CRISPR/Cas9 β-globin gene targeting in human haematopoietic stem cells

Abstract

The β-haemoglobinopathies, such as sickle cell disease and β-thalassaemia, are caused by mutations in the β-globin (HBB) gene and affect millions of people worldwide. Ex vivo gene correction in patient-derived haematopoietic stem cells followed by autologous transplantation could be used to cure β-haemoglobinopathies. Here we present a CRISPR/Cas9 gene-editing system that combines Cas9 ribonucleoproteins and adeno-associated viral vector delivery of a homologous donor to achieve homologous recombination at the HBB gene in haematopoietic stem cells. Notably, we devise an enrichment model to purify a population of haematopoietic stem and progenitor cells with more than 90% targeted integration. We also show efficient correction of the Glu6Val mutation responsible for sickle cell disease by using patient-derived stem and progenitor cells that, after differentiation into erythrocytes, express adult β-globin (HbA) messenger RNA, which confirms intact transcriptional regulation of edited HBB alleles. Collectively, these preclinical studies outline a CRISPR-based methodology for targeting haematopoietic stem cells by homologous recombination at the HBB locus to advance the development of next-generation therapies for β-haemoglobinopathies.

Extended Data Figure 1. High tropism of rAAV6 for CD34⁺ HSPCs, and viability and specificity assessment of gene editing in CD34⁺ HSPCs.

(A) CD34⁺ HSPCs were transduced with a scAAV6 expressing GFP from an SFFV promoter at multiplicities of infections (MOIs) of 10,000 vg/cell or 100,000 vg/cell for 48hrs and then analyzed for percent GFP expression by flow cytometry using a non-transduced sample to set the GFP⁺ gate at <0.1% GFP+ cells. scAAV was used because it eliminates second strand synthesis as a confounder of actual transduction. Results are from two independent experiments from at least two donors and error bars represent S.D. ABM: Adult Bone Marrow; mPB: Mobilized Peripheral Blood; CB: Cord Blood. (B) CD34⁺ HSPCs were electroporated with the HBB CRISPR system (mRNA or RNP delivery) or without (AAV only), and then transduced with HBB rAAV6 donor vectors at an MOI of 100,000. Day 4 post electroporation, cells were analyzed by flow cytometry and live cells were gated in high forward scatter (FSC) and low side scatter (SSC). Percent cells in FSC/SSC gate is shown relative to that of Mock-electroporated cells. Each dot represents a unique CD34⁺ HSPC donor. (C) Upper panel: sgRNA target sequences at the HBB on-target site and a highly complementary off-target site (Chr9:101833584-101833606) are shown. PAM sequences are underlined and red sequence highlights the 3 mismatches of the off-target site. Lower panel: HSPCs were electroporated with either the “All RNA” or RNP-based CRISPR system, and 4 days post electroporation gDNA was extracted and analyzed for INDEL frequencies using TIDE at the on-target HBB and the off-target site. Results are graphed as a ratio of on to off-target activity highlighting the increased specificity of the RNP system. Averages from three different CD34⁺ HSPC donors are shown and error bars represent S.E.M. ** p < 0.01, **** p < 0.0001, ns = p ≥ 0.05, unpaired Student’s t-test. (D) INDEL frequencies for the data presented in (C) * p < 0.05, paired Student’s t-test. (E) Representative FACS plots showing stable GFP rates at Day 18 post-electroporation in donor-nuclease mismatch experiments. Mismatching nuclease and donor (red box) leads to infrequent end-capture events compared to on-target HR events observed with matched nuclease and homologous rAAV6 donor (green box). HSPCs were electroporated with 15μg Cas9 mRNA and either HBB MS sgRNA or IL2RG MS sgRNA, then transduced with HBB-GFP rAAV6 donor followed by 18 days of culture. (F) End-capture experiments were performed in three replicate experiments each in three unique CD34+ HSPC donors. ns = p ≥ 0.05, paired Student’s t-test. (G) Activity of the IL2RG CRISPR was confirmed by quantification of INDELs at the IL2RG target site using TIDE analysis

Extended Data Figure 2. Schematic of targeting rAAV6 E6V homologous donor to the HBB locus.

