INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Adeno-associated virus (AAV) has been engineered for use as a vector in human gene therapy. Replication-defective versions of AAV devoid of all viral open reading frames can be isolated. AAV vectors are capable of efficient and prolonged transgene expression in a number of tissues including skeletal muscle (2). Clinical applications are being considered for the treatment of primary neuromuscular disorders, such as limb-girdle muscular dystrophy, and diseases where muscle serves as a site for secretion of a therapeutic protein such as factor IX for hemophilia B and erythropoietin (EPO) for anemia (6, 9, 14, 17, 19).

For AAV to be useful in the treatment of chronic diseases, it will be necessary to achieve durable expression for the life of the individual. The experience with AAV vectors in skeletal muscle of both small and large animals has been encouraging, with levels of expression persisting for 1 to 2 years, which is a substantial improvement over previous vector systems. Two aspects of AAV biology contribute to the prolongation of transgene expression. First, it appears that the vector genome replicates as a large concatemer, which either exists stably as a large episome or integrates randomly into the host genome (3). In addition, the vector avoids activation of destructive cytotoxic T-lymphocyte responses to antigenic transgene products, a problem previously observed with naked DNA and adenoviral vectors. This may be explained by the restricted tropism of AAV and the resistance of dendritic cells to inadvertent AAV transduction, which is felt to be important in the initial activation of T cells (10).

The stability of transgene expression obtained with AAV vectors in skeletal muscle is impressively prolonged, although it is not permanent; most studies have shown little diminution in transgene expression (i.e., two- to fivefold) over a 1- to 2-year period. The reason for this shallow but steady decline in expression is unclear. Treatment of a chronic disease with AAV vectors over the life of the patient, therefore, will require readministration of vector. A number of investigators have detected antibodies to AAV capsid proteins following vector administration in preclinical models (8, 11). This study evaluates the nature of the humoral immune response to AAV vector in both mice and nonhuman primates and suggests a strategy to allow efficient and repeated readministration of vector.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Animals and specimen collection. Wild, caught juvenile rhesus monkeys were purchased from Southwest Foundation for Biomedical Research (San Antonio, Tex.) and underwent full quarantine. The monkeys weighed approximately 3 to 4 kg and were serologically negative for simian immunodeficiency virus, simian T-cell leukemia virus, other simian retroviruses, and human adenovirus. Animals were bled for immunological analyses at 7% of the body weight over a 3-week period. C57BL/6, BALB/c, nude BALB/c, RAG1 KO, and CD40LKO mice were purchased from Jackson Laboratories (Bar Harbor, Maine). Mice deficient for the murine CD4 gene and transgenic for human CD4 (HuCD4) were bred in a specific-pathogen-free facility under contract at Charles River (Wilmington, Mass.). HuCD4 mice have previously been described (5). The protocols were approved by the Infection Control Committee of The Hospital of the University of Pennsylvania and the Environmental Health and Safety Office, the Institutional Biosafety Committee, and the Institutional Animal Care and Use Committee of the University of Pennsylvania.

Vectors. AAV type 2 (AAV2) vectors expressing 1 antitrypsin (1AT) were constructed. Under the control of a cytomegalovirus promoter (AAV-1AT), AAV green fluorescent protein (AAV-GFP), rhesus monkey EPO (AAV-EPO), and rhesus monkey growth hormone (AAV-GH) were made by transfection-infection protocols described earlier (7, 16). Briefly, the cis plasmid (with AAV-ITR), the trans plasmid (with the AAV rep gene and cap gene), and a helper plasmid (pF13; with an essential region from the adenovirus genome) were cotransfected into 293 cells in a ratio of 1:1:2 by calcium phosphate precipitation. pF13 has an 8-kb deletion in the adenovirus E2B region and most of the late genes. The cells were harvested 96 h posttransfection and subsequently purified by two rounds of CsCl gradient. For the generation of recombinant AAV (rAAV) based on AAV2, p5E18 was used as the trans plasmid since it greatly improved the rAAV yield. Large-scale production of the virus was according to good manufacturing practices, and virus was obtained from the Human Applications Laboratory of the Institute for Human Gene Therapy.

