|Home | About | Journals | Submit | Contact Us | Français|
Latent infection with wild-type (wt) adeno-associated virus (AAV) was studied in rhesus macaques, a species that is a natural host for AAV and that has some homology to humans with respect to the preferred locus for wt AAV integration. Each of eight animals was infected with an inoculum of 1010 IU of wt AAV, administered by either the intranasal, intramuscular, or intravenous route. Two additional animals were infected intranasally with wt AAV and a helper adenovirus (Ad), while one additional animal was inoculated with saline intranasally as a control. There were no detectable clinical or histopathologic responses to wt AAV administration. Molecular analyses, including Southern blot, PCR, and fluorescence in situ hybridization, were performed 21 days after infection. These studies indicated that AAV DNA sequences persisted at the sites of administration, albeit at low copy number, and in peripheral blood mononuclear cells. Site-specific integration into the AAVS1-like locus was observed in a subset of animals. All animals, except those infected by the intranasal route with wt AAV alone, developed a humoral immune response to wt AAV capsid proteins, as evidenced by a ≥fourfold rise in anti-AAV neutralizing titers. However, only animals infected with both wt AAV and Ad developed cell-mediated immune responses to AAV capsid proteins. These findings provide some insights into the nature of anti-AAV immune responses that may be useful in interpreting results of future AAV-based gene transfer studies.
Adeno-associated virus type 2 (AAV) is a nonpathogenic parvovirus that generally requires coinfection with a helper virus, such as an adenovirus (Ad) or herpesvirus, to undergo productive infection (3, 4, 7). In cultured cells infected with wild-type (wt) AAV in the absence of helper virus, AAV establishes a latent infection (5, 11, 22) and often integrates site specifically into a sequence located on the q arm of human chromosome 19, termed the AAVS1 site (19, 20, 29–31, 35). AAV proviral DNA remains in this latent state until rescued by a helper virus. Studies on the mode of viral integration suggest that tandem copies of the viral DNA are inserted into the host cell chromosome in a head-to-tail orientation via the inverted terminal repeats of the virus (28).
Due to its lack of pathogenicity and its ability to establish persistent infections in human cells, AAV has gained acceptance as a potential vector for human gene therapy (18). Recombinant AAV (rAAV) vectors mediate stable in vivo expression in a wide range of different tissues including the lungs (16), muscles (12, 13, 26, 40), brain (2, 24, 27, 39), spinal cord (34), retinas (14, 32, 43), and liver (36). The use of AAV vectors has not been associated with significant toxicity in animal models. In addition, two phase I trials of recombinant AAV vectors have been undertaken in patients with cystic fibrosis (15, 38).
Despite the growing body of data regarding the biology of AAV latency in vitro, very few studies have examined this phenomenon in vivo. Epidemiologic studies have shown a high prevalence of AAV2 seropositivity (6, 8, 9). However, it is not known whether seropositivity is indicative of previous productive infection or of latent infection. Although infectious wt AAV has been cultured from samples from the respiratory and gastrointestinal tract in association with productive helper virus infection (9), latent AAV DNA has been found only in peripheral blood mononuclear cells (PBMCs) (21). Since humans and monkeys are the only species known to possess the AAVS1 integration sequence (35), they are the only two in vivo models which would be expected to faithfully reflect the wt AAV latency pathway. One previous study examined the interactions between rAAV vectors, wt-AAV and Ad in a rhesus macaque model and demonstrated that the productive phase of the wt AAV life cycle could be reproduced in that model (1).
In this study, we have examined latent wt AAV infection in that same rhesus model. Additional studies investigated the cell specificity of persistence, the integration state of latent viral genomes, and the immune responses to viral infection. In this setting, persistence was observed, although site-specific integration was infrequent. wt AAV genomes were detectable both at the site of administration and in circulating PBMCs. Neutralizing antibodies directed against wt AAV were observed in serum 21 days postinfection in animals infected intramuscularly or intravenously with wt AAV alone, but cell-mediated immune responses were elicited only in the presence of helper Ad.