(A) The human HBB locus on chromosome 11 is depicted at the top of the schematic and consists of three exons (black boxes) and two introns. The rAAV6 E6V donor includes the glutamic acid (E) to valine (V) mutation at codon 6, which is the amino acid change causing sickle cell disease. Other SNPs (all SNPs are capitalized) were introduced to PAM site (blue) and sgRNA binding site (bold) to prevent recutting following HR in HSPCs. To analyze targeted integration frequencies in HSPCs, a 2-step PCR was performed. First, a 3400bp In-Out PCR (green) was performed followed by a nested 685bp PCR (purple) on a gel-purified fragment from the first PCR. This 2^nd PCR fragment was cloned into TOPO vectors, which were sequenced to determine the allele genotype (WT, INDEL, or HR). (B) The sequence of a wild-type HBB allele aligned with the sequence of an allele that has undergone HR. (C) Representative INDELs from the data represented in Figure 1d. The HBB reference sequence is shown in green.

Extended Data Figure 3. Linear regression model shows that the Day 4 GFP^high population is a reliable predictor of targeting frequencies.

Day 4 GFP^high percentages (x-axis) were plotted against Day 18 total GFP⁺ percentages (y-axis), and linear regression was performed. Data was generated from experiments including a total of 38 different CD34⁺ HSPC donors, treated with either 15µg or 30µg Cas9 RNP to generate data points with a wider distribution of targeting frequencies.

Extended Data Figure 4. Overview of PCR genotyping of methylcellulose colonies with HR of the GFP and tNGFR donor at the HBB locus.

(A) The HBB locus was targeted by creating a DSB in exon 1 via Cas9 (scissors) and supplying a rAAV6 GFP donor template. Alleles with integrations were identified by PCR (red, 881bp) on methylcellulose-derived colonies using an In-Out primer set. Wildtype alleles were identified by PCR (green, 685bp) using primers flanking the sgRNA target site. (B) Representative genotyping PCRs showing mono- and bi-allelic clones as well as a clone derived from Mock-treated cells. NTC = non-template control (see Suppl. Fig. 1a for uncropped gel). (C) Representative Sanger sequence chromatograms for junctions between right homology arm (in blue) and insert (in green) or genomic locus (in white) highlighting seamless homologous recombination. (D) The HBB locus was targeted by creating a DSB in exon 1 via Cas9 (scissors) and supplying a rAAV6 tNGFR donor template. Genotypes were assessed by a 3-primer genotyping PCR on methylcellulose-derived colonies using an In-Out primer set (red, 793bp) and a primer set flanking the sgRNA target site (green, 685bp). Note that the two forward primers are the same. (E) Representative genotyping PCRs showing a WT/unknown, mono-, and bi-allelic clone (see Suppl. Fig. 1b for uncropped gel). (F) Representative Sanger sequence chromatograms for junctions between left homology arm (in blue) and insert (in green) or genomic locus (in white) highlighting seamless homologous recombination.

Extended Data Figure 5. Hematopoietic progenitor colony-forming unit (CFU) assay and targeting in different HSPC subpopulations.

