Correction of F8 intron 1 inversion in hemophilia A patient-specific iPSCs by CRISPR/Cas9 mediated gene editing

Introduction: Hemophilia A (HA) is the most common genetic bleeding disorder caused by mutations in the F8 gene encoding coagulation factor VIII (FVIII). As the second predominant pathogenic mutation in hemophilia A severe patients, F8 Intron one inversion (Inv1) completely splits the F8 gene into two parts and disrupts the F8 transcription, resulting in no FVIII protein production. The part which contains exon 2-exon 26 covers 98% of F8 coding region. Methods: We hypothesized that in situ genetic manipulation of F8 to add a promoter and exon one before the exon two could restore the F8 expression. The donor plasmid included human alpha 1-antitrypsin (hAAT) promoter, exon one and splicing donor site (SD) based on homology-mediated end joining (HMEJ) strategy was targeted addition in hemophilia A patient-derived induced pluripotent stem cell (HA-iPSCs) using CRISPR/Cas9. The iPSCs were differentiated into hepatocyte-like cells (HPLCs). Results: The hAAT promoter and exon one were targeted addition in HA-iPSCs with a high efficiency of 10.19% via HMEJ. The FVIII expression, secretion, and activity were detected in HPLCs derived from gene-targeted iPSCs. Discussion: Thus, we firstly rescued the 140 kb reversion mutation by gene addition of a 975 bp fragment in the HA-iPSCs with Inv1 mutation, providing a promising gene correction strategy for genetic disease with large sequence variants.


Introduction
Hemophilia A (HA) is an X-linked recessive genetic bleeding disorders with the incidence of one in 5,000 male births (Berntorp et al., 2021;Ragni, 2021). Affected males suffer from spontaneous soft-tissue, muscle and joint bleeding symptoms, the severe patients (coagulation factor VIII (FVIII) activity <1% of the normal value) even experience life-threatening intracranial hemorrhage (Song et al., 2021). HA is caused by the deficiency of functional FVIII, encoded by F8 gene, which is one of the largest genes spanning 186 kb on Xq28 (Lassalle et al., 2020).
Traditionally, HA is treated by FVIII protein replacement. Owing to the short half-life of FVIII (14-19 h), the HA patients need repeat injections of the FVIII, resulting in huge economic burden on the patients and their families (Batty and Lillicrap, 2019). In recent decades, HA gene therapy was developed and made breakthrough (Perrin et al., 2019;Rodriguez-Merchan et al., 2021;Ozelo et al., 2022). The codonoptimized BDD-F8 was transduced into hepatocytes using AAV5 vectors and the median FVIII coagulation activity was maintained 20 IU/dL in the patients treated with high dose AAV (6 × 10 13 vg/kg) in a 3-year follow-up study (Pasi et al., 2020) and the European Commission granted conditional marketing authorization to valoctocogene roxaparvovec gene therapy on 24 August 2022. However, the transduced F8 via AAV vector wasn't integrated into the genome and the F8 coding sequence (7 kb) far exceeds the packaging capacity of AAV (4.7 kb) (Tornabene and Trapani, 2020;Marrone et al., 2022). Even the B domain deleted F8 version cannot be easily packaged into the AAV vector. Considering these issues and some HA patients with AAV antibodies (Verdera et al., 2020), gene therapy strategy via nonviral system was developed actively. Since the strategy of in situ gene repair for HA enables retention of the main F8 gene regulatory elements, the strategy has been extensively investigated.
F8 Intron one inversion (Inv1) is the second predominant pathogenic mutation in severe HA patients. In human genome, the reverse repeat of a 1,041-bp sequence within F8 intron one is located in 140 kb telomeric to the F8 gene, and this repeat may induce intrachromosomal recombination during male meiosis and cause large inversion (Inv1) (Fahiminiya et al., 2021). This large inversion completely splits the F8 gene into two parts and disrupts the F8 transcription, resulting in no FVIII protein production. Notably, the part which contains exon 2-exon 26 covers 98% of F8 coding region. Thus, we hypothesized that in situ genetic manipulation of F8 to add a promoter and exon 1 (146 bp) before the exon two might represent a therapeutic strategy for restoring the reading frame for all HA patients with F8 Inv1mutations.
In this study, we performed a targeted addition of human alpha 1antitrypsin (hAAT) promoter and exon one before the exon two in HA patient-derived induced pluripotent stem cell (HA-iPSCs) using CRISPR/Cas9 and donor plasmid. To achieve a higher integration efficiency, we constructed a donor plasmid with the homologous arms flanking with two same sgRNA4 sites to excise the backbone sequences based on homology-mediated end joining (HMEJ) strategy (Yao et al., 2017;Li et al., 2021;Yuan et al., 2021). Meanwhile, a donor plasmid for classic homologous recombination (HDR) and a donor plasmid for non-homologous end joining (NHEJ) were constructed as control. The integration efficiency in HMEJ group was 10.19%, higher than that in HDR group (6.25%) and NHEJ group (0.99%). The F8 transcript and FVIII secretion were rescued in the hepatocyte-like cells (HPLCs) derived from gene targeted iPSCs. Our findings provide an in situ genetic addition strategy which is promising for the clinical translation in gene therapy for HA involving large sequence variants.

