AAV and short ssODN both allow precise editing in human HSPCs.

A) Experimental design. B) HDR donor configurations. The SRSF2 P95H AAV donor is shown above and short and long ssODN donors are shown below with features indicated. Annotated sequences are shown in Supplementary Methods. C) HDR integration efficiency by AAV dose. Cells were edited with 30.5 pmol RNP (or not as indicated) with indicated multiplicities of infection (MOI) of AAV donor. Bars show mean values and points show measurements for individual cords. Male cords are shown as triangles and female as circles. D) Viable cell number by AAV dose. Hemocytometer counts at the time of harvest are shown for each sample from (C). E) HDR integration efficiency for short and long ssODN donors. Donor DNA amounts are shown in pmol. F) Viable cell number by ssODN dose. False-discovery rate (FDR) corrected paired t-test significance values are shown in Table S1.

Small molecule-mediated inhibition of DNA-PK and optimal donor design substantially improve precise editing efficiency.

A) AZD7648 and M3814 improve HDR efficiency in primary human HSPC. Cells were edited with 30.5 pmol RNP (or not as indicated) with 400 MOI of AAV donor and small molecules added as indicated (in μM). Bars show mean values and points show measurements for individual cords. Male cords are shown as triangles and female as circles. B) Viable cell numbers with AZD7648 and M3814 addition. Hemocytometer counts at the time of harvest are shown for each sample from (A). C) HDR efficiency with combinations of AZD7648, p53 siRNA and RS-1. Cells were edited with 30.5 pmol RNP (or not as indicated) with 400 MOI of AAV donor in the presence of the indicated additives. AZD7648 was used at 5 μM, p53 siRNA at 20 fmol, and RS-1 at 15 μM. D) Viable cell numbers with additive combinations. Hemocytometer counts at the time of harvest are shown for each sample from (C). E) Technical factors associated with high sample number is associated with decreased HDR efficiency. HDR efficiency is shown for all 30.5 pmol RNP, 400 MOI AAV, 20 fmol p53 siRNA, and 5 μM AZD7648 samples by the number of conditions processed in a given experiment. A linear fit is indicated as a red line. The R2 is indicated, and overall p value was <<0.001. F) Alternative designs for ssODN donors with key features indicated. Annotated sequences are shown in Supplementary Information. G) Silent mutations allow ssODN donors to achieve similar efficiencies to AAV. All edits were performed with 0.5 μM AZD7648, 20 fmol p53 siRNA, 50 pmol ssODN or 400 MOI AAV as indicated. Donor types are shown as their logos from (1B, 2G). H) No observable off-target mutations at predicted target sites even with the addition of AZD7648. The overall percent of reads containing exclusively reference allele, or any substitutions, deletions, or insertions that overlap with the predicted off-target cut sites is shown for 3 individual cords across the top 3 cut sites. Cells from each individual cord were split into an unedited control, and cells edited with the silent mutation containing ssODN for the SRSF2 locus under either standard conditions (ie. no p53siRNA or AZD7648) or with our optimal editing protocol (ie. with p53siRNA and 0.5 μM AZD7648). False-discovery rate (FDR) corrected paired t-test significance values are shown in Table S1.

Editing has a minimal impact on HSPC function and hierarchy.

A) Integration efficiency is equivalent across phenotypically defined progenitor compartments. All edits were performed with 0.5 μM AZD7648, 20 fmol p53 siRNA, and 50 pmol of silent mutation ssODN. Values show the difference in precise edit efficiency for each phenotypic subset compared to bulk assessment within that cord. Bars show mean values and points show measurements for individual cords. Male cords are shown as triangles and female as circles. All populations show no significant difference from bulk. B) Progenitor phenotypes are minimally altered across the hierarchy. The % of CD34+ for each sub-population is shown. Mean values are indicated as lines. A slight but significant decrease was present for late progenitors (CD34+CD45RA+) associated with donor addition (but not different with editing). C) An example image of a well of colonies, and example colonies. D) Total colonies are decreased by the addition of donors, and further by editing. Total CFC per 1000 CD34+ cells is shown for each cord. Lines indicate mean values. E) No changes were observed in the frequency of colonies of each type. As before points are individual cords and lines show mean values. F) Colonies showed a preponderance of homozygous editing. Mean homozygous, heterozygous edited, and unedited cells are shown from 36 analyzed colonies across 3 independent cords. False-discovery rate (FDR) corrected paired t-test significance values are shown in Table S1. G) No change in the dynamics of colony emergence from single-LT-HSCs in LTC-IC. The presence or absence of an obvious colony in each well (initially sorted with a single LT-HSC) was scored weekly over the first 6 weeks of the LTC-IC assay, and again at week 8. Clonal outputs are shown as lines with unedited in black and edited in blue. H) Example colonies at 8 weeks. At 8 weeks, clones were scored as negative (no colony at any point), transient (previous colony without a colony at endpoint), low proliferation (>50 cells, but below confluence), and highly proliferative (confluent). Example images of negative, low proliferation, and highly proliferative clones are shown. Scalebars (white) show 1 mm. The low proliferation colony is circled in red. I) Highly proliferative clones are not lost from the LT-HSC population in the editing process. The frequency of clones of the indicated types is shown per 100 phenotypic LT-HSC either without editing or following optimal editing. Error bars represent 95% confidence intervals. Frequencies, p-values, and error bars were calculated using Extreme Limiting Dilution Analysis, based on colony numbers measured from 3 independent experiments (each with a different cord donor). Numbers for each clone per donor, the total number of clones analyzed for that donor, and donor sex are indicated below the relevant bar.

Zygosity can be tuned using a mixture of mutant and silent donors.

A) Experimental design. Green bars represent the proportion of mutant donor. Specific amounts of mutant and silent donor are shown underneath for each condition. All edits were performed with 0.5 μM AZD7648, 20 fmol p53 siRNA, and indicated amounts of each donor. B) Overall mutant integration efficiency varies linearly with the proportion of mutant donor. Individual cords are shown as points. Male cords are shown as triangles and female as circles. A linear fit is indicated as a red line. The R2 is indicated, and overall p value was <<0.001. C) Mean clonogenic frequencies are consistent across donor proportions. Data are a mean of two independent cords with a total of 30 cells for each cord at each dose analysed except the 0% condition which only had 20 cells per cord. D) Zygosity can be adjusted by inclusion of silent donor. Mean frequencies of homozygous mutant, heterozygous and homozygous silent donors are shown within all clones with any observed editing. A total of 52 clones with some degree of editing across 2 independent cords were analyzed.