Delivery of Full-Length Factor VIII Using a piggyBac Transposon Vector to Correct a Mouse Model of Hemophilia A

Viral vectors have been used for hemophilia A gene therapy. However, due to its large size, full-length Factor VIII (FVIII) cDNA has not been successfully delivered using conventional viral vectors. Moreover, viral vectors may pose safety risks, e.g., adverse immunological reactions or virus-mediated cytotoxicity. Here, we took advantages of the non-viral vector gene delivery system based on piggyBac DNA transposon to transfer the full-length FVIII cDNA, for the purpose of treating hemophilia A. We tested the efficiency of this new vector system in human 293T cells and iPS cells, and confirmed the expression of the full-length FVIII in culture media using activity-sensitive coagulation assays. Hydrodynamic injection of the piggyBac vectors into hemophilia A mice temporally treated with an immunosuppressant resulted in stable production of circulating FVIII for over 300 days without development of anti-FVIII antibodies. Furthermore, tail-clip assay revealed significant improvement of blood coagulation time in the treated mice.piggyBac transposon vectors can facilitate the long-term expression of therapeutic transgenes in vitro and in vivo. This novel gene transfer strategy should provide safe and efficient delivery of FVIII.


Introduction
Hemophilia A is a congenital bleeding disorder caused by a deficiency of procoagulation Factor VIII (FVIII). Repeated intravenous injection of recombinant FVIII protein can prevent bleeding events, but alternative treatments could potentially improve patients' quality of life; therefore, hemophilia A is an attractive target disease for the application of gene therapy. Significant advances in both preclinical animal models and phase I/II human clinical trials have been reported [1].
Due to the large size of human FVIII cDNA (7.0 kb in total), it cannot be effectively delivered by most existing viral vectors; this technical issue is one of the biggest obstacles to the gene augmentation approach. Consequently, most gene therapy studies, including those aimed at clinical applications, use truncated FVIII that lacks the non-essential B-domain, so called BDD FVIII. However, the B-domain contains several glycosylation sites and is involved in intracellular interactions that regulate protein quality control and secretion; in particular, this domain is presumed to be involved in the clearance of FVIII from plasma [15]. Whether the B-domain is beneficial for FVIII expression in the context of gene therapy remains to be determined.
Recently, non-viral gene-transfer technologies, such as the sitedirected integration approach, have been significantly improved [16,17]. Although conventional plasmid vector systems are inefficient at stably integrating into the target genome, transposon vectors have emerged as attractive gene-delivery tools because of their ability to stably integrate into the genome and achieve efficient and prolonged transgene expression both in vitro and in vivo. In fact, the Tc1/mariner family member Sleeping Beauty has been utilized to deliver BDD FVIII into the hemophilia A mouse model [18].
More recently, the piggyBac DNA transposon vector was derived from cabbage looper Trichoplusia ni. This vector is active not only in insect cells [19], but also in mammalian cells [20]. In HEK293 cells and human primary T cells, piggyBac vectors have higher transposition activity than other widely used transposon vector systems, such as Sleeping Beauty or Tol2 [21,22]. Importantly, the piggyBac transposon has a larger cargo size, and can deliver up to 9.1 kb of foreign sequence without significant loss of integration efficiency [20]. Moreover, a recent paper showed that a 100-kb DNA fragment from the HPRT gene locus could be inserted into mouse ES cells using piggyBac, although the efficiency was low and selective pressure was required [23]. In addition, other groups have shown that piggyBac vectors can deliver a reporter gene, such as lacZ or luciferase, into mice via hydrodynamic tail-vein injection [24], suggesting that piggyBac vectors could be utilized as a non-viral vector for in vivo gene therapy applications.
In this study, we investigated the ability of a piggyBac vector to deliver full-length FVIII, both in cultured cells and in a mouse model. Injection of piggyBac vector into hemophilia A mice resulted in stable and sustained expression of full-length FVIII and improvement of the bleeding phenotype.

