Therapeutic Levels of Functional Human Factor X in Rats After Retroviral-Mediated Hepatic Gene Therapy

HEMOPHILIAS RESULT in spontaneous and injury-induced bleeding due to a deficiency of a coagulation factor.1 Expression of a normal gene by some tissue of these patients could ameliorate the clinical manifestations of hemophilia by achieving continuous production of the appropriate coagulation factor. Liver is the optimal target organ to express coagulation factors because it is the major site of their synthesis and has direct contact with the blood. Gene therapy for hemophilia has been attempted by delivering plasmid DNA or viral vectors to various cell types using ex vivo or in vivo methods.2-5 Delivery of adenoviral vectors or plasmid DNA to the livers of mice, rats, or dogs has resulted in therapeutic levels of factor VIII6,7 (FVIII) or factor IX^8-12 (FIX), but levels decreased rapidly in normal animals. Retroviral vectors have resulted in long-term expression of FIX from hepatocytes13 or myoblasts14 15 in vivo, but the levels were subtherapeutic.

We are investigating the use of retroviral vector-mediated hepatic gene therapy to treat blood protein deficiencies. Recently, we determined that the human α₁-antitrypsin (hAAT) promoter led to higher expression of hAAT protein from a retroviral vector than from several other promoters tested. These included the long terminal repeat (LTR), albumin, apolipoprotein AI, phosphoenolpyruvate carboxykinase, or the large subunit of RNA polymerase II promoters.16-18 We hypothesized that therapeutic levels of a coagulation factor might be achieved by using a retroviral vector containing the hAAT promoter. Coagulation factor X (FX) was chosen as the protein to study for developing gene therapy for hemophilia. Activated FX plays a pivotal role in coagulation by converting prothrombin to thrombin. Because the incidence of FX deficiency is low at 1 per 500,000, concentrated human FX (hFX) preparations have not been developed.1 Patients with severe hFX deficiency might be more appropriate for initial trials of gene therapy than patients with hemophilia A or B, for which safe preparations of purified factors are available. In addition, the half-life of FX in humans is longer (40 hours) than the half-life of FIX (20 hours) or FVIII (10 hours),1 which could facilitate achieving therapeutic levels at low biosynthetic rates. We therefore generated a high-titer retroviral vector with the hFX cDNA and delivered it to regenerating rat livers in vivo using a technique that selectively modifies hepatocytes.16 This retroviral vector resulted in expression of stable and therapeutic levels of functional hFX in vivo. This high-level long-term expression of therapeutic levels of a coagulation factor in animals overcomes one of the primary obstacles to successful application of gene therapy in humans.

Construction of retroviral vectors.The previously described retroviral vector RS8391¹⁹ was digested with Bgl II and ligated with annealed oligonucleotides A (5′ GATCTGCGGCCGC 3′) and B (5′GATCGCGGCCGCA 3′) to generate RS8391-BN, which has a unique Bgl II site upstream of a unique Not I site. A 1.5-kb hFX cDNA20 was obtained from Dr Earl Davie (University of Washington, Seattle, WA). It was modified to contain 35 nt of 5′ untranslated and 53 nt of 3′ untranslated sequence and Not I sites and was cloned into the Not I site of RS8391-BN to generate LTR-FX. The hAAT promoter (−347 to +56) was cloned into the Bgl II site of LTR-FX as described17 to create hFX-514. hAAT-540B contains the 600-bp apolipoprotein E enhancer, the 402-bp hAAT promoter, the 1.3-kb hAAT cDNA, and the internal ribosome entry site (IRES)-mutant dihydrofolate reductase (*DHFR) cassette. It was previously designated ApoE(−)hAAT-LTR.18 The nomenclature was changed to avoid confusion between the hAAT promoter and the hAAT cDNA.

Generation of the retroviral packaging cell lines.The amphotropic GP+envAM12,21 NIH 3T3 murine fibroblast,22 and Hepa1A²³ cells were maintained as described previously17 in medium containing 10 μg/mL vitamin K (Sigma Chemical, St Louis, MO). CaPO₂-mediated transfection and selection of GP+envAM12 cells with 250 nmol/L methotrexate (Sigma Chemical) were performed as described previously.16 To screen colonies, 1 mL of conditioned medium was used to infect NIH 3T3 cells. The supernatant from infected NIH 3T3 cells was tested for hFX by enzyme-linked immunosorbent assay (ELISA) after a 4-day collection.

