Ex vivo factor VIII‐modified proliferating human hepatocytes therapy for haemophilia A

Abstract Ex vivo gene manipulation in human hepatocytes is a promising therapeutic strategy in the treatment of inherited liver diseases. However, a major limitation is the lack of a highly efficient and safe genetic manipulation system for transplantable primary human hepatocytes (PHHs). Here, we reported that proliferating human hepatocytes (ProliHHs) cultured in vitro showed high susceptibility to lentivirus‐mediated genetic modification and maintained cellular phenotypes after lentiviral infection. Human factor VIII expression was introduced through F8‐Lentivirus‐mediated transduction of ProliHHs followed by xenotransplantation into immunocompromised haemophilia A mice. We demonstrated that these F8‐modified ProliHHs could effectively repopulate the mouse liver, resulting in therapeutic benefits in mouse models. Furthermore, no genotoxicity was detected in F8‐modified ProliHHs using lentiviral integration site analysis. Thus, this study demonstrated, for the first time, the feasibility and safety of lentiviral modification in ProliHHs to induce the expression of coagulation factor VIII in the treatment of haemophilia A.


| INTRODUCTION
Haemophilia A is a severe X-linked recessive inherited bleeding disorder caused by a deficiency in coagulation factor VIII (FVIII), a critical protein secreted by liver endothelial cells that is involved in the blood coagulation cascade. 1 Haemophilia A is a prevalent bleeding disorder (1:5000 male births) and represents 80%-85% of the entire haemophilia population. 2 Based on residual factor activity, haemophilia A can be classified as severe (<1% normal FVIII activity), moderate (1%-5%), or mild (5%-50%). 3 Patients usually experience spontaneous bleeding into joints and soft tissue, eventually resulting in disabling hemophilic arthropathy and early death, especially in cases of severe haemophilia.
Currently, the mainstay of haemophilia A therapy is protein replacement of deficient FVIII using either conventional or extended half-life FVIII to achieve sufficient levels for prophylaxis or on-demand therapy in response to bleeding. 4 Despite significant progress, replacement therapy requires repeated administration and may induce severe adverse events, such as blood virus infections (HBV, HCV and HIV) and the generation of neutralizing antibodies against FVIII. 5 Therefore, further improvements are warranted in the treatment of patients with haemophilia A.
Haemophilia A is an ideal target disease for gene or cell therapy, given that the quality of life can be significantly improved in patients with circulating FVIII activity >5%. Furthermore, gene therapy provides an alternative approach-eliminating the disadvantages of protein replacement-and has the potential of being a one-time, life-long, disease-altering therapy. In the last four decades, numerous efforts and remarkable advances have been made in the treatment of haemophilia A. Previous studies have shown that Adeno-associated virus (AAV) vectors injected into the liver or muscle can produce FVIII at a therapeutic level and ameliorate disease symptoms in mouse and dog animal models of haemophilia A. 6,7 Encouraged by these findings, researchers designing clinical trials have focused on in vivo clotting factor gene transfer to cure haemophilia using AAV delivery. According to recent clinical experiments, a wide range of FVIII activity (12%-250%) was observed in seven patients treated with AAV5-FVIII. 8,9 However, the levels of F8 gene expression in treated patients were found to gradually decrease over 4 years (2016-2020), and whether these levels are in the final steady state remains to be determined. In addition, AAV-mediated gene therapy is still hindered by problems; in particular, some patients can possess pre-existing antibodies to AAV and the lack of clarity on the effects in paediatric patients with haemophilia remain unclear. Therefore, other research groups are investigating gene-modified cell therapy approaches, which could avoid the critical obstacles of AAV therapy in vivo. F8-modified haematopoietic stem cells and FVIII-corrected endothelial cells derived from haemophilia A patient-induced pluripotent stem cells also provide long-term phenotypic correction of haemophilia A. [10][11][12] These results suggest that ex vivo gene-modified cell therapy is a promising strategy for haemophilia A treatment.
Hepatocytes can engraft and repopulate the liver post-transplantation. Therefore, hepatocyte transplantation is a promising alternative to liver transplantation in the treatment of various genetic liver diseases. 13,14 Although liver sinusoidal endothelial cells, but not hepatocytes, are the primary cells of FVIII secretion, [15][16][17] hepatocytes can   still be considered as an ideal target cell for F8 expression, as they are   the predominant host cells for gene therapy (AAV5-hFVIII derived fac-tor VIII protein expression). 9 However, little progress has been made in gene-modified hepatocyte therapy owing to the complicated and inefficient process of gene manipulation in primary human hepatocytes (PHHs). 18 In this study, we examined the potential of F8-modified human hepatocytes to treat haemophilia A in a mouse model. The results showed that proliferating human hepatocytes (ProliHHs) were more susceptible to lentivirus infection than PHHs. F8-modified ProliHHs effectively secreted FVIII protein and maintained a cell state similar to that of untreated ProliHHs. Importantly, we demonstrated that F8-modified ProliHHs transplantation could restore coagulation function to wild-type (WT) mice, with implications for effective treatment.
Furthermore, analysis of the integration profile indicated no genotoxicity with these lentiviral vectors (LVs). Thus, these findings provide proof of concept for the potential use of F8-modified human hepatocytes in the treatment of haemophilia A.