piggyBac vector construction
A schematic representation of the piggyBac transposon vectors used in this study is shown in Figure 1. In brief, to construct the PB-EF1a-EiP vector, the 59 and 39 terminal repeat (TR) regions were derived from the PB-MSCII plasmid, based on Trichoplusia ni IFP2 piggyBac transposon [kind gift from Dr. Knut Woltjen]. Human EF1a promoter was derived from PL-sin-EF1a-EiP [25] as a HindIII-NcoI fragment, and a DNA fragment containing an internal ribosome entry site (IRES) derived from encephalomyocarditis virus and the puromycin resistance gene was obtained by NcoI-ClaI digestion from the PL-sin-EF1a-EiP plasmid. For construction of PB-EF1a-GWiP, the NcoI (blunted with Klenow)-XbaI fragment of the EF1a promoter and the SacII (blunted by Klenow)-HindIII fragment of the Gateway destination cassette (attR1-Cm R -ccdB-attR2) were introduced into the XbaI-HindIII sites of the PB-EF1a-EiP plasmid. For construction of PBaseII, we digested the pCyL-43 plasmid [kind gift from Dr. Knut Woltjen] with SalI and KpnI, blunted it with Klenow, and self-ligated to remove the PGK/puromycin-resistance cassette. The 1.8 kb fragment of HyAcPBase gene [26] with optimized codon usage to human was chemically synthesized (GeneScript). To generate pCAG-HyAcPBase, the gene fragment was cloned into the SacI-NheI site of pCAGGS vector [27].

piggyBac vector transfection in vitro
Human embryonic kidney 293T cells were maintained in DMEM media supplemented with 10% fetal bovine serum (FBS), ampicillin, and streptomycin. Lipofectamine 2000 (Life Technologies, CA, USA) was used to transfect 293T cells as described in the manufacturer's protocol. Human iPS cells (201B7 [30]) were maintained in ReproCELL Primate ES Cell media (ReproCELL, Japan) or Knockout-DMEM media (Life Technologies) supplemented with 15% Knockout serum replacement, 2 mM Gluta-MAX, 2 mM MEM non-essential amino acids, 0.1 mM 2mercaptoethanol, and 10 ng/ml recombinant human basic-FGF. Puromycin-and hygromycin-resistant SNL cells (SNL-PH) treated with mitomycin C were used as a feeder layer for human iPS cell cultures. To transfect human iPS cells, we treated the cells with ROCK inhibitor Y-27632 (final 10 mM) for 1 hour, and then dissociated them with CTK solution (0.25% trypsin, 0.1 mg/ml collagenase IV, 1 mM CaCl 2 , and 20% KSR) to remove feeder cells. Human iPS cell colonies were dissociated into single cells using Trypsin-EDTA solution, and then washed twice with Opti-MEM media. We used either the Neon system (Life Technologies) with pulse voltage = 1200 V, pulse width = 40 ms, and 1 pulse, or the NEPA21 electroporation system (NepaGene Inc., Japan) with poring pulse voltage = 125 V and poring pulse width = 5 ms. Transfection efficiencies of the piggyBac vectors with fluorescence reporter were monitored by flow cytometry as previously described [25].

DNA or RNA extraction and qRT-PCR
For genomic DNA extraction, cell pellets were lysed with 200 mg/ml Proteinase K solution (50 mM Tris-HCl, 20 mM EDTA, 100 mM NaCl, 1%w/v SDS) at 55uC for 3-18 hours, and genomic DNA was extracted by phenol-chloroform-isoamyl extraction method. For RNA, cell samples were lysed using the Trizol reagent (Life Technologies), and total RNA was extracted as described in the manufacturer's protocol. One microgram of RNA was reverse transcribed into cDNA using random primers (9 mer) and Oligo dT (20 mer) with ReverTraAce (Toyobo Inc., Japan).
Real-time quantitative PCR was performed on a StepOnePlus thermal cycler (Life Technologies) with SYBR Select Master Mix (Life Technologies). For genomic copy number analysis, genomic DNA with 3 copies of piggyBac vector insertion sites was used as a standard curve, and amount of input genomic DNA was normalized with Ct value by the primers to specifically amplify the NANOG gene region. For cDNA quantification, diluted plasmid DNA (PB-EF1a-EiP) was used to establish a standard curve for the IRES primers, and cDNA input was normalized with the GAPDH primers. Primer sequences are listed in Table S1. Data are presented as means 6 standard deviation. The statistical  Figure 1A). doi:10.1371/journal.pone.0104957.g001 difference was determined by two-sided Student's t test, and P, 0.05 was considered as significant.

