Chromatin context-dependent regulation and epigenetic manipulation of prime editing

Abstract

We set out to exhaustively characterize the impact of the cis-chromatin environment on prime editing, a precise genome engineering tool. Using a highly sensitive method for mapping the genomic locations of randomly integrated reporters, we discover massive position effects, exemplified by editing efficiencies ranging from ∼0% to 94% for an identical target site and edit. Position effects on prime editing efficiency are well predicted by chromatin marks, e.g., positively by H3K79me2 and negatively by H3K9me3. Next, we developed a multiplex perturbational framework to assess the interaction of trans-acting factors with the cis-chromatin environment on editing outcomes. Applying this framework to DNA repair factors, we identify HLTF as a context-dependent repressor of prime editing. Finally, several lines of evidence suggest that active transcriptional elongation enhances prime editing. Consistent with this, we show we can robustly decrease or increase the efficiency of prime editing by preceding it with CRISPR-mediated silencing or activation, respectively.

Keywords: DNA damage repair; cis x trans interactions; cis-chromatin effect; epigenome editing; genome engineering; prime editing.

Figure 1.. Efficient genome-wide mapping of integration sites of synHEK3 reporters.

A) Measuring cis-chromatin effects on prime editing efficiency. A library of synHEK3 reporters is randomly inserted throughout the genome. Genomic locations of individual reporters are determined with a T7-assisted reporter mapping method. The cis-chromatin contexts of mapped reporters are used to model prime editing outcomes as measured from each reporter. B) Genome browser view of a read pileup pinpointing the precise coordinates of a synHEK3 reporter integration. Barcode sequence, orientation and coordinates of the reporter are annotated. C) Motif enrichment analysis of 20-bp windows surrounding synHEK3 integration sites. D) Coverage plot of unique synHEK3 reporters identified in the bottlenecked pool (n = 4,273). Vertical bar lengths correspond to read counts. E) UpSet plot of genomic annotations of synHEK3 integration sites.

Figure 2.. Chromatin context has a major impact on prime editing efficiency.

A) Left: workflow of experiment. Right: density plot of CTT insertion frequency in all uniquely barcoded synHEK3 reporters (n = 4,273). Red line indicates prime editing efficiency (17%) at endogenous HEK3 locus in K562 cells. B) Heatmap of fractions of highly editable (>25%) sites in synHEK3 sites stratified by chromatin feature scores. SynHEK3 reporters are binned into 10 equally sized bins with increasing chromatin feature scores. The chromatin features are ordered left-to-right by their correlation coefficient (Spearman’s ρ) with prime editing efficiency. C) Scatter plot of observed vs. predicted prime editing efficiencies using reporters in a holdout test set. Points colored by the number of neighboring points. D) Scatter plot of Spearman’s ρ between chromatin feature scores and prime editing efficiencies, calculated separately for intragenic and intergenic reporters. E) Prime editing efficiency for gene-proximal reporters. Distance was scaled by gene length and binned. Negative values refer to synHEK3 sites located upstream of TSS. Values >100% refer to synHEK3 sites located downstream of transcription termination site (TTS). Points colored based on expression levels (log₁₀) of the genes. TPM, transcripts per million. F) Genome browser views of the 4 most highly editable sites. Sites of integration and measured editing efficiencies are shown as a dot plot at top and aligned with selected epigenetic tracks. For each synHEK3 insertion, editing efficiency, number of reads with edit (numerator), and total number of reads (denominator) are annotated. The dashed vertical lines mark locations of the insertions. G) Scatter plot of sequence-based prediction (DeepPrime, x-axis) vs. normalized editing rate (y-axis, log₁₀ scale) for epegRNAs designed for prime editing at endogenous genomic sites. H) Scatter plot of chromatin-based prediction (our model, x-axis) vs. normalized editing rate (y-axis, log₁₀ scale). I) Scatter plot of combined score (x-axis) vs. normalized editing rate (y-axis, log₁₀ scale).

Figure 3.. Comparison between Cas9 and prime editing, leveraging a common set of integrated reporters.

A) Comparison of editing efficiencies for Cas9 (Day 1, 2 or 4 after transfection) vs. prime editing (Day 4 after transfection) at an identical set of synHEK3 reporters. Plots colored based on the number of reporters assigned to each bin. 1-D histograms of editing efficiencies plotted at top and right. B) Hierarchical clustering of synHEK3 reporters based on prime and Cas9 editing efficiencies. C) Density plot of prime and Cas9 editing efficiencies for 14 groups of synHEK3 reporters, ordered by mean prime editing efficiency. D) Bar graph of the log₂ ratio between number of intragenic vs. intergenic sites in each of the 14 groups. E-G) Comparison of properties of intragenic sites in selected groups. P-value: two-sided Kolmogorov–Smirnov test. E) Boxplot of ATAC-seq scores of selected groups. F) Expression levels of the overlapping genes in TPM of selected groups (x-axis; log₁₀ scale). G) Distance (bp) between the synHEK3 reporters in selected groups and the nearest TSS (x-axis; log₁₀ scale).

Figure 4.. Dissecting chromatin context-dependent regulation of prime editing using a modified sci-RNA-seq3 workflow.