wt AAV was prepared with 293 cells grown as monolayers in Dulbecco modified Eagle medium, supplemented with 10% fetal bovine serum and 100 U of penicillin-streptomycin per ml, at 37°C under humidified air containing 5% CO2. The cells were grown to confluency on Cell Factories (Nunc) and infected with wt AAV seed stock at a multiplicity of infection (MOI) of 1. The cells were coinfected with Ad5 at an MOI of 3. After 48 h, the cultures were harvested, resuspended in phosphate-buffered saline (PBS), and frozen and thawed three times. The crude lysate was heated at 56°C for 15 min to inactivate Ad and then centrifuged for 5 min at 9,000 × g to remove cellular debris. The supernatant’s volume was raised to 10 ml with PBS and loaded onto a HiTrap heparin affinity column (Pharmacia Biotech) at a rate of 1 ml per min with a peristaltic pump (Bio-Rad). The column was then washed with 20 ml of 0.01 M sodium phosphate buffer (pH 8) at a rate of 1 ml per minute. The flowthrough was retained and later analyzed. Virus was eluted at a rate of 1 ml per min (15 ml total volume) with a salt gradient (0 to 1 M NaCl) in 0.01 M sodium phosphate buffer (pH 8). Fifteen fractions were collected and analyzed by PCR and sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). PCR-positive fractions were pooled, and CsCl was added to the pooled sample to a density of 1.41 g per ml (refractive index, 1.3710). The sample was then loaded onto an ultracentrifuge tube (Beckman) and centrifuged in an SW50 swinging-bucket rotor (Beckman) at 35,0000 rpm at 4°C for 24 h. The contents of the tube were then fractionated (500-μl fractions) and dialyzed against PBS by using an ultraconcentrating tube with a molecular mass cutoff of 50 kDa (Amicon). The dialyzed fractions were then analyzed for the wt AAV genome by PCR. The positive fractions were pooled and analyzed by SDS-PAGE.
Viral capsid proteins were analyzed by PAGE. A 5-μl volume from each fraction was denatured, mixed with 1× loading dye (final concentrations, 12.5 mM NaH2PO4, 35 mM Na2HPO4, 0.5% SDS, 0.5% β-mercaptoethanol, 75 μg of bromophenol blue per ml, and 3 M urea), and loaded onto a 10% polyacrylamide gel (Bio-Rad). Samples were electrophoresed at 100 V for 2 h at room temperature with a MiniProtein electrophoresis apparatus (Bio-Rad). The gel slab was then fixed, stained with Coomassie blue (2.5 mg/ml in 45% methanol–10% acetic acid) at room temperature overnight, and destained (in 30% methanol–6% acetic acid) for several minutes until the background cleared, to visualize protein bands.
The physical titer was assessed by a PCR-based protocol. A 1-μl sample was taken from the purified stock and treated with 10 U of DNase (Boehringer Mannheim) in 10 mM MgCl2–50 mM Tris-HCl (pH 7.5) (total volume, 100 μl) for 1 h at 37°C. The sample was then treated with proteinase K (Boehringer Mannheim) at a final concentration of 0.2 μg/ml for 1 h at 37°C, using the manufacturer’s recommended buffer conditions. Viral DNA was then purified by two phenol-chloroform extractions and one phenol extraction. The sample was precipitated with ethanol and then centrifuged at 10,000 × g for 15 min. The supernatant was carefully discarded. The DNA pellet was resuspended in 10 μl of distilled H2O. The sample was then serially diluted. A PCR cocktail containing 1 μl of serially diluted viral sample and different amounts of the internally deleted competitor template was prepared. The PCR mixture consisted of 50 mM KCl, 10 mM Tris-HCl (pH 9.0), 1.5 mM MgCl2, 200 mM each dATP, dGTP, dCTP, and dTTP, 0.5 U of Taq polymerase (Promega), and 5 pmol of each amplification primer. The primers used for quantitative competitive (QC)-PCR were as follows: the 5′ primer sequence was 5′-TGGCCCACCACCACCAAAGCCCGCA-3′ hybridizing to wt AAV nucleotides (nt) 2283 to 2308, and the 3′ primer sequence was 5′-TGGCCCGCCTTTCCGGTTCCCGAGG-3′ hybridizing to wt AAV nt 2668 to 2693. Thirty-five cycles of PCR were performed with the following program: 96°C for 1 min, 72°C for 1 min, and 60°C for 1 min. The products were analyzed on a 1.5% agarose gel stained with ethidium bromide. These primers generated a 410-bp product from the wt AAV sequence and a 360-bp product from the standard template. Quantification was performed by comparing the PCR bands of the known standard template to the unknown concentration of the competitor.
