Establishment of an efficient CRISPR/Cas9 approach
To optimize genome-editing conditions, we used human erythroleukemia (HEL) cells, a cell line carrying normal HBB/HBB genotype . First, to test the electroporation efficacy, the DNA sequence encoding sgRNA in vivo for HBB gene was amplified by PCR and labeled with Cy3 fluorescent reporter. The optimal electroporation conditions for HEL cells were determined (Additional file 1: Figure S3). Subsequently, to assess CRISPR efficacy, the Cas9 mRNA and DNA sequence encoding sgRNA in vivo, specific for enhanced green fluorescent protein gene (EGFP), were introduced by electroporation into stably EGFP-expressing HEL cells. The resultant EGFP gene silencing was quantified by flow cytometry (Additional file 1: Figure S4). To verify RT-PCR genotyping of cell HbS and HBB genes, standard curves were established using genomic DNA from the whole blood of SCD patients and healthy donors at different ratios (Additional file 1: Figure S5). In addition, sgRNA specific for HBB was also produced by in vitro transcription (Additional file 1: Figure S6) and introduced into HEL cells by electroporation with an homology-directed repair (HDR) template to convert HBB to HbS. Target genome-editing efficacy was evaluated by genotyping and compared to that induced by the DNA sequence encoding sgRNA in vivo (Additional file 1: Figure S7). Furthermore, the ratio of sgRNA to HDR template used for electroporation was optimized (Additional file 1: Figure S8). To achieve the highest genome-editing efficiency, HDR templates targeting different cutting loci in HBB (Additional file 1: Figure S9) and HDR templates with different lengths (Additional file 1: Figure S10) were investigated.
For further validation, we created a SCD cell model by applying the optimized CRISPR/Cas9 approach as previously described. Step-by-step genome-engineering was conducted, and the resultant SCD HEL cells carrying homozygous HbS/HbS genotype were cloned (Additional file 1: Figure S11A). Genotyping and sequencing analysis showed 9.6 and 9.4% genome-editing efficacy in each CRISPR step (Additional file 1: Figure S11B). Finally, to mimic disease treatment, SCD HEL cells were genome-engineered to convert HbS/HbS to HBB/HbS, corresponding to the conversion of SCD to SCT genotype (Additional file 1: Figures S12 and S13). The aforementioned validation studies demonstrated reproducibility, efficiency, and feasibility of this CRISPR/Cas9 approach for clinical studies (Additional file 1: Figure S1).
Development of a clinically practicable approach for genome-editing therapy of SCD
For target genome editing of HbS by the optimized CRISPR/Cas9 method, we used small amounts of the peripheral blood from SCD patients (2–3 mL). Immunophenotyping analysis of the patient blood revealed a scant number of HSPCs that were positive for CD34 expression, but negative for CD45, CD19, or CD2 (Fig. 1a). The HSPCs were then isolated using the MACS CD34+ Progenitor Cell Isolation Kit and confirmed by immunophenotyping analysis (Fig. 1b). For morphology evaluation, cytospin was performed with Wright-Giemsa staining, and the cells were examined under the microscope. In comparison to blood-nucleated cells, the isolated HSPCs were uniform in size and shape with very scant cytoplasm (Fig. 1c). The HSPCs were cultured for 7 days in Stemspan SFEM II medium, resulting in an average eightfold increase in cell number (Fig. 1d).
For CRISPR genome editing, the HSPCs were subjected to electroporation for intracellular delivery of Cas9 mRNA, and then cultured for 6 h to allow expression of cellular Cas9 protein. Subsequently, cells were repeatedly subjected to electroporation to introduce sgRNA specifically targeting HbS and the 127nt HDR. To correct the V6G mutation in HbS protein, nucleotides 69T and 70G in HbS were substituted with 69A and 70A, respectively, to encode glutamic acid at position 6 in HBB* of the expressed HbA* protein identical to natural HbA (Additional file 1: Figure S1A). Notably, to distinguish genome-edited HBB* from natural HBB, the point mutations 70G>A and 58C>T, which do not alter the encoded amino acid residues, were introduced as hallmarks of HBB*. For cell colony formation, the electroporated HSPCs were seeded in Methocult H4030 medium and cultured for 13 days. The formed erythroid progenitor colonies (E-colonies), which had a light reddish color, and the granular/monocytic progenitor colonies (G/M-colonies), which appeared semi-transparent, were detected under the microscope (Fig. 1e), revealing an average 47 ± 7 E- and 27 ± 5 G/M-colonies per milliliter of the SCD patient blood (Fig. 1f). Giemsa-Wright staining showed the morphology of the erythroid progenitor cells (Additional file 1: Figure S14). To evaluate genome-editing status, genomic DNA was prepared from the half of cloned cells, and RT-PCR genotyping screening was performed using TaqMan MGB probe-reporter system specific for HBB* and HbS as described in “Materials” and “Methods.” Simultaneously, target gene sequencing analysis was performed; equal amounts of HbS and engineered HBB* containing nucleotide 69A and the tracking nucleotides 58T and 70A were detected (Fig. 2a). As an experiment internal control, homozygous HbS was detected in cloned cells from the same patient HSPCs that had not undergone genome editing. The overall genome-editing efficacy of HSPCs in E-colonies was 9.0%, which is similar to that detected in the validation studies of SCD HEL cell model. Moreover, according to the chromatogram of Sanger sequencing, the other allele of corrected single E-colony and G/M-colony cells does not carry indels near the cleavage site (Additional file 1: Figure S15), as the results in the corrected SCD HEL cells (Additional file 1: Figure S13). The genome-editing status of G/M-colonies was also evaluated by both RT-PCR genotyping and sequencing analysis, revealing 8.6% efficacy, which is nearly identical to that found in E-colonies derived from HSPCs from the same patient (Fig. 2b).