FVIII activity measured by activated partial thromboplastin time (aPTT) assay
To confirm FVIII secretion in vitro, supernatant from cell cultures (seeded at 2610 5 cells per one well of 12-well plate for 293T cells) was collected 3 days after medium change, centrifuged at 11,0006g for 5 minutes to remove cell debris, and then stored at 280uC until being subjected to the aPTT assay as previously reported [10][11][12]. In brief, samples were diluted 10-fold by Owren's Veronal buffer (Sysmex Corporation, Japan) and mixed with FVIII-deficient plasma and aPTT reagent (Sysmex Corporation, Japan). After 3 minutes of incubation at 37uC, the coagulation reaction was activated by addition of 0.02 M CaCl 2 solution. Coagulation time was measured using a STart4 hemostasis analyzer (Roche Diagnostics). Recombinant human FVIII (Kogenate-FS, Bayer HealthCare Pharmaceuticals, NJ, USA) was used to prepare the standard curve.

FVIII gene transfer studies in hemophilia A mice
All animal procedures were reviewed and approved by the Nara Medical University Animal Care Committee (Permit Number: 10728). Hemophilia A knockout mice, in which the FVIII gene is disrupted by insertion of the neomycin gene into Exon 16, were purchased from the Jackson Laboratory (stock number 004424) and back-crossed with C57BL/6 mice. Mice were used at 6-8 weeks of age. Plasmid DNA was diluted in a 10% body-weight volume of lactated Ringer solution and injected via the tail vein within 7 seconds [31]. Because human FVIII is inherently immunogenic in hemophilia A mice, some of the hemophilia A mice also received intraperitoneal injection of cyclophosphamide (20 mg/kg per injection) on the day of vector injection, and then biweekly for 4 weeks. Mouse blood was first collected 1 week after injection from the saphenous vein under isofluorane/oxygen anesthesia, using 10% buffered citrate as an anticoagulant. To generate platelet-poor plasma, collected blood samples were centrifuged at 11,0006g for 5 minutes and stored at 280uC prior to testing.
Chromogenic assay for FVIII activity, FVIII antigen ELISA, and detection of anti-FVIII antibodies A chromogenic assay was used to measure plasma FVIII functional activity (FVIII: C) as previously described [10][11][12]. Normal pooled human plasma was used to establish the FVIII standard curve; the sensitivity of this assay is 10 mU/mL. In some mouse plasma samples, FVIII antigen (FVIII:Ag) was determined by ELISA (Affinity Biologicals Inc., Ancaster, ON, Canada). Anti-FVIII inhibitory antibodies were measured by the Bethesda assay as previously described [11].

Tail-clip challenge test
At the termination of the experiments, phenotypic correction was tested in both untreated and vector-treated hemophilia A mice, as previously reported [12]. In brief, mice were anesthetized with isoflurane, and their tails were clipped off at a position near the end of the tail where the cross-sectional diameter was 2.0 mm. The mice were then observed to determine the bleeding time.