The infectious-center assay was used as a means of quantifying the infectivity of purified wt AAV. The assay measures the ability of the virus to infect, uncoat, and replicate. 293 cells were plated in a 96-well plate at 50%. At 24 h later, the wells were infected at serially diluted amounts of the purified wt AAV and with Ad at an MOI of 10. The cells were then incubated for 24 h and harvested with trypsin-EDTA solution. The cells from individual wells were suspended in 5 ml of PBS and vacuum filtered onto wet nylon membrane filters (Whatman). They were lysed by placing the membrane for 5 min (cell side up) on filter paper saturated with 10% SDS, processed, and hybridized at 60°C with an AAV [32P]DNA probe specific for wt AAV (AAV2) radiolabeled by random priming (Boehringer Mannheim). Individual spots corresponding to infectious centers were visualized by autoradiography and counted manually.
The recombinant AAV, rAAV-UF5, expressing the humanized green fluorescent protein (hGFP) transgene driven by a cytomegalovirus promoter was packaged as previously described by Zolotukhin et al. (43) with Ad5-infected 293 cells cotransfected with a helper plasmid (to provide rep and cap from an ori construct) and a vector plasmid containing the cDNA flanked by AAV inverted terminal repeats. The vector was purified by two successive CsCl ultracentrifugation steps.
Eleven female rhesus macaques ranging from 2 to 3 years of age and weighing between 2.8 and 4.2 kg were obtained (Covance Research). Sera from the animals were assayed for AAV antibodies prior to purchasing. The animals were housed at the University of Florida Animal Facility. The animals were sedated during all procedures by administration of 10 mg of Ketamine per kg intramuscularly. wt AAV (1010 IU) was administered intravenously (into the right femoral vein) to two animals in a 3-ml suspension of bacteriostatic 0.9% sodium chloride. Three animals received 1010 IU of wt AAV into the left quadriceps muscle (6.5 cm from the patella) in a 500-μl suspension of bacteriostatic 0.9% sodium chloride. Three other animals received 5 × 109 IU of wt AAV in a 250-μl suspension of bacteriostatic 0.9% sodium chloride into each nostril (for a total dosage of 1010 IU). Two animals received coadministrations of wt AAV and a mutant form of Ad. These two animals received 5 × 109 IU of wt AAV in a 250-μl suspension of bacteriostatic 0.9% sodium chloride into each nostril (for a total wt AAV dosage of 1010 IU) plus 5 × 107 PFU of AdHR405 per nostril (for a total AdHR405 dosage of 108 PFU). AdHR405 is a host range mutant form of Ad selected for growth on monkey cells (1, 10). The control animal was given 250 μl of bacteriostatic 0.9% sodium chloride into each nostril and was sedated regularly along with the other animals. The animals were bled every 7 days throughout the study.
High-molecular-weight DNA was extracted from animal tissue by using QIAamp tissue kits (Qiagen). The DNA concentration was determined by spectrophotometric analysis of the optical density at 260 nm. The DNA (30 μg) was digested for 24 h with KpnI (New England BioLabs) under conditions recommended by the manufacturer. The DNA was then separated by agarose gel electrophoreses (1% agarose) in TBE buffer (10 mM Tris borate, 2 mM EDTA [pH 8]). The agarose gel was acid treated for 20 min with 0.2 N HCl and denatured for 15 min with 1.5 M NaCl–0.5 M NaOH. The agarose gel was then neutralized with 3 M NaCl–0.5 M Tris and then blotted via capillary forces by using 20× SSC (1× SCC is 0.15 M NaCl plus 0.015 M sodium citrate) (Sigma) onto nylon membranes. The nylon membrane was then baked for 2 h at 80°C under vacuum. The membranes were hybridized at 60°C with an AAV [32P]DNA probe specific for wt AAV (AAV2) radiolabeled by random priming (Boehringer Mannheim). The hybridization solution contained 6× SSC, 0.5% SDS, 1× Denhardt’s solution, 20 μg of herring sperm DNA per ml, and 0.01 M EDTA. The membrane was washed in large volumes of 2× SSC–0.1% SDS at 60°C, dried, placed in an X-ray cassette (Kodak), and exposed to X-ray film (Kodak) for several days.
