Gene editing without ex vivo culture evades genotoxicity in human hematopoietic stem cells

Abstract

Gene editing the BCL11A erythroid enhancer is a validated approach to fetal hemoglobin (HbF) induction for β-hemoglobinopathy therapy, though heterogeneity in edit allele distribution and HbF response may impact its safety and efficacy. Here, we compare combined CRISPR-Cas9 editing of the BCL11A +58 and +55 enhancers with leading gene modification approaches under clinical investigation. Dual targeting of the BCL11A +58 and +55 enhancers with 3xNLS-SpCas9 and two single guide RNAs (sgRNAs) resulted in superior HbF induction, including in sickle cell disease (SCD) patient xenografts, attributable to simultaneous disruption of core half E-box/GATA motifs at both enhancers. Unintended on-target outcomes of double-strand break (DSB) repair in hematopoietic stem and progenitor cells (HSPCs), such as long deletions and centromere-distal chromosome fragment loss, are a byproduct of cellular proliferation stimulated by ex vivo culture. Editing quiescent HSPCs bypasses long deletion and micronuclei formation and preserves efficient on-target editing and engraftment function.

Keywords: BCL11A; enhancers; fetal hemoglobin; genotoxicity; hematopoietic stem cells; sickle cell disease; therapeutic gene editing.

Figure 1.. Robust HbF induction by combined editing of the +58 and +55 BCL11A erythroid enhancers in HSPCs

(A) ATAC-seq following BCL11A enhancer editing in in vitro erythroid differentiated progeny. (B) HbF induction in in vitro erythroid differentiated progeny. Mean ± SEM with Dunnett’s multiple comparisons test comparing sg1617+sg1450 to other groups. (C) HbF level from individual clonal liquid primary erythroid cultures. Median with range with Dunnett’s multiple comparisons test comparing sg1617+sg1450 to other groups. (D) Fraction of sickled cells from two SCD donors and two technical replicates per donor. Mean ± SEM with Dunnett’s multiple comparisons test comparing sg1617+sg1450 to other groups. (E) Human chimerism 16 weeks post xeno-transplantation. Grand median with Dunnett’s multiple comparisons test comparing sg1617+sg1450 to other groups. n = 11–13 primary recipients. (F) Indel frequencies for sgHBG1/2–115, sg1617 and sg1450 and overall editing by ddPCR for sg1617+sg1450. Median with range and with unpaired two-tailed Student’s t-test. n = 3 donors for input, n = 11–13 primary recipients for engrafted. (G) HbF levels in engrafted erythroid cells. Grand median with Dunnett’s multiple comparisons test comparing sg1617+sg1450 to other groups. n = 11–13 primary recipients.

Figure 2.. Potent disruption of TGN_7–9WGATAR motifs after combined editing of the BCL11A erythroid enhancers

(A) Schema of ddPCR assays. (B) Frequencies of edit allele types by ddPCR assays in engrafted compared to input cells. n=3 healthy and 2 SCD donors for input, 26 primary recipients and 15 secondary recipients for sg1617+sg1450 editing. n = 1 healthy donor for sg1617+sg1449 input and n = 3 primary recipients for engrafted. n = 3 healthy donors for sg1618+sg1450 input and n = 10 primary recipients for engrafted. Median with range. (C) Frequencies of edit allele types by ddPCR assays for bulk cells corresponding to colonies in Figures 2E and S3K–S3N. (D) Schema of sg1617+sg1450, sg1618+sg1450, sg1617+sg1449 editing and observed edit alleles that preserve TGN_7–9WGATAR motifs. (E) TGN_7–9WGATAR motif disruption and edit types compared to HbF levels from colonies subject to combined BCL11A enhancer editing with indicated sgRNA pairs. Grand median, with simple linear regression. (F) ATAC-seq accessible chromatin reads with intact or disrupted TGN_7–9WGATAR motifs. Mean±SEM, n = 3 healthy donors.

Figure 3.. Off-target analysis of combined BCL11A +58 and +55 enhancer editing

(A) Off-target sites by in silico, in vitro or in cellulo nomination methods with verification attempt by pooled amplicon sequencing for sg1617, sg1618, sg1450 and sg1449. (B) Mean difference between indels in edited and control samples. The threshold of indel frequency for true-positive significant off-target editing is set as 0.1% increase in the edited sample as compared to in the paired control sample in the mean of informative donors (dotted line).