| Human hepatocyte culture
We used cryopreserved human hepatocytes (Lot: MRW, JFC) from two individuals provided by Celsis In Vitro Technologies (Blatimore, MD). The culture protocol was adapted from our previously reported system with modifications. 19 The cryopreserved PHHs were thawed in the 37 C-water bath in Dulbecco's Modified Eagle Medium with 10% fetal bovine serum. The thawed hepatocytes were resuscitated and cultured in 37 C, Hypoxia incubator (5% CO 2 , 1% O 2 ). The medium was changed to human HM about 24 h after seeding and every 2 days thereafter.

| Mice
The Fah À/À Rag2 À/À Il2rg À/À (FRG) mice were presented from Dr. Wang Xin's Lab. FRG mice are on the hybrid strain of C57BL/6J and 129S6/ SvEvTac. The mice were fed with drinking water containing 7.5 mg/L NTBC (Synthesized by Capot Chemical, China). A 8 to 12 weeks old male and female mice were used for experiments and we observe no sex bias differences were detected. FVIII À/À mice were purchased from Shanghai Biomodel Organism Center, Inc.
None of the animals used in our study had been subjected to prior procedures and was drug and test naive. All animals were housed in a temperature-and light-controlled (12-h light/dark cycle) specific pathogen-free animal facility, in individually ventilated cages always with companion mice. All animal experiments were performed basis on protocols approved by the institutional animal care and use committee at the Shanghai Institute of Biochemistry and Cell Biology.

| Polymerase Chain Reaction (PCR)
Total RNA was isolated from cells and mice liver by Trizol (Invitrogen).  Table S1.
After transplantation, NTBC was transiently put on for 4 days when mice lost 10%-20% of their body weights. Mice were sacrificed at 5 months after transplantation. Integration site sequence data were controlled for IS that were aligned to more than one definite location (multiple hits) and collision reads (identical IS found in different patients). The tool ISOT 1.0 (inhouse Waker Bioscience) generated a list of all unique IS with precise loci, gene names and sequence counts.

| Evaluation of integrated cell cloning
Illumina MiSeq second generation sequencing platform could semiquantitatively evaluate the integrated cell cloning by determining the number of sequences (search frequency) of single vector-genome integration sites. The relative sequence count of all detected ISs is calculated based on all sequences that can be mapped to specific locations in the genome. For each sample, analyse the top 10 most significant ISs (ranking from 1 to 10, from high to low).

| Analysis of clone diversity
We developed a method based on the property of Rényi entropy to evaluate the diversity of each sample cell clone. The algorithm calculates α Take the H value under two extremes: richness and evenness, and construct a cloning plane. The distance ratio between theoretical maximum polyclonal degree and monoclonal degree defines the polyclonal monoclonal distance, which is a diversity measurement developed internally and can accurately evaluate the sample richness and uniformity.