Genomic DNA samples from peripheral blood mononuclear cells (PBMCs) and from the sites of virus administration were purified with QIAamp blood kits (Qiagen). High-molecular-weight DNA was extracted from animal tissue by using QIAamp tissue kits. PCR was carried out with 100 ng of genomic DNA (or 1 μl of fractioned material) added to 50 μl of total PCR cocktail (ingredients and primers are given above). Thirty-five cycles of PCR were performed with the following program: 96°C for 1 min, 72°C for 1 min, and 56°C for 1 min. The products were analyzed on a 1% agarose gel, stained with ethidium bromide, transferred to nitrocellulose, and hybridized with a wt AAV-specific probe radiolabeled by random priming (Boehringer Mannheim).
To detect site-specific integration, DNA samples that were positive for AAV sequences by the internal PCR described above were also analyzed by the PCR dot blot methods described by Yang et al. (41), in which the 5′ PCR primer was chosen from within the 3′ end of the wt AAV sequence (5′-ATAAGTAGCATGGCGGGTTA-3′) and was directed outward from the proviral insert while the 3′ primer was chosen from within the AAVS1 site (5′-GCATAAGCCAGTAGAGCTCA-3′). Homology between the primer sequence and the rhesus sequence was confirmed by sequence alignment. PCR products were immobilized on nylon membranes by using a Schleicher and Schuell Minifold II dot blot manifold, and a random-primed 32P-labeled probe from the end of the AAV genome [the 180-bp PvuII-XbaI fragment from pSub201(+)] was used for the hybridization. The specificity of this signal for AAV-chromosomal junctions was confirmed by comparison with a control in which only the internal AAV probe was used.
Metaphase chromosomes and interphase nucleus preparations were prepared by a mitotic shakeoff method (25). Hypotonic fixation and slide preparation were performed by standard cytogenetic methods. A wild-type AAV probe was labeled and hybridized to each preparation as previously described (25). Photomicrographic images of nuclear signals were acquired by using a cooled charge-coupled device camera under the control of the Metamorph software package.
293 cells were plated in a 96-well plate at 50 to 75% confluency (5 × 103 cells per well). The cells were cultured overnight at 37°C in humidified air containing 5% CO2. The following day, serial dilutions of animal serum (day 0 and day 21) were incubated with 105 IU (equivalent to an MOI of 10) of recombinant AAV expressing the hGFP transgene (rAAV-UF5). The dilutions were performed in Hanks balanced salt solution a 100-μl total volume. The sample was then incubated at 37°C for 1 h. After 1 h, the medium from the previously plated cells was removed and 100 μl of Dulbecco modified Eagle medium supplemented with 20% heat-inactivated fetal bovine serum and 200 U of penicillin-streptomycin per ml containing 2 × 105 PFU of Ad was added. Additionally, 100 μl of the serially diluted serum plus rAAV-UF5 solution was added to the wells. The cells were cultured for 24 h at 37°C in humidified air containing 5% CO2. After 24 h, the transgene product was visualized under a fluorescence microscope. The end point was defined as the dilution of serum which inhibited the transgene efficiency by at least 10-fold.
Heparinized whole blood (5 ml) was collected and diluted 1:1 in Hanks buffered salt solution in a conical centrifuge tube. Ficoll-Hypaque (5 ml; Pharmacia) was slowly layered at the bottom of the conical tube. The tube was then centrifuged for 30 min at 500 × g at room temperature. The layer above the clear layer was carefully removed with a sterile transfer pipette. The removed material was transferred to a centrifuge tube containing 10 ml of Hanks buffered salt solution and centrifuged for 10 min at 500 × g at room temperature. The supernatant was removed, and the cell pellet was washed again with 10 ml of Hanks buffered salt solution and recentrifuged for 10 min at 500 × g at room temperature. The supernatant was discarded, and the cell pellet was resuspended in 2 ml of RPMIC+ medium (CellGrow). The cells were counted by a Trypan blue exclusion method. β-Mercaptoethanol was added at 2 μl per ml of cell suspension (adjusted to account for 106 cells per ml). Two 96-well plates were set up for every animal, one for the antigen (VP3 capsid proteins) and a second for the mitogen phytohemagglutinin as a positive control. Cells were plated on the wells, and the respective agent was added and incubated for 3 days at 37°C in humidified air containing 5% CO2. On day 3, 20 μl of a 1:20 [3H]thymidine dilution was added to the mitogen-treated plate. On day 4, the mitogen-treated cells were removed and the level of radioactivity was determined. On day 5, 20 μl of a 1:20 [3H]thymidine dilution was added to the antigen-treated plate. After 24 h, the plate was harvested and the level of radioactivity was determined by liquid scintillation counting.