Experimental workflow of the pooled shRNA screen. A) The two monoclonal K562 lines used in this experiment stably expressed PE2 and reverse tetracycline transactivator (rtTA), and together contained 50 unique synHEK3 reporters. Cells were transduced with the TRE-shRNA library at a high multiplicity of infection (MOI) and treated with doxycycline. On Day 2, cells were transfected with pegRNAs to introduce random 6-bp insertions at synHEK3 reporters. After 3–4 days, nuclei were extracted and fixed. TRE: tetracycline response element. B) Fixed nuclei were subjected to IST with T7 polymerase (pink circle) to produce transcripts from synHEK3 and shRNA constructs. C) Nuclei were distributed to 96-wells for indexed RT. In each well, a cocktail of three indexed RT primers were used: oligo-dT primers, and synHEK3- and shRNA-specific primers. D) After RT, nuclei were pooled and redistributed into 96-well plates for indexed hairpin ligation. Then, they were pooled and split to final 96-well plates. After second-strand synthesis, nuclei were lysed and the resulting lysates were split to two plates. One plate underwent Tn5 tagmentation and indexed PCR to generate a transcriptome library. The other plate was used for indexed enrichment PCR targeting the synHEK3 and shRNA transcripts. E) For each synHEK3 reporter, editing outcomes were computed and compared between cells with vs. without a specific shRNA.

Figure 5.. Effects of perturbing MMR-related genes on prime editing.

A) Q-Q plot of statistical significance (-log₁₀) of synHEK3-shRNA pairs in clones 3 (left) and 5 (right). Candidate shRNAs (green) and control shRNAs (gray) are plotted separately. B) Plots of adjusted p values (-log₁₀) of all synHEK3-shRNA pairs. Target genes with high statistical significance are annotated. Points colored by editing efficiency changes caused by corresponding shRNAs. C) Effects of shRNAs targeting MMR-related genes in clone 5. Log₂ fold-changes of prime editing efficiencies of synHEK3-shRNA pairs are plotted and colored by their corresponding adjusted p values (-log₁₀). D) Effects of shRNAs against MLH1 and PMS2. Pink lines: editing frequencies in cells with individual shRNAs; red line: mean editing frequencies of the gene-targeting shRNAs; light blue lines: control editing frequencies for individual shRNAs (not visible because low variance relative to mean line, shown in blue); blue line: mean control editing frequencies.

Figure 6.. Chromatin context-specific response to HLTF inhibition.

A) Violin plot of fold-changes of editing efficiency of synHEK3 sites in response to inhibition of HLTF, MLH1 and PMS2. Points colored by shRNA identity. B) Heatmap of synHEK3 reporters (row) and their responses to shRNAs against HLTF (left: clone 5; right: clone 3). Leftmost bar annotates the overlapping status of synHEK3 reporters with GRCh38 gene annotations. Second left bar indicates the expression status of the overlapping or nearest gene in TPM. Third left bar indicates distances (bp) to corresponding TSS of gene-overlapping synHEK3 reporters or nearest genes for synHEK3 reporters outside genes. Middle heatmap was generated using scaled chromatin feature scores of synHEK3 reporters and clustered by column. Right line plot shows effects of shRNAs against HLTF. Pink lines: editing frequencies in cells with individual shRNAs; red line: mean editing frequencies of the gene-targeting shRNAs; light blue lines: control editing frequencies for individual shRNAs (not visible because low variance relative to mean line, shown in blue); blue line: mean control editing frequencies. Dashed lines: sites showing differential response to HLTF inhibition. C) Bar plot of synHEK3 reporter counts based on their responsiveness to HLTF inhibition and overlapping status with GRCh38 gene annotations. P value: Fisher’s exact test. D) Expression of genes overlapping or proximal (within 10 kb) to a synHEK3 reporter. Genes above the dashed line (TPM = 3) are considered expressed. P value: two-sided Kolmogorov–Smirnov test.

Figure 7.. Modulating prime editing outcomes by epigenetic conditioning.

A) Workflow of the CRISPRoff experiment. B) Scatter plots of mean prime editing efficiencies of synHEK3 reporters in cells transfected with CRISPRoff gRNAs targeting USP7, METTL2A and LRRC8C promoters. SynHEK3 reporters in corresponding target genes are labeled. Error bars correspond to standard deviation of measured editing efficiencies. C) Bar plot of prime editing efficiency changes of synHEK3 reporters in CRISPRoff experiment. Control editing efficiencies are predicted efficiencies using linear models trained in synHEK3 reporters that are not in the corresponding CRISPRoff target genes as shown in panel B. D) Workflow of the CRISPRa experiments. E) Prime editing efficiency (%) at endogenous gene targets in K562 cells, with or without epigenetic editing via CRISPRa. Green bars show mean prime editing efficiencies in a wild-type K562 cell line which received only PEmax and (e)pegRNA. Gray bars show mean prime editing efficiencies when control promoters were activated via CRISPRa. Blue bars show mean prime editing efficiencies when target gene promoters were activated. Fold-changes are calculated between the CRISPRa (blue) and control (gray) groups. Inset shows a zoomed view of the first three genes. F) Prime editing efficiency (%) at endogenous gene targets in a human iPSC line, with or without epigenetic editing via CRISPRa. Gray bars show mean prime editing efficiencies when control promoters were activated via CRISPRa. Blue bars show mean prime editing efficiencies when target gene promoters were activated. G) Scatter plots of mean prime editing efficiencies measured in control vs. CRISPRa cells in exons of IL2RB. Points colored by edit type. x- and y-axes in log₁₀ scale. The pink and red lines indicate 2-, 5- and 10-fold differences between the CRISPRa and control groups. H) Boxplot of prime editing efficiency fold-change for all variants. The dashed line indicates a fold-change of 2.