Immunohistochemical analysis of FVIII in multiple organs
Immunohistochemical analysis of FVIII expression in both untreated and vector-treated hemophilia A mice were performed as previously reported [11,12] with minor modifications. In brief, after tail-clip challenge tests, the mice were sacrificed, multiple organs liver, spleen, kidney, heart, and lung) were harvested, fixed with 10% formalin, embedded in paraffin and sectioned for staining. To assess transgene expression in tissues, specimens were stained with Cy3-labeled human anti-FVIII IgG [32] using the FluoroLink Ab labeling kit (Amersham Biosciences, Buckinghamshire, UK). Staining images were captured with FV300 confocal laser scanning microscope (Olympus Co., Tokyo, Japan).

Stable expression of FVIII by piggyBac vector in multiple cell types
In order to achieve high-level transgene expression, we first optimized the internal promoter of our piggyBac vector. We constructed EGFP-expressing piggyBac vectors containing various promoters including those of PGK, CAG, and Elongation factor 1a (EF1a or EEF1A1). The EF1a promoter was the strongest among those we tested ( Figure S1). Next, we optimized the amount of the piggyBac vector DNA and transposase-expressing plasmid. In agreement with previous reports [33], we found that a higher amount of piggyBac vector and a lower amount of transposase (400 mg/ml and 50 mg/ml, respectively) were optimal for transduction of our piggyBac vectors ( Figure S2). We also tested a hyperactive version of the piggyBac transposase [24,26], but observed no significant improvement in transduction efficiency relative to the wild-type transposase (data not shown). To examine the effect of vector size on transduction efficiency, we constructed mCherry-expressing piggyBac vectors containing cDNAs of various sizes. Although the transduction efficiencies were reciprocally correlated with the vector sizes, the full-length FVIII cDNA (10.6 kb) could be transduced as efficiently as the BDD FVIII cDNA ( Figure S3). In fact, when we transfected piggyBac vectors expressing EGFP, BDD FVIII, or full-length FVIII ( Figure 1A), we observed that full-length FVIII vector could integrate into host chromosomes as efficiently as BDD FVIII vector ( Figure 1B).
Next, to evaluate the expression of FVIII cDNAs by piggyBac vectors, we transfected these vectors into 293T cells and human iPS cells (clone 201B7); the former cell type is easy to transfect, whereas the latter type retains a normal karyotype while retaining unlimited self-renewal capacity and pluripotency [34]. We then extracted total RNA from these cells and quantitated transcription levels from piggyBac vectors by qRT-PCR using IRES primers. Even though the transcription levels of the FVIII vectors were lower than those of the of EGFP control vectors, comparable levels of transcription were detected from BDD and full-length FVIII vectors (Figure 2A,B).
To assess the level of secreted and functional FVIII protein, we measured the FVIII coagulation activity in the culture supernatant by activated partial thromboplastin time (aPTT) assay. We observed higher coagulation activity with full-length FVIII than BDD FVIII ( Figure 2C,D). We also observed lower secretion of FVIII protein in undifferentiated iPS cells than in 293T cells, partly due to the lower transduction efficiency of iPS cells. The undifferentiated state of iPS cells may be associated with immaturity of the secretory pathways involved in FVIII production.