(A) GFP^high HSPCs were single cell-sorted into 96-well plates containing methylcellulose. Representative images from fluorescence microscopy show lineage-restricted progenitors (BFU-E, CFU-E, CFU-GM) and multipotent progenitors (CFU-GEMM) with GFP expression. (B) Colony forming units (CFUs) derived as described above were counted 14 days post sort and shown relative to the total number of cells sorted (% cloning efficiency) (N=2 different HSPC donors). (C) Colonies from above were scored according to their morphology: 1) CFU-Erythroid (CFU-E), 2) Burst Forming Unit-Erythroid (BFU-E), 3) CFU-Granulocyte/Macrophage (CFU-GM), and 4) CFU-Granulocyte/Erythrocyte/Macrophage/Megakaryocyte (CFU-GEMM). (N=2 different HSPC donors) (D) 500,000 HSPCs isolated from mobilized peripheral blood (mPB), adult bone marrow (ABM), or cord blood (CB) were electroporated with RNP and transduced with GFP rAAV6 donor. At Day 4 post-electroporation, cells were phenotyped by flow cytometry for the cell surface markers CD34, CD38, CD90, and CD45RA (Supplemental Figure 2). Percent GFP^high cells in the indicated subpopulations are shown (data points represent unique donors, N=3 per HSPC source), **** p < 0.0001, paired Student’s t-test. (E) CD34+ or CD34+/CD38-/CD90+ cells were sorted directly from freshly isolated cord blood CD34+ HSPCs, cultured overnight, and then electroporated with RNP and transduced with GFP rAAV6. Bars show average percent GFP+ cells at Day 18 post-electroporation. Error bars represent S.E.M. (N=3 from different HSPC donors), ** p < 0.01, paired Student’s t-test. (F) MPP (CD34⁺/CD38^-/CD90^-/CD45RA^-) and HSC (CD34⁺/CD38^-/CD90⁺/CD45RA^-) populations were sorted from fresh cord-blood-derived CD34+ HSPCs and immediately after sorting, cells were transduced with scAAV6-SFFV-eGFP at an MOI of 100,000 vg/cell along the bulk HSPC population. scAAV6 was used because it eliminates second strand synthesis as a confounder of actual transduction, though the activity of the SFFV promoter may not be equivalent in each population thus potentially underestimating the degree of transduction of MPPs and HSCs. Two days later, transduction efficiencies were measured by flow cytometric analysis of eGFP expression using non-transduced cells (Mock) to set the GFP⁺ gate. Error bars represent S.E.M., N=4, two different HSPC donors, ns = p ≥ 0.05, ** p < 0.01, unpaired t test with Welch's correction.

Extended Data Figure 6. Analysis of human engraftment.

(A) Representative FACS plot from the analysis of the bone marrow of a control mouse not transplanted with human cells. Mice were sacrificed and bone marrow was harvested from femur, tibia, hips, humerus, sternum, and vertebrae. Cells were subject to Ficoll density gradient to isolate mononuclear cells, which were analyzed for human engraftment by flow cytometry. Human engraftment was delineated as huCD45/HLA-ABC double positive. 4 out of 157,898 cells were found within the human cell gate. (B) Representative FACS plots showing gating scheme for analyses of NSG mice transplanted with human cells and analyzed as described in (A). Representative plots are from one mouse from the RNP + rAAV6 experimental group. As above, human engraftment was delineated as huCD45/HLA-ABC double positive. B cells were marked by CD19 expression, and myeloid cells identified by CD33 expression. GFP expression was analyzed in total human cells (2.4%), B-cells (1.9%) and myeloid cells (2.8%). The GFP brightness in B cells is lower than in myeloid cells suggesting that the SFFV promoter is not as active in the B-cell lineage compared to the myeloid lineage (see also Fig 4a). (C) Overview of engraftment for RNP+AAV and RNP+AAV GFP^high experimental groups. Average engraftment frequencies and percent GFP⁺ human cells +/-S.E.M are shown. Total number of cells transplanted was the same (500,000) for all mice in the RNP group, whereas in the GFP^high group, one mouse was transplanted with 100,000 cells, two mice with 250,000 cells, and three mice with 500,000 cells. The total number of HSCs transplanted per mouse (+/-S.E.M.) was calculated based on the frequencies of GFP⁺ cells in the CD34⁺/CD38^-/CD90⁺/CD45RA^- subset analyzed by flow cytometry (see Fig 3d) directly before injection. The total number of modified human cells in the bone marrow (BM) at Week 16 post transplant per mouse (+/-S.E.M.) was estimated based on calculations presented in the materials and methods. This shows that the enrichment not only resulted in a higher percentage of edited cells (column 3) but also resulted in an absolute higher number (column 6) of edited cells as well.

Extended Data Figure 7. Genome-edited human HSPCs in the bone marrow of NSG mice at Week 16 post transplantation.

