Adeno-Associated Virus Vector Genomes Persist as Episomal Chromatin in Primate Muscle

Abstract

Recombinant adeno-associated virus (rAAV) vectors are capable of mediating long-term gene expression following administration to skeletal muscle. In rodent muscle, the vector genomes persist in the nucleus in concatemeric episomal forms. Here, we demonstrate with nonhuman primates that rAAV vectors integrate inefficiently into the chromosomes of myocytes and reside predominantly as episomal monomeric and concatemeric circles. The episomal rAAV genomes assimilate into chromatin with a typical nucleosomal pattern. The persistence of the vector genomes and gene expression for years in quiescent tissues suggests that a bona fide chromatin structure is important for episomal maintenance and transgene expression. These findings were obtained from primate muscles transduced with rAAV1 and rAAV8 vectors for up to 22 months after intramuscular delivery of 5 × 10¹² viral genomes/kg. Because of this unique context, our data, which provide important insight into in situ vector biology, are highly relevant from a clinical standpoint.

Long-term transgene expression is achieved following recombinant adeno-associated virus (rAAV) vector-mediated gene transfer to skeletal muscle. After conversion of the single-stranded rAAV genome into double-stranded DNA (dsDNA), the vector genome is concatemerized (23, 57, 67) and circularized (12, 20, 21, 35, 57, 64), processes that have been well described in vivo. In rodent skeletal muscle, rAAV vector genomes are maintained mainly as extrachromosomal forms, as demonstrated by a sensitive PCR-based assay (53, 54); thus, gene expression derives predominantly from the episomal forms in this tissue. Similar data were reported for rAAV vector persistence in liver (37, 43), even though the integration rate observed in this tissue is higher (38). The mechanism of how rAAV vector episomes persist and maintain long-term expression of a therapeutic transgene in muscle and other tissues is of great interest to the field.

Numerous episomal viruses are organized in a chromatin-like structure during their life cycle. The circular genomes of papovaviruses, Simian virus 40 (SV40), and polyoma virus exist as minichromosomes composed of cellular histones organized in nucleosomes (1, 2, 13). The structure and assembly of SV40 chromatin have been studied extensively as a model of cellular chromatin (49, 56). The Abutilon mosaic geminivirus, a single-stranded-DNA-containing virus of plants (48) that resembles animal papovaviruses, also assembles into minichromosomes in situ. Other viruses, such as the duck hepatitis B virus, an avian hepadnavirus (45), and the latent genomes of alphaherpesviruses, such as herpes simplex virus type 1, and of gammaherpesviruses, such as Epstein-Barr virus and Kaposi's sarcoma-associated herpesvirus, are maintained as episomal chromatin (14, 34, 55). The parvoviruses AAV and minute virus of mice (MVM) have been observed to assimilate with histones within hours after infection of cells in culture (3, 36). However, conflicting reports for wild-type AAV (5) and MVM (15) exist for the actual nucleoprotein compositions of these genomes in vitro.

For recombinant AAV vectors, transgene expression in cell lines harboring a single copy of an rAAV-lacZ vector stably integrated into the cellular genome can be inactivated over time, but the expression can be rescued in vitro by treatment of the cells with trichostatin, a histone deacetylase (HDAC) inhibitor (9, 10). In parallel, Okada et al. were able to enhance rAAV transduction when tumor cell lines were treated in vitro with an HDAC inhibitor at the time of vector infection (46). HDAC inhibitors, such as butyric acid, trichostatin, sodium phenylbutyrate, and valproic acid, modify the acetylation of histones, resulting in chromatin rearrangements that allow gene expression (39). Nevertheless, the direct demonstration of rAAV genomes associated with histones has not been shown in vivo.

In this report, we characterize the rAAV vector genome structure in nonhuman primate (NHP) skeletal muscle when the vector copy number has reached a steady state, i.e., several months following gene transfer. The vector genomes persist as episomal monomeric and concatemeric circles in a structure consistent with that of chromatin. Administration of sodium phenylbutyrate to primates previously transduced intramuscularly with rAAV vectors does not increase transgene expression, indicating that the majority of persistent vector genomes are maintained in a chromatin configuration that accounts for long-term transgene expression. The structure of the rAAV genomes assimilated with histones may begin to explain the long-term persistence of the vector genome episomes and transgene expression in nondividing tissues.

MATERIALS AND METHODS

Production of AAV vectors.