Antibody. Clenoliximab, a primatized monoclonal antibody (MAb) to human CD4, contains the variable domains from a MAb generated in cynomolgus macaque and human immunoglobulin G4 (IgG4) constant domains. It also contains two single-residue substitutions, the first in the hinge region and the second within the C_H domain, designed to enhance heavy chain dimer formation and reduce Fc receptor binding (A. Truneh and M. Reddy, unpublished data). This MAb has no C1q binding or complement-fixing activity and has a dramatically reduced Fc receptor binding activity. The preparation of purified Clenoliximab used in this study was provided by the Department of Pharmaceutical Technologies, SmithKline Beecham Pharmaceuticals, King of Prussia, Pa. In one experiment, a human IgG1 isotype antibody of irrelevant specificity was used as a control.

LPR assays. Lymphoproliferative (LPR) responses were performed for AAV antigens, using methods optimized for adenovirus antigens described earlier (4, 5). Peripheral blood mononuclear cells (PBMC) harvested from heparinized blood from rhesus monkeys and splenocytes harvested from mice were isolated following Ficoll-Hypaque density gradient centrifugation, washed in phosphate-buffered saline (PBS), and resuspended in RPMI 1640 supplemented with 10% fetal calf serum, penicillin, streptomycin, and 10⁵ M 2-mercaptoethanol. Triplicate cultures of PBMC (100 µl of medium at 10⁶ cells/ml) were cultured with either rAAV (multiplicity of infection in particles = 100) supplemented with 100 ng of staphylococcus enterotoxin B (SEB; Toxin Technologies, Sarasota, Fla.)/ml or medium alone. Proliferation was measured by a standard 16-h [³H]thymidine (1 µCi/well) pulse on a liquid scintillation counter (Wallach, Gaithersburg, Md.). Results are presented as stimulation indexes, which denote the ratio of counts per minute of stimulated cultures to counts per minute of unstimulated cultures.

Cytokine release assays. PBMC from rhesus monkeys or splenocytes from mice were cultured with or without antigen (i.e., rAAV) for 48 h in a 24-well plate. Cell-free supernatants were collected and analyzed for the presence of interleukin-2 (IL-2), IL-4, gamma interferon (IFN-), and IL-10 by commercial enzyme-linked immunosorbent assay (ELISA) kits (BioSource, Camarillo, Calif.) using the manufacturer's protocols. The sensitivities of the kits for human cytokines were as follows: IL-2, 5 pg/ml; IFN-, 4 pg/ml; IL-4, 2 pg/ml; IL-10, 5 pg/ml; those for murine cytokines were as follows: IL-2, 13 pg/ml; IFN-, 1 pg/ml; IL-4, 5 pg/ml; IL-10, 13 pg/ml.

AAV2-specific Igs. Serum samples from mice and rhesus monkeys were analyzed for AAV2-specific, isotype-specific Igs (IgM, IgG1, IgG2, and IgG4 for rhesus monkeys; IgM, IgG1, IgG2a, and IgG3 for mice) by ELISA as described earlier (5). For the ELISA, 96-well flat-bottom, high-binding ELISA plates (Costar, Cambridge, Mass.) were coated with 100 µl of AAV2 vector antigen (5 × 10¹⁰ particles/ml) in PBS overnight at 4°C, washed four times with PBS containing 0.05% Tween, and blocked in PBS containing 1% bovine serum albumin for 1 h at 37°C. Appropriately diluted samples were added to antigen-coated plates and incubated for 4 h at 37°C. Plates were washed four times with PBS containing 0.05% Tween and incubated with peroxidase-conjugated goat anti-human IgM, IgG1, IgG2, or IgG4 (for rhesus monkey samples, the dilution was 1:2,000; Sigma Chemical Co., St. Louis, Mo.) or biotin-conjugated rat anti-mouse IgM, IgG1, IgG2a, IgG3 (for mice, the dilution was 1:1,000; PharMingen, San Diego, Calif.) for 2 h at 37°C. For the mouse ELISA, plates were washed and a 1:20,000 dilution of alkaline phosphatase-conjugated avidin was added. Plates were washed as described above, and ABTS [2,2'-azinobis(3-ethylbenzthiazolinesulfonic acid)] substrate for peroxidase (Kirkegaard and Perry, Gaithersburg, MD) or p-nitrophenyl phosphate in diethanolamine buffer substrate for alkaline phosphatase (Sigma) was added. Optical densities were read at 405 nm on a microplate reader (Dynatech Laboratories, Chantilly, Va.).