In vivo transduction of rat hepatocytes.Adult male Lewis rats (Sprague Dawley, Indianapolis, IN) weighing ∼200 g received standard institutional care. Twelve milliliters of conditioned medium was collected from each of 24 confluent 15-cm plates of packaging cells grown at 32°C²⁴ and concentrated by ultracentrifugation.16 The pellets were suspended in medium with 8 μg/mL polybrene and injected into the portal vein of a rat that received a 70% hepatectomy 24 hours previously.18 Retroorbital blood was anticoagulated with a 1/10 vol of 3.8% trisodium citrate.

FX ELISA.Mouse monoclonal antibodies no. 1 and 1066 were obtained from Dr J. Miletich (Washington University, St Louis, MO). ELISA25 was performed using antibody no. 1 as the first antibody and horseradish peroxidase (HRP)-coupled antibody no. 1066 as the second antibody. Standards were created by diluting a pool of normal human plasma (George King Biomedicals, Overland Park, KS) that was assumed to have an hFX concentration of 8 μg/mL.1 All samples obtained at 5 months or earlier were analyzed in the same ELISA. Samples obtained at later time points were analyzed in a separate assay using different aliquots from the same set of standards.

ELISA for anti-hFX or anti-hAAT antibodies.ELISA plates were coated with 100 μL containing 1 μg/mL of hFX or hAAT and then blocked with 100 mmol/L NaCl, 50 mmol/L Tris[hydroxymethyl]amino-methane-HCl (pH 7.5) with 5% dry milk (Schnuck's Grocery, St Louis, MO) (TBS-milk). Plasma samples were diluted at ≥1:100 in TBS-milk and incubated on the plates for 2 hours at 25°C and washed with TBS-0.05% Tween 20. An affinity-purified goat antirat IgG antibody (Organon Teknika Corp, Durham, NC) coupled to HRP was added at a 1:1,000 dilution, the washed plate was developed with 3,3′,5,5′-tetramethylbenzidine dihydrochloride, and the optical density (OD) was read at 450 nm.25 Samples were considered positive if the OD was at least twice as high as that observed in the 1:100 dilution of the pretreatment sample for that rat.

Immunoprecipitation/chromogenic substrate assay.Fifty microliters containing 250 μg of 0.87-μm diameter polystyrene beads (IDC Spheres, Portland, OR) coupled to 5 μg of antibody no. 1 was incubated with 5 or 20 μL of rat plasma or 2.5 to 40 μL of human plasma for 90 minutes at 25°C. The beads were washed twice with TBS and resuspended in 100 μL of HBS (10 mmol/L HEPES, pH 7.4, 100 mmol/L NaCl, 1 mg/mL polyethylene glycol [8,000 gm/mol], and 1 mg/mL bovine serum albumin) containing 5 mmol/L CaCl₂ and 50 nmol/L Russsell's viper venom (Sigma Chemical). After 30 minutes at 37°C, 100 μL of HBS with 5 mmol/L CaCl₂ and 200 μmol/L spectrozyme Xa (American Diagnostica, Greenwich, CT) was added. After 15 minutes at 25°C, 40 μL of 1 mol/L acetic acid was added, and the samples were centrifuged for 1 minute at 12,000 rpm. The OD of the supernatant was read at 405 nm. The activity in rat plasma was determined by comparison with the human plasma standard curve. Blanks did not receive any plasma and were subtracted from all samples.

DNA and RNA analysis.DNA and RNA were isolated and analyzed as described previously.16 A small liver biopsy sample was homogenized in a guanididium-HCl solution. Half of the sample was used to prepare DNA by proteinase K digestion, and the other half was used to prepare RNA by acid phenol extraction. For DNA analysis, a 30-cycle multiplex polymerase chain reaction (PCR) technique was employed, using one primer set that amplified proviral (IRES) DNA and one primer set that amplified rat genomic (liver fatty acid binding protein [LFABP]) DNA, as described previously.16 Amplified DNA was electrophoresed through a 2% agarose/1× Tris-ammonium acetate gel, transferred to an Optitran nitrocellulose membrane, and hybridized with a 217-bp HindIII/Kpn I fragment of the IRES labeled by random primer extension. After quantitation, the membrane was stripped and reprobed with a radiolabeled 620-bp LFABP probe. Radioactivity was quantitated on a Betascope 630 two-dimensional Beta counter (Betagen, Waltham, MA).