The vector plasmid used for rAAV production was generated by cloning the LEA29Y (belatacept) sequence in a pZA backbone harboring the AAV serotype 2 (AAV2) inverted terminal repeats (ITRs) (a gift from J. M. Wilson's laboratory, Philadelphia, PA) between the Rous sarcoma virus (RSV) promoter and a Woodchuck hepatitis virus posttranscriptional regulatory element (WPRE). The LEA29Y sequence was obtained by oligonucleotide-directed mutagenesis, which leads to the substitution of two amino acids in CTLA4-Ig (abatacept) (kindly provided by B. Vanhove, INSERM U643, Nantes, France): L104E and A29Y. rAAV stocks were produced by cotransfection of either pDP1 or pDP8 (respectively, for AAV vector production of serotype 1 or 8) and of the pZA RSV-LEA29Y-WPRE-pA vector plasmid in 293 cells. rAAV RSV-LEA29Y-WPRE-pA particles were subsequently purified with a CsCl density gradient. The titration of rAAV stocks was typically in the range of 10¹³ viral genomes (vg)/ml.

Intramuscular injection and tibialis anterior muscle ablation of NHPs.

Five captive-bred cynomologus macaques were purchased from the Centre de Recherches Primatologiques, Ferney, France. Anesthesia was performed with ketamine (10 mg/kg of body weight) before intramuscular rAAV delivery. The total volume per injection site was 170 to 300 μl, and each animal received 5 × 10¹² vg/kg of rAAV2/1 (macaques Mac 1, Mac 2, and Mac 9) or rAAV2/8 (Mac 5 and Mac 6) RSV-LEA29Y-WPRE-pA distributed at three injection sites in a tibialis anterior muscle (or at six injection sites in the two tibialis anterior muscles for Mac 9). The protocol was approved by the Institutional Animal Care and Use Committee of the University of Nantes. The integrality of injected tibialis anterior muscles was taken off at 10 to 22 months postinjection under isoflurane anesthesia. Muscles were immediately transferred in cold phosphate-buffered saline (PBS; pH 7.5) and preserved for a maximum of 24 h at +4°C for chromatin studies.

Separately from the five individuals used in the present study, Mac 13, whose results are presented in Fig. 1, was described previously and was followed for up to 5 years after that study (11). Briefly, Mac 13 received by intramuscular injection 6.7 × 10¹¹ vg/kg of an AAV2/1 vector encoding the cynomologus macaque erythropoietin (cmEPO) under the control of the rtTA2^s-M2 doxycycline-regulatable system (61). Doxycycline induction cycles were described in reference 11.

FIG. 1.

Long-term expression of a transgene after intramuscular administration of recombinant AAV. An NHP transduced intramuscularly with an AAV1-cmET-Des-M2 construct expresses cynomologus macaque EPO under the control of the Tet_ON promoter after induction with doxycycline.

Total DNA extraction from macaque muscle.

Total DNA was isolated from the first step of the procedure described below under “Isolation of nuclei from primate muscle.” Briefly, tissues were minced finely, the pellet was resuspended in 2 ml of urea buffer (8 M urea, 1% sodium dodecyl sulfate [SDS], 10 mM EDTA, pH 8, 300 mM NaCl, 10 mM Tris, pH 7.5) per gram of muscle, and proteinase K was added at a concentration of 2 mg/ml. Samples were incubated for 3 h at 56°C by homogenizing them from time to time. A second run of digestion was realized overnight at 56°C by adding 1 volume of urea buffer and the same quantity of proteinase K. DNA was extracted two times with phenol-chloroform (pH 7.5) and precipitated with sodium acetate and absolute ethanol.

Quantitative PCR.

Recombinant AAV genome copy number was quantified in 10 ng of total DNA extracted from NHP-injected muscles using a real-time TaqMan PCR system (LightCycler 2.0; Roche Applied Science). Primers and probes used for amplification of the vector-specific sequence and of the endogenous macaque sequence situated in the ɛ-globin gene, as well as quantitative PCR conditions, are described in reference 60.

Southern blot analysis.