Anti-AAV2 NAbs. Neutralizing antibody (NAb) titers were analyzed by determining the ability of serum antibody to inhibit transduction of reporter virus AAV-GFP into E4-expressing 293 cells (84-31 cells) cells as described earlier (7). Various dilutions of antibodies preincubated with the reporter virus for 1 h at 37°C were added to 90% confluent 84-31 cell cultures. The lowest dilution used in the assay was 1:20. Cells were incubated for 16 h, and expression of GFP was analyzed by fluoroimaging (Molecular Dynamics, Sunnyvale, Calif.). The neutralizing titer of antibody was calculated as the highest dilution with which 50% of the cells turned green.

Human 1AT assay. The concentration of human 1AT in mouse serum was measured using ELISA (16). The coating antibody was rabbit anti-human 1AT (Sigma), and goat anti-human 1AT (Sigma) was used as the detection antibody. The sensitivity of this assay was around 0.3 ng/ml.

Study design for HuCD4 mice. Particles (10¹¹) of AAV-GFP vector were administered intramuscularly in a volume of 25 µl to HuCD4 mice. For this purpose, animals were anesthetized and a skin incision was made on the leg. Muscles were exposed, and vector was injected into the tibialis anterior using a 27-gauge needle. The skin incision was then closed using a 4-0 vicryl suture material. Clenoliximab was administered intraperitoneally in a volume of 100 µl as several doses, at 2 mg/dose/mouse, on days 3, 1, 4, 8, 15, 22, and 28 as described earlier (5). AAV-1AT was readministered on day 56 or day 180. Blood was drawn by retro-orbital bleeding on days 56 and 180 for AAV-NAb analyses, on days 70 and 194 for determining 1AT levels, and on day 56 for Ig isotype analyses.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Humoral immune responses to AAV in mice. Figure 1A shows that C57BL/6 mice administered AAV vectors intramuscularly induce NAbs, which peak on day 29 and persist through 360 days. A more detailed characterization of the nature and relevance of AAV-NAbs is presented in Fig. 2. AAV-GFP or PBS was administered to different strains of mice on day 1, followed by readministration of AAV encoding human 1AT on day 28. Analysis of serum for human 1AT levels on day 56 provided a quantitative readout of vector readministration. Figure 2 shows the 1AT levels (on day 56) and AAV-NAb titers (on day 28) in these mice. Immunodeficient RAG1 KO mice, CD40LKO mice, and nude mice failed to generate AAV-NAbs, and AAV-1AT was successfully administered to immunized animals at levels equivalent to that achieved in naive animals (i.e., animals that initially received PBS). AAV-1AT gene transfer was diminished relative to gene transfer in naive mice but was still detectable in BALB/c (4-fold decrease) and C57BL/6 (10-fold decrease) mice when mice were challenged with vector 1 month after immunization with AAV-GFP, at a time when significant NAbs are present in serum. These observations indicate that the AAV-NAb response following intramuscular administration of AAV vectors is T cell dependent and that NAbs substantially inhibit, but do not completely block, vector uptake.

View larger version (22K): [in this window] [in a new window]   FIG. 1.   Development of NAbs to AAV. (A) C57BL/6 mice were injected with 1011 particles of AAV-GFP into the tibialis anterior. Serum samples drawn on various days were analyzed for AAV-NAbs. Each point represents a mean ± standard deviation for five to eight animals per group. diln, dilution. (B) Rhesus monkeys were administered AAV-EPO (animals 93B644 and RQ1582) intramuscularly as described in Materials and Methods. Sera were obtained on various days, and NAbs were measured. The neutralizing titers of the sera were calculated as the highest dilutions with which 50% of the cells turned green. The dashed lines indicate the lowest dilution in the assays (1:20).