| Relationship between integration sites and tumour related genes
We compiled a well-defined list of cancer genes from the cancer census database (http://cancer.sanger.ac.uk/census). The selection of these genes is based on how mutations in these genes promote can-

| Statistical analyses
The number of biological and technical replicates and animals are indicated in figure legends and text. All data are presented as the mean ± SD. For most statistic evaluation, an unpaired Student's t test was applied for calculating statistical probability in this study. p Values were calculated by two-tailed test. Only for survival analysis in Figure 5B,G, the Mantel-Cox log-rank test was applied. Statistic calculation was performed using GraphPad Prism9 (GraphPad).

| ProliHHs are highly susceptible to lentiviral infection
Gene manipulation of PHHs is challenging and inefficient, which greatly limits the potential application of gene-modified human hepatocytes for cell therapy and in vitro modelling of liver diseases. 18,21,22 Recently, we established a hepatocyte medium (HM) culture system that allowed 10,000-fold expansion of PHHs.
These ProliHHs have a bi-phenotypic 'intermediate' status between mature hepatocytes and liver progenitor cells. 19 Therefore, we speculate that ProliHHs may be more amenable to gene manipulation for various applications.
Lentiviruses are generally the most convenient and commonly used tools for genetic manipulation in the basic research and clinical treatment of life science, such as gene overexpression, knockdown, and cell marking. 23 Here, we used PHHs as a negative control and

| F8-lentiviral modification maintains features of ProliHHs
To test the hypothesis that F8-modified ProliHHs have therapeutic potential for the treatment of haemophilia A, we first confirmed that PHHs and ProliHHs do not express F8 through analysis of cellular transcriptional profiling ( Figure S1A), which was a prerequisite for the F8-modified ProliHH therapy. Next, we constructed a new third- generation self-inactivating LV carrying an EF1α promoter, a codonoptimized DNA encoding a B domain-deleted (BDD) human FVIII protein, and a reporter GFP to indicate the infection efficiency ( Figure S1B).
The packaging capacity of LVs readily accommodated these expression cassettes, and high-titre F8-lentiviral particles were produced using a lentiviral packaging system. Using untreated cells as a negative control, we found that the efficiency of F8-lentivirus-infected ProliHHs (F8-ProliHHs) was approximately 65%, which was lower than that of empty lentivirus-infected ProliHHs (EV-ProliHHs), likely due to the large size of the F8 gene (4375 bp) (Figure 2A). To investigate whether F8-ProliHH was functional, F8 gene expression and protein secretion were separately tested in F8-modified ProliHHs. We found that F8 showed a significant increase as determined by F8-qPCR and ELISA ( Figure 2B,C). In addition, we analysed the characteristics of ProliHHs before and after F8 modification, including the mRNA expression and protein immunofluorescent staining of hepatic-function-and progenitorassociated signature genes ( Figure 2E,G) and found that these characteristics were not affected by F8 modification. Together, these findings suggested that F8-ProliHHs gained the ability to secrete FVIII proteins, and F8 modification did not affect the bi-phenotypic status of ProliHHs.

| Transplantation of F8-modified ProliHHs results in functional rescue in haemophilia A mice
We examined whether F8-ProliHHs were viable and functional in the treatment of a mouse model of haemophilia A. For this purpose, we generated a quadruple knockout mouse model of haemophilia A that allowed the engraftment and expansion of human hepatocytes.
With the established animal model, we then transplanted F8-ProliHHs into FRGF8 mice to examine their therapeutic efficacy ( Figure 3A). EV-ProliHHs were used as the negative control.
The results showed that five of the six FRGF8 mice transplanted with EV-ProliHHs died of surgical bleeding caused by intrasplenic cell transplantation. However, all the mice treated with F8-ProliHHs transplantation survived, implying that an immediate, effective therapeutic effect had been achieved after F8-ProliHH transplantation ( Figure 3B). We found that the repopulation of genetically modified ProliHHs in mouse liver was promoted by cyclical NTBC withdrawal and restoration. The levels of human albumin in the plasma were gradually increased after transplantation of gene-modified ProliHHs, reaching a level of 4 mg/mL after 5 months ( Figure 3C). This indicated that gene-modified ProliHHs efficiently engrafted and repopulated mouse livers after transplantation. Moreover, the concentration of secreted human FVIII was improved steadily in mice transplanted with F8-ProliHHs and was close to the normal FVIII concentrations in the human plasma after 5 months ( Figure 3D). By contrast, the human FVIII protein was not detected in the EV-ProliHH-transplanted group.
Next, we analysed the relationship between human albumin (ALB) and FVIII concentration. The plot (ALB against FVIII; Figure 3E

