Correction of β-thalassemia mutant by base editor in human embryos

Abstract

β-Thalassemia is a global health issue, caused by mutations in the HBB gene. Among these mutations, HBB −28 (A>G) mutations is one of the three most common mutations in China and Southeast Asia patients with β-thalassemia. Correcting this mutation in human embryos may prevent the disease being passed onto future generations and cure anemia. Here we report the first study using base editor (BE) system to correct disease mutant in human embryos. Firstly, we produced a 293T cell line with an exogenous HBB −28 (A>G) mutant fragment for gRNAs and targeting efficiency evaluation. Then we collected primary skin fibroblast cells from a β-thalassemia patient with HBB −28 (A>G) homozygous mutation. Data showed that base editor could precisely correct HBB −28 (A>G) mutation in the patient’s primary cells. To model homozygous mutation disease embryos, we constructed nuclear transfer embryos by fusing the lymphocyte or skin fibroblast cells with enucleated in vitro matured (IVM) oocytes. Notably, the gene correction efficiency was over 23.0% in these embryos by base editor. Although these embryos were still mosaic, the percentage of repaired blastomeres was over 20.0%. In addition, we found that base editor variants, with narrowed deamination window, could promote G-to-A conversion at HBB −28 site precisely in human embryos. Collectively, this study demonstrated the feasibility of curing genetic disease in human somatic cells and embryos by base editor system.

Keywords

β-thalassemia HBB −28 (A>G) base editor human embryo

Download fulltext PDF

Introduction

The explosive growth of human genomic data has revealed unprecedented numbers of disease-causing point mutations. Repairing such mutations may offer the best, and in some cases, only cure for genetic diseases. We and other groups have sought to correct disease mutant by combining CRISPR/Cas9 and homology directed repair (HDR) in human tripronulcear zygotes and diploid zygotes. However, low efficiency, mosaicism, off-target cleavage, and unintended homologous recombination (between target site and endogenous homologous genomic DNA sequence) remain obstacles that hamper the clinical applications of such approaches (Kang et al., 2016; Liang et al., 2015; Tang et al., 2017). In a recent report, it was found that diploid human zygotes, distinct from pluripotent cells, tends to repair DNA double strand break (DSB) using endogenous homologous sequence (Ma et al., 2017), consistent with what we have found in human tripronuclear zygotes (Liang et al., 2015). In the study, highly efficient repair of the mutant allele was achieved using the wild-type (WT) allele in heterozygous human zygotes through CRISPR/Cas9 (Ma et al., 2017). However, homozygous mutant embryos could not be repaired in way because of the lack of WT alleles. Additionally, recombination may occur with similar but not identical endogenous sequences, leading to unexpected mutations, as we found in human tripronuclear zygotes in which HBB recombined with HBD (Liang et al., 2015). Using base editors to directly repair point mutations may represent an efficient and highly specific alternative.

The base editor is a RNA-protein complex, adapted from the CRISPR/Cas9 system and cytidine deaminase (Komor et al., 2016). The effector protein is composed of cytidine deaminase (rAPOBEC1), Cas9, and uracil DNA glycosylase inhibitor (UGI). It can deaminate cytidine (C) to uridine (U) without inducing DNA DSB, and finally result in C-to-T (or G-to-A) conversion in the target DNA sequence (Hohmann, 2017; Komor et al., 2016; Liang et al., 2015). Efficient base editing at single-base resolution has been reported in plant, yeast, human cells, mouse zygotes, and human tripronuclear zygotes (Chen et al., 2017; Kim et al., 2017b, c; Komor et al., 2016; Li et al., 2016, 2017a, b; Liang et al., 2017; Lu and Zhu, 2016; Ren et al., 2017; Zhou et al., 2017); Zong et al., 2017). Intriguingly, mouse embryos and pups with 100% point mutation efficiency (free of mosacism), as well as human tripronuclear zygotes has been generated (Kim et al., 2017c; Li et al., 2017a; Liang et al., 2017). However, whether base editors can repair homozygous T>C (or A>G) disease mutant in human embryos remains to be tested.

