Sustained correction of disease in naive and AAV2-pretreated hemophilia B dogs: AAV2/8-mediated, liver-directed gene therapy

Hemophilia B, an X-linked severe bleeding disorder caused by the deficiency of blood clotting factor IX (FIX), is one of the most extensively studied disease models for gene therapy. Long-term correction of the bleeding disorder has been achieved in the murine and dog models of hemophilia B using adeno-associated viral (AAV) vectors.¹  There have also been 2 clinical trials carried out using vectors based on AAV serotype 2.² 

Muscle and liver are the 2 main targets for AAV-mediated gene therapy for hemophilia B. Liver-targeted delivery seems attractive for administration of AAV2-FIX vectors, because of the higher levels of FIX achieved due to more efficient gene transfer and reduced immune response to the transgene.^3,4  Several groups have shown that hepatic delivery of AAV2-FIX vectors results in long-term high levels of FIX in normal and hemophilic mice, and therapeutic levels in hemophilia B dogs and primate models.^5-10  Inhibitor formation against FIX by this route of administration in mice is very rare, although it could happen in some strains of mice, and it correlated with low FIX expression.¹¹  Among the hemophilia B dogs treated with AAV vectors via the liver approach, only one dog with the null mutation developed anti–canine FIX (cFIX) antibodies that resulted in transient FIX expression.⁹  A phase 1/2 clinical trial for liver-directed AAV2-mediated gene transfer for severe hemophilia B has shown that vectors were well tolerated.¹²  Patients treated with low vector doses showed no vector-related toxicity but also failed to achieve FIX levels above baseline. Transient FIX expression at the range of 5% to 12% was detected in a patient who received 5 × 10¹² vg/kg, but after a transient elevation of liver transaminases, FIX levels decreased to less than 1%. A second patient at the same cohort with higher pre-existing neutralizing antibody (NAB) titer against AAV2 (1:17) had only a transient FIX expression at 1% to 3% for 2 weeks. These data suggest that (1) the high prevalence of pre-existing immunity to AAV2 in the human population could interfere with AAV transduction and would preclude efficient in vivo gene delivery in clinical trials; and (2) in vivo transduction efficiency of AAV2 is still relatively low. At low dose, there is no toxicity, but also no gene transfer. At high dose, there is gene transfer, however, hepatotoxicity was observed. Thus, an improved vector with higher efficiency and less pre-existing immunity in humans would be highly desirable.

Alternative serotypes of AAV could circumvent these drawbacks. We have recently identified an expanding family of AAVs from human and nonhuman primate tissues of which at least 3 are serologically different from serotypes 1 to 6; these are called AAV7, 8, and 9.¹³  Pseudotyping strategies have been developed to cross package vector with AAV2 inverted terminal repeats (ITRs) with capsids from other serotypes.¹⁴  Increased in vivo gene transfer efficiency with vectors of other AAV serotypes has been reported.^14-19  Our initial evaluation of these vectors has shown that AAV8 is especially efficient in transduction of mouse liver cells (1-2 logs more efficient than other AAV serotypes).¹⁹  Preimmunization with other AAV serotypes in C57BL/6 mice did not block further transduction by AAV2/8 vectors. Neutralizing antibody screen against AAV8 detected only low titer (1:20) in 3 of 52 healthy human subjects, whereas substantially higher titers were observed in up to 20% of human subjects for AAV2.¹⁹  Thus, AAV8 appears to be an attractive candidate for liver-directed hemophilia gene therapy. Preclinical evaluation of pseudotyped AAV vectors for hemophilia A gene therapy in mice has shown 100% correction of plasma FVIII activity by AAV2 vectors pseudotyped with AAV8 capsid (AAV2/8 vectors).²⁰ 

In this study, we demonstrate the long-term efficacy and safety of AAV2/8 vector both in naive and in AAV2-pretreated hemophilia B dogs. Stable expression of 10% and 26% of normal levels of cFIX has been achieved in the 2 naive hemophilia B dogs for more than a year, and 15% of normal level for 2 years in the dog that was previously treated with an AAV2 vector. No significant liver toxicity was observed. Finally, clotting functions have also been greatly improved, and no bleeding episode has occurred for more than a year since AAV2/8 vector treatment.