Successful long-term phenotypic correction of hemophilia A mice by hydrodynamic injection with piggyBac vectors
As a proof-of-concept that piggyBac vectors are applicable to gene therapy applications, we injected FVIII-expressing piggyBac vectors into hemophilia A model mice. Because FVIII is mainly  produced in the liver, we targeted the liver by hydrodynamic injection via the tail vein. Human FVIII is inherently immunogenic in hemophilia A mice; consequently, all recipient mice that did not receive immunosuppresive treatment developed inhibitory anti-human FVIII antibodies, and there was no difference in the titer levels of FVIII antibodies between BDD and full-length FVIII vectors ( Figure 3A, B). Therefore, under transient immunosuppression with cyclophosphamide, we injected either BDD or fulllength FVIII piggyBac vectors (25 mg) with various amounts of transposase (0 mg, 6.25 mg, or 25 mg). As expected from the in vitro results (Supplemental Figure 2), we observed higher FVIII level in plasma of animals that received lower amounts of transposase. The levels of FVIII in plasma were 25-40 mU/ml (2.5-4.0% of normal level) in all hemophilia A mice injected with 6.25 mg of transposase ( Figure 3C, D). Furthermore, we observed that the activity levels of full-length FVIII ( Figure 3D) were comparable to those of BDD FVIII ( Figure 3C), despite the size difference between the two proteins. In addition, the levels of FVIII antigen determined by human-specific FVIII ELISA corresponded with FVIII activity, as revealed by chromogenic assay in the same plasma samples (data not shown). The plasma levels of FVIII were stably maintained for more than 300 days.
In an attempt to boost the production of FVIII, we tested the effect of multiple injections of piggyBac vector. Surprisingly, when we injected the vector at 24-hour intervals, no FVIII production was observed ( Figure 3E). Rather, we observed the production of FVIII inhibitors in response to the injections ( Figure 3F). On the other hand, multiple vector injections at 4-week intervals boosted the levels of FVIII expression in hemophilia A mice without causing development of FVIII antibody ( Figure 3E, F).
Next, to determine whether the piggyBac vector injection corrected the bleeding diathesis phenotype, we subjected the mice to tail-clip challenge. The tails of treated and untreated hemophilia A mice from Figure 3D were transected at a position corresponding to a specific diameter, and then bleeding times were monitored. Whereas untreated hemophilia A mice (n = 6) tended to lack stable clot formations, all hemophilia A mice treated with piggyBac vector (n = 5) formed a stable clot after tail clipping (Mean bleeding time; Untreated hemophilia A mice: 18 min 24 sec, Vector-treated hemophilia A mice: 6 min 13 sec, respectively) as shown in Figure 4A.
Finally, the mice were sacrificed, and multiple recovered organs were examined for FVIII expression by immunohistochemical staining. Among the organs we examined, only liver expressed detectable level of human FVIII protein ( Figure 4B). We also confirmed the expression of human FVIII transcripts in homogenized liver by qRT-PCR ( Figure 4C). Thus, even though FVIII expression was controlled by the ubiquitously active EF1a promoter, the expression pattern achieved by the hydrodynamicinjection method is largely restricted in liver.
Overall, these data indicate that piggyBac vectors can mediate efficient and sustained expression of the full-length FVIII gene, and thereby rescue the bleeding phenotype of hemophilia A mice.