Twenty micrograms of total DNA extracted from primate muscles was digested twice for 3 h at 37°C with 2.5 U/μg of the restriction enzyme(s) XbaI, EcoRI, or BamHI-ClaI (Roche Applied Science), engendering zero, one, or two cuts, respectively, in the vector. The samples were run on a 0.8% agarose gel (SeaKem LE agarose; Lonza) in Tris-acetate-EDTA (TAE) for 16 h at 35 V at room temperature or on a 0.5% agarose gel prepared with multipurpose QA-agarose (MP Biomedicals) in TAE for 24 h at 35 V and at +4°C. Gels were then stained with ethidium bromide and photographed. After depurination in HCl (0.25 N), DNA was transferred onto a nylon membrane (MP Biomedicals) under denaturing conditions. Southern blotting was performed by using the vector-specific WPRE-pA probe or the RSV probe, which correspond, respectively, to the 867-bp BamHI/ClaI fragment or to the 646-bp BglII fragment of the pZA RSV-LEA29Y-WPRE-pA vector plasmid. The endogenous cmEPO probe corresponds to a 1,226-bp sequence of the macaque EPO gene amplified by PCR with the primers Mac5 (5′-GGTGGAGGTGGGAAGCTAG) and Epo3′ (5′-GTGTCAGCAGTGATGGTTCGGAG). Probes were randomly labeled by using the Rediprime II DNA labeling kit (Amersham Biosciences). Twenty-five nanograms of each fragment was labeled by adding 5 μl of Redivue [α-³²P]dCTP at 6,000 mCi/mol (Amersham Biosciences) and incubating the mixture for 1.5 h at 37°C. The probes were then purified on a Nick column (GE Healthcare), and their specific activities were calculated based on scintillation counting. DNA hybridization conditions were 65°C for 16 h in buffer containing 5× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate), 5× Denhardt's reagent, 0.5% SDS, and 100 μg of sonicated herring sperm DNA per ml. Membranes were washed under high-stringency conditions with from 2× SSC, 0.1% SDS at room temperature to 0.1× SSC, 0.1% SDS at 60°C. The storage phosphor signal was detected with a Typhoon 9410 Imager (GE Healthcare).

PS digestion.

For plasmid-safe DNase (PS-DNase) treatments, 20 μg of total DNA was digested with XbaI, an enzyme that does not cut in the vector, as described above. DNA was then incubated twice for 16 h at 37°C in 33 mM Tris acetate, pH 7.8, 66 mM potassium acetate, 10 mM magnesium acetate, 0.5 mM dithiothreitol (DTT), and 1 mM ATP with 5 U/μg of ATP-dependent PS-DNase (Epicentre Biotechnologies). An ethanol precipitation step was added between the two PS-DNase digestion runs. PS-DNase was inactivated by a 30-min incubation at 70°C. After ethanol precipitation, samples were resuspended in H₂O and loaded onto a 0.8% agarose gel.

Detection of rAAV-cellular genome junctions by LAM-PCR.

Linear-amplification-mediated (LAM)-PCR was used to isolate sequences contiguous with AAV genomes in samples of total DNA extracted from macaque muscles (52). The AAV sequence was first amplified from 10 ng of genomic DNA using one of the following biotinylated primers corresponding to the 5′ region of the RSV promoter: RSV1bio, 5′-bio-CTCAGCGACCTCCAACACAC, or RSV2bio, 5′-bio-GAGCAGATACTGGCTTAAC. The reaction conditions were as follows: 83.5 fmol of the primer, a 200 μM concentration of the deoxynucleoside triphosphates (dNTPs), and 1.25 U Taq DNA polymerase with the Taq buffer (Qiagen). The cycling conditions were denaturation at 95°C for 2 min followed by 50 cycles of 95°C for 45 s, 57°C for 45 s, and 72°C for 1 min, with a final extension of 72°C for 5 min. This cycle was repeated after the addition of 1.25 U of fresh Taq DNA polymerase for a total of 100 cycles of linear amplification.