Characterization of HA-iPSCs
We previously generated an iPSC line (HA-iPSCs) derived from the urine cells of a HA patient with F8 Inv1 (Hu et al., 2015).
Here the HA-iPSCs were identified via immunofluorescence. The HA-iPSCs maintained pluripotency according to immunofluorescence. The HA-iPSCs expressed Oct4, Nanog and SSEA-4, while SSEA-1 wasn't expressed (Supplementary Figure S1A). To further evaluate the pluripotency in vivo, HE staining of teratomas was performed and the results showed that teratomas contained ectoderm, endoderm and mesoderm (Supplementary Figure S1B). Meanwhile, the HA-iPSCs maintained a normal karyotype (Supplementary Figure S1C).

Generation of CRISPR/Cas9 and donor template for in situ gene addition
Inv1 of F8 splits the F8 gene into two parts and disrupts the F8 transcription, resulting in no FVIII protein production. The part which contains exon 2-exon 26 covers 98% of F8 coding region. So the Inv1 mutation could be rescued by gene addition of a promoter and the coding sequences of exon 1 (Figure 1). We then designed six single-guide RNAs (sgRNAs) F8-sg1, F8-sg2, F8-sg3, F8-sg4, F8-sg5, and F8-sg6 mapping to target sites in intron 1 ( Figure 2A) and constructed and verified the cleavage activity via T7 Endonuclease I (T7E1) ( Figure 2B). The cleavage frequency of F8-sg4 was 56.53% and was used for targeted addition. The donor plasmids were designed ( Figure 2C), constructed and verified with Sanger sequencing ( Figure 2D).

CRISPR/Cas9 and donor plasmid mediated targeted addition
The HA-iPSCs were nucleofected with the plasmids expressing the CRISPR/Cas9 complex and F8-sg4 along with the donor plasmid F8-NHEJ, F8-HDR, F8-HMEJ, respectively. The single-cell clone was screened using primers across homology arms 5F/R ( Figure 3A) and 3F/R ( Figure 3B), and the sequences were verified via Sanger sequencing ( Figure 3C). The targeting efficiency was 10.19% with the donor plasmid F8-HMEJ, higher than that in the F8-HDR group (6.25%) and the F8-NHEJ group (0.99%) ( Table 1). Two targeted addition clones (T-26 and T-73) generated from the HA-iPSCs were then used for further research. The primers 5F and 3R were used to detect the purity of the single-cell clone ( Figure 3D). The immunofluorescence showed that T-26 and T-73 maintained pluripotency ( Figure 3E). The HE staining of teratomas further confirmed the pluripotency in vivo ( Figure 3F), and T-26 and T-73 maintained a normal karyotype ( Figure 3G). To evaluate the off-target effect of CRISPR/Cas9, the potential offtarget sites (≤4 mismatches) of F8-sg4 predicted using CHOPCHOP (http://chopchop.cbu.uib.no/) were amplified and sequenced. No off-target indels in the potential off-target sites were observed comparing the sequences in HA-iPSCs with that in T-26 and T-73 (Supplementary Figure S2).
Considering a modified hAAT promoter was used in the gene addition, we detected the transcription of the hAAT gene and found that the hAAT gene was transcribed in the iPSCs ( Figure 4A). Then the F8 transcription was detected in T-26 Frontiers in Genetics frontiersin.org and T-73 via reverse transcription PCR (RT-PCR), while no F8 transcript was detected in HA-iPSCs ( Figure 4B). The sequencing results revealed that the promoter and the exon one were successfully inserted into F8 in T-26 and T-73 with the F8 transcription restored ( Figure 4C), demonstrating the F8 expression was rescued in the gene targeting group.