The AAV vector plasmid pAAV-LSP-cFIX-WPRE was described previously.⁸  All AAV vectors used in this study were made by the Vector Core of the University of Pennsylvania. AAV vectors were produced by 3 plasmid cotransfection methods as described by Xiao et al,²¹  with modifications. A pseudotyping strategy was used to produce AAV vectors packaged with AAV5 and AAV8 capsid proteins by using AAV trans plasmid containing AAV2 rep and capsid from AAV5 (packH) or AAV8 (p5E18-VD2/8).^17,19  A total of 50 15-cm plates of semiconfluent 293 cells were cotransfected with 650 μg AAV vector plasmid, 650 μg AAV trans plasmid, and 1300 μg Ad helper plasmid pAdΔF6 by standard calcium phosphate method. Cells were harvested 3 days after transfection. AAV vectors used in this study were purified by 3 rounds of cesium chloride gradient centrifugation, buffer-exchanged with PBS, and concentrated using Amicon Ultra 15 centrifugal filter devices-100K (Millipore, Bedford, MA).

All AAV vectors were subjected to 3 routine quality control assays including genome copy (GC) titration by real-time polymerase chain reaction using primer/probe set corresponding to the polyA region of the vector and linearized plasmid standard; infectious center assay (ICA)²²  on B50 cells²³ ; and endotoxin assay using QCL-1000 Chromogenic LAL Test Kit (Cambrex Bio Science, Walkersville, MD).

Hemophilia B dogs used in this study were produced at the Francis Owen Blood Research Laboratory at the University of North Carolina, Chapel Hill. All animals were treated according to the standards set in the Guide for the Care and Use of Laboratory Animals.²⁴  All procedures were in accordance with institutional guidelines under approved protocols at the University of North Carolina. All dogs were placed under isoflurane (2%-5%) anesthesia. Under sterile conditions, a midline laparotomy was performed. A balloon-tipped catheter was advanced into the portal vein under direct vision. The recombinant AAV (rAAV)–cFIX vector diluted with sterile phosphate-buffered saline was then infused directly into the portal vein with the inflated balloon. The infusion generally took from 30 to 60 minutes, after which the catheter was removed. Canine plasma was used as a source of FIX before and briefly after surgery to prevent hemorrhage.

The whole blood clotting time (WBCT) and the activated partial thromboplastin time (aPTT) were performed as previously described.²⁵  Canine FIX antigen levels in dog plasma were determined by enzyme-linked immunosorbent assay (ELISA) as described with modifications.²⁶  Polyclonal sheep anti–canine FIX antibody (Enzyme Research Laboratories, South Bend, IN) was used as capture antibody (1:1000 dilution), rabbit anti–canine FIX antibody (Enzyme Research Laboratories) was used as secondary antibody (1:1000 dilution), and goat anti–rabbit immunoglobulin G labeled with horseradish peroxidase at a dilution of 1:2000 was used for detection (Santa Cruz Biotechnology, Santa Cruz, CA). Dog serum samples were also analyzed by immunocapture assay based on an ELISA technique for the presence of anti-cFIX antibodies as described.²⁷  To test if the expressed cFIX was fully gamma carboxylated, 10% BaSO₄ treatment was performed as described on samples from 2 different time points for each dog as well as the normal dog plasma (Sigma, St Louis, MO), which served as a control.²⁸  Both treated and untreated samples were assayed at the same time for cFIX levels by ELISA.

Canine blood chemistries (gamma-glutamyltransferase [GGT], aspartate aminotransferase [AST], alanine aminotransferase [ALT], alkaline phosphatase, and total bilirubin) were analyzed in an automated clinical laboratory.

AAV2, 2/5, and 2/8 neutralizing antibody titers in canine serums were assayed by the ability of serums to inhibit transduction of 84-31 cells by reporter viruses (AAVCMVEGFP) of the respective serotypes as described.¹⁹  Specifically, the reporter virus AAVCMVEGFP of each serotype (at multiplicity of infection equal to 10⁴ genome copies/cell) was preincubated with 2-fold serially diluted heat-inactivated serum from dogs collected at different time points. After 1 hour of incubation at 37°C, viruses were added to 84-31 cells in 96-well plates for 24 hours (for AAV2 and 2/8 vectors) or 48 hours (for AAV2/5 vector) depending on the virus serotype. The number of green fluorescent protein–expressing cells was assessed by fluorescent microscopy. Neutralizing antibody titers were reported as the highest serum dilution that inhibited transduction by 50% of that seen with serum from a naive animal. The lowest dilution was 1:20.