View larger version (23K): [in this window] [in a new window]   FIG. 2.   Readministration of AAV in mice. Various strains of mice were administered either PBS or 1011 particles of AAV-GFP into the tibialis anterior of the left leg on day 1. C57BL/6 and BALB/c mice represent immunocompetent mice; BALB/c nude mice and CD40LKO mice have B cells but lack functional T cells; RAG1 KO mice lack both mature T and B cells. On day 28, these mice were readministered AAV-1AT into the tibialis anterior of the right leg. 1AT levels in sera (A) were measured on day 56 (28 days following the second vector administration) by ELISA. NAbs in sera (B) obtained on day 28 were measured. Each bar represents a mean and standard deviation for three to five animals per group. The dashed line indicates the lowest dilution in the assay (1:20).

View larger version (15K): [in this window] [in a new window]   FIG. 3.   AAV vector-specific Ig isotypes in sera following intramuscular administration of adenovirus vectors to mice. Sera obtained from mice administered AAV-GFP were analyzed for the presence of AAV-specific Ig isotypes (IgM, IgG1, IgG2a, and IgG3) by ELISA, as described in Materials and Methods. Results are expressed as optical densities (O.D.). Each bar represents the mean and standard deviation for three to five animals per group.

Humoral immune responses to AAV in nonhuman primates. AAV-mediated gene transfer to rhesus monkey skeletal muscle was studied with a vector that expresses rhesus monkey EPO from a cytomegalovirus promoter (AAV-EPO). This provides a quantitative assessment of transgene engraftment by direct measure of serum EPO by ELISA and the resulting elevation in hematocrit. Intramuscular administration of AAV-encoding EPO to rhesus monkeys resulted in prolonged expression of the transgene, as evidenced by supraphysiological serum EPO concentrations (Fig. 4A) in the setting of a persistently elevated hematocrit (Fig. 4B).

View larger version (33K): [in this window] [in a new window]   FIG. 4.   Serum EPO and hematocrit levels following intramuscular injection of AAV vectors in rhesus monkey muscle. Rhesus monkeys RQ1582 and 93B644 were administered AAV-EPO intramuscularly on day 1. Rhesus monkeys 93B662 and RQ1830 were administered AAV-GH on day 1 and readministered AAV-EPO on day 151. The AAV-NAb titers of these animals are shown in Fig. 1B and 5A. Sera were analyzed for the presence of EPO (A and C) by ELISA, and for hematocrits (B and D) by centrifugation of whole blood. Rhesus monkeys were bled for therapeutic reasons when hematocrits increased over 65%.

View larger version (22K): [in this window] [in a new window]   FIG. 5.   Readministration of AAV2 in nonhuman primates. Rhesus monkeys were administered AAV-GH (animals 93B662 and RQ1830) intramuscularly as described in Materials and Methods on day 1 and readministered AAV-EPO on day 151. Animals RQ1542 and 93B644 represent animals expressing EPO following administration of AAV-EPO to naive animals. Sera were obtained from rhesus monkeys on various days. (A) Serum AAV-NAb levels following administration of AAV-GH (on day 1) and AAV-EPO (on day 151) are shown as reciprocal dilutions (diln) of sera. (B) Serum EPO levels measured 60 days after administration of AAV-EPO to either naive rhesus monkeys or AAV-GH-treated monkeys, as measured by ELISA.

View larger version (24K): [in this window] [in a new window]   FIG. 6.   AAV vector-specific Ig isotypes in sera following intramuscular administration of AAV vectors in nonhuman primates. Sera obtained from rhesus monkeys administered AAV-GH (93B622 and RQ1830) or AAV-EPO (93B644 and RQ1582) intramuscularly were analyzed for the presence of AAV-specific Ig isotypes (IgM, IgG1, IgG2, IgG4) by ELISA, as described in Materials and Methods. Results are expressed as optical densities (O.D.).

