Development and IND-enabling studies of a novel Cas9 genome-edited autologous CD34+ cell therapy to induce fetal hemoglobin for sickle cell disease

Abstract

Sickle cell disease (SCD) is a common, severe genetic blood disorder. Current pharmacotherapies are partially effective and allogeneic hematopoietic stem cell transplantation is associated with immune toxicities. Genome editing of patient hematopoietic stem cells (HSCs) to reactivate fetal hemoglobin (HbF) in erythroid progeny offers an alternative potentially curative approach to treat SCD. Although the FDA released guidelines for evaluating genome editing risks, it remains unclear how best to approach pre-clinical assessment of genome-edited cell products. Here, we describe rigorous pre-clinical development of a therapeutic γ-globin gene promoter editing strategy that supported an investigational new drug application cleared by the FDA. We compared γ-globin promoter and BCL11A enhancer targets, identified a potent HbF-inducing lead candidate, and tested our approach in mobilized CD34⁺ hematopoietic stem progenitor cells (HSPCs) from SCD patients. We observed efficient editing, HbF induction to predicted therapeutic levels, and reduced sickling. With single-cell analyses, we defined the heterogeneity of HbF induction and HBG1/HBG2 transcription. With CHANGE-seq for sensitive and unbiased off-target discovery followed by targeted sequencing, we did not detect off-target activity in edited HSPCs. Our study provides a blueprint for translating new ex vivo HSC genome editing strategies toward clinical trials for treating SCD and other blood disorders.

Keywords: CRISPR-Cas9; HSCs; autologous cellular therapy; ex vivo; fetal hemoglobin; gene therapy; genome editing; sickle cell disease; γ-gamma globin promoter editing.

Graphical abstract

Figure 1

HbF is more effectively induced by editing of −115 BCL11A binding motif than −196 ZBTB7A binding motif in the γ-globin promoters (A) Schematic of SpCas9-3xNLS genome editing target sites in the γ-globin promoters highlighting ZBTB7A (−196) and BCL11A (−115) binding motifs in red (boxed). Blue arrows with dotted line show the predicted SpCas9 cleavage position at −196 and −115 sites. Single-guide RNA (sgRNA) design locations are shown as a black line with an SpCas9 protospacer adjacent motif highlighted in a blue line (NGG) for respective sites. 13 nt deletion associated with hereditary persistence of fetal hemoglobin (HPFH) mutation is indicated with a black dotted line. (B) Indels for WT SpCas9 and HiFi SpCas9 (R691A) determined by NGS after editing at −196 and −115 γ-globin promoter targets using Lonza 4D nucleofector under optimal conditions. Dot plots shown as mean with solid line. (C) Percent of HbF estimated in erythroid differentiated cultures at day 21 by ion-exchange high-performance liquid chromatography (IE-HPLC). Dot plots show the results as a mean; adjusted p values for all samples are <0.0001∗∗∗∗ compared with controls (ordinary one-way ANOVA, Dunnett’s multiple comparison tests). Adjusted p values for −196 vs. −115 γ-globin targets with WT and HF are 0.0002∗∗∗ and 0.0024∗∗ (two-way ANOVA, Šídák’s multiple comparisons test). (D) On-target indels determined by NGS after editing at days 4 and 17 weeks of post-transplantation into NBSGW mice for different targets, including the BCL11A enhancer, −196 γ-globin promoter, and −115 γ-globin promoter (“ns” represents statistically not significant for input and 17 weeks, p values 0.374, 0.991, 0.347 and −115 HF γ-globin promoter (∗∗∗∗p < 0.0001) (two-way ANOVA, Šídák’s multiple comparisons test). Each dot represents data from an independent mouse at 17 weeks post-transplantation (n = 5). (E) Percent of HbF in human erythroid cells for control or edited CD34⁺ cells, measured by IE-HPLC at 17 weeks after xenotransplantation. Data shown as mean with solid line and adjusted p < 0.0001∗∗∗∗ (n = 5) compared with controls with each value shown as dots (ordinary one-way ANOVA, Dunnett’s multiple comparison tests) and adjusted p values for BCL11A enhancer vs. −115 with WT, −196 vs. −115 with WT, and −115 with WT vs. HF are 0.0169∗, <0.0001∗∗∗∗, and 0.0001∗∗∗ (ordinary one-way ANOVA, Tukey’s multiple comparison tests).

