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.