Some preparations of RNA were hybrid selected with an oligo-(dT)-sepharose column (Collaborative Research Inc, Bedford, MA). RNA was electrophoresed on a 1% agarose gel in the presence of formaldehyde and transferred to an Optitran nitrocellulose membrane (Schleicher and Schuell, Keene, NH). Membranes were hybridized with the 1.5-kb hFX cDNA probe labeled to a specific activity of 3 × 10⁹ cpm/μg DNA by random primer extension. For both Southern and Northern blots, the final wash was performed at 60°C in 0.1× SSC.

Production of a retroviral packaging cell line.The retroviral vector designated as hFX-514 is shown in Fig 1. hFX-514 was used to create an amphotropic packaging cell line with a titer of 7 × 10⁵ ± 1.8 colony-forming units (cfu)/mL on NIH 3T3 cells. Pools of NIH 3T3 and Hepa1A cells with one copy of the retrovirus per cell produced 246 and 276 ng of hFX/10⁶ cells/24 hours, respectively. More than 90% of the transcripts from packaging cells, transduced NIH 3T3 cells, or transduced Hepa1A cells were 5.1 kb and thus initiated from the LTR promoter (data not shown).

In vivo expression of functional hFX.hFX-514 was delivered to regenerating rat hepatocytes in vivo. As controls, some rats were transduced with hAAT-540B, which contains the hAAT cDNA. Plasma obtained at various times after transduction was tested for hFX or hAAT by immunoassay. hFX levels in hFX-514–transduced rats increased gradually over the first several weeks after transduction for reasons that are unclear. At 4 weeks or later after transduction, all hFX-transduced rats had hFX levels of 0.8 to 3.5 μg/mL, whereas hAAT levels were zero, as expected (Fig 2A). Two hFX-514–transduced rats have exhibited stable expression of hFX for 10 months to date. Two other rats maintained hFX expression for 6 weeks, but died during liver biopsy at that time. Because there was no evidence of bleeding or thrombosis at autopsy and we have previously observed a mortality rate of ∼25% at liver biopsy after transduction with other retroviral vectors, it is unlikely that the deaths were related to the fact that these animals exhibited the highest level of hFX expression. hAAT-540B–transduced rats had no detectable hFX, but exhibited stable expression of hAAT at 0.2 to 0.6 μg/mL (Fig 2B). hAAT levels reached their peak at 1 week after transduction, which is consistent with previous studies.

The hFX levels achieved in the hFX-514–transduced rats would be sufficient to correct the clinical manifestations of hFX deficiency if it were functional. However, the presence of rat FX in normal rats complicates a functional assay using rat plasma. We therefore immunoprecipitated hFX with a human specific antibody before performing an assay measuring proteolysis of an FX chromogenic substrate. The hFX-514–transduced rat plasma had 10% to 54% as much FX activity as did normal human plasma (Table 1). The hAAT-540B–transduced samples had no detectable FX activity, demonstrating the specificity of the assay for human FX. The functional activity correlated well with antigen levels for individual rats, suggesting that the majority of the hFX in hFX-514–transduced rats was functional.

