In situ gene editing of hematopoietic stem cells via AAV-delivered CRISPR guide RNAs

Abstract

Hematopoietic stem cells (HSCs) are self-renewing, multipotent, and engraftable precursors of all blood cells. Efficient delivery of therapeutic gene products and gene editing machinery to correct disease-causing gene variants in endogenous HSCs while they remain in the body holds exciting potential to leverage HSC potency for the treatment of monogenic blood disorders. Toward this goal, we used adeno-associated virus (AAV) to deliver CRISPR guide RNAs (gRNAs) to edit HSC genomes in situ in Ai9;SpCas9-EGFP transgenic mice carrying a Cas9-activatable Lox-STOP-Lox-tdTomato reporter cassette together with a constitutive SpCas9-2A-EGFP. Using a variety of conditions and vector designs, we tested whether systemic administration to these mice of AAVs carrying SpCas9-compatible gRNAs designed to cut DNA upstream and downstream of the STOP cassette would induce tdTomato expression in HSCs. Our findings identify self-complementary AAVs (scAAVs) and increased ratio of guide to Cas9 as parameters facilitating higher editing efficiency. Of note, we find preserved multilineage output and engraftability of HSCs upon scAAV-gRNA editing. In an example application of this technology, we explore the potential for in situ HSC gene editing by dual AAV-CRISPR delivery and demonstrate robust gene modification, concurrent with induction of therapeutic fetal hemoglobin, in a sickle cell disease mouse model modified to express SpCas9. In summary, this work offers a sensitive and adaptable platform that allows robust modification of HSC genomes in situ.

Graphical abstract

Figure 1.

Impact of vector design on CRISPR-mediated gene editing of HSCs and more mature hematopoietic lineages in situ in adult Ai9;SpCas9-EGFP mice. (A) Schematic of Ai9 transgene in Rosa26 locus. Double-stranded break induction and excision of LoxP-flanked STOP cassette can be detected by robust tdTomato expression. Triangles: LoxP sites. (B) Vector design. ssAAV or scAAVs carrying gRNAs driven by U6, H1, or 7SK promoters. ssAAV-1xgRNAs and scAAV-1xgRNAs carry 1 copy of each gRNA, whereas scAAV-2xgRNAs encode 2 copies of each gRNA. (C-E) Frequency of tdTomato⁺ cells among BM HSCs in adult (5 to 8-week-old) animals receiving IV injections of vehicle or AAV and analyzed 5 weeks after injection. Animals received vehicle only or 5 × 10¹¹ vg or 5 × 10¹² vg per animal of ssAAV-1xgRNAs (C), vehicle only or 5 × 10¹² vg of either ssAAV-1xgRNAs or scAAV-1xgRNAs (D), or vehicle only or 5 × 10¹² vg per animal of either scAAV-1xgRNAs or scAAV-2xgRNAs (E). One-way analysis of variance (ANOVA) with Brown-Forsythe and Welch correction. ∗P < .05; ∗∗∗P < .001. Data compiled from 2 (panels C-D) or 4 (panel E) independent experiments. Females, square; males, triangles. (F) Frequency of tdTomato⁺ cells among different cellular subsets in the spleen, BM, or thymus of Ai9;SpCas9-EGFP mice injected with vehicle or 5 × 10¹² vg per mouse scAAV-2xgRNAs. B cell, B220⁺; T cell, CD3⁺; myeloid, B220⁻CD3⁻, EryA, CD71^highTer119⁺; EryB, CD71^intermediateTer119⁺, EryC, CD71⁻Ter119⁺; CD4 SP, CD3⁺CD4⁺CD8⁻; DP, CD3⁺CD4⁺CD8⁺; DN, CD3⁺CD4⁻CD8⁻; CD8 SP, CD3⁺CD4⁻CD8⁺. Unpaired t test with Welch correction. ∗P < .05; ∗∗P < .01; ∗∗∗P < .001; ∗∗∗∗P < .0001. Data compiled from 3 independent experiments. dITR, defective inverted terminal repeat; DN, double negative; DP, double positive; ITR, inverted terminal repeat; EryA, basophilic erythroblasts; SP, single positive. Panels A and B created with biorender.com. Karimzadehfard A. (2025) https://biorender.com/1fldmz.