Data are presented as a mean ± standard deviation. Statistical analysis of aPTT data was performed using the SigmaStat 3.1 program (SPSS, Chicago, IL). Statistical significance was set at P ≤ .001 and a statistical power more than .80 was required. For each dog, 10 time points after vector treatment or vector readministration were randomly selected and analyzed by one-way repeated measures analysis of variance (ANOVA) using the Holm-Sidak test to identify differences between pretreatment and posttreatment aPTT values.

At the dose of 5.25 × 10¹² GC/kg, 2 naive hemophilia B dogs received a single intraportal infusion of AAV2/8 LSP-cFIX-W. G43, a 7.6-month-old male dog with a weight of 16.5 kg, received a total of 8.7 × 10¹³ GC AAV2/8 vector. H12, a 4-month-old female dog with a weight of 8.8 kg, received a total of 4.6 × 10¹³ GC AAV2/8.

For both dogs, whole blood clotting times (WBCTs) quickly decreased from more than 60 minutes before injection to close to normal range (8-12 minutes) after injection (Figure 1A). The reduction of WBCT for the first 3 weeks could partly result from the infusion of normal dog plasma in the first 3 to 4 days after surgery; but after 3 weeks, the reduction of WBCT is likely due to the gene transfer. In G43, the average WBCT remained stable at 13.5 ± 2.1 minutes from week 3 to 15.7 months; and in H12, the average WBCT was 14.5 ± 1.9 minutes from week 3 to 14.3 months (Figure 1A). Activated partial thromboplastin time (aPTT), a common clinical parameter, was also shortened significantly from 101.9 seconds before injection to average 48.0 ± 6.5 seconds in G43 and from 98.2 seconds to 55.6 ± 6.3 seconds in H12 (P < .001, ANOVA). These values have persisted for the first year after treatment (Figure 1B). aPTT for normal dogs is 24 to 32 seconds. Surprisingly, the cFIX antigen levels reached high peaks within 4 to 5 days after vector administration in both dogs, which have not been reported in the literature on AAV serotype 2–treated hemophilia B dogs. At day 4, cFIX levels in G43 peaked at 1.25 μg/mL and were maintained through day 9, then levels gradually decreased to 466 ng/mL at day 17, followed by a gradual increase to 1.58 μg/mL at day 77; levels stabilized to approximately 1.3 ± 0.2 μg/mL through the duration of the experiment (15.7 months). This level is about 26% of the normal level of cFIX, 5-fold of the therapeutic levels of FIX (Figure 1C). In H12, a higher peak of cFIX expression at 3.8 μg/mL was achieved at day 5; then it gradually decreased to 210 ng/mL at day 21. The cFIX expression levels then slowly increased to 458 ng/mL at day 64 and were subsequently maintained at 468 ± 110 ng/mL for 14.3 months. This level is almost twice the therapeutic level of FIX for hemophilia B. In addition, no spontaneous bleeding episodes have occurred in either dog for more than one year since the treatment with AAV2/8 vectors. The hemophilia B dogs from the University of North Carolina (UNC) Chapel Hill colony have on average 6 bleeds per year that require treatment with normal canine plasma.²⁵ 

As one of the gamma-carboxylated plasma proteins, plasma factor IX can efficiently bind to barium sulfate through the multiple gamma-carboxylated glutamic acid residues localized in the amino-terminal region. This unique property has been used for purification of various gamma-carboxylated proteins or to eliminate the interference in activity by endogenous animal factor IX in the serum added to the culture media in tissue culture experiments.²⁸  To test if the expressed cFIX was fully carboxylated, barium sulfate precipitation was performed on plasma samples from 2 time points for each dog. In the day-144 and day-529 samples of G43, 43.7% and 25.2%, respectively, were retained in the supernatant after barium sulfate treatment. And in the day-148 and day-485 samples of H12, 42.3% and 58.6%, respectively, were retained in the supernatant. In the same assay, 37.7% of cFIX was retained in the normal dog plasma (Sigma) after barium sulfate treatment. Thus, except for in the day-529 sample of G43, cFIX in the AAV2/8-treated dogs was less than fully gamma carboxylated.