β-Thalassemia, a common genetic disease in Mediterranean countries, North Africa, the Middle East, India, Central Asia, and Southeast Asia, is a major problem of global health (Cao and Galanello, 2010; Galanello and Origa, 2010; Weatherall, 2010). Genetic mutations, which will lead to the reduction of hemoglobin β chain (β-globin) and erythrocytes, finally cause oxygen shortage, bone deformity, organ dysfunction and even organ failure in many parts of the human body (Cao and Galanello, 2010). Based on the severity of the disease, β-thalassemia can be classified into β-thalassemia minor (also called β-thalassemia carrier), β-thalassemia intermedia, and β-thalassemia major (Cooley’s anemia) (Cao and Galanello, 2010). Without treatment, patients with β-thalassemia major usually die before age 5. Thalassemia major patients require lifelong blood transfusion and iron chelation treatment to survive, often accompanied by numerous complications, including arrhythmia, congestive heart failure, hypothyroidism, hypoparathyroidism, hypogonadism, diabetes, osteoporosis, liver cirrhosis, and infection (Chern et al., 2007; Wu et al., 2017). To date, allogeneic bone marrow transplantation (BMT) is the only curative therapy, but BMT is limited by human leukocyte antigen (HLA) compatibility. β-Thalassemia is mainly caused by mutations in the HBB gene, of which −28 (A>G) mutation is a common defect reducing the transcription of HBB (Orkin et al., 1983). Patients with homozygous or compound heterozygous −28 (A>G) mutation may develop severe anemia or intermedia anemia (Cao and Galanello, 2010; Orkin et al., 1983). Correcting the −28 (A>G) mutation by base editing should help to ameliorate anemia. Here, we report the efficient correction of −28 (A>G) mutation in human primary cells and human embryos by base editors.

Results

Correcting HBB −28 (A>G) mutation in human cell line by base editor

Of the two base editors (BE), BE2 (rAPOBEC1-dCas9-UGI) and BE3 (rAPOBEC1-nCas9-UGI), BE3 showed higher editing efficiency (Kim et al., 2017a). We therefore decided to repair HBB −28 (A>G) mutation using BE3. HBB −28 (A>G) mutation, in which the wild-type A at position −28 (A−28) is replaced with G in patients (G−28), locates in the ATA box upstream of the first exon of HBB (Fig. 1A) (Orkin et al., 1983). Three gRNAs targeting this mutant HBB allele were designed to convert C (on the complementary strand) to T (Figs. 1A and S1). We found that G at position −25 (G−25) might also be converted to A by these gRNAs (Fig. 1A). To test the deamination activity of these three gRNAs, we cloned the DNA fragment surrounding the HBB −28 (A>G) mutation into a lentiviral vector for stable integration in 293T cells. After selection with puromycin, three different cell clones were picked and verified by PCR (Fig. 1B). PCR primers (FP1 & RP1), that could specifically amplify this exogenous HBB −28 (A>G) mutant fragment, were designed (Fig. 1B). Sanger sequencing of this PCR amplicons indicated a clear G at HBB position −28 in these cell clones (Fig. 1B).

Figure 1
CorrectingHBB−28 (A>G) mutation in human cell line. (A) Schematic of *HBB* −28 (A>G) mutation. The exons are labeled with blue boxes. −28 (A>G) mutation was in red and indicated with red line (G−28). The −25 (G), next to G−28, was in blue and indicated with blue line (G−25). And gRNAs were labeled with black arrow. (B) Generation of *HBB* −28 (A>G) mutant stable cell lines. A fragment of *HBB* gene, containing the −28 (A>G) mutation, was cloned into a lentiviral vector. Packaged lentivirus was used to infect 293T. Virus-infected cells were selected by puromycin. 7 days after selection, single clones of cells were picked. The up panel showed the design of the recombined lentivirus vector. *HBB* gene fragment containing −28 (A>G) mutation was labeled with green box. LTR (long terminal repeat) region of lentiviral vector was labeled with blue arrowhead. PCR primer used to specifically amplify *HBB* fragment from integrated provirus were showed. The down panel showed the results of one wild-type 293T cells and three clones, amplified using FP1 and RP1. Representative sequencing chromatographs of the PCR amplicons of #3 clone were shown. The mutant base (G−28) was indicated by red arrowheads. (C) Precise repairing of *HBB* −28 (A>G) mutation by base editor 3 in the *HBB* −28 (A>G) mutant stable cell line. TA cloning sequencing showed clear G>A conversion at the target site. The frequency of each allele is shown. (D) Deep sequencing to detect on-target and off-target deamination at 10 potential off-target sites in *HBB* −28 (A>G) mutant stable cell line. Bars represent mean ± SEM (n = 3). Significance was calculated using a two-tailed unpaired t test (*P < 0.05, **P < 0.01)