Figure 2.

Functional analyses of in situ gene-edited HSCs. (A) Frequency of tdTomato⁺ cells among HSCs (CD150⁺CD48⁻LSK) or among the indicated subsets of BM progenitors in adult injected animals 5 weeks after receiving 5 × 10¹² vg per animal scAAV-2xgRNAs. One-way ANOVA with Brown-Forsythe and Welch correction. ∗P < .05; ∗∗P < .01; ∗∗∗∗P < .0001. Data compiled from 4 independent experiments. Females, square; males, triangles. (B) Frequency of primitive progenitors within tdTomato⁻ and tdTomato⁺ subsets of Lin^loKit⁺ BM progenitors at 5 weeks after injection of adult animals with 5 × 10¹² vg per animal scAAV-2xgRNAs. One-way ANOVA with Brown-Forsythe and Welch correction. ∗P < .05. Data compiled from 4 independent experiments. (C) Experimental design. For CFU assay, HSCs from animals injected with vehicle or NT vector (all tdTomato⁻), and tdTomato⁺ or tdTomato⁻ HSCs from animals receiving 5 × 10¹² vg per animal scAAV carrying targeting gRNAs, were sorted 5 weeks after injection and plated in methylcellulose. Colonies were scored visually at day 12. For transplant, Lin^loKit⁺ HSPCs from the tdTomato⁺ or tdTomato⁻ fractions of the BM were sorted and transplanted into sublethally irradiated NSG recipients. Peripheral blood was analyzed by FACS every 4 weeks, and mice were euthanized >1 week after last bleed for BM chimerism analysis (Table 1; supplemental Figure 6 for results). (D) Percentages of CFU-GEMM, BFU, and CFU-GM formed from HSCs sorted from animals injected with vehicle or NT vector, or from tdTomato⁺ or tdTomato⁻ HSCs sorted from animals injected as adults with 5 × 10¹² vg per animal scAAV carrying targeting gRNAs. Total number of colonies scored is as follows: vehicle, 288; NT vector, 264; tdTomato⁻, 279; and tdTomato⁺, 225. Individual colony numbers provided in supplemental Table 2. One-way ANOVA with Brown-Forsythe and Welch correction. ∗P < .05. Data compiled from 2 independent experiments. BFU-E, burst-forming unit-erythroid; CFU, colony-forming unit; GEMM, granulocyte, erythrocyte, macrophage, megakaryocyte; GM, granulocyte, macrophage; NT, nontargeting. Panel C created with biorender.com. Karimzadehfard A. (2025) https://biorender.com/mz1lx3u.

Figure 3.