Subsequently, the other half of cells from single E-colonies was transferred to an erythroid progenitor medium for clonal cell amplification , and the cells were cultured for 10 days. The erythroid progenitor cells, cloned from single E-colonies, were amplified 19-fold on average and showed an enhanced red color, indicating synthesis of cellular hemoglobin (Fig. 3a). The amplified cells were stained with anti-CD34, CD36, CD45, CD71, and CD235a antibodies, and the nuclei were stained with DRAQ-5 dye. Flow cytometry demonstrated that the cloned cells had an immunophenotyping profile consistent with erythroid progenitor cells; in particular, they expressed CD36, CD235a, and CD71; lacked CD34 and CD45; and possessed nuclei (Fig. 3b) .
To investigate changes in cellular protein expression, the cloned cells were lysed, and hemoglobin types were assessed by a clinically viable HPLC method (8.3 × 103 cells/test) as described in “Materials” and “Methods” (Additional file 1: Figure S15) . Due to genome editing resulting in heterozygous HBB*/HbS, a new peak for HbA* protein was detected in the cloned erythroid progenitor cells at a similar level to residual HbS protein (Fig. 3c). In the experiment internal control, the cloned cells from the same patient HSPCs that did not undergo genome editing expressed the HbS protein only. For further comparison, the HPLC elution profiles of blood samples from normal people, SCT carriers, and SCD patients are shown in Fig. 3d. Notably, although HBB* contains two different nucleotides as compared to HBB, it encodes the same amino acid residues, and thus, HbA* appears at the same retention time with natural HbA in HPLC assay.
For cellular function study, the amplified erythroid progenitor cells were further cultured in erythroid terminal maturation medium  to induce in vitro differentiation and RBC maturation as described in “Materials” and “Methods.” Nuclear staining with Hoechst fluorescent dye was initially used to assess the maturation states of cells. Microscopic examination revealed presence of nucleated RBCs and mature RBCs that did not have nuclei (Fig. 4a). In addition, cytospin with Wright-Giemsa staining confirmed in vitro RBC differentiation at different maturation stages, including nucleated RBCs, RBCs with nuclear contraction, and mature RBCs with no nuclei (Fig. 4b) . To evaluate cellular function changes resulting from genome editing, cell sickling tests were performed. To induce hypoxia, in vitro differentiated RBCs were suspended in sodium metabisulfite buffer, loaded on a glass slide, and immediately covered with a cover slip. The time course of the resultant sickle cell formation was recorded under a microscope at 1-min intervals for 30 min. As shown in Fig. 4c and Additional files 2, 3, and 4: Movies S1–S3, the genome-edited RBCs were more resistant to hypoxia and did not display sickling formation. In contrast, under the same hypoxia condition, sickle cell formation was observed in the in vitro differentiated RBCs from the same patient HSPCs that did not undergo genome editing, i.e., experiment internal control. The peripheral blood from SCD, SCT, and healthy donors were also tested as control. Some RBCs from SCT sickled in the test but much more sickled RBCs presented in the specimen from SCD (Fig. 4c). For quantitative analysis, 500 RBCs from each specimen were examined and the percentage of sickle cells vs. time lapse was calculated (Fig. 4d). These findings demonstrated that genome editing of HSPCs from a SCD patient reinstituted cellular function of the offspring RBCs, eliminated sickling formation under hypoxia, and confirmed feasibility of gene therapy to treat SCD. Notably, because a number of the in vitro differentiated RBCs contained nuclei (Figs. 4a, b) and their cellular shape could not change, the RBCs in the experiment internal control group showed a lower percentage of sickle cell formation than that observed for peripheral blood RBCs from the SCD patient under the same hypoxia conditions (Fig. 4d). Thus, the sickling of in vitro differentiated RBCs from the genome-edited SCD patient HSPCs may also be underestimated.
Additional file 2: Movie S1. Sickling assay of genome-edited RBCs. (WMV 4447 kb)
Additional file 3: Movie S2. Sickling assay of RBCs without genome editing. (WMV 4213 kb)
Additional file 4: Movie S3. Sickling assay of RBCs from the SCD patient blood. (WMV 4629 kb)
Validation of a Cas9/sgRNA ribonucleoprotein system for SCD genome editing
To reduce the number of electroporation steps, the newly reported Cas9/gRNA ribonucleoprotein (RNP) delivery system was tested and compared to the method used in this study (Fig. 5a). A genome-editing RNP complex was formulated by mixing the Cas9 protein with two synthetic RNA oligonucleotides, a CRISPR targeting RNA (crRNA) duplexed to a trans-activating crRNA (tracrRNA) as described in “Materials” and “Methods.” The formed Cas9/RNP complex and HDR template, used to convert HbS to HBB*, were introduced by electroporation into HSPCs isolated from the SCD patient blood. CRISPR-treated HSPCs were then cultured to form cell colonies, and genome-editing status of the resultant cells cloned from individual colonies was evaluated by RT-PCR genotyping and gene sequencing. Single-step electroporation with Cas9/RNP complex resulted in a significant improvement in cell colony formation of HSPCs, namely 56 and 73% increase in formation of E- and G/M-colonies, respectively (Fig. 5b). However, the resultant genome-editing efficacy of HSPCs with these two approaches showed no statistical difference in both E- and G/M-colonies (Fig. 5c); although, a slight increase was found with Cas9/RNP method.