The ability to reinject AAV to hemophilia B dog liver was studied with a dog that received AAV2/8-cFIX 2.7 years after the initial administration of AAV2-cFIX-W. D39, a female hemophilia B dog that was previously treated with 2.8 × 10¹² GC/kg of AAV2 LSP-cFIX-W via intraportal injection at the age of 2.8 months, has sustained expression of low levels of cFIX at 34.2 ± 9.8 ng/mL since vector injection.⁸  Although both WBCT and aPTT values were shortened after the first treatment, a total of 6 spontaneous bleeding episodes that require plasma infusion have occurred during the 2.7 years since the first AAV2 vector treatment. At day 995 after the first injection, D39 received a second intraportal injection with a total of 1.5 × 10¹⁴ GC AAV2/8 LSP-cFIX-W vector (vector dose = 9.28 × 10¹² GC/kg). WBCT decreased quickly from 19 minutes before injection to 8 minutes at day 6, then increased to 14.5 minutes at day 14 and remained stable at an average of 12.7 ± 1.6 minutes for more than 2 years since the second injection (Figure 2A). aPTT was corrected to normal levels transiently, from 57.2 seconds before injection to 28.9 seconds at day 6, and gradually increased and stabilized at an average of 48.6 ± 5.1 seconds for more than 2 years after the readministration (Figure 2B). The reduction of aPTT value after readministration of AAV2/8 vector was significant (P ≤ .001, ANOVA). The pattern of cFIX expression correlated well with FIX functional assays. cFIX antigen rose quickly to 9.5 μg/mL at day 3, and 10.4 μg/mL at day 6, twice the normal level of cFIX in dogs. The cFIX level decreased after day 6 to 1.4 μg/mL at day 14 and stabilized at an average of 785 ng/mL (16% of normal level) for more than 2 years (Figure 2C).

AAV serotype 5 has been shown to have improved hepatic gene transfer efficiency in mice.¹⁸  To test if it is also the case in dogs and to compare the gene transfer efficiency of AAV2/8 and AAV2/5, we evaluated AAV2/5 LSP-cFIX-W in an AAV2-pretreated hemophilia B dog. C52, a male hemophilia B dog, received an intraportal injection of a low dose of AAV2 LSP-cFIX vector (2.8 × 10¹¹ GC/kg) at the age of 6.5 months and expressed only very low levels of cFIX, at 3.6 ng/mL.⁸  WBCT and aPTT have been maintained at an average of 16.3 ± 3.4 minutes and 81.7 ± 17.8 seconds, respectively, and 3 bleeding episodes have occurred over the 3.2 years since injection. At day 1180 after the first injection, C52 received a second intraportal injection with AAV2/5 LSP-cFIX-W vector at a dose of 2.3 × 10¹³ GC/kg. WBCT decreased quickly from 21 minutes before injection to 11.5 minutes at day 4, and remained stable at an average of 13.1 ± 1.7 minutes for 2½ years since the second injection (Figure 2D). aPTT was corrected to normal levels transiently, from 66.5 seconds before injection to 29.3 to 32 seconds between day 7 and day 10, then gradually increased and sustained at an average of 47.6 ± 6.3 seconds for 2½ years after the readministration (Figure 2E). The reduction of aPTT value after readministration of AAV2/5 vector was significant (P ≤ .001, ANOVA). An early peak of cFIX antigen level was reached at day 7 (4.6 μg/mL), close to the normal level of cFIX in dogs. The cFIX level decreased after day 7 to 1.7 μg/mL at day 13 and stabilized at an average of 799 ng/mL (16% of normal level) for 2½ years (Figure 2C).

Barium sulfate precipitation was also performed on plasma samples from 2 time points for each dog to test if the expressed cFIX was fully carboxylated. In the day-156 and day-849 samples of D39 following AAV2/8 vector administration, 58.3% and 51.3%, respectively, were retained in the supernatant after barium sulfate treatment. And in the day-139 and day-944 samples of C52, 54.5% and 54.8%, respectively, were retained in the supernatant. In the same assay, 37.7% of cFIX was retained in the normal dog plasma (Sigma) after barium sulfate treatment. These results suggested that cFIX in the AAV2/8- and AAV2/5-treated dogs was less than fully gamma carboxylated.