The biotinylated PCR products were captured using 200 μg of streptavidin-coated magnetic beads (Dynal Dynabeads kilobase binder kit) in conjunction with a magnetic particle concentrator (Dynal) according to the manufacturer's protocol. After binding, the beads were washed twice with H₂O and incubated with 2 U of the Klenow fragment (Roche Applied Science), a 250 mM concentration of the dNTPs, and a random-hexanucleotide mix (Roche Applied Science) in a 20-μl reaction volume for 1 h at 37°C to create dsDNA. Following incubation, the DNA-beads were washed twice as described above and incubated for 1 h at 65°C with 4 U of the restriction endonuclease Tsp509I, which generates a 5′ overhang. After restriction enzyme digestion, the beads were washed and a cohesive, double-stranded Tsp509I adaptor (40 pmol) generated by hybridization of two oligonucleotides (LC1, GACCCGGGAGATCTGAATTCAGTGGCACAGCAGTTAGG, and LC3, AATTCCTAACTGCTGTGCCACTGAATTCAGATC) was ligated to the restricted DNA for 5 min at room temperature in a 10-μl volume containing 2 U of Fast-link DNA ligase, 1 mM ATP, and the accompanying Fast-link buffer (Epicentre). The DNA-beads were washed, and a 1:5 dilution of the eluate was used as the template in a nested PCR. The first round of PCR used 8.35 pmol of each primer, a 200 μM concentration of the dNTPs, and 2.5 U of Taq DNA polymerase (Qiagen), with the corresponding reaction buffer using the same cycling conditions as described above with the exception that only 35 cycles were performed. The adaptor and AAV-specific primers used in the first-round PCR were as follows: LCI (5′-GACCCGGGAGATCTGAATTC) and RSV4Bbio (5′-CTGAGAGTGCACCATACGGATC). The biotinylated PCR products were purified as explained above, and a 1:10 dilution of the eluate was used as the template for the nested PCR with the same reaction conditions as used in the first-round PCR but with the following primers: LCII (5′-GATCTGAATTCAGTGGCACAG) and ITR1bio (5′-GTCGAGATCTTCCAGAGCATG) or RSV5bio (5′-GATAAGCTGTCAAACATGACG). PCR products were purified with the PCR QIAquick kit (Qiagen) and cloned into the pCR2.1 TOPO-TA cloning vector (Invitrogen). The sequencing of the clones obtained by LAM-PCR was performed by GATC Biotech.

Isolation of nuclei from primate muscle.

Nuclei were isolated from freshly excised tibialis anterior muscles of macaques. Muscle was immediately placed in cold 1× PBS, pH 7.5, and kept on ice. All manipulations were performed on ice. Muscle was sliced and washed three times with 1× PBS, pH 7.5. The connective tissue, the fat, and the nerves were removed. The muscle pieces intended for nucleus isolation or total DNA extraction were distributed randomly. Fragments of muscle were weighed and quickly placed in 15 ml of buffer A (15 mM Tris HCl, pH 7.5, 300 mM sucrose, 60 mM KCl, 15 mM NaCl, 2 mM EDTA, pH 8, 0.5 mM EGTA, pH 8, 0.5 mM spermidine, 0.15 mM spermine, 0.5 mM DTT, protease inhibitor cocktail [EDTA-free Complete; Roche Diagnostics]) for native chromatin preservation. The volumes are given for 1 g of muscle (about 10⁶ nuclei). Muscle was minced using a scalpel and a sharp forceps and homogenized with 10 supplementary ml of buffer A using an Ultra-Turrax T25 homogenizer. After a centrifugation of 10 min at 700 × g at +4°C, the supernatant of tissue remnants was discarded. The pellet was gently resuspended in 15 ml of buffer B (15 mM Tris HCl, pH 7.5, 300 mM sucrose, 60 mM KCl, 15 mM NaCl, 3 mM MgCl₂, 0.5 mM DTT, protease inhibitor cocktail) plus 0.5% NP-40 for cytoplasmic membrane lysis. After 5 min of incubation on ice, the samples were filtered through 16 layers of cheese cloth (17 threads) and centrifuged for 10 min at 700× g at +4°C to pellet the nuclei. Nuclei were washed in buffer B and quantified by trypan blue staining.

Micrococcal nuclease digestion assay of myocyte nuclei.

An average of 5 × 10⁵ freshly prepared nuclei were resuspended in 200 μl of digestion buffer (15 mM Tris HCl, pH 7.5, 60 mM KCl, 15 mM NaCl, 3 mM MgCl₂, 0.25 mM sucrose, 0.5 mM DTT, protease inhibitor cocktail). A stock solution of 5 U/μl micrococcal nuclease (MNase) (Roche Applied Science) was made in 5 mM Tris HCl, pH 7.5, 25 μM CaCl₂. One MNase digestion condition required two aliquots of 200 μl of the nucleic suspension. Aliquots were preheated for 2 min at 37°C. MNase was added to nuclei at a range of 0 to 50 U/ml. The nucleus suspension was incubated at 37°C for 1 to 15 min in the presence of 10 mM CaCl₂, and the reaction was stopped by adding 800 μl of a stop solution (15 mM EDTA, pH 8, 0.625% SDS). The samples were then deproteinized by an overnight incubation at 37°C in the presence of proteinase K (Eurobio) at a final concentration of 40 μg/ml. The purification of DNA was performed by diluting samples in TNM buffer (10 mM Tris HCl, pH 7.5, 10 mM NaCl, 3 mM MgCl₂), followed by sequential phenol (pH 8)-chloroform extraction and isopropanol precipitation in the presence of sodium acetate and glycogen. The DNA pellet was washed with 70% ethanol and resuspended in a Tris-EDTA (10:1) overnight at room temperature. Twenty micrograms of each MNase digestion condition mixture was run on a 1.4% agarose gel containing ethidium bromide for 24 h at 40 V at +4°C in Tris-borate-EDTA (TBE) buffer. The gel was photographed under UV light. After transfer under denaturing conditions, the membrane was hybridized with a radiolabeled probe specific for the WPRE-pA vector region, stripped with 0.5% SDS, and hybridized with the RSV promoter probe. The same membrane was finally stripped with 0.5% SDS and hybridized with the genomic ³²P-labeled probe cmEPO. The size of the mononucleosome was determined from the following digestion conditions: 20 U/ml for 5 min at 37°C. The average unit length is calculated from the 1-, 2- and 3-mer fragments on each blot after rAAV- or cell-specific hybridization.