Tissues from the lungs, nasal passages, trachea, thymus, bronchial lymph nodes, heart, liver, spleen, pancreas, kidney, jejunum, mesenteric lymph nodes, gonads, brain, and muscles were isolated aseptically and placed in 4% paraformaldehyde for 24 h at 4°C. The tissues were then embedded in paraffin, and 10-μm sections were made. The sections were then stained with hematoxylin and eosin, coverslipped, and photographed with a Zeiss Axioskop upright microscope.
Since one of the primary goals of this study was to characterize immune responses to AAV in both latent and productive AAV infections, it was essential to eliminate contaminating proteins that could serve as adjuvants to an anti-AAV host response. To purify wild-type AAV free of Ad proteins and other contaminants, an affinity purification method based on the recently described discovery of heparan sulfate proteoglycan as an attachment receptor for AAV2 was developed (37). A cleared lysate of Ad5- and AAV2-infected 293 cells was loaded directly onto a heparin affinity column, washed in low-salt buffer, and eluted with a continuous NaCl gradient, ranging in concentration from 0 to 1 M. Fifteen successive fractions were analyzed for viral proteins by SDS-PAGE with silver staining (Fig. (Fig.1A)1A) and for viral genomes by PCR (Fig. (Fig.1B).1B). AAV genomes were detected in fractions 1 through 10, while the majority of cellular proteins were eluted in fractions 9 through 15. While this is not quantitative, more intense PCR signals were observed in fractions 6 to 9. These fractions corresponded to a concentration of 0.4 to 0.6 M NaCl in the elution buffer, indicating that this would be the optimal salt concentration for eluting the bound viral particles. The positive fractions were further purified by CsCl density gradient ultracentrifugation, and the resultant CsCl gradient fractions were analyzed for wt AAV DNA by PCR. PCR-positive fractions (Fig. (Fig.2A)2A) were observed at a refractive index of 1.3715 (ρ = 1.41 g/cm3). Peak fractions were pooled and examined by PAGE with silver staining (Fig. (Fig.2B).2B). In the final material, only three protein bands were detectable, and these migrated at apparent molecular masses identical to those predicted for AAV capsid proteins (62, 73, and 87 kDa). No contaminating proteins were detectable.
To determine the yield and infectivity of virus, we independently assessed the physical and biological titers of wt AAV in the preparation. The physical titer was determined by genome quantitation via QC-PCR. By using this technique, it was estimated that there were 1013 particles of AAV per ml of solution (Fig. (Fig.3).3). To determine the biological titer of this virus preparation, a cell-based infectious-center assay was used. The infectious titer was 1011 IU per ml (Fig. (Fig.3).3). This particle-to-IU ratio of 100 compares favorably with that reported by others for wt AAV (20 to 200) and is superior to that achieved with recombinant vectors under standard conditions.
To validate the use of rhesus macaques (Macaca mulatta) as a model of AAV latency, we cloned and sequenced the rhesus AAVS1 locus from rhesus genomic DNA (data not shown). After sequence alignment with the human sequence, it was determined that there was moderate sequence homology between the human and rhesus loci at the two key sites within this locus, the Rep binding element and terminal resolution site (trs), while the intervening sequence was identical. Based on these findings, an in vivo model of latent AAV infection was established. Each of eight rhesus macaques was infected with 1010 IU by one of several routes (intranasal, intravenous, or intramuscular). Two additional animals had a productive infection with wt AAV established by intranasal coadministration of wt AAV and Ad2HR405, a host range mutant Ad selected for growth on monkey cells (10). One additional animal was inoculated with isotonic saline intranasally as a control. All the animals were analyzed for antibodies to AAV capsid proteins prior to experiments.
To determine whether infection with wt AAV resulted in persistence of AAV DNA at the site of inoculation, DNA was isolated from each site and examined by Southern blot analysis. Based on a previous study with recombinant AAV indicating that systemically delivered vector is distributed mostly to the liver (17), liver tissue was taken from animals infected intravenously. Muscle tissue was used from animals infected intramuscularly, and nasal tissue was used from animals infected intranasally with and without helper viruses. The abundance of viral DNA was below the level of detection by Southern blotting at each of the sites of infection (Fig. (Fig.4A),4A), despite having a sensitivity of <0.1 copy of AAV DNA per cell. This was unexpected, given that we delivered at least 103 to 104 viral genome copies per cell at the site of administration in the muscle, assuming that between 107 and 108 nuclei would be present in the region of muscle subtended by a 500-μl injection. The same samples were then analyzed by a PCR assay for internal wt AAV sequences (sensitive to 0.001 copy per cell), and several were found to be positive for AAV DNA (Fig. (Fig.4B).4B).