Serum chemistry panels including liver enzyme levels were closely monitored following vector infusion. In G43 and H12, the 2 naive hemophilia B dogs injected with 5.25 × 10¹² GC/kg AAV2/8 vectors, alanine aminotransferase (ALT), aspartate aminotransferase (AST), and total bilirubin remained within the normal range for more than a year after vector infusion (Figure 3A-B; and data not shown). D39 had elevated AST (2.5 times upper level of normal) at day 2 and day 3 following infusion with 9.28 × 10¹² GC/kg AAV2/8 vector, and its ALT was slightly more than the normal range at day 3 (121 U/L) (Figure 3C-D). C52 had elevated ALT (within 1.5 times upper level of normal) during the first 2 to 4 days after readministration with 2.3 × 10¹³ GC/kg AAV2/5 vector, and its AST was slightly more than the normal range at day 2 (88 U/L) (Figure 3E-F). None of the 4 dogs experienced surgical complications and apparently tolerated vector infusion well with transient hypotension that responded to intravenous fluids. No antibodies against cFIX as assayed by ELISA were detected in any of the 4 dogs (data not shown).

The 2 naive hemophilia B dogs that received 5.25 × 10¹² GC/kg AAV2/8 vector intraportally had robust humoral response to AAV8 capsid (Figure 4A). In G43, NAB to AAV8 quickly rose from less than 1:20 before injection to a titer of 1:320 at day 4 and peaked on day 8 at a high titer of 1:5120. The peak then gradually dropped 2- to 8-fold after week 6 and persisted through the experiment (one year). The peak level of AAV8-NAB response in H12 was 4- to 8-fold less than it was in G43.

D39 developed a peak NAB response to AAV2 on day 4 after the initial administration with 2.8 × 10¹² GC/kg AAV2-cFIX; the titer decreased about 1-log within a month and was maintained at that level for the first year; then the titer further decreased 2- to 4-fold and persisted through the second year. By the time of the second injection (day 994), the NAB titer to AAV2 was less than 1:20 (Figure 4B). After the readministration with AAV2/8-cFIX, AAV8-NAB titer quickly rose from less than 1:20 before injection to 1:1280 on day 6 and peaked at 1:5120 on day 14. The titer gradually dropped about 1-log during the 2 years following vector readministration. Surprisingly, concurrent with the AAV8-NAB response, D39 displayed a similar pattern of AAV2-NAB response after the readministration of AAV2/8 vector. The peak levels of AAV2-NAB that lasted from day 8 to day 27 after readministration of AAV2/8 vector were 10-fold higher than its initial AAV2-NAB response following the first vector administration with AAV2 vector.

The AAV2-NAB response in C52, which received a low dose of AAV2 vector (2.8 × 10¹¹ GC/kg), was relatively weak and short lived. At 6 weeks after the injection, AAV2-NAB titers were below detection and remained such in the following 3 years. On day 1179, the dog was readministrated with a high dose of AAV2/5-cFIX vector (2.3 × 10¹³ GC/kg). High titer of AAV5-NAB appeared 7 days after the readministration and persisted during the following 2 years (Figure 4C). AAV2-NAB remained undetectable after readministration.