Robust in situ HSC gene editing in neonatally injected mice. (A) Percentage TdTomato⁺ among BM HSCs at 8 weeks after injection of P1 to P3 neonates. Animals received vehicle control or the indicated dose (vg) of ssAAV-1xgRNAs. One-way ANOVA with Brown-Forsythe and Welch correction. ∗P < .05. Data compiled from 3 independent experiments. (B) Comparison of percentage tdTomato⁺ HSCs in animals receiving vehicle or 1 × 10¹² vg per mouse of either ssAAV-1xgRNAs or scAAV-1xgRNAs or scAAV-2xgRNAs. One-way ANOVA with Brown-Forsythe and Welch correction. Data compiled from 4 independent experiments. (C) Frequency of TdTomato⁺ cells among different cell lineages in the spleen, BM, and thymus of animals injected with vehicle or 5 × 10¹² vg per animal scAAV-2xgRNAs. B cell, B220⁺; T cell, CD3⁺; myeloid, B220⁻CD3⁻; EryA, CD71^highTer119⁺; EryB, CD71^intermediateTer119; EryC, CD71⁻Ter119⁺; CD4 SP, CD3⁺CD4⁺CD8⁻; DP, CD3⁺CD4⁺CD8⁺; DN, CD3⁺CD4⁻CD8⁻; and CD8 SP, CD3⁺CD4⁻CD8⁺. Unpaired t test with Welch correction. ∗P < .05; ∗∗P < .01; ∗∗∗P < .001; ∗∗∗∗P < .0001. Data compiled from 4 independent experiments. (D) Frequency of TdTomato⁺ cells among downstream BM progenitors in animals that received 1 × 10¹² vg scAAV-2xgRNAs. One-way ANOVA with Brown-Forsythe and Welch correction. Data compiled from 4 independent experiments. (E) Design of dual AAV vectors delivering separated CRISPR editing components. (F) Frequency of TdTomato⁺ HSCs in single or dual AAV delivery groups in neonatally injected Ai9;SpCas9-EGFP at 8 weeks after injection. Neonates were injected systemically with vehicle or with 2 × 10¹¹ vg per animal of either scAAV-5′ gRNA only or scAAV-3′ gRNA only or scAAV-5′ gRNA and scAAV-3′ gRNA together or scAAV-2xgRNAs (ie, scAAV delivering 5′ gRNA and 3′ gRNA in 1 vector). One-way ANOVA with Brown-Forsythe and Welch correction. ∗P < .05. Data compiled from 2 independent experiments. Females, square; males, triangles. dITR, defective inverted terminal repeat; ITR, inverted terminal repeat. Panel E created with biorender.com. Karimzadehfard A. (2025) https://biorender.com/3vjs8fl.

Figure 4.

Efficient HSC editing at the HBG1 promoter in Townes;SpCas9-EGPF animals.HBB^A/A and HBB^S/S animals were IV injected as adults (A,C,E) or neonates (B,D,F) with vehicle or scAAV-3xHBG1.gRNA and analyzed 12 weeks after injection. Approximately 4 × 10¹² vg per animal was used for adults and ∼2 × 10¹¹ vg per animal was used for neonates. HBG1 promoter modification in HSCs of adult injected (panel A) or neonatally injected (panel B) animals. HBG1 promoter modification in BM Ery progenitors from adult injected (panel C) or neonatally injected (panel D) animals. HBG1 promoter modification in whole blood from adult injected (panel E) or neonatally injected (panel F) animals and analyzed 12 weeks after injection. One-way ANOVA with Brown-Forsythe and Welch correction. ∗P < .05; ∗∗P < .01; ∗∗∗P < .001; ∗∗∗∗P < .0001. Data compiled from 4 (panels A,C,E) or 3 (panels B,D,F) independent experiments. Females, square; males, triangles. A/A, healthy HBB^A/A; BM Ery, BM erythroid; S/S, sickle homozygote HBB^S/S.

Figure 5.

HbF induction in HBG1 promoter edited animals. (A-B) Flow cytometric analysis of the percentage of RBCs containing HbF (F+) in adult injected (A) and neonatally injected animals (B) at the terminal time point. (C-D) HPLC analysis of percent HbF in blood lysates collected at terminal time point from adult injected (C) and neonatally injected animals (D). (E-F) CBC analysis at the terminal time point for hemoglobin levels (gram/deciliter) in adult injected (E) or neonatally injected animals (F). (G-H) CBC analysis at the terminal time point for RBC count (×10⁶ cells per μL) in adult injected (G) or neonatally injected mice (H). One-way ANOVA with Brown-Forsythe and Welch correction. ∗P < .05; ∗∗P < .01; ∗∗∗P < .001; ∗∗∗∗P < .0001. Data compiled from 4 (panels A,C,E) or 3 (panels B,D,F) independent experiments. Females, square; males, triangles. A/A, healthy HBB^A/A; S/S, sickle homozygote HBB^S/S.