Discussion
The continuous improvement of gene-transfer vectors has broad implications for genetic studies and gene-therapy applications. In the gene-therapy field, host immunoresponse to viruses pose a major challenge for virus-mediated gene transfer, especially in the case of adenoviral or AAV vectors [2,3,6,14,35,36]. In addition, some hemophilia patients still suffer from AIDS due to the contamination of HIV in clotting factor concentrates that were used for the treatment. To avoid the risk of recombination between viral vectors and wild-type viruses, the use of relevant lentiviral vectors [8][9][10][11] should be avoided in patients who have a history of infection with HIV. Another limitation of viral vectors relates to the costly and cumbersome manufacturing process, which poses significant regulatory hurdles. Therefore, there is a need to develop safe and efficient non-viral gene delivery approaches that diminish the immunoresponse against viral components, overcome some of the manufacturing and regulatory hurdles, and allow for stable expression of the therapeutic gene(s).
Currently available non-viral plasmids integrate inefficiently, and gene expression from these plasmids typically declines within days after transfection. To overcome these issues, we have attempted to establish non-viral gene delivery approaches that take advantage of the latest transposon technology [19,20]. Because piggyBac transposons integrate into the target-cell genome, they have the potential for long-term expression of therapeutic genes.
Here, we performed hydrodynamic injection of piggyBac vectors into hemophilia A mice, resulting in stable and long-term expression of BDD and full-length FVIII in plasma. Furthermore, the bleeding phenotype of hemophilia A mice was almost completely abolished after transduction with piggyBac vectors. Although hydrodynamic transfection is not readily applicable to a clinical setting, our study demonstrates proof-of-principle that piggyBac vectors could be used for hemophilia gene therapy, because these vectors enjoy several advantages relative to previously developed viral vector-mediated gene transfer approaches.
First, we took advantage of the large cargo size of piggyBac vectors to deliver full-length FVIII cDNA. Despite the difference in the sizes of the two vectors (8.1 kb and 10.6 kb), we observed comparable transduction efficiencies between BDD and full-length FVIII vectors. The B-domain of FVIII is not essential for coagulation activity; however, previous reports have suggested that it plays a role in protein secretion and quality control [15]. Such a function of the B-domain may influence the overall FVIII activity under some circumstances. Our piggyBac vector should provide a powerful platform for investigating the effect and biochemical properties of the B-domain and its variants.
Second, piggyBac vector-mediated gene transfer offers a simple and cost-effective gene delivery system, because gene transfer can be achieved simply by co-transfection of vector DNA and transposase. This is in contrast to viral vector preparation, which requires a packaging cell line, several helper plasmids, concentration, and purification. The simplicity of the production procedure and lower cost should facilitate the manufacture of piggyBac vectors for clinical applications.  On the other hand, several problems must be solved before transposon-based vectors can be used in therapeutic applications. Similar to gammaretroviruses or lentiviruses, piggyBac vector preferentially integrate their DNA sequences into host chromosomes near transcriptional start sites [21]; therefore, insertional mutagenesis and genotoxicity remain significant concerns. Indeed, insertion of non-self-inactivating retroviral vectors has resulted in the activation of oncogenes in human patients [37,38]. More recently designed self-inactivating retroviral vectors are less likely to activate surrounding genes via their LTR promoters; however, the possibility of trans-activation by internal enhancers/promoters must still be taken into account [39]. The degree of genotoxicity mediated by our piggyBac vector remains to be determined.
Hemophilia A patients require a continuous supply of active FVIII protein; therefore, for gene-therapy applications, it is preferable to achieve stable and sustained expression of FVIII in order to achieve consistent therapeutic outcomes and reduce the risk of developing inhibitory antibodies. Our approach to in vivo delivery by hydrodynamic injection cannot control the integration sites or targeted cells. However, the ex vivo gene therapy approach using tissue stem cells, BOECs [11,12], or patient-derived iPS cells enables us to pre-determine the integration sites of the vectors. The ex vivo approach allows us to select for cell clones that effectively secrete active FVIII and the vector has integrated into a ''safe harbor'' in the human genome, however, suitable cell type(s) for ex vivo targeting need to be determined.
Another limitation of the piggyBac vector is its delivery method. Hydrodynamic injection is a well-established and efficient delivery system in small animals [31,40], but not readily applicable to a human clinical setting for hemophilia patients as the method carries a risk of liver damage and bleeding. This warrants the development of alternative in vivo DNA delivery approaches. More controlled (smaller buffer volume, but with similar hydrodynamic pressure for purposes of gene transfer) and local hydrodynamic injection methods using catheter are preferred [41,42]. Transposon and transposase constructs can be delivered to target cells by chemical transfection with carrier molecules that facilitate their entry. Physical transfection methods such as electroporation have been used for ex vivo gene delivery of transposon/transposase constructs, resulting in long-term and efficient transgene expression. Furthermore, recent advances in tissue-specific DNA delivery methods based on nanotechnology [43] or chemical modifications may accelerate the delivery of piggyBac vectors in vivo.
In conclusion, we successfully achieved long-term therapeutic expression of full-length FVIII gene in vitro and in vivo by nonviral piggyBac vectors. The present study is the first report demonstrating that piggyBac vectors can mediate sustained FVIII expression in vivo, and provides evidence that piggyBac is a promising tool for various in vivo gene-transfer applications. We believe that the present study will encourage the development of piggyBac-based vectors, and contribute to future in vivo gene and cell therapy for genetic diseases including hemophilia.