Figure 2

Editing of −115 γ-gamma globin promoter efficiently induces fetal hemoglobin and functionally reduces sickling in erythroid progeny of edited CD34⁺ HSPCs from SCD patients Plerixafor-mobilized CD34⁺ HSPCs derived from normal donor and three SCD donors electroporated with WT-Cas9-3xNLS and unedited controls were transplanted into mice and analyzed after 17 weeks. Each donor is annotated with different colors (orange, red, purple, pink) and each dot in the graph represents an individual mouse with a mean indicated by solid line. (A) Percentage of human engraftment (hCD45⁺) evaluated in mouse BM after 17 weeks by flow cytometry (p values are 0.992, 0.855, 0.948, and 0.977 and “ns” indicates statistically insignificant, two-way ANOVA, Šídák’s multiple comparisons test). (B) Percentages of human donor-derived lineages shown as human T (CD3⁺), B (CD19⁺), and myeloid (CD33⁺) cells in human CD45⁺ population in mouse BM after 17 weeks and erythroid lineage is shown as the %CD235a⁺ cells within the mouse and human CD34⁻ population. p values are >0.999, 0.959, 0.862, 0.9755 and statistically not significant compared with controls (two-way ANOVA, Šídák’s multiple comparisons test). (C) On-target indels measured by high-throughput sequencing 4 days after editing (infusion product/input) or after xenotransplantation into mice at 17 weeks (p values for normal, HbSS 1, and HbSS 2 are 0.993, 0.177, 0.723, not significant, and SCD donor 3 is 0.021∗ (two-way ANOVA, Šídák’s multiple comparisons test). (D) Percentages of HbF in human erythroid cells isolated from mouse BM measured by IE-HPLC. Adjusted p values are <0.0001∗∗∗∗ and significant (two-way ANOVA, Šídák’s multiple comparisons test). (E) In vitro differentiated human erythroid cells derived from edited, and unedited control cells were sorted for enucleated cells and incubated for 8 h in 2% O₂. Enumerated percent sickled reticulocytes from the images shown in Figure S4A and more than 300 cells/image from 3 technical replicates counted by 2 blinded observers. ∗∗∗∗p < 0.0001, ∗∗∗∗p < 0.0001, and ∗∗p < 0.0075 for HbSS, HbSS, and HbSC donors (two-way ANOVA, Šídák’s multiple comparisons test).

Figure 3

Editing the 115 γ-globin promoter site broadly induces fetal hemoglobin as measured by complementary single-cell analyses Analysis of CD235a⁺ cells from transplanted mouse BM derived from three SCD donors at 17 weeks. (A) Representative scatterplots of HbF-expressing cells by flow cytometry for control and Cas9-edited cells. F cell percentages are shown as mean (SD), with p values <0.0001∗∗∗∗ and significant for all three donors (two-way ANOVA, Šídák’s multiple comparisons test). (B) Scatterplots of α-globin (shown on the x-axis) and γ-globin (shown on the y-axis) comparing control and Cas9-3xNLS edited cells by scWestern analysis. (C) Kernel density plots showing the distribution of the percentage of HBG transcripts (HBG1+HBG2/HBG1+HBG2+HBB) in Cas9-edited erythroblasts compared with unedited controls. (D) UMAP plot showing annotated cell clusters at different stages of erythroid maturation from pooled erythroid samples derived from mouse BM. Terminal erythroid differentiation stages are classified as: proerythroblast/basophilic erythroblast (ProE/BasoE), early and late stages of polychromatophilic erythroblast (PolyE), and early and late stages of orthochromatic erythroblast (OrthoE). (E) Dot plot showing the percentage of each cluster at different stages of erythroid differentiation between control and Cas9 edited cells. Differences between control and Cas9 were not significant (i.e., ns) (two-way ANOVA, Šídák’s multiple comparisons test). (F) Volcano plot showing differentially expressed genes. Blue dots indicate the expressed genes with FDR ≤ 0.01, red dots indicate upregulated genes with both FDR ≤ 0.01 and log₂FC ≥ 1, and gray dots indicate not significant.