Preclinical studies using AAV2 vectors for liver-directed gene transfer in hemophilia B dogs were encouraging. Snyder et al reported sustained expression of 0.2% to 2% of normal levels of cFIX in the UNC hemophilia B dogs with AAV2-MFG-cFIX at a dose of approximately 2 × 10¹¹ vg/kg,⁷  while Mount et al achieved 5% in the same colony using AAV2-(ApoE)₄/hAAT-cFIX at a dose of 8 × 10¹¹ vg/kg.⁹  Using the same vector by the latter group in the Auburn hemophilia B dogs with a null mutation, 4% and 12% of normal levels of cFIX were generated in 2 dogs injected with vectors at doses of 1.2 × 10¹² and 1.6 × 10¹² vg/kg, respectively, while the third dog had only transient expression and developed neutralizing cFIX inhibitors.⁹  Using a different synthetic liver-specific promoter, we have previously treated a hemophilia B dog (C55) with a single intraportal infusion of AAV2-LSP-cFIX-W at a dose of 4.6 × 10¹² GC/kg; this resulted in sustained expression of therapeutic levels of cFIX⁸  (218 ng/mL, 4% of normal), which have been stable for 5 years, as well as improvements in clotting functions (L.W., I.M.V., T.C.N., unpublished data, November 2004). The other 2 dogs (C52 and D39) that received lower vector doses expressed only less than 1% of normal levels of cFIX (Table 1). In this study, we packaged the same vector construct with AAV8 capsid (AAV2/8) and evaluated its efficacy and safety for hepatic gene transfer in hemophilia B dogs. The reasons we are interested in AAV2/8 vectors for liver-directed hemophilia B gene therapy are (1) AAV2/8 has shown extraordinary liver tropism in mouse models.^19,20,29  (2) There is less pre-existing AAV8 immunity in the human population compared with AAV2.¹⁹  Among the 91 human serum samples we tested for NAB against AAV2 and AAV8, 21 samples (23.1%) had AAV2 NAB titer of 1:20 or higher, while only 8 (8.8%) had AAV8 NAB titer of 1:20 or higher. Among the 21 positive samples for AAV2, 9 (42.9%) had titer of 1:1280 or higher, while only 1 (12.5%) of 8 AAV8-positive samples had NAB titer at 1:1280. (3) The results from the recent phase 1/2 clinical trial on AAV2-mediated, liver-directed gene transfer for hemophilia B indicated that pre-existing antibodies to AAV2 may block transduction when the vector is administrated systemically and affect the outcome of gene transfer.¹²  (4) In this same trial, therapeutic expression of FIX was only short-lived possibly due to pre-existing host immunity to AAV2 as evidenced by transient vector induced hepatitis.

As summarized in Table 1, the 3 hemophilia B dogs, 2 naive dogs (G43 and H12), and 1 AAV2-pretreated dog (D39) all demonstrated long-term expression of functional cFIX at the levels ranging from 10% to 26% of normal levels for more than 1 to 2 years since intraportal infusion of the AAV2/8 vector. The vector doses ranged from 5.25 × 10¹² GC/kg to 9.28 × 10¹² GC/kg, equivalent to or 2-fold higher than the dose we used to treat a hemophilia B dog with AAV2 vector previously (4.6 × 10¹² GC/kg).⁸  The highest expression came from G43, the only male dog among the 3, even though it received a lower vector dose than D39. Whether the sex plays a role in influencing the transduction of canine liver by AAV2/8 remains to be investigated by increasing the sample numbers. Davidoff et al reported 5- to 13-fold higher gene expression in male mice by AAV2- or AAV5-mediated hepatic gene transfer and indicated the role played by androgens.³⁰  After readministration with AAV2/8 vector, the AAV2-pretreated dog D39 showed persistent expression of 16% of normal, suggesting that pre-exposing to AAV2 does not prohibit AAV2/8 gene transfer; although at the time of readministration, the NAB to AAV2 was below the detection level. Figure 5 depicted the cFIX expression levels in the 5 hemophilia B dogs normalized to ng/mL/10¹² GC/kg. Compared with the dog treated with AAV2 vector at a dose of 4.6 × 10¹² GC/kg (C55) in our previous study,⁸  the gene transfer efficiency of AAV2/8 is about 1.5- to 4.3-fold greater than AAV2 in hemophilia B dogs. A study done in hemophilia B mice showed AAV2/8 was about 8-fold more efficient than AAV2 vector for liver-directed gene transfer (L.W. and J.M.W., manuscript in preparation). C52, an AAV2-pretreated dog, has had a stable expression of 16% of normal level since the readministration of a high dose of AAV2/5 vector. This level in C52 was close to the level in D39 which received a 2.5-fold less vector dose of AAV2/8. Normalized to ng/mL/10¹² GC/kg, the cFIX level in C52 was slightly lower than it was in C55, and 2.4- to 7.1-fold less than it was in AAV2/8-treated dogs (Figure 5). Based on data from these dogs, we conclude that AAV2/8 is about 1.5- to 4.3-fold more efficient than AAV2, and 2- to 7-fold more efficient than AAV2/5 for hepatic gene transfer in dogs.