Differentiation of targeted iPSCs into hepatocyte-like cells
To evaluate the F8 expression in hepatocyte, we differentiated the HA-iPSCs, the gene-corrected iPSCs T-26 and T-73, the normal hiPSCs (N-iPSCs) into the hepatocyte-like cells (HA-iHPLCs, T-26-iHPLCs, T-73-iHPLCs, and N-iHPLCs) as the diagram in Figure 5A. During the differentiation, the cells went through four stages. The cell morphology was gradually changed from iPSC clone to epithelioid cell morphology ( Figure 5B) and identified the cell marker with immunofluorescence. All cells in the first stage expressed the definitive endoderm cell markers SOX17 and FOXA2; 5 days later, the AFP signal was positive in hepatoblast-like cells; after 5 days of culture, followed by 11 days of culture in Hepatocyte Culture Medium (HCM) contained 20 ng/mL oncostatin M (OsM), the hepatocyte-like cells expressed ALB ( Figure 5C). In addition, differentiated HPLCs on Day 25 with characteristic functions of mature hepatocytes could store glycogen and metabolize indocyanine green (ICG) ( Figure 5D, E).
And the FVIII coagulation activities in the supernatants from the T-26-iHPLCs were detectable by FVIII activity assay ( Figure 6C). More importantly, the FVIII was expressed in T-26-iHPLCs, T-73-iHPLCs via immunofluorescence staining, but no FVIII was detected in HA-iHPLCs due to the interrupted F8 gene ( Figure 6D). These results suggest that FVIII expression was restored in gene-corrected iPSCs derived HPLCs.

Discussion
Considering the regulatory elements of the F8 gene can be retained to the greatest extent, the strategy of in situ gene correction for HA was designed and performed. Wu et al. inserted the coding sequences of exon 23-26 into the exon 22intron 22 junctions to correct the intron 22 inversion mutation. F8 expression was restored in mesenchymal stem cells (MSCs) and endothelial cells (ECs) differentiated from the gene-corrected iPSCs (Wu et al., 2016). The targeting efficiency was 62.5% with the Neo selection cassette. Park et al. reverted the F8 intron one or intron 22 inversion mutations in HA patient derived iPSCs using CRISPR/ Cas9, with a frequency of 6.7% (Park et al., 2015). The ECs derived from the gene corrected iPSCs were transplanted into the HA mice and functionally rescued the FVIII deficiency. We previously deleted the coding sequences of B domain in F8 gene precisely to rescue the HA with pathogenic mutations in B domain of F8. The F8 expression and secretion were validated in the ECs derived from the B domain deleted iPSCs in vitro and in vivo (Hu et al., 2019;Hu et al., 2022). In this study, we firstly corrected the Inv1 mutation of F8 via in situ gene addition strategy in HA-iPSCs with an efficiency up to 10.19% without any screening.
This work suggests a feasible therapeutic gene addition strategy for HA involving large sequence variants. Luo et al. reported the F8 expression and FVIII deficiency were rescued via injection of the AAV carrying CRISPR/SaCas9 and the donor plasmid with a promoter and the coding sequence of exon one into the HA mice with deletion of the promoter region and exon one of F8 (Luo et al., 2021). However, the strategy wasn't validated in cells or HA mice with Inv1 of F8. Here we firstly rescued the 140 kb Frontiers in Genetics frontiersin.org reversion mutation by gene addition a 975 bp fragment in the HA-iPSCs with Inv1 mutation, providing a promising gene correction strategy for other genetic birth defects with large sequence variants.
Researchers have made many attempts in CRISPR-based strategies for targeted gene correction and made tremendous advances. The error-prone NHEJ repair pathway is the main DNA repair pathway in non-dividing cells and occurs during the whole cell cycle, whereas the HDR only occurs in S/G2 phase in dividing cells (Lau et al., 2020). The HMEJ strategy is based on both the targeted genomic site and the donor vector with homology arms flanking the recognition sequence of CRISPR/ Cas9 cleaved via CRISPR/Cas9. It has been reported that a higher site-specific gene integration efficiency can be achieved by HMEJ based strategy than the classic HDR strategy. In this study, the NHEJ, HDR, and HMEJ based donor plasmid were designed and constructed. Consistent with previous reports, the gene integration efficiency was 10.19% in HMEJ group, higher than that in HDR group (6.25%) and NHEJ group (0.99%) without any drug screening.
Although the FVIII is mainly synthesized in liver sinusoidal endothelial cells (LSECs) under physiological conditions (Shahani et al., 2014;Hayakawa et al., 2021), many studies demonstrated that the ectopic expression of FVIII in hepatocytes was efficient and the FVIII deficiency in HA mice and HA patients were rescued (Bunting et al., 2018;Chen et al., 2019;Zhang et al., 2019). Successful amelioration of hemophilia A has been reported by several groups targeting the liver-expressed mouse Alb locus in vivo using AAV vectors to transfer the transgene into the liver (Sharma et al., 2015;Chen et al., 2019;Zhang et al., 2019). By integrating the promoter-less BDD-F8 into this locus, the transgene is expressed under the robust Alb promoter, achieving therapeutic levels of plasma FVIII for up to 7 months after injection. Conversely, this study integrated a foreign promoter to the Frontiers in Genetics frontiersin.org endogenous F8 coding sequence, using smaller integrated fragments (<1 kb) to repair F8 compared to the more extensive BDD-F8 (over 4 kb) in previous studies. Our data demonstrated that FVIII expression and secretion were rescued in hepatocytes derived from gene corrected iPSCs. The iPSCs with hAAT-promoted F8 cassette provide an adequate cell source for therapeutic hepatocytes. As the endogenous endothelial-expressed F8 promoter was lost in cells of HA patient with Inv1 mutation during the chromosomal reversion and the integrated hAAT promoter is liver-specific, we did not differentiate the gene corrected iPSCs into LSECs as a control in this study. In our further Frontiers in Genetics frontiersin.org studies, we will integrate an endothelial-specific promoter to correct the iPSCs and differentiate them into LSECs, and the comparison of FVIII expression in HPLCs and LSECs will be investigated. In summary, we performed a CRISPR/Cas9 mediated HMEJ in HA-iPSCs with Inv1 by targeting gene addition of the hAAT promoter and F8 exon one at the intron 1 with a high efficiency up to 10.19%. Both F8 transcription and FVIII secretion were rescued in the hepatocytes derived from gene corrected iPSCs. This is the first report of an efficient in situ genetic addition strategy in HA-iPSCs with Inv1 mutation, while further in vivo  Frontiers in Genetics frontiersin.org experiments need to perform to evaluate the effectiveness. Hopefully, our findings suggest a feasible and promising in situ genetic addition strategy for HA involving large sequence variants.

