An RPS19-edited model for Diamond-Blackfan anemia reveals TP53-dependent impairment of hematopoietic stem cell activity

Diamond-Blackfan anemia (DBA) is a genetic blood disease caused by heterozygous loss-of-function mutations in ribosomal protein (RP) genes, most commonly RPS19. The signature feature of DBA is hypoplastic anemia occurring in infants, although some older patients develop multilineage cytopenias with bone marrow hypocellularity. The mechanism of anemia in DBA is not fully understood and even less is known about the pancytopenia that occurs later in life, in part because patient hematopoietic stem and progenitor cells (HSPCs) are difficult to obtain, and the current experimental models are suboptimal. We modeled DBA by editing healthy human donor CD34+ HSPCs with CRISPR/Cas9 to create RPS19 haploinsufficiency. In vitro differentiation revealed normal myelopoiesis and impaired erythropoiesis, as observed in DBA. After transplantation into immunodeficient mice, bone marrow repopulation by RPS19+/− HSPCs was profoundly reduced, indicating hematopoietic stem cell (HSC) impairment. The erythroid and HSC defects resulting from RPS19 haploinsufficiency were partially corrected by transduction with an RPS19-expressing lentiviral vector or by Cas9 disruption of TP53. Our results define a tractable, biologically relevant experimental model of DBA based on genome editing of primary human HSPCs and they identify an associated HSC defect that emulates the pan-hematopoietic defect of DBA.


91
Transducing normal donor CD34 + cells with LVs expressing short hairpin RNAs to deplete RP 92 genes causes impaired erythropoiesis in vitro and has been a useful tool in DBA research (10, 93 33, 34). However, it is difficult to achieve precise haploinsufficiency to recapitulate a disease that 94 is ostensibly dependent on the dosage of the affected genes (27). We and others have previously 95 demonstrated an erythroid defect in induced pluripotent stem cells derived from patients with DBA 96 (27, 35), but these cells are currently of limited utility for studying HSC properties. Therefore, we sought to model DBA and study the effects of RP haploinsufficiency on adult-type definitive HSCs.
RPS19 protein to be reduced by approximately 40% (supplemental Figures 3A-B). Transduction 123 of RPS19-targeted HSPCs with RPS19-GFP LV followed by Western blot analysis showed 124 efficient cleavage at the P2A site and expression of LV-encoded RPS19, which is fused to the 20 125 amino acid P2A peptide (supplemental Figure 3C). Deficiency of RPS19 impairs processing of 126 the 18S rRNA precursor, resulting in an increased ratio of 21S/18SE rRNA intermediates (39).

131
During in vitro erythroid differentiation of RNP-treated CD34 + cells, AAVS1 indel frequencies were 132 stable over time, whereas RPS19 indels declined, suggesting dropout of RPS19-deficient cells 133 (Figure 1C). The RPS19 indel frequency also decreased over time in myeloid culture, although to 134 a lesser extent than in erythroid culture ( Figure 1E-F), suggesting that erythroid cells are more 135 sensitive to RPS19 loss than are myeloid cells in vitro. All three RPS19 sgRNAs produced similar 136 results, indicating that impaired erythropoiesis is due to RPS19 disruption and not off-target 137 effects. The sgRNA RPS19.1 was used in subsequent experiments because it is predicted to 138 have fewer off-target DNA cleavage sites (40). Compared to AAVS1 targeting, RPS19 disruption 139 resulted in >50% fewer live cells at 3 days after electroporation (P < 0.001) (Figure 1G) and 140 generated >80% fewer burst-forming unit erythroid (BFU-E) colonies (P < 0.01) ( Figure 1H). This 141 could be partly due to some cells having biallelic RPS19 disruption, which are likely non-viable 142 (41). Reducing the dose of RNP from 0.4 to 0.04 mg/mL Cas9 component resulted in improved 143 survival of HSPCs, most likely due to fewer bi-allelic edits ( Figure 1G), an increased number of 144 approximately 50% reduction in cell number after 14 days in erythroid culture (P < 0.001) but no 151 reduction in cell number in myeloid culture (Figures 2A-B). Transducing RPS19-targeted HSPCs 152 with RPS19-GFP LV rescued the defect in erythroid cell expansion but had no effect on the growth

237
To investigate whether the repopulation defect in RPS19 +/− HSCs is TP53 dependent, CD34 + 238 HSPCs were edited with RNP targeting one or both genes then analyzed by xenotransplantation.

239
After editing RPS19 alone, the indel frequency dropped by 95% from 27.9%±2.2% in input cells 240 to below the limit of detection (1.2%±0.7%), as expected ( Figure 6A). After RPS19/TP53 multiplex 241 editing, the RPS19 indel frequency declined by only 55%, which represented an approximately 242 16-fold increase in RPS19 indels, as compared to those in cells treated with RPS19 RNP alone 243 (P < 0.01) ( Figure 6A; supplemental Figure 9A). The TP53 indel frequencies were similar in input 244 cells and at 16 weeks post-transplantation, suggesting that TP53 disruption alone conferred no major selective advantage to HSPCs ( Figure 6B). However, the TP53 indel frequency increased erythroid, B-lymphocyte and HSPC lineages ( Figure 6C), although the percentage of human cell 250 lineages was unchanged between the groups (supplemental Figure 9B).         with an RPS19 indel frequency of less than 25% were graded as unedited wildtype cells, those 420 with an indel frequency between 25% and 75% were graded as heterozygously edited cells, and 421 those with an indel frequency greater than 75% were deemed to have undergone homozygous For xenotransplantation, 4 × 10 5 to 5 × 10 5 HSPCs were washed and resuspended in phosphate-427 buffered saline (PBS) with 2% fetal bovine serum (FBS) then injected into the tail veins of 5-to 8-

428
week-old female NSGW mice. Mice were euthanized and analyzed at 16 weeks after 429 xenotransplantation. Recipient bone marrow cells were incubated with mouse and human 430 lineage-specific antibodies (supplemental Table 3 Table 1). Droplets were read using a QX200 Droplet Reader (Bio-Rad), and 488 data were analyzed using QuantaSoft software v.1.7.4.0917 (Bio-Rad). VCNs per diploid genome 489 were calculated as the ratio of Psi copies to every 2 copies of RPP30.

492
RNA was extracted from RNP-treated cells by using a RNeasy Mini Kit (Qiagen), and the 493 concentration of RNA was determined using a NanoDrop spectrophotometer (Thermo Scientific).

494
The change in CDKN1A expression was determined by using the One-Step RT-ddPCR Advanced

655
Targeted disruption of the ribosomal protein S19 gene is lethal prior to implantation.

825
Linear mixed-effects model approach was used to test for statistical significance. Asterisks