One previous study (21) had shown evidence of AAV DNA persistence in peripheral blood cells after naturally occurring infections. To determine whether this could have occurred in our animals, genomic DNA was isolated from lymphocytes and amplified by PCR with wt AAV primers. DNA was isolated from lymphocytes 21 days after viral infection and amplified with wt AAV primers (see Materials and Methods) under optimized conditions. Lymphocytes isolated from both of the intravenously infected animals were positive for AAV DNA (Fig. (Fig.5,5, lanes 1 and 2). Of the intramuscularly infected animals, only one (95B005) was positive for AAV DNA (lane 5). Additionally, one of the intranasally infected animals was positive for AAV DNA (95B032) (lane 7). Neither of the intranasally infected animals given helper virus was positive for AAV DNA (lanes 8 and 9).
To determine whether site-specific integration had occurred within the rhesus AAVS1-like locus, all organ DNA samples that had scored positive by PCR for internal AAV sequences were also assayed by a junction PCR dot blot assay described by Yang et al. (41). Junction sequences were amplified with a 5′ primer within the AAV “tail” (3′ end) directed outward from the genome sequence (5′-ATAAGTAGCATGGC-GGGTTA-3′) and a 3′ primer from the AAVS1 site (5′-GCATAAGCCAGTAGAGCTCA-3′). A dot-blot hybridization was used to detect amplified products, since previously published data indicated that this amplification produces a heterogeneous mix of products that are not easily distinguished by agarose gel electrophoresis with ethidium staining. To distinguish tail-to-tail junctions between two viral genomes from bona fide AAV-cell DNA junctions, the reactions were also performed with a single internal AAV primer in the reaction. This primer alone would be expected to amplify inverted tandem (tail-to-tail) forms whether or not they are integrated.
The positive control for this assay was genomic DNA extracted from a culture of IB3-1 cells (CF bronchial epithelial cell line) infected with wt AAV2 at an MOI of 5. This cell line showed both a strong signal with the double primer pair (Fig. (Fig.6,6, bottom row) and a weaker signal with the single internal AAV primer, consistent with a tail-to-tail junction between two viral genomes. The negative control was genomic DNA from uninfected IB3-1 cells. Evidence of site-specific integration in both the nasal epithelium and the PBMCs from one of the animals that had received vector intranasally was observed by this assay (Fig. (Fig.6).6). Two of the animals that had received intravenous injections of AAV showed very faint site-specific integration signals in hepatocyte DNA. The specificity of this signal for AAV-cell DNA junctions as opposed to AAV-AAV junctions was confirmed by the lack of amplification with the single AAV PCR primer. DNA sequencing of such junctions is a subject for future studies.
FISH was performed on cell cultures isolated either from solid organs at the site of administration (muscle or nose), from the skin fibroblasts from the intravenously infected animals, or from PBMCs from wt AAV-infected animals 21 days postinfection. Of particular note, none of the muscle, nose, or skin samples were positive by FISH, while numerous PBMCs were positive. These FISH data were consistent with the Southern blot data indicating a low copy number of AAV DNA at the site of administration. FISH preparations of lymphocyte interphase nuclei showed wt AAV DNA signals in animals infected intranasally with helper virus (Fig. (Fig.7A)7A) and animals infected intramuscularly (Fig. (Fig.7B).7B). No signals were observed in lymphocyte interphase nuclei of the control animal (Fig. (Fig.7C).7C). Additionally, the signal numbers were greater in the lymphocytes isolated from the animal infected intranasally with wt AAV and helper virus (Table (Table1).1). Metaphase nuclei were examined by FISH for wt AAV DNA in all the groups of animals, and none were positive. However, due to the low mitotic index of these primary cultures, there were only one to five metaphase nuclei available for scoring on most culture samples. This very small number of metaphase spreads present would make it difficult to detect integration by this method.