In all 4 dogs, an early peak expression of cFIX was detected between day 4 to day 7 after vector administration, ranging from 1.3 μg/mL (G43, days 4-9), 3.8 μg/mL (H12, day 5), 4.6 μg/mL (C52, day 7), to 10.4 μg/mL (D39, day 6). The normal canine plasma infusions before and after the surgery could contribute to some part of these cFIX peak expressions; however, the majorities are likely derived from the AAV vectors based on the calculation of the amount of plasma infused, the half-life of cFIX, and the duration curve of FIX after intravenous administration.³¹  Rapid uncoating has been recently proposed as a mechanism for efficient liver transduction in mouse liver by AAV8 vectors.³²  Similar mechanisms may also apply to AAV8 or AAV5 transduction in canine liver, since C52, which received AAV2/5 vector, also had the peak expression at day 7. The peak expression could be derived from the annealing of the complementary plus and minus ss genomes, which have been detected in murine hepatocytes³³ ; however, this form of DNA might not be very stable in majority of the hepatocytes, and thus cause a rapid fall in FIX levels. Early peak of transgene expression has also been observed in nonhuman primates following AAV-mediated hepatic gene transfer in our lab (J.M.W. and G. Gao, unpublished data).

In all 4 treated dogs, WBCT has been corrected close to normal ranges, especially in C52; a further reduction of WBCT was observed after the readministration with AAV2/5 vector. aPTT has also been reduced in all 4 dogs, and further reductions of aPTT have been observed in both D39 and C52 after the second injection. Based on the results of the barium sulfate precipitation experiments, cFIX in the circulation generated by AAV vector treatment was less than fully gamma carboxylated. It is possible that some hepatocytes are transduced with high copies of vector, which leads to overexpression of FIX in excess of the capacity of the cell to fully accomplish the extensive posttranscriptional modifications essential for biogenesis of mature FIX. G43 had almost twice the antigen level than D39 assayed by ELISA, yet its average aPTT value was only slightly lower than D39. Nevertheless, none of the 4 dogs has had any spontaneous bleeding event since the injection of AAV2/8 or 2/5 vectors.

The reactivation of anti-AAV2 NAB in D39 following AAV2/8 vector delivery was unexpected. One possibility is that due to the high homology between the capsid protein of AAV2 and AAV8 (83% similarity), there might be some common epitopes between these 2 serotypes. Following readministration of AAV2/8 vector in D39, some AAV2-memory B cells might be reactivated by the input AAV8 capsids. Cross neutralization between AAV2 and AAV8 is also possible, as we have seen in sera from rabbits immunized with AAV2 or AAV2/8 vectors; however, the equivalent high titers to AAV2 and AAV8 in D39 make this explanation less likely since usually the titers for the “crossed” serotypes are significantly reduced (> 16-fold). This observation should be further investigated in more dogs or other animal models. Reactivation of anti-AAV2 NAB was not observed in C52 following AAV2/5 vector administration. Compared with AAV8, AAV5 shares less sequence homology with AAV2 (56% identical)³⁴  and is more distantly related to AAV2 in phylogeny analysis.¹³  Moreover, cross neutralization between AAV2 and AAV5 was not detected in immunized rabbit sera.¹³ 

While these experiments were not formal toxicology studies using vectors made under Good Manufacturing Practices (GMP), they do provide some information regarding safety. There were no apparent clinical sequela and a survey of blood hematology and clinical pathology revealed no vector-related abnormalities other than a transient elevation of transaminases in a few animals at the time of the intraportal injection, which could have been procedure related. Of note is the absence of a syndrome observed in the 2 research subjects administered intrahepatic AAV2-expressing factor IX (ie, delayed hepatitis concordant with loss of transgene expression).¹²  It is interesting that neither naive nor AAV2-pretreated dogs exhibited this kind of syndrome.

In summary, we demonstrated for the first time the improved hepatic gene transfer efficacy of AAV2/8 vector and long-term cure by this vector in hemophilia B dogs, both in naive and AAV2-pretreated animals. The procedure was safe, without significant vector-related toxicities. AAV2/8 vector is a promising gene therapy vector for applications in humans to treat hemophilia B and other genetic diseases.

We appreciate the support from the Vector Core and Bioassay Core of the Gene Therapy Program at the University of Pennsylvania. We thank Dr M. Limberis for assistance with statistical analysis, and the support staff at the Francis Owen Blood Research Laboratory (Chapel Hill, NC) for their excellent care and handling of the animals.