| Maturation of gene-modified ProliHHs after repopulation in vivo
Gene-modified ProliHHs are bi-phenotypic cells, they retained some of the features of mature hepatocytes and expressed liver progenitorassociated markers. That is, they do not exhibit fully mature hepatic functions. 19 Here, we further examined whether gene-modified cells could transform into mature hepatocytes after transplantation. Firstly,  Figure 4D). Together, these data suggested that genemodified ProliHHs underwent maturation with human-specific metabolism in vivo.

| Integration profile of F8-modified ProliHHs
In our previous study, no tumours were found in ProliHH-transplanted FRG mice after transplantation. 19 Nevertheless, whether there are potential genotoxic effects associated with the lentivirus-modified ProliHHs remains unclear. Therefore, the livers of F8-modified ProliHH-transplanted FRGF8 mice were examined 5 months after transplantation. We did not observe macroscopic liver tumours in any of the treated mice.
To investigate the risk of oncogenesis, we evaluated the clonal nature of engrafted F8-modified ProliHHs in transplanted FRGF8 mice. We sequenced the pre-transplanted cells (F8-modified ProliHHs) and two samples of the post-transplanted humanized liver using ligation target amplification PCR (LTA-PCR) to collect the genome-wide lentiviral integration profile for clone analysis. [27][28][29] After mapping to specific locations in the genome using the ISover-Time tool, we identified 9902, 3355 and 1587 unique integration sites (UISs) in F8-modified ProliHHs and two liver samples of engrafted F8-modified ProliHHs, respectively ( Figure 5A). We then performed a global analysis of integration site distributions. The results of the gene mapping analysis showed that integration events were widely distributed throughout the human chromosomes and were not exactly consistent with a random distribution of UISs using in silico data ( Figure 5B,C). To identify integration site preferences within the gene, integrations were mapped to exons, introns, and intergenic regions.
We found that 75.8% of the UISs occurred within genes in which integrations were mainly located in introns ($72%) ( Figure 5D Comparison of gene expression of mature hepatic markers, such as phase I, phase II enzymes, and transporters genes in PHHs, EV-ProliHHs, repopulated EV-ProliHHs in mice liver, and F8-ProliHHs, repopulated F8-ProliHHs in mice liver. RNA was extracted from repopulated livers. Human-specific primers were used in qPCR.
susceptible to lentiviral infection than PHHs. ProliHHs effectively secreted F8 proteins and maintained their bi-phenotypic status after F8-lentivirus modification. Importantly, F8-modified ProliHHs showed a remarkable capability to repopulate and significantly corrected coagulation function in adult mice with haemophilia A after in vivo transplantation. As a proof of concept, our study provided a new approach for the treatment of haemophilia A.
A major goal of cell therapy is to replace deficient functions with healthy cell transplantation. Liver sinusoidal endothelial cells are known to be the main producers of FVIII, [15][16][17] and several studies have shown that endothelial cells repopulated the liver endothelium and corrected the bleeding phenotype of haemophilia A mice. 30,31 Unfortunately, liver sinusoidal endothelial cell transplantation has not been successfully achieved in clinical treatments, owing to the uncertainty regarding safety and effectiveness of cell transplantation. In fact, clinically feasible cell transplantation therapies may be applied to haemophilia treatment, such as haematopoietic stem cell and hepatocyte transplantation. 10   In conclusion, our study provides a new proof of concept that ex vivo gene therapy of human ProliHHs by lentiviral infection is feasible. Furthermore, we evaluated the safety and efficacy of the modified ProliHHs using animal models. In addition, these modified human hepatocytes may be used for extensive studies of cell therapy in vivo and liver disease modelling in vitro.