A prime editing strategy to rewrite the γ-globin promoters and reactivate fetal hemoglobin for sickle cell disease Open Access

Screening of epegRNAs and ngRNAs targeting the HBG1/2 promoters in K562 cells. (A) Schematic representation of the β-globin locus on chromosome (chr) 11, including the HBG1 and HBG2 genes and their promoters (prom) (top). HPFH or HPFH-like mutations (bold) disrupt LRF or BCL11A repressor BSs (red boxes) or generate KLF1, TAL1, or GATA1 activator (yellow boxes) BSs. PAM-disrupting mutation (PAMm) and BCL11A BS-deletion (Δ2B: ΔCC, Δ5B: ΔTGACC) present in pegRNA1/epegRNA1 and pegRNA10/epegRNA10.1/epegRNA10.2 are displayed below the promoter sequence. Schematic representation of epegRNA5 (bottom left) and epegRNA10.1 (bottom right) targeting the HBG1/2 promoter region. The epegRNA is composed (5′-3′) of a spacer, a scaffold, a primer binding site (PBS), a reverse transcription template (RTT) containing the edits (ie, base substitutions and deletion indicated in colors and black), and the tevopreQ₁ motif. Upon annealing of the spacer to the target DNA, the Cas9n induces a DNA single-strand break (ie, nick, arrowhead) liberating the opposite strand (3′ flap), which anneals to the PBS of the epegRNA and primes the reverse transcription of the RTT. The interconversion of the original 5′ flap by the newly synthesized, edited 3′ flap and the degradation of the 5′ flap are essential steps to install the DEs into the target region. (B-C) Frequency (%) of InDels generated by transfecting plasmids coding for (B) pegRNAs or (C) ngRNAs and Cas9 (pegRNA1 and ngRNA1.1) or Cas9-SpRY (pegRNA2 to pegRNA9 and ngRNA5.1 to ngRNA5.9) nucleases. Although pegRNA1 was designed to install KLF1, TAL1, and PAMm mutations (to avoid retargeting of the region), pegRNA2 to pegRNA5 have a shorter RTT to install only the KLF1 BS. (D) Percentage of NGS reads containing the DE (3 mutations: PAMm, KLF1, and TAL1 BSs; PAMm_K_T), the partial edits (1 or 2 mutations; K or PAMm_K), or InDels generated using epegRNA1, ngRNA1.1, and PE3max or PE5max in the top 30% green fluorescent protein–positive (GFP⁺) cells. (E) Percentage of NGS reads containing the DE (2 mutations: KLF1 and TAL1; K_T), the partial edit (1 mutation; K), or InDels generated using epegRNA5, with ngRNA5.5 (left) or ngRNA5.6 (right) and PE3max or PE5max (in their PAM-less SpRY version) in the top 30% GFP⁺ cells. (F-G) Frequency (%) of InDels generated by transfecting plasmids coding for (F) pegRNAs or (G) ngRNAs and Cas9 (pegRNA10 and ngRNA10.1 to ngRNA10.5) or Cas9-SpRY (pegRNA11 to pegRNA14) nucleases. (H) Percentage of NGS reads containing the DE (3 modifications: GATA1 BS, BCL11A-5-nt BS deletion [ΔTGACC from –114 to –118 upstream of HBG1/2 TSS], and KLF1 BS; G_Δ5B_K), DE with mutations, or InDels induced by epegRNA10.1 with ngRNA10.3 or ngRNA10.5 in the top 30% GFP⁺ cells (left). The sequences of the PBS and RTT of epegRNA10.1 are aligned on the most frequent NGS reads. The KLF1 and GATA1 BSs are indicated in gray and bold, respectively, and the desired base conversions are in lower case. The 5 base-pair (5-bp) deletion of the edited 3′-flap facilitates the annealing of its GCC trinucleotide (underlined) to the closest complementary motif (red square) leading to DE with mutations (pink) with or without scaffold incorporation (pink and bold) (right). (I) Percentage of NGS reads containing the DE (3 modifications: GATA1 BS, BCL11A-BS 2-nt deletion [ΔCC –117/–118 upstream of HBG1/2 TSS], and KLF1 BS; G_Δ2B_K), DE with mutations, the partial edit (G_Δ2B) or InDels induced by epegRNA10.2 with ngRNA10.3 or ngRNA10.5 in the top 30% GFP⁺ cells (left). The sequences of the PBS and RTT of epegRNA10.2 are aligned on the most frequent NGS reads. The KLF1 and GATA1 BSs are indicated in gray and bold, respectively, and the desired base conversions are in lower case (right). (J) Percentage of NGS reads containing the DE (2 modifications: GATA1 and KLF1 BSs; G_K), the partial edits (GATA1 or KLF1 BSs), or InDels induced by epegRNA10.3 with ngRNA10.3 or ngRNA10.5 in the top 30% GFP⁺ cells (left). The sequences of the PBS and RTT of epegRNA10.3 are aligned on the most frequent NGS reads. The KLF1 and GATA1 BSs are indicated in gray and bold, respectively, and the desired base conversions in lower case (right). The long 3′-flap facilitates the annealing of its GCC trinucleotide (underlined) to the expected complementary motif, preventing the formation of the DE with unintended mutations. (K) Percentage of NGS reads containing the DEs, the partial edits, or InDels and frequency (%) of 4.9-kb deletion measured by droplet digital PCR (ddPCR) (as previously described⁶ using the primers and probes described in supplemental Table 2) generated using epegRNA10.2 or epegRNA10.4 with ngRNA10.5 into the HBG1/2 promoters in the top 30% GFP⁺ cells. (L-M) Top 5 (L) epegRNA10.4- and (M) ngRNA10.5-dependent off-target DNA sites, as identified by GUIDE-seq⁶ in K562 cells. epegRNA and ngRNA were coupled with a Cas9 nuclease equivalent to the Cas9n included in the PE except for carrying the WT amino acid allowing nuclease activity. The protospacer targeted by each epegRNA or ngRNA and the PAM are reported in the first line, followed by the on- and off-target sites and their mismatches with the on-target (highlighted in color). For each target, the abundance (ie, the total number of unique alignments associated with the target site), chromosomal coordinates (Human GRCh38/hg38), type of region, gene and targeted strand, and number of mismatches with the epegRNA or ngRNA protospacer are reported. In total, we identified 14 off-targets for epegRNA10.4 including only 5 with an abundance >3, and 44 off-targets for ngRNA10.5 including 27 with an abundance >3. Most of them have a high number of mismatches (≥5) and map to nonexonic regions. (B-C,F-G) Samples mock-transfected with Tris-EDTA buffer were used as controls. PCR products were subjected to Sanger sequencing (as previously described⁶ using the primers and probes described in supplemental Table 2), and InDels were measured using the TIDE software.⁷ Bars represent the mean ± standard deviation (SD) of 3 biologically independent replicates. Statistical significance was assessed between mock and treated samples (comparison between mock and each of the treated condition) using an unpaired Kruskal-Wallis test with multiple comparisons (Dunn’s Correction) and displayed for ∗∗P < .01, ∗P < .05, or not significant. (D-E,H-K) NGS analysis was performed as previously described⁶ using the primers and probes described in supplemental Table 2. A customized Python pipeline was used to align NGS reads to a reference amplicon sequence and count desired and partial edits, and other InDels. The “Desired edit” and “Partial edits” categories do not contain InDels and are stacked. InDels are located at the nick induced by the epegRNA or the ngRNA. Bars represent the mean ± SD of n = 3 for the panels D,E,H-J or n = 3 to 4 for the panels K-L biologically independent replicates. Statistical significance was assessed between the different groups using unpaired Kruskal-Wallis test with multiple comparisons (Dunn’s correction): data are not significantly different. WT, wild-type.

BCL11A BS disruption combined with generation of GATA1 and KLF1 BSs in the HBG1/2 promoters reactivates HbF in erythroid cells differentiated from SCD HSPCs. (A) Relative messenger RNA (mRNA) expression of genes encoding the main factors involved in MMR in K562 cells and peripheral blood CD34⁺ HSPCs from 3 donors with SCD, as measured by quantitative reverse transcription (qRT)-PCR as previously described.¹²MSH2, MSH6, MLH1, and PSM2 mRNA expression was normalized to the expression of ACTB mRNA. Bars represent the mean ± SD of n = 3 biologically independent replicates for K562 cells and n = 3 donors for SCD HSPCs. Statistical significance was assessed between K562 and HSPCs for each gene using an unpaired Mann-Whitney test with single comparisons (ie, each comparison stands alone): data are not significantly different. (B) Experimental protocol used for prime editing experiments in peripheral blood CD34⁺ HSPCs from donors with SCD. The PEmax or PEnmax mRNA (6.4 μg of mRNA in vitro transcribed as previously described,⁶ using the AM1334 or AM1345 kit, Thermo Fisher Scientific), epegRNA (200 pmol, Integrated DNA Technologies), with or without the ngRNA (200 pmol, Synthego), and GFP mRNA (0.5 μg, Tebubio), used here as an experimental control for transfection, were cotransfected (P3 Primary Cell 4D-Nucleofector X Kit [Lonza] and the CA137 program [Nucleofector 4D]) in SCD HSPCs 24 hours after cell thawing and culture in preactivation medium (StemSpan [STEMCELL Technologies] supplemented with penicillin/streptomycin [Thermo Fisher Scientific] and L-glutamine [Thermo Fisher Scientific], 750 μM of StemRegenin1 [STEMCELL Technologies], and the recombinant human cytokines (PeproTech), human stem cell factor (300 ng/mL), Flt-3L [300 ng/mL], thrombopoietin [100 ng/mL], and interleukin-3 [60 ng/mL]). After transfection, bulk HSPCs (ie, without any selection) were either cultured in the preactivation medium or plated in a semisolid medium to induce erythroid differentiation (colony-forming assay). Prime editing efficiency was evaluated in bulk HSPCs 6 days posttransfection and in BFU-Es (pools of 25 colonies and single colonies) by NGS (Illumina, as previously described¹⁵ using the primers and probes described in supplemental Table 2) and a customized Python pipeline that aligns NGS reads to a reference amplicon sequence and counts precise and partial edits and InDels (at the pegRNA or ngRNA nick sites). The presence of the 4.9-kb deletion was evaluated by ddPCR and PCR. Globin mRNA expression was quantified in single BFU-E by qRT-PCR and Hb expression in BFU-E pools (25 colonies) by high-performance liquid chromatography, as previously described.¹⁴ (C) Percentage of NGS reads containing the DEs (3 modifications: GATA1 BS, BCL11A-BS 2-bp deletion [ΔCC –117/–118 upstream of HBG1/2 TSS], and KLF1 BS; G_Δ2B_K), partial edit (G_Δ2B), or InDels in HSPCs from 3 donors with SCD using epegRNA10.4 with or without ngRNA10.5 and PE3max or PEnmax. (D) Frequency (%) of 4.9-kb deletion measured by ddPCR generated using epegRNA10.4 with or without ngRNA10.5 in HSPCs from 3 donors with SCD. The gray area indicates background 4.9-kb deletion levels (ie, observed in the control-treated samples). (E) Agarose gel images showing a band corresponding to the PCR product (HBG1/2 junction) obtained as described previously⁶ using the primers described in supplemental Table 2 surrounding the HBG1/2 junction (generated upon the deletion of the 4.9-kb region) in control (ctrl) SCD HSPCs or SCD HSPCs treated with PEmax or PEnmax (n = 3 donors). For the positive control, we used DNA from K562 cells with 60% of 4.9-kb deletion (ctrl⁺). For the negative control, we performed the PCR reaction in the absence of DNA (ctrl^–). (F) The sequences of the PBS and RTT of epegRNA10.4 are aligned on the most frequent NGS reads. The KLF1, GATA1, and BCL11A BSs are indicated in gray, bold, and purple, respectively; the desired base conversions in lower case; and the unexpected deletions with hyphens. Micro-homology motifs involved in deletions are also indicated. InDels included deletions disrupting the BCL11A BS or RTT insertions, both potentially reactivating HbF. (G) Frequency (%) of BFU-E and CFU-GM derived from ctrl^– and PE-transfected HSPCs. Results are represented as frequency of colonies obtained from 500 plated HSPCs and shown as mean ± SD. ns, not significant, unpaired Kruskal-Wallis test with multiple comparisons (Dunn’s correction). (H) Percentage of NGS reads containing the DEs (3 modifications: GATA1 BS, BCL11A-BS 2-bp deletion [CC –117/–118 upstream of HBG1/2 TSS], and KLF1 BS; G_Δ2B_K), partial edit (G_Δ2B), or InDels in SCD bulk BFU-Es using epegRNA10.4 with or without ngRNA10.5 and PE3max or PEnmax. Pools of 25 BFU-E colonies were analyzed for each condition. (I) Frequency (%) of 4.9-kb deletion measured by ddPCR in SCD bulk BFU-Es using epegRNA10.4 with or without ngRNA10.5 and PE3max or PEnmax for colonies shown in panel H. The gray area indicates background 4.9-kb deletion levels (ie, observed in the control-treated samples). (J) HbF and sickle Hb expression measured by high-performance liquid chromatography¹⁴ in SCD BFU-E pools treated with epegRNA10.4 with or without ngRNA10.5 and PE3max or PEnmax for colonies shown in panel H. (K) γ-Globin mRNA expression (qRT-PCR, normalized on α-globin mRNA as previously described¹⁴) in individual BFU-Es derived from SCD HSPCs (1 donor). HSPCs were mock-transfected or transfected with PEn and epegRNA10.4 or BE and gRNAs and cultured in a colony forming cell assay. Each single BFU-E was classified depending on its genotype (determined by NGS). Of note, deviation from expected 25% editing intervals (from 1-4 promoters) indicate that editing occurred over multiple progenitor divisions, as previously reported.