The overall incidence of FISH-positive PBMCs was also analyzed. Of 100 PBMC interphase nuclei examined from each study group, 3% of the PBMCs from animals infected with AAV alone intranasally were positive, compared with 15% from the intramuscularly infected animals and 6% from animals coinfected with AAV and Ad (Fig. (Fig.8).8). The 1% of lymphocytes that apparently show positive FISH signals represent the background on this assay.
To assess whether rhesus monkeys infected with wt AAV developed humoral immune responses, sera were collected prior to and 21 days after infection and assayed for in vitro anti-AAV neutralizing activity. Neutralizing activity was found in most of the preinfection sera. A significant increase in neutralizing-antibody response was defined as a fourfold increase in neutralizing titer between the pre- and postinfection samples. Most of the infected animals did develop a significant neutralizing-antibody response (Fig. (Fig.9).9). This included animals that were infected with AAV alone by the intravenous or intramuscular route and animals infected intranasally with both AAV and Ad. In contrast, animals infected nasally with wt AAV alone did not develop a significant increase in anti-AAV neutralizing antibodies (Table (Table2).2). To determine whether 21 days was sufficient to detect a response, one of these animals (96B026) was monitored for over 7 weeks and bled on two additional occasions, and it still did not have a detectable response. This animal was later immunized with 100 μg of purified VP3 as a positive control and was then found to develop neutralizing antibodies within 21 days after administration (data not shown).
To determine whether cell-mediated immune responses to AAV had occurred, lymphocytes from wt AAV-infected animals were exposed ex vivo to AAV capsid antigen and the extent of antigen-specific stimulation of lymphocyte proliferation was assessed. The stimulation index was calculated by dividing the number of cpm of [3H]thymidine incorporated into lymphocyte cultures in the presence of the specific antigen (with 1, 5, or 10 μg of the antigen) by the number of cpm incorporated into parallel cultures grown in the absence of antigen (28). Based on previous norms for this assay, a positive response was defined as a stimulation index of 3. The viability of each of these PBMC cultures was confirmed by PHA stimulation of parallel wells, and each showed a stimulation index of ≥3.0.
At baseline, all animals were negative in this assay. On day 26, animals infected with wt AAV intravenously (95B002 and 95B003) were negative in this assay (stimulation index, <1), as were the animals infected intramuscularly (stimulation indices, 1.8 and 2.7) and intranasally (stimulation indices, 2.1 and 1.9) (Fig. (Fig.10).10). In contrast, animal 95B039, which was infected intranasally with both wt AAV and Ad, had a stimulation index of 3.8. Thus, only the animal with productive helper virus infection developed a specific cell-mediated immune response to AAV capsid proteins.
Histological examination was performed to determine if inflammation or cellular infiltration had occurred in response to wt AAV infection at the site of infection. Tissues isolated from muscle, liver, kidney, and lungs from animals infected intramuscularly and intravenously were histologically examined. No morphological abnormalities were observed in any case compared to matched tissue samples from the control animal (Fig. (Fig.11).11). Tissues from the nasal cavity and trachea of intranasally infected animals were also examined (Fig. (Fig.12).12). No abnormalities were observed in the animals infected intranasally with wt AAV alone. However, animal 95B039, which was coinfected with wt AAV and Ad demonstrated goblet cell hyperplasia (Fig. (Fig.12B)12B) in the nasal epithelium. The control animal had no apparent increase in goblet cell number.
Despite the high prevalence of AAV infection in humans, relatively little data is available with regard to the latency of wt AAV in vivo. Current models of AAV latency are derived from experiments performed with immortalized and primary cell lines. Our findings indicate that viral DNA persists both at the site of administration and in peripheral blood cells. Notably, the only published data about isolation of latent AAV DNA from humans was from peripheral blood cells in a pattern consistent with our findings (21). The data presented here did not allow a precise determination of the in vivo integration frequency. Although site-specific integration was apparently present in some instances, the frequency of this process appears to be rather low.
Previous findings with recombinant AAV in the respiratory tract also indicated infrequent integration (1). However, previous studies with recombinant AAV vectors in muscle have indicated that vector genomes persist in muscle either as integrated proviral genomes or as high-molecular-weight concatemers (12, 13, 40). There are several potential reasons why our studies might have underestimated the integration frequency somewhat. One possibility is that sampling of the sites of administration may have been imprecise. It is also possible that despite having integrated into host cells, these cells were eliminated by the immune response. This seems very unlikely, however, since elimination of infected cells by cytotoxic T lymphocytes would typically be associated with positive findings on the antigen-specific lymphocyte proliferation assay and with histological changes at the site of delivery. Finally, it is possible that rhesus monkeys simply do not represent a suitable host for AAV. The fact that productive infections have been achieved in this model (1) and that it possesses an AAVS1-like site with moderate homology to the human sequence argue against that possibility. However, there are significant limitations to the rhesus model, both as a model of productive infection and as a model of latency. The host range mutant Ad strains are less efficient for replication in culture than are wt Ad strains and thus may not faithfully reproduce Ad infection in vivo. Furthermore, the AAVS1-like sequences we identified have not been proven to be functional for AAV integration.