Test for anti-hFX antibodies.Rat plasma was tested for anti-hFX antibodies. Initial experiments showed that all pretransduction plasma samples had a low activity in the anti-hFX antibody assay when tested at a 1:100 dilution, whereas most samples had detectable activity above background when assayed at a 1:50 dilution. We therefore tested all posttransduction samples at a 1:100 dilution and performed additional serial dilutions if the initial value was positive. Two hFX-514–transduced rats (rats no. 115 and 119) had no detectable anti-hFX antibodies for 6 weeks or 10 months after transduction, as shown in Fig 2C. Rat no. 114 had anti-hFX antibodies that appeared at 22 days after transduction and were present at a titer of ≤1:400 at the time of his demise at 42 days. Rat no. 113 had a low-titer antibody that appeared at day 36 and fluctuated slightly over time, but never exceeded 1:400. The antibody had no apparent effect on the serum level of hFX for rat no. 113, which remained stable. At all time points for up to 10 months after transduction, both of the hAAT-540B–transduced rats were negative or were positive at a dilution of 1:100 in the anti-hFX antibody assay, as shown in Fig 2D. We conclude that significant titers of anti-hFX antibodies did not develop in the hFX-514–transduced rats. Neither hFX-h514– nor hAAT-540B–transduced rats had any detectable antibody against hAAT protein at any time point when plasma was tested at a 1:100 dilution (data not shown).

DNA and RNA analysis of livers from transduced rats.The expression of hFX in plasma depends on both the copy number and expression per copy of the retroviral vector. PCR-based analysis of rat liver DNA (Fig 3A) showed that 0.3% to 1.5% of cells from hFX-514–transduced rats contained the retrovirus. This value is slightly lower than the fraction of cells that were transduced in previous studies after in vivo delivery of a similar number of retroviral particles.16-18 The expression of hFX was normalized for the transduction efficiency, as shown in Table 1. hFX-514–transduced rats expressed an average of 2.13 μg of hFX/mL/1% transduction efficiency.

To determine which retroviral promoter was used in vivo, RNA was analyzed by Northern blot using a human FX probe (Fig 3B). Normal human liver contained a 2.2-kb band that is similar to the reported size of the human FX mRNA.26 hFX-514–transduced NIH 3T3 have a major band at 5.1 kb that represents LTR-initiated transcripts and a minor band at ∼2.2 kb that probably represents a splicing event between the 5′ splice site of the retroviral vector and a 3′ splice site in the hFX coding sequence. The smaller RNA does not represent endogenous FX, because the same band was present when RNA from hFX-514–transduced NIH 3T3 cells were probed with an IRES-specific probe and was absent from RNA from nontransduced NIH 3T3 cells that were probed with an hFX-specific probe (data not shown).

All rat samples had a 2.2-kb RNA that probably represents the endogenous rat FX mRNA and can be used to estimate the loading efficiency for each sample. Samples from hFX-514–transduced rats had additional bands that were absent in control rats. The 5.1-kb band that comigrates with the major band in hFX-514–transduced NIH 3T3 cells represents LTR-initiated transcripts. The 3.7-kb band is the expected size of the hAAT promoter-initiated transcript. Because hAAT-initiated transcripts were more abundant than the LTR-initiated transcripts, the hAAT promoter is probably the major promoter used in vivo. Retroviral mRNA levels for individual rats did not correlate perfectly with the protein levels that were observed on the day of liver biopsy, although they did correlate well with the calculated DNA copy number for each rat. This discrepancy may be due to the fact that retroviral vector transduction is not uniform throughout the liver (Y.Z. Lin and K.P.P., unpublished data), and a relatively small sample of the liver was obtained for isolation of RNA and DNA.

This study shows that 0.8 to 2 μg/mL of functional hFX (10% to 25% of levels contained in human plasma) can be achieved for more than 10 months in rat plasma by using retroviral vector-mediated hepatic gene therapy. These levels would correct most or all of the bleeding manifestations in human patients with hFX deficiency, because patients with greater than 10% of normal activity have few hemorrhages.1 We believe that the hAAT promoter is responsible for the high-level expression observed, because we previously showed that it is much stronger than several other promoters from a retroviral vector in vivo. hFX expression will probably be maintained, because stable expression of hAAT protein has been observed in rats for 2 years after transduction with similar retroviral vectors.5 Although the number of rats analyzed in this study was small, four additional Lewis rats have exhibited stable expression of hFX for up to 7 months after transduction.