Representative FACS plots from the analysis of NSG mice from the (A) Mock or (B) RNP+AAV experimental group at Week 16 post transplantation. Mice were sacrificed and bone marrow was harvested, PBMCs were isolated via Ficoll density gradient, after which human CD34⁺ cells were enriched by magnetic-activated cell sorting (MACS), and finally cells were stained with antiCD34, antiCD38, and antiCD10 antibodies to identify human GFP⁺ cells in the CD34⁺/CD10^- and CD34⁺/CD10^-/CD38^- populations (note that CD10 was included as a negative discriminator for immature B cells) (C) Collective data from the analysis of GFP⁺ cells in the human CD34⁺/CD10^- population from the RNP+AAV (N=11) and RNP+AAV GFP^high (N=6) experimental groups. For the RNP+AAV GFP^high group, cells from all six mice were pooled before analysis and thus, no error bar is available. Error bar on RNP group represents S.E.M.

Extended Data Figure 8. Correction of the sickle cell mutation in patient-derived CD34+ HSPCs.

(A) Schematic overview of the sequence of the sickle allele aligned with the sequence of an allele that has undergone HR using the corrective SNP donor. The E6V mutation in sickle cell patients (A->T) is highlighted in yellow. The sgRNA recognition sequence, the PAM site, and the cut site (scissors) are shown. The donor carries synonymous nucleotide changes between the sickle nucleotide and the cut site to avoid premature cross-over during HR. Synonymous changes are also added to the PAM and an early nucleotide in the sgRNA target site to avoid subsequent re-cutting and potential inactivation of the corrected allele. (B) HSPCs from two different sickle cell patients were targeted with the corrective SNP donor and seeded in methylcellulose. After 14 days, In-Out PCR amplicons from a total of 38 clones were sequenced and genotypes were extracted from sequence chromatograms.

Extended Data Figure 9. Edited HSPCs from sickle cell patients differentiate into erythrocytes that express Glycophorin A.

CD34⁺ HSPCs derived from sickle cell patients were edited with HBB Cas9 RNP and either the corrective SNP donor or the cDNA donor. 4 days post-electroporation, cells edited with the cDNA donor were sorted for tNGFR⁺ cells. This population as well as the populations edited with the corrective SNP donor and Mock-electroporated cells were subjected to a 21-day erythrocyte differentiation protocol, followed by staining for Glycophorin A (GPA). All data points within experimental groups are derived from experiments in cells from different sickle cell patients, N=3 (Mock) and N=2 (SNP and cDNA donor).

Fig 1. CRISPR/Cas9 and rAAV6-mediated targeted integration at the HBB locus in human CD34⁺ hematopoietic stem and progenitor cells (HSPCs).

a) Schematic of targeted genome editing at the HBB locus using CRISPR/Cas9 and rAAV6. Site-specific double strand breaks (DSBs) are created by Cas9 (scissors) mainly between nucleotide 17-18 of the 20bp target site, which is followed by the ‘NGG’ PAM (red). A DSB stimulates homologous recombination (HR) using rAAV6 homologous donor as repair template. White boxes: HBB exons, blue boxes: homology arms, orange boxes: SFFV-GFP-polyA expression cassette b) HSPCs were electroporated with all RNA or RNP CRISPR system and INDELs were analyzed via TIDE software (N=number of data points within group, all from different mPB or CB donors). c) HSPCs electroporated as above and transduced with HBB-specific rAAV6s were analyzed by flow cytometry 18-21 days post-electroporation when GFP levels were found to be constant. Left panel shows percentage of HSPCs. Right panel shows representative FACS plots (N=number of data points within group, all from different donors). d) HSPCs were treated as above but targeted with rAAV6 E6V donor. Frequencies of allele types were quantified by sequencing of a total of 600 clones from TOPO-cloned In-Out PCRs (N=6, all from different CB or BM donors)

Fig 2. Enrichment of HBB-targeted HSPCs using FACS and magnetic bead-based technologies.

a) Left panel: representative FACS plots highlight the GFP^high population (red gate) generated by the addition of Cas9 RNP. Right panel: HBB-targeted HSPCs from GFP^high (red), GFP^low (green), and GFP^neg (blue) fractions were sorted and monitored for GFP expression. Error bars represent S.E.M. (N=11, all from unique mPB or CB donors) b) PCR was performed on methylcellulose colonies from GFP^high HSPCs to detect targeted integration at the 3’ end.c) Left panel: representative FACS plots highlight the tNGFR^high population (red gate) generated by the addition of Cas9 RNP. Right panel: tNGFR^high (red) HSPCs were enriched using anti-CD271 (LNGFR) magnetic microbeads and cultured for 18 days while monitoring tNGFR expression. Error bars represent S.E.M. (N=5, all from unique CB donors). d) PCR was performed on tNGFR^high-derived methylcellulose colonies to detect targeted integrations at the 5’ end.