FISH analysis of PBMCs also suggested that wt AAV sequences persist in these cells, as evidenced by signals on interphase nuclei. However, no signals were detectable on metaphase chromosomes. Since less than 15% of cells were positive in the interphase nuclei in every case, however, one would not expect the examination of such a small number of metaphase spreads (five or fewer per sample) to show integration even if it occurred in every case where AAV DNA was persistent. The accessibility of these rather short target sequences to hybridization with the FISH probes would also be expected to be greater with the unwound chromatin of interphase nuclei than with that observed in condensed metaphase chromosomes. There was also a notable lack of FISH signals on nuclei from primary cells isolated from the site of vector administration. While our group has found positive FISH signals from bronchial epithelial cells harvested from the site of rAAV-CFTR administration (1), the possibility remains that the process of establishing primary culture somehow selects for a population of cells that are less likely to be latently infected. Recent in vivo data with rAAV indicate that terminally differentiated cells may be more permissive for AAV infection. If this is the case, the process of establishing a primary culture, which favors the growth of less differentiated cells, could substantially underestimate the actual frequency of integration in vivo.
Previous studies indicated that 60% of adults are seropositive for AAV and that this seroconversion occurs early in life (6). It has never been known whether this humoral immune response to AAV capsid was elicited primarily by productive infection or by latent infection. The evidence presented here suggests that nonhuman primates can develop a neutralizing-antibody response to wt AAV in the absence of helper virus if infected parenterally but that mucosal exposure without helper virus does not elicit such a response. These results are compatible with data generated from experiments with rAAV, in which administration to the maxillary sinus did not elicit an anti-capsid antibody response, while intramuscular administration did so in several cases (12, 33, 40).
It is also notable that cell-mediated immune responses to a single dose of rAAV were observed only in the presence of active AAV replication, i.e., in the presence of helper virus. This is also consistent with previous results with rAAV. These data are consistent with the findings of Joos et al. (23), which indicated that antigen-presenting cells are relatively resistant to AAV infection. Alternatively, the helper virus may simply be providing an adjuvant effect. It is important to point out that these studies represent a single-exposure paradigm and that repeated dosing might have resulted in more vigorous responses.
To limit any adjuvant effects from contaminants in our wt AAV preparations, careful purification and quality control assays were used. We used properties of the recently described AAV receptor to purify large quantities of wt AAV. The use of a heparin affinity column prior to isopycnic centrifugation yielded a large viral particle number. Additionally, the viral stock appeared to be free of Ad proteins and other contaminating proteins. Viral genomes were quantified by QC-PCR, and the infectivity of the virus was quantified by the infectious-center assay. Thus, the physical titer was 1013 particles per ml, compared to the biological titer of 1011 IU per ml, for a particle-to-IU ratio of 100, which is nearly optimal for a DNA virus. The method described herein is similar to one recently described by Zolotukhin et al. (42).
In summary, the rhesus macaque was used as a model of latent wt AAV infection. In this model, wt AAV persisted both at the site of administration and in peripheral blood cells and in some instances integrated within an AAVS1-like site. Furthermore, latent AAV infection was capable of eliciting humoral immune responses but not cell-mediated immunity. While the immune response data is quite consistent with the results of previous reports with wt AAV, the lack of site-specific integration calls into question the relevance of such findings in cell cultures ex vivo. Additional studies by methods more sensitive for detecting low-frequency integration are required before a definite conclusion can be drawn about the capacity of AAV to integrate site specifically in vivo.
This work was supported by a grant from the National Institute for Diabetes, Digestive, and Kidney Diseases (DK51809).
Many thanks to David Muir for assistance with column chromatography techniques, to Kye Chesnut, Barry Byrne, and Nick Muzyczka for advice on this work, and to Mark Atkinson for assistance with immune response assays.