^6,16 BFU-Es treated with PEn/epgRNA10.4 (n = 53 colonies) carry both DE (G_Δ2B_K) and InDels in variable proportions. For these colonies, we reported on the x-axis either the % of total edits (DE + InDels; red) or only the % of DEs (blue) as a function of γ-globin level for each BFU-E. In the blue group, some colonies have low editing frequency and a high variability of γ-globin level, which can be explained by the presence of InDels that leads to variable, unpredictable γ-globin expression (see InDel group, black line) and by some variability typical of the assay.^6,16,17 We also identified a single colony (black arrow) containing 24% of DEs with minimal InDels. This BFU-E expressed the highest γ-globin level, confirming the efficacy of our approach in reactivating γ-globin. In the case of control BFU-Es, we reported mock-transfected colonies (displaying minimal editing due to background NGS errors), colonies containing only InDels obtained from the PEn/epgRNA10.4-treated samples (“InDels”; n = 46 colonies; black; InDels’ frequency were indicated), and colonies containing the GATA1 BS and a disrupted BCL11A BS (A>G at positions –113 and –116 of the HBG promoter, respectively; n = 15 colonies; “+G/–B”; burgundy) or the KLF1 BS (T>C at position –198 of the HBG promoter; n = 16 colonies; “+K”; green), generated using the ABE8-13m¹⁸ and published gRNAs.^13,17 For BE-treated colonies, we reported the frequency of base conversion (conv.). InDels’ BFU-Es enable to evaluate whether DE in addition to InDels further boost γ-globin expression, whereas +G/–B and +K colonies allow us to compare whether multiple HPFH/HPFH-like mutations (ie, presence of the DE) enhanced γ-globin expression compared to individual mutations. The gray area indicates basal γ-globin expression level (ie, observed in the mock-treated sample). Linear regressions were determined for each category, except the mock. Lines were extended to start at x = 0, and they end exactly at the last data point. The R² of each linear regression is indicated on the graph. Statistical significance between 2 curves was assessed using linear regression: ∗∗P < .01, ∗P < .05, or not significant. (C-D,G-J) Bars represent the mean ± SD of n = 1 to 3 technical replicates per donor (3 donors for C-D and 2 donors for G-J). Tris-EDTA buffer-transfected samples and cells transfected with PEmax or PEnmax and no pegRNA/ngRNA were used as controls (ctrls). (C,H) The DE and partial edit categories do not contain InDels and are stacked. InDels are located at the nick induced by the epegRNA or ngRNA.

Comparison of prime- and base-editing inducing HPFH/HPFH-like mutations in SCD HSPCs and their erythroid progeny. (A) Schematic representation of the β-globin locus on chromosome (chr) 11 including the HBG1 and HBG2 genes and their promoters (prom). The light and dark gray bars denote the gRNAs used with BEs and their PAM, respectively^6,13,17-21 (supplemental Table 1). In the –200 region, gKLF198 induces the T>C (A>G) HPFH mutation at the –198 site generating a KLF1 motif, with a bystander at position –199. gTAL175-NG and gTAL175 generate a TAL1 motif by introducing the HPFH T>C (A>G) mutation at the –175 site, using NG (dotted line) and NGG (solid line) PAMs, respectively. In the –115 region, gKLF123/124 induces T>C (A>G) mutations at the –123 and –124 sites, creating a KLF1 BS. gBCL113 creates an A>G (T>C) mutation at the –113 site, generating GATA1 BS and A>G mutation at the –116 site disrupting the BCL11A BS and bystander edits at the –112 and/or –110 sites. (B) Experimental protocol used for prime and base editing experiments in peripheral blood (donor 1) or bone marrow (donors 2 and 3) CD34⁺ HSPCs from SCD donors. HSPCs were transfected with PEs using the protocol described in the legend to Figure 2B (supplemental Table 1). For base editing, cells were transfected with 3 μg of in vitro-transcribed BE mRNA (as previously described⁶) and 1 or 2 guide RNAs (100 pmol each; Integrated DNA Technologies; supplemental Table 1). Control (ctrl) samples were either mock-transfected or transfected with the PEnmax or the ABE8-13m mRNA alone. After transfection, bulk HSPCs (ie, without any selection) were either cultured in preactivation medium or differentiated in mature RBCs following the previously described method^6,14 or plated in a semisolid medium to induce erythroid differentiation (colony-forming assay). The editing efficiency was evaluated in bulk HSPCs and erythroid precursors 6 days posttransfection