Primatized anti-CD4 antibodies inhibit AAV-mediated NAbs and permit vector readministration. In an attempt to prevent the B-cell response to capsid proteins we studied transient inhibition of T-cell function at the time of vector administration. A primatized blocking antibody to human CD4, called Clenoliximab, was tested in a biologically compatible murine model, the HuCD4 mouse (5). The endogenous murine CD4 gene has been functionally replaced in this mouse with a HuCD4 gene. Animals were treated with AAV-GFP in skeletal muscle in the presence or absence of a short course of the CD4 antibody and injected 2 or 6 months later with AAV-1AT (Table 3). Engraftment of the second vector was diminished when the vector was administered on day 56 (threefold reduction) and day 180 (twofold reduction) after immunization with AAV-GFP. Treatment of animals with CD4 antibody at the time of administration of the first vector prevented the formation of NAbs and allowed efficient vector readministration (equivalent to that seen in naive animals) long after the initial doses of CD4 antibody had waned (Table 3). Analyses of Ig secretion patterns, at the time of AAV-1AT administration on day 56, are also presented in Table 3. HuCD4 mice treated with AAV-GFP alone on day 1 (group A) generated IgM, IgG2a, and IgG1 responses. IgG3 responses were not significantly induced. Clenoliximab treatment had no effect on IgM responses, but IgG1 and IgG2a responses were significantly inhibited.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Durable gene expression will be required to effectively treat many chronic diseases such as inherited diseases manifested during childhood. This can be accomplished through the combination of very prolonged gene engraftment in combination with efficient vector readministration when the previous treatment wanes.

Expression of AAV-encoded genes in skeletal muscle persists for at least 1 to 2 years in a variety of animal models including mice, dogs, and nonhuman primates (6, 9, 14, 17, 19). Quantitative analysis of these models has shown modest declines in transgene expression over this period, indicating that vector readministration may be necessary at intervals of several years. Mechanisms responsible for the steady decline in transgene expression have yet to be determined, although a number of hypotheses have been entertained including promoter shutoff, degradation of the vector genome (if it indeed persists as an unintegrated concatemer), and turnover of the transduced muscle fiber (15). The last mechanism is likely if in vivo transduction is restricted to muscle fibers as opposed to the satellite cells, which are progenitors of the muscle fibers.

Our study characterized several features of humoral immunity to AAV in skeletal muscle relevant to the challenge of vector readministration. A vibrant humoral response ensues; it declines over a year, but remains substantial. Experiments with mice genetically deficient in T-cell function indicated that the B-cell response is T cell dependent despite an inability to demonstrate significant activation of CD4 T cells exposed to AAV antigens in vitro. Immunocompetent mice developed a strong NAb response. The differences in the Ig isotypes and NAb titers in C57BL/6 and BALB/c mice suggest that some of the anti-AAV antibody made by BALB/c mice was not neutralizing, although direct comparisons between these two assays are difficult. Isotype analyses of Clenoliximab-treated mice suggest that the T-cell-dependent isotypes, IgG1 and IgG2a, contribute to the NAb response. Readministration of vector is possible in mice and nonhuman primates in the presence of maximal NAb responses, although it is diminished 4- to 20-fold with significant strain-specific variation in mice and animal-to-animal variation in nonhuman primates.

The strategy for blocking the humoral response emerged from an understanding that it is T cell dependent and that presentation of capsid antigens by antigen-presenting cells would be short-lived. Sher and coworkers suggested that activated dendritic cells turn over in 7 to 10 days and that therefore the window for inhibition of T-cell activation can be restricted to a 2-week period following vector administration (13). Our experience with an antibody to HuCD4 supports this hypothesis (5). Application of Clenoliximab to gene therapy will be greatly facilitated by the extensive safety data being generated on its application for patients with a variety of autoimmune diseases, such as rheumatoid arthritis and psoriasis.

In summary, readministration of AAV to skeletal muscle is indeed possible in the face of high NAb levels, although efficiencies are strongly diminished. It is possible that NAbs may eventually wane to noninhibiting levels over an interval, measured in years, that is less than or equal to the duration of therapeutic gene expression. In this scenario, inhibiting the B-cell response is not necessary. We believe it is prudent to anticipate that the interval of gene expression will be less than the duration of NAbs, requiring a strategy to eliminate or substantially reduce the humoral response. A course of treatment with Clenoliximab, a nondepleting anti-CD4 antibody, is one way that such a strategy could be evaluated in humans. Realize, however, that this has been modeled in recipients who are naive to vectors at the time of gene transfer, which will apply to two-thirds of humans before the absence of preexisting NAb and T-cell responses to AAV2. Different strategies may be required to modify secondary responses.