Figure 4.. Gene editing HSPCs without ex vivo culture evades micronucleation

(A) Cell size without (0 h) and with 24 h and 48 h of cytokine culture. (B) HSPCs in G0, G1, S and G2/M immediately after thawing or selection or indicated pre-stimulation culture from three cryopreserved healthy donors mobilized with G-CSF and one fresh SCD patient donor mobilized with plerixafor. Mean ± SEM. (C) CDKN1A expression, by RT-qPCR, in HSPCs after RNP electroporation (EP). Unpaired two-tailed Student’s t-test. n = 3 healthy donors. (D) Micronucleus analysis by flow cytometry (on the top). Mean ± SEM with the unpaired two-tailed Student’s t-test. n = 3–5 healthy donors. (E) Design of ddPCR assays recognizing chr2p (telomeric to cleavage site) and chr1 as reference autosome. Enrichment of chr2p in sorted nuclei (left) and micronuclei (right). Mean ± SEM, analyzed with the unpaired two-tailed Student’s t-test. n = 3–4 healthy donors.

Figure 5.. Gene editing HSPCs without ex vivo culture evades nonprogrammed long deletions or rearrangements

(A) Frequency of nonprogrammed long deletions or rearrangements by ddPCR. Mean ± SEM with the unpaired two-tailed Student’s t-test. n = 3 – 5 healthy donors. (B) Frequency of long deletions or rearrangements by ddPCR. Mean± SEM with the unpaired two-tailed Student’s t-test for HSC and HPC. Dunnett’s multiple comparisons test comparing cell cycle groups. n = 3 technical replicates. (C) Frequency of long deletions or rearrangements by ddPCR following ELANE exon 2 editing in HPSCs. Mean± SEM with Dunnett’s multiple comparisons test comparing groups. n = 3 independent donors and two replicates for each donor. (D) Frequency of long deletions or rearrangements by ddPCR following Safe-chr4q33 editing. Mean± SEM with Dunnett’s multiple comparisons test comparing groups. n=3 technical replicates from one donor. (E) Agarose gel following amplification of 6 kb around BCL11A +58 to +55 enhancers. (F) Unique nonprogrammed long deletions by long read sequencing. Dotted lines show short indel edit window of cleavage site +/− 100 bp. (G) Unique nonprogrammed long deletions by long read sequencing. Reads downsampled to normalize sequencing depth.

Figure 6.. Ex vivo editing without cytokine culture produces potent engraftment, preserves gene edits, and reactivates HbF

(A) Experimental schema. (B) Cell viability after selection, thaw, or electroporation as indicated. n = 1 healthy and 1 SCD donor for fresh and. n = 2 healthy donors for cryopreserved HSPCs. Brown: mock, blue: sg1618+sg1449, red: sg1617+sg1450. (C) Frequenciy of edit allele types by ddPCR Mean ± SEM, n = 2 healthy and 1 SCD donor. (D) HbF levels in in vitro erythroid differentiated progeny. Mean ± SEM, n = 2 healthy and 1 SCD donor. (E) Human chimerism 16 weeks post xeno-transplantation. Grand median, n = 7–13 primary recipients. (F) Frequency of edit allele types by ddPCR. Median with range and unpaired two-tailed Student’s t-test. n = 7–12 primary recipients. (G) Frequency of edited alleles with motif editing. Mean ± SEM, for input n=3 techinical replicates from SCD3, for engrafted n=2–3 primary recipients for each condition. (H) Unique nonprogrammed long deletions by long read sequencing. Dotted lines show short indel edit window of cleavage site +/− 100 bp. (I) Unique nonprogrammed long deletions by long read sequencing. Reads downsampled to normalize sequencing depth. (J) Diversity index by analyzing major edit junction amplicon sequencing. n = 3 technical replicates for input, n = 2 – 3 recipients. (K) HbF levels in engrafted erythroid cells. Grand median. ns: nonsignificant. n = 6–13 primary recipients.