Figure 4

Development of a cGMP, clinical scale process for −115 γ-globin promoter editing (A) Streamline for Cas9-3xNLS purification under GMP conditions. (B) Schematic version of Cas9 protein with c-Myc nuclear localization signal (NLS) at the N-terminal and both SV40 NLS and bipartite NLS tags at the C-terminal. Reverse Phase UPLC analysis of purified Cas9-3xNLS fraction and purity of Cas9 evaluated by different methods. (C) Illustration of sgRNA sequencing of GMP HBG-sgRNA using SMARTer smRNA-seq technology. Bar and scatterplots show the percentage of reads mapped to full-length targeted sgRNA (shown on the y-axis) and insertions and deletions in the sgRNA sequence (shown on the x-axis). (D) Size-exclusion chromatography for RNP complexes and free sgRNA residues at different molar ratios of Cas9:sgRNA ranging from 1:0.5 to 1:3. Bar plot represents the percentages of free sgRNA at different Cas9-3xNLS:sgRNA ratios. Surface plot of optimization conditions through DOE studies using JMP, looking at editing efficiency with (E) Cas9 concentration and RNP ratio and (F) Cas9 concentration and cell concentration. (G) Measurement of indels in all hematopoietic lineages by NGS analysis before transplant (input) and after transplant in bulk BM, CD34⁺, CD33⁺ (myeloid), CD19⁺ (B cell), and CD235a⁺ (erythroid) using MaxCyte GTx (n = 10). (H) IE-HPLC analysis for HbF induction in erythroid cells after 17 weeks post transplantation (n = 10). Adjusted p values are <0.0001∗∗∗∗ for control vs. cas9 (two-way ANOVA, Šídák’s multiple comparisons test).

Figure 5

Characterization of potential on- and off-target genotoxicities associated with γ-globin promoter editing (A) Schematic representation of designated ddPCR probes and primers at HBG1 and HBG2 gene promoters. Simultaneous Cas9-induced DNA DSBs (red arrows) result in loss of 4.9 kb deletion represented in blue dotted lines. PacBio long-range PCR primers for the 14 kb region shown as purple dotted lines. (B) Measurement of 4.9 kb deletion by ddPCR in edited human CD34⁺ bulk HSPCs (pre-GMP engineering runs, n = 3, plerixafor-mobilized normal donors) at day 5. (C) Coverage of large deletions evaluated by long-range PCR-based PacBio sequencing in edited human HSPCs at day 5. (D) Manhattan plots of CHANGE-seq (n = 3) detected on- and off-target sites. Intended on-target site highlighted with arrows (pink) and possible off-target sites shown as bar heights. The x-axis represents chromosome location and the y-axis represents number of CHANGE-seq read counts. (E) Genomic features for off-target sites identified by CHANGE-seq and in silico predicted sites by CasOFFinder. TTS, transcription termination site; UTR, untranslated region; TSS, transcription start site. (F) Alignment of intended target site (top line) with CHANGE-seq detected on- and off-target sites (top to bottom by read count) for the Cas9:sgRNA (RNP) complex. The mismatches column indicates the number of mismatches relative to intended target site, where 0 indicates the on-target identified by CHANGE-seq. The coordinates column indicates the genomic coordinate for the on- and off-target sites identified by CHANGE-seq. Note: output is truncated to the top sites. (G) Genome-wide off-target activity evaluated by multiplex targeted sequencing (rhAmp-seq, IDT) based on on- and off-target sites nominated by CHANGE-seq and in silico tools such as CasOFFinder, CRISPRme, and CALITAS. The x-axis represents chromosomal location, and the y-axis indicates indels for control and Cas9 (pre-GMP engineering, n = 3 and engineering run, n = 1) (multiple paired t-tests comparing indel frequencies between edited and unedited cells using two-tailed paired t-tests, controlling for false discovery rate using the procedure of Benjamini, Krieger, and Yekutieli). (H) Circos plots representing genomic re-arrangements for the −115 γ-globin promoter and the −115 γ-globin promoter + BCL11A enhancer as a control by UDiTaS method.