Next, we co-transfected the gRNA and the BE3 expression vectors into clone #3. Cells transfected with GFP were included as a control. After 48 h, the cells were harvested. Target sites were amplified with FP1 and RP1 primers. Sanger sequencing of the PCR amplicons revealed obvious G>A conversion using the three gRNAs (Fig. S2). TA cloning and sequencing further confirmed active conversion in these cells (Fig. 1C). The conversion efficiency was 46.7% (14/30) for gRNA-1 (Fig. 1C). And consistent with previous findings in human cells and mouse embryos, we found proximal-site deamination using gRNA-2 (Fig. 1C) (Liang et al., 2017). Off-target deamination could be a concern in base editing, so we further investigated off-target deamination in this HBB −28 (A>G) mutant cell line. We again co-transfected BE3 together with either gRNA-1 or gRNA-2 into clone #3. GFP transfected cells were used as a control. The cells were harvested for genomic DNA extraction 48 h after transfection. The exogenously integrated HBB DNA fragment and 10 potential off-target sites were PCR amplified for deep sequencing. We found 16.3% and 26.0% G>A conversions at the target sites for gRNA-1 and gRNA-2 respectively, significantly higher than the rate of 1.2% in GFP control cells (Fig. 1D). And in line with data in Fig. 1C, we found that both the G−28 and G−25 at the target region could be deaminated by BE3 (Fig. 1D). We found higher G>A conversion efficiency at G−28 and G−25 using gRNA-2 (Fig. 1D). Moreover, we did not found any off-target deamination at the 10 potential off-target sites examined for both gRNAs, indicating high specificity (Fig. 1D). Taken together, these results clearly indicate the feasibility of repairing HBB −28 (A>G) in human cells in situ by base editing.

Correcting HBB −28 (A>G) mutation in primary skin fibroblast cells of a β-thalassemia patient by base editing

Inspired by the high efficiency and specificity of repairing HBB −28 (A>G) mutation by base editing, we sought to correct HBB −28 (A>G) mutation in patient’s cells. We isolated and cultured the skin fibroblast cells from a homozygous −28 (A>G) mutant patient (Fig. 2A and 2B). After transfection of BE3 and gRNA-1 into these cells by nucleofection, we achieved 80%–90% transfection efficiency (Fig. S3). At 48 h after transfection, the cells were used for single cell sorting (Fig. 2C). The sorted cells were whole genome amplified by multiplex displacement amplification (MDA), and then the HBB locus was PCR amplified (Fig. 2C). Here, we also observed efficient repairing of the homozygous mutation to heterozygotes or WT bases as shown by Sanger sequencing.

Figure 2
CorrectingHBB−28 (A>G) mutation in primary skin fibroblast cells of beta thalassemia patient. (A) Sanger sequencing to detect the genotype of the patient. Genomic DNA from the patient’s cells was extracted for PCR amplification of the target region. PCR amplicons were then sequenced by Sanger sequencing. *HBB* −28 (A>G) mutation were labelled with red arrowhead. (B) Primary skin fibroblast cells from the *HBB* −28 (A>G) mutant patient. (C) Schematic of base editing in *HBB* −28 (A>G) homozygous mutant skin fibroblast cells and single cell genotyping. Skin fibroblast cells were transfected with BE3 and gRNA-1. 48 h after transfection, single cell was isolated and whole genome amplified. The genomic DNA was then used as the template for PCR amplification of *HBB* site. The PCR product was sequenced by Sanger sequencing. (D) Representative sequencing chromatographs of homozygous mutant cells (G−28G−25/G−28G−25), heterozygous cells (A−28G−25/G−28G−25), and wild-type cells (A−28G−25/A−28G−25). (E) A summary of the base editing efficiency in homozygous skin fibroblast cells from the patient. A total of 30 single cells were whole-genome amplified. And 28/30 cells were successfully amplified by PCR. Both G−28 and G−25 were converted to A (A−28and A−25 respectively). *PCR amplification failed in 2 cells

We found 2 wild-type cells (2/28, 7.1%) with the genotype of A−28G−25/A−28G−25, proving precise repair of both mutant alleles (Fig. 2D and 2E). Additionally, only one mutant allele (A−28G−25/G−28G−25) was repaired in 3/28 (10.7%) cells, resulting in heterozygosity (Fig. 2E). Consistent with our previous data using human cell lines (Fig. 1D), we also found G>A conversion at G−25 of the target site in 3/28 (10.7%) cells, highlighting the need for developing base editor variants with a narrower deamination window to improve the precision of base editing (Figs. 1D and 2E). Here, these data showed that 5/28 (17.8%) cells was repaired precisely, demonstrating the feasibility of repairing HBB −28 (A>G) mutation in situ.