Fig 3. HBB gene targeted HSPCs display long-term and multi-lineage reconstitution in NSG mice.

a) 16 weeks post-transplantation, mouse bone marrow was analyzed for human cell chimerism and GFP expression by flow cytometry. Top panel: Representative FACS plot from a mouse transplanted with RNP+AAV GFP^high HSPCs showing engrafted human cells in the red gate. Bottom panel: Representative FACS plot showing GFP-expressing human cells (red gate). CD19⁺ B cells and CD33⁺ myeloid cells are backgated and shown in blue and green, respectively. b) Human engraftment in NSG mice from all experimental groups. Three different HSPC donors were used for engraftment studies (N=number of data points within group), **** p < 0.0001, ns = p ≥ 0.05, one-way ANOVA and Tukey's multiple comparison test. Bars represent median. c) Percent GFP⁺ cells in the total human population (red), CD19⁺ B cells (blue), and CD33⁺ myeloid cells (green) (N=number of data points within group), * p < 0.05, **** p < 0.0001, one-way ANOVA and Tukey's multiple comparison test for total human cells and unpaired t test with Welch's correction for B and Myeloid cells. Bars represent median. d) 12-14 weeks post-secondary transplantation human cell chimerism and GFP expression was analyzed by flow cytometry. Left panel: Representative FACS plot from a secondary mouse transplanted with RNP+AAV (top) or RNP+AAV GFP^high (bottom) cells showing engrafted human cells in the red gate. Right panel: Percent GFP⁺ human cells in the BM of secondary recipients. e) Gel images of In-Out PCRs on sorted human cells from secondary recipients to analyze on-target integrations at the 5’ and 3’ ends. Input control PCR was performed on the human CCR5 gene. Positive control is an HSPC sample targeted at HBB with SFFV-GFP-PolyA. f) 80 million mPB-derived CD34⁺ cells were electroporated with HBB-RNPs and transduced with HBB AAV6s Bulk HSPCs or HSPCs enriched for targeting (by FACS or bead-enrichment) were transplanted into the tail vein of sublethally irradiated mice. 16 weeks post transplant human cell chimerism was analyzed by flow cytometry (N=number of data points within group). g) Percent GFP⁺ and tNGFR⁺ cells in the human population was analyzed by flow cytometry (N=number of data points within group), bars represent median.

Fig 4. Correction of the E6V mutation in SCD patient-derived HSPCs.

a) gDNA from HBB RNP-treated SCD-HSPCs was harvested and INDELs were analyzed via TIDE software (N=4 different SCD patient donors). b) HBB-targeted SCD-HSPCs were analyzed for GFP expression by flow cytometry (N=4 different SCD patient donors) c) SCD HSPCs were targeted with rAAV6 corrective SNP donor. HBB allele types were analyzed by sequencing of TOPO-cloned PCR fragments derived from In-Out PCR. 50-100 TOPO clones were analyzed from each of three different HSPC donors (N=3 different SCD patient donors). d) SCD HSPCs were targeted with anti-sickling HBB cDNA-EF1-tNGFR correction donor. Frequencies of tNGFR⁺ cells were analyzed by flow cytometry (N=3 different SCD patient donors). e) SCD Mock HSPCs and sorted SCD tNGFR^high HSPCs were differentiated into erythrocytes in vitro. Representative FACS plots from Day 21 of differentiation show cell surface markers associated with erythrocytes (GPA⁺/CD45^-/ CD71⁺/CD34^-) f) HbS, HbA, and HbAS3 mRNA expression was quantified by RT-qPCR in erythrocytes differentiated from HBB-edited or Mock SCD HSPCs. All mRNA transcript levels were normalized to the RPLP0 input control (N=2-3 different SCD patient donors).