addition of the desired dN concentration to the growth media. Symbols and error bars represent the mean ± SD, respectively, of n = 3 independent replicates. (n=3). Arrow indicates the concentration of dNs [50 μM] used in subsequent experiments (b,c). Evaluation of editing efficiency for various prime editor variants (PE2 system) under different conditions (co-delivery of VPX, SAMHD1 and dNs [50 μM]) for insertion of a 20-bp PE-tag sequence at the HEK4 site in HEK293T (b) and U2OS cells (c) (unsaturated conditions: transfection with 1000ng prime editor + 330ng pegRNA). For b and c, bars and error bars represent the mean ± SD, respectively of n = 3 independent replicates.

Extended Data Fig. 5:. Comparison of prime editing mediated by various prime editor variants at FANCF (PE2) and HEK3 (twin-PE) under different conditions in fibroblasts

(a) Cell viability assay for determining the optimal dN dose in growth media for use with fibroblasts. Cell viability was monitored every 24 hours following addition of the desired dN concentration to the growth media. Symbols and error bars represent the mean ± SD, respectively, of n = 3 independent replicates. (n=3). Arrow indicates the concentration of dNs [30 μM] used in subsequent experiments (b) Evaluation of editing efficiency for various prime editor variants (PE2 system) under different conditions (co-delivery of VPX, SAMHD1 and dNs [30 μM]) for FANCF (+5 G to T) base conversion in fibroblasts (electroporation with 1000ng prime editor mRNA + 50pmol pegRNA). (c) Evaluation of editing efficiency for various prime editor variants (PE2 system) under different conditions (co-delivery of VPX, SAMHD1 and dNs [30 μM]) for Twin-PE in fibroblasts for the installation of the 38-bp Bxb1 attB site at HEK3 site. For b and c, bars and error bars reflect the mean ± SD, respectively, of n = 3 independent replicates from targeted amplification deep-sequencing.

Extended Data Fig. 6:. Comparison of 53-bp PE-tag+Flag-tag sequence insertion at HEK4 site mediated by various prime editor variants under different conditions in HEK293T and U2OS cells

(a,b) Evaluation of editing efficiency for various prime editor variants (PE2 system) under different conditions (co-delivery of VPX, SAMHD1 and dNs [50 μM]) for insertion of a 53-bp tag sequence at the HEK4 site in HEK293T (a) and U2OS cells (b) with various prime editor variants mediated PE2 approach under different conditions (unsaturated conditions: transfection with 1000ng prime editor + 330ng pegRNA). For a and b, bars and error bars reflect the mean ± SD, respectively, of n = 3 independent replicates.

Extended Data Fig. 7:. Delivery of VPX increases the intracellular dNTP pool by down regulating SAMHD1 expression in fibroblasts and human primary T cells.

(a,b) Measurement of intracellular dNTP pool in (a) fibroblasts and (b) human primary T cells using long synthetic oligonucleotides and EvaGreen dye based assay. For perturbation of intracellular dNTP levels VPX, VPX mutant (Q76A) or SAMHD1 mRNA are electroporated into patient derived fibroblasts (a) or primary T cells (b). Cells are harvested 48 hours post treatment and the intracellular dNTP concentration was determined. For a and b, bars and error bars reflect the mean ± SD, respectively, of n = 2 biological replicates with 3 technical replicates each. SAMHD1 protein levels in fibroblasts (n=1) (c) and primary T cells (n=1) (d) electroporated with VPX, VPX mutant or SAMHD1 mRNA was detected by immunoblotting with GAPDH used as a loading control.

Extended Data Fig. 8:. PEmax** and co-delivery of VPX protein improves prime editing rates in resting T cells.

a) PEmax-mRNA mediated PE3 editing at FANCF (+5 G to T) in resting human primary T cells with co-delivery of synthetic pegRNA/sgRNA and increasing amounts of VPX protein (n=3 different donors). Two-way ANOVA was used to compare the intended edit and unintended edits across all the groups for each graph, and the no VPX group was used as a control column for multiple comparisons (Supplementary table); ns indicates P > 0.05, * indicates P ≤0.05, ** indicates P ≤0.01 and **** indicates P ≤0.0001. (b) PEmax, PEmax*, and PEmax** mRNA mediated PE3 editing at FANCF (+5 G to T) in resting human primary T cells with co-delivery of synthetic pegRNA/sgRNA and 0.5 μg VPX protein (n=3 different donors). Two-way ANOVA was used to compare the intended edit and unintended edits across all the groups for each graph, and the no VPX group was used as a control column for multiple comparisons (Supplementary table); ns indicates P > 0.05, * indicates P ≤0.05, ** indicates P ≤0.01. (c) PEmax and PEmax* mRNA-mediated PE3 editing at FANCF (+5 G to T) in resting human primary T cells with co-delivery of synthetic pegRNA/sgRNA and VPX or VPX Q76A mRNA (n=3 different donors). One-way ANOVA was used to compare the intended edits across all the groups for multiple comparisons (Supplementary table); ns indicates P > 0.05, * indicates P ≤0.05, ** indicates P ≤0.01 and **** indicates P ≤0.0001.

Extended Data Fig. 9:. SC-islets are glucose responsive and express appropriate markers.