Correcting HBB −28 (A>G) mutation in cloned human embryos by BE3

Next, we tested the feasibility of repairing HBB −28 (A>G) mutation in human embryos. To model disease embryos, we generated cloned human embryos by nuclear transfer (Fig. 3A). The 1st polar body (PB1) and spindle of the in vitro matured oocytes were removed, and then the oocytes were fused with lymphocyte cells from peripheral blood of the patient. The reconstructed oocytes were activated and cultured until the appearance of pronucleus (PN). Approximately 5–6 h later, BE3 mRNA (200 ng/μL) and gRNA-1 (100 ng/μL) were injected into the cytoplasm after the appearance of pronucleus (Fig. S4). Of the 30 embryos injected, 26 survived (Fig. 3B). 48 h later, the HBB site of each embryo was PCR amplified individually. And then the PCR products were detected by Sanger sequencing and deep sequencing. HBB site was successfully amplified in 22/26 embryos (Fig. 3B). Interestingly, in these cloned embryos, we found high point mutation repairing efficiency, which was between 7.0% and 25.9% among the repaired embryos (Figs. 3C, S5 and Table S1). Analysis of the data showed that G−28 was converted to either A or C in 45.4% (10/22) of the injected embryos (Fig. 4B). In embryo #17, G−28 was converted to C. In the other 9 embryos, G−28 was converted to A, representing precise mutation repairing (Fig. 3B and 3C). Furthermore, we did not find deamination at G−25, indicating highly efficient and specific point mutation repairing in these embryos (Fig. 3C).

Figure 3
EffectiveHBBgene correction in human embryos by BE3. (A) Schematic of repairing *HBB* −28 (A>G) in cloned human embryos by BE3 and gRNA-1. Cloned *HBB* −28 (A>G) mutant homozygous human embryos were generated by fusing lymphocyte cell, from peripheral blood of the patient, with *in vitro* matured oocytes. And the BE3 mRNA and gRNA mixture was injected after the appearance of pronucleus. *HBB* site from each embryo was amplified by PCR and deep sequenced. PB1, the 1st polar body. PN, pronucleus. ZP, zonapellucida. (B) Summary of base editing-mediated point mutation repairing by BE3 in cloned human embryos. The repaired embryo contains G>A conversion at the *HBB* −28 site. *, The target G at the *HBB* −28 site was converted to C instead of A. (C) Deep sequencing to detect successful repairing by BE3 in human embryos. Target site PCR amplicons from these embryos were deep sequenced

Figure 4
Improving the precision of gene correction in human embryos by YEE-BE3. (A) Schematic of repairing *HBB* −28 (A>G) in cloned human embryos by YEE-BE3 and gRNA-1. Firstly, cloned *HBB* −28 (A>G) mutant homozygous human embryos were generated by fusing skin fibroblast cell from the patient with *in vitro* matured oocytes. And YEE-BE3 mRNA and gRNA mixture was injected after removing PB1. And 1 h later, the injected oocytes were fused with skin fibroblast cells. Then the fused embryos were activated and cultured for another 48 h. Single blastomere was isolated and MDA amplified. Then *HBB* site was amplified and sequenced. PB1, the 1st polar body. PN, pronucleus. ZP, zonapellucida. (B) Summary of base editing-mediated point mutation repairing by YEE-BE3 in cloned human embryos. The numbers of homozygous mutant blastomere (G−28G−25/G−28G−25), heterozygous blastomeres (A−28G−25/G−28G−25), and wild-type blastomeres (A−28G−25/A−28G−25) were calculated. #, 4 embryos did not develop into 2-cell stage. *, *HBB* site failed to be amplified by PCR. (C) Sanger sequencing to detect successful repairing by YEE-BE3 in each blastomere. Representative sequencing chromatographs of homozygous mutant blastomeres, heterozygous blastomeres and wild-type blastomeres

Effective HBB −28 (A>G) mutation repair in cloned human embryos by YEE-BE3

Although we did not find off-target deamination at G−25, we could not rule out the possibility of off-target deamination at G−25 in human embryos according to the data in human cells (Figs. 1D and 2E). We therefore turned to YEE-BE3, a BE3 variant with a smaller deamination window (Kim et al., 2017d). We injected gRNA-1 and YEE-BE3 mRNA before fusing the skin fibroblast cell with oocytes in which spindle and PB1 had been removed. Injecting YEE-BE3 mRNA before fusion will leave more time for protein translation and deamination before cell division. At about one hour after fusion, the reconstructed embryos were activated and cultured for another 48 h, when embryos were at 4–8 cell stage (Fig. 4A). The zona pellucidas of these embryos were removed, and 73 blastomeres were isolated from 20 embryos (Fig. 4B). Then the blastomeres were MDA-amplified individually (Fig. 4A). The HBB site was PCR amplified from these MDA products and sequenced by Sanger sequencing (Fig. 4C). We successfully amplified the HBB loci in 48 blastomeres (48/73, 65.8%) (Fig. 4B), and found that 37 blastomeres were still homozygous mutants (G−28G−25/G−28G−25), while the other 11 blastomeres (11/48, 22.9%) had been repaired (Fig. 4B and Table 1). A total of 3 out of 11 (6.3%) repaired blastomeres were heterozygous, and the other 8 (16.7%) were WT with both mutant alleles repaired perfectly (Table 1). More importantly, no off-target deamination at G−25 was observed, suggesting highly precise deamination at G−28.

文章分类：科技论文

分享到：