and in single BFU-Es by NGS (Illumina, as previously described⁶ using the primers and probes described in supplemental Table 2). The presence of the 4.9-kb deletion was evaluated by ddPCR and PCR as described in the Figure 2 legend. PE and BE efficiency was evaluated following the method described in the Figure 2B legend and the CRISPResso2 webtool,²² respectively. Globin mRNA expression was measured in erythroid precursors and single BFU-Es using qRT-PCR. Hemoglobin, erythroid markers, and enucleation were analyzed by flow cytometry in erythroid precursors and mature RBCs. Sickling assay was performed on mature RBCs. (C-D) Percentage of NGS reads in the HBG promoters in (C) HSPCs and (D) erythroid precursors. Prime editing modifications were reported as indicated in Figure 2C. Base editing was performed using the gRNAs displayed in panel A and ABE8-13m or ABEmax-NG. In the –115 region, +G, –B, and +K refer to GATA1 BS insertion (–113), BCL11A BS (–116) disruption, and KLF1 BS insertion (–124/123), respectively. In the –200 region, +K and +T create the KLF1 (–198) and TAL1 (–175) BSs, respectively. Bystanders (Byst.) were found in positions –110, –112, and –199. Control (ctrl) samples were either mock-transfected or transfected with the PEnmax or the ABE8-13m mRNA alone. (E) NGS reads from the HSPCs of donor 3, ABE8-13m, and gBCL113 generate a new GATA1 motif (blue) and disrupts the BCL11A BS (red). Editing with gKLF123/124 leads to the generation of a KLF1 motif (orange). Multiplex base editing using both gRNAs showed the presence of dual editing events (dual edit) and a high rate of InDels, as shown in the nucleotide distribution around the gRNAs. At each base in the reference amplicon, the percentage of each base as observed in sequencing reads is shown (A = green, C = orange, G = yellow, and T = purple). Black bars show the percentage of reads for which that base was deleted. Brown bars between bases show the percentage of reads having an insertion at that position (top). ABE8-13m and gKLF198 create a KLF1 motif (green). Editing with ABE8-13m and gTAL175 induces the generation of a TAL1 motif (pink). Both gRNAs together allow dual editing (bottom). (F) Frequency (%) of the 4.9-kb deletion measured by ddPCR in HSPCs and erythroid precursors from 3 donors with SCD following editing with PEs or BEs. Control (ctrl) samples were either mock-transfected or transfected with the PEnmax or the ABE8-13m mRNA alone. The gray area indicates background 4.9-kb deletion levels (ie, observed in the control-treated samples). (G) Agarose gel image showing a band corresponding to the PCR product (HBG1/2 junction) obtained using primers surrounding the HBG1/2 junction (generated upon deletion of the 4.9-kb region) in control (ctrl; mock-transfected or transfected with the PEnmax or ABE8-13m mRNA alone) SCD HSPCs or SCD erythroid precursors and treated with PE or BEs (donor 3). For the positive control, we used DNA from K562 cells displaying 60% of 4.9-kb deletion measured by ddPCR (ctrl⁺). For the negative control, we performed the PCR reaction in the absence of DNA (ctrl^–). (H) Frequency of CD36⁺, CD71⁺, CD235a⁺, and enucleated RBCs at day 13 (light gray) and 20 (dark gray) of erythroid differentiation, as measured by flow cytometry analysis of CD36, CD71, and CD235a erythroid markers and by DRAQ5 staining for donors 1 and 2. Control (ctrl) samples were mock-transfected. (I) Globin mRNA expression in erythroid precursors from 3 donors was measured by qRT-PCR. We reported the editing frequency below each graph. Control (ctrl) samples were mock-transfected. (J) Flow cytometry histograms showing the frequency of HbF⁺ cells in erythroid precursors from donors 1 and 2. For each sample, we also reported the editing frequency. Control (ctrl) samples were mock-transfected. (K) Frequency of sickling RBCs after 2-hour incubation under hypoxic conditions (0% O₂) for donor 1. We reported the editing frequency below each graph. Control (ctrl) samples were mock-transfected. (L) Globin mRNA expression in single BFU-Es from donor 1 was measured by qRT-PCR. We reported the editing frequency below each graph. Control (ctrl) samples were transfected only with the PEnmax mRNA. Statistical significance was assessed using an unpaired Kruskal-Wallis test with multiple comparison (Dunn’s correction) and displayed as ∗∗∗∗P < .001, ∗∗P < .01, and ∗P < .05.