(a) Glucose-stimulated insulin secretion (GSIS) was performed to assess in vitro function of SC-islets at time of nucleofection. The GSIS index, calculated by dividing the amount of insulin secreted following incubation in high glucose (20 mM) by that following incubation with basal glucose (2 mM), was 3.7. Bars and error bars reflect the mean ± SD. Each point represents an independent sample (n=4). *** indicates P ≤0.001 and **** indicates P<0.0001, one-way ANOVA (Supplementary table). (b) Representative flow cytometry staining of dispersed SC-islet clusters for differentiation and proliferation markers at time of nucleofection, with representative scatter plots (left) and data from replicate samples (right, n=2). Bars and error bars reflect the mean ± SD.

Figure 1:. PEmax** and VPX improve prime editing efficiency.

(a) Fold-change in PE2 and Twin-PE editing efficiency for prime editor variants compared to PEmax in HEK293T cells, U2OS cells and fibroblasts (n = 18 comparisons across multiple target sites; Extended Data Figs. 4, 5 & 6). P values are from one-way ANOVA. (b) In vitro biochemical analysis of PEmax and PEmax** RNP activity under different dNTP concentrations assessing 20 nt tag incorporation at the HEK4 target site in purified gDNA (qPCR, n=3). P values are from two-way ANOVA. (c) PE3 precise editing efficiency for prime editor variants (editor protein or mRNA, pegRNA, sgRNA) delivered by electroporation to iMyoblasts or primary T cells (n=3). P values are from unpaired two-sample t-test or one-way ANOVA. (d) Comparison of PEmax and PEmax** RNP editing efficiency at different doses when introducing R841W at the tek locus in zebrafish embryos (n=3, Methods). Codons with synonymous or non-synonymous mutations are indicated by bold underline. P values are from one-way ANOVA. (e) Violin plots of fold-change in PE2 or Twin-PE editing efficiency of prime editor variants compared to PEmax in HEK293T cells, U2OS cells and fibroblast cells under different conditions (standard, with VPX or with dN supplementation). Horizontal white lines indicate quartiles. Medians are indicated by the black horizontal lines. P values are from two-way ANOVA with standard PEmax editing as the control for multiple comparisons. (f) Co-delivery of VPX mRNA increases precise editing rates. Left: PEmax, PE6d and PEmax** mediated PE3 editing and Cas9-NG-PEmax and Cas9-NG-PEmax** mediated Twin-PE editing to install the CCR5^delta32 HIV-1 resistance mutation in activated human primary T cells (n=3 from 2 different donors, Methods). Right: PE3b prime editing in SC-Derived islet cells (n=3). P values are from two-way ANOVA. For c, d and f, data and error bars indicate mean ± SD. The PAM sequence is boxed, protospacer sequence is underlined, edited bases are indicated in red lowercase, and deleted sequence is indicated in blue letters. For all statistical tests, * indicates P ≤ 0.05, ** indicates P ≤0.01, *** indicates P≤ 0.001, **** indicates P≤ 0.0001, and ns indicates not significant (Methods).

Figure 2:. VPX enhances therapeutic prime editing in a variety of human cell lines

(a) PE3b editing mediated by PEmax, PE6d and PEmax** at FANCF (+5 G to T) in fibroblasts, human bronchial epithelial (HBE) cells and EbvB cells without (blue bars) or with (orange bars) co-delivery of VPX mRNA. (b~d) PEmax, PE6d and PEmax** mRNA-mediated PE3b editing without (blue bars) or with (orange bars) co-delivery of VPX mRNA at (b) MECP2 (+2+4+5 CTG to TCC) to revert the T158M mutation in patient derived fibroblasts, (c) CFTR (insert CTT) to correct the ΔF508 mutation in HBE cells, and (d) HEXA (delete TATC) to correct the 4 bp microduplication in EbvB cells. (e) Comparison of the relative precise prime editing rate with or without co-delivery VPX mRNA across multiple endogenous genomic loci and human cell types (summarized data from fig a~d). White lines indicate quartiles. Medians are indicated by the black horizontal lines. P values are from two-way ANOVA. (f~h) Co-delivery of VPX mRNA improves the prime editing outcomes: (f) Ratio of precise edits to undesired edits, (g) ratio of pegRNA scaffold insertions to precise edits, and (h) ratio of indels (not including scaffold insertions) to precise edits following prime editing using PEmax, PE6 or PEmax** with or without co-delivery of VPX across multiple loci and cell lines (PE3/PE3b editing in T cells, fibroblasts, HBE and EbvB cells). Dots show individual replicate values (n=24). P values are from two-way ANOVA. (i) Schematic overview of the combination of methods used to boost prime editing efficiency. For a-d, data and error bars indicate mean ± SD (n=3). Two-way ANOVA was used to calculate whether the VPX addition (column factor) makes significant difference across all the PE variants as a group (row factor). The PAM sequence is boxed, protospacer sequence is underlined, edited bases are indicated in red lowercase, and inserted bases are indicated in blue lowercase. The protospacer for the PE3b nicking sgRNAs is indicated on bottom strand. For all statistical tests, * indicates P ≤ 0.05, ** indicates P ≤0.01, *** indicates P≤ 0.001, **** indicates P≤ 0.0001, and ns indicates not significant (Methods).