Karyotype analysis of iPSCs
G-banding analysis of chromosomes was performed. The cells were incubated with 0.1 μg/mL colcemid (Sigma-Aldrich #D7385, St. Louis, MO, United States) for 4 h, followed by trypsinization, and hypotonic treatment with 0.075 M KCl for 10 min at 37°C. Then the cells were fixed with Carnoy fixative, and the metaphase chromosomes were spreaded using an air-drying method, then treated with Giemsa (Sigma-Aldrich #48900) and analyzed.

Analysis of potential off-target sites
The potential off-target sites of sgRNA4 were searched using CHOPCHOP (http://chopchop.cbu.uib.no/) (Montague et al., 2014), seven potential sites were predicted for mismatches of up to four nucleotides. Then the regions encompassing the seven potential sites in gene-edited clones were PCR amplified, followed by Sanger sequencing (Supplementary Table S1). Different indels in HA-iPSCs and genetically edited iPSCs were used to evaluate off-target effects. Primer sequences are shown in Supplementary Table S2.

RT-PCR
Total RNA isolated with TRIzol reagent (Sigma-Aldrich #T9424) was digested with DNase for 30 min, followed by reverse transcribed via HiScript II 1st Strand cDNA Synthesis Kit (Vazyme #R212). The primers were based on exons one and four to detect the F8 transcripts. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was amplified to represent an endogenous control. Primer sequences are shown in Supplementary Table S3.
For periodic acid schiff stain, the hepatocytes on Day 25 were stained with the Periodic Acid Schiff Stain Kit (Solarbio #G1280, Beijing, China) according to the manufacturer's instructions. The periodic acid schiff stain was detected by microscopy.
For indocyanine green (ICG) uptake assay, hepatocytes on Day 25 were treated with 1 mg/mL ICG (Sigma-Aldrich #1340009) for 30 min, then washed with DPBS thoroughly and cultured in fresh Hepatocyte Culture Medium. The cells were detected by microscopy. After 12 h, the cells were observed using microscopy.

FVIII ELISA
Culture supernatants of mature hepatocytes from 12-well plates were harvested in triplicate after medium replacement for 24 h. ELISA was performed with paired antibodies for ELISA-Factor VIII: C (Cedarlane #CL20035K, Burlington, ON, Canada) according to manufacturer instructions. The standard curves were constructed using serial dilutions of normal pooled plasma, with a correlation coefficient (R2) greater than 0.990 using a semilog fit.
Frontiers in Genetics frontiersin.org

FVIII activity assay
For the FVIII activity assay, 24-hour-old culture supernatants of mature hepatocytes were collected. The activated partial thromboplastin time (aPTT) was detected using a Destiny Max hemostasis analyzer (Tcoag, Lemgo, Germany) according to the manufacturer's instructions.

Statistical analysis
GraphPad Prism 8.0 was used for data analysis. Data were analyzed using ANOVA for more than two groups.

Data availability statement
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding authors.

Ethics statement
The animal study was reviewed and approved by the Institutional Review Board of the School of Life Sciences, Central South University of China.

Author contributions
DL, MZ, ZH, and YW designed the study. ZH, MZ, and RX performed experiments and collected the data. ZH, YW, RX, JZ, and YC assembled and analyzed the data. ZH prepared original draft. DL, MZ, YW, and LW reviewed and edited the manuscript. ZH and MZ provided financial support. All authors read and approve the final manuscript. All authors contributed to manuscript revision, read, and approved the submitted version.