Previous studies attempting gene therapy have failed to achieve stable expression of therapeutic levels of coagulation factors. In vivo transduction of hepatocytes of FIX-deficient beagles with a canine FIX-expressing retroviral vector resulted in stable FIX levels of ∼5 ng/mL (0.1% of normal). The low expression was probably partially due to the use of the LTR promoter, which is weak in hepatocytes in vivo,16 although a low transduction efficiency or differences in expression of FIX as compared with FX may have played a role. Similarly, ex vivo transduction of primary myoblasts with a retroviral vector containing a muscle-specific enhancer upstream of the CMV¹⁴ or β-actin15 promoter resulted in stable but subtherapeutic FIX levels of ∼10 ng/mL. In contrast, adenoviral vectors have resulted in 0.3 to 1 μg/mL of FVIII (150% to 500% of normal)6,7 or 0.35 to 100 μg/mL of FIX (7% to 2,000% of normal)8-11 in plasma of animals, whereas plasmid DNA resulted in up to 35% of normal FIX levels.12 Although therapeutic, expression has been transient due to the immunologic rejection of adenoviral-transduced cells and the inability of plasmid vectors to integrate into the chromosome. Immunosuppressant therapy slowed but did not prevent the inexorable loss of expression from adenoviral vectors in most animals.8 27 

Absence of significant levels of anti-hFX antibodies in Lewis rats.Patients with hemophilia could develop antibodies directed against a coagulation factor protein that is expressed de novo in their blood. This might limit the efficacy of the gene therapy procedure or lead to immune-complex disease. In this study, two hFX-514–transduced Lewis rats had no detectable anti-hFX antibodies. Two other Lewis rats had low-titer antibodies that did not increase over time and had no effect on the serum level of hFX. We conclude that anti-hFX antibody production was not a significant problem in the Lewis rats. In contrast, all of the 10 Sprague-Dawley rats that received the same retroviral vector developed anti-hFX IgG antibodies whose titer was ≥1:6,400 in the assay described here (data not shown). Furthermore, although the expression of hFX was high at 1 to 4 μg/mL at 1 week after transduction, it decreased to undetectable levels by 1 month after transduction. Intriguingly, retroviral vector DNA was absent from the Sprague-Dawley rats at 2 months after transduction, suggesting that retroviral-transduced cells were rejected by an as yet undefined immunologic mechanism. This shows that the genetic background can affect the immunologic response to novel proteins. Further experiments are in progress to study the development of anti-hFX antibodies in Sprague-Dawley rats.

Implications for human gene therapy.Retroviral vector-mediated hepatic gene therapy has been limited in humans because of subtherapeutic protein expression and the surgical risks associated with the genetic modification of hepatocytes. The first problem, that of low-level expression, has been overcome in the current study by using a strong liver-specific promoter. The second major obstacle is that in vivo and ex vivo approaches require a major surgical procedure to induce hepatocyte replication or to harvest hepatocytes, respectively.5 

Although partial hepatectomy effectively facilitated retroviral vector transduction in this study, implementation of retroviral vector-mediated gene therapy for the treatment of hemophilia will require a safer method for retroviral vector transfer to hepatocytes. A major focus of this laboratory is to identify safer methods for the in vivo delivery of retroviral vectors to the liver. We have recently determined that portal branch occlusion is nearly as effective as partial hepatectomy at facilitating retroviral vector transduction in vivo and has a lower mortality.28 Portal branch occlusion could be performed in larger animals by minimally invasive techniques and has been used safely to treat human patients with liver cancer. Portal branch occlusion might ultimately be used in humans to facilitate retroviral vector transduction in vivo for the treatment of genetic diseases. Alternatively, it was recently reported that an adenoviral vector that encodes a toxic protein could facilitate retroviral vector transduction of the liver.29 We are currently exploring the possibility that therapeutic levels of hFX can be achieved after induction of hepatocyte replication by one of these nonsurgical methods without significant toxicity. If this approach is successful and safe in large animals, in vivo retroviral vector-mediated gene therapy would be a very attractive method for permanently correcting the clinical manifestations of patients with severe hemophilia due to hFX deficiency. Methods that are successful for hFX deficiency should be efficacious for hemophilia B, because FIX and FX are highly homologous proteins.

We thank Drs J. Miletich and A. Rudolph for antibodies and assistance in FX assays, Dr E. Davie for the hFX cDNA, and Dr P. Majerus for helpful comments on the manuscript.