Treatment with an HDAC inhibitor.

Mac 9 received sodium phenylbutyrate (Ammonaps), an HDAC inhibitor, administered per os for 2 months at a dose of 540 mg/kg/day and then for 1 month at a dose of 1 g/kg/day. Sodium phenylbutyrate was started 5.5 months after vector administration and fractionated in three equivalent doses over the course of a day, with careful intake monitoring. Serum was collected under ketamine-induced anesthesia, and the LEA29Y level was measured by an enzyme-linked immunosorbent assay described in Toromanoff et al. (60).

RESULTS

Long-term rAAV-mediated protein expression.

We previously published 2 year's worth of expression data on NHPs that received an rAAV Tet_ON promoter-regulated EPO expression vector intramuscularly (11). In Fig. 1, we show that the AAV1-cmET-Des-M2 vector genome is functionally stable for at least 5 years; these results are similar to those of another study using the rapamycin-inducible system expressed from an AAV vector delivered intramuscularly to NHPs (51). In the present study, we evaluate the structure of rAAV vector genomes at late time points in five primates that were administered a single-stranded rAAV vector of serotype 2/1 (macaques Mac 1, Mac 2, and Mac 9) or serotype 2/8 (Mac 5 and Mac 6) in their tibialis anterior muscles as described in a recent study (60). The vector encodes LEA29Y (belatacept), a mutant form of the human immunosuppressive molecule CTLA4-Ig (abatacept) under the control of the constitutive RSV promoter, followed by the WPRE sequence. The immunosuppressive properties of LEA29Y have been demonstrated in a preclinical primate model of renal allografting (32) and in patients, in whom it prevented rejection after renal transplantation (63). The AAV-RSV-LEA29Y-WPRE-pA-injected animals exhibited the classic expression profile, characterized by an increase in the first 3 months, followed by a slower decrease and a steady state achieved from 6 months on (60).

To evaluate in NHP skeletal muscle the status of the rAAV vector in the steady-state phase, injected muscles were surgically excised 10 to 22 months after injection and total DNA was prepared for analysis. The average number of vector genome copies per diploid cellular genome at the injected site was 34.3 ± 2.5 copies or 8.3 ± 1.5 copies for the NHP that received the AAV1 or the AAV8 vector, respectively (data not shown).

rAAV genomes exist as monomers and HMW concatemers in NHP muscle.

In order to assess the molecular status of rAAV, we first performed a Southern blot analysis using several restriction enzymes and a vector-specific probe (WPRE-pA). We determined that the rAAV genomes persist in monomers and high-molecular-weight (HMW) concatemers (Fig. 2, lane 1). Monomers were observed in a relaxed or supercoiled form. Supercoiled monomers were characterized by further experiments described below. When digested with a single cutter (EcoRI) (Fig. 2, lane 3), the HMW forms were converted to head-to-head or head-to-tail junction fragments. The 2,911-bp fragment generated from linear molecules corresponds to a very weak signal, suggesting that, several months following transduction, the rAAV genomes are mainly circular. When nondigested DNA (Fig. 2, lane 1) is run on a classical agarose gel, a portion of the vector signal remains trapped in the well by chromosomal DNA. The migration of this signal into the gel is facilitated when the cellular chromosomal DNA is digested with XbaI, an enzyme that does not cut in the rAAV genome (Fig. 2, lane 2). Furthermore, the trapping phenomenon is prevented when the DNA is separated on a 0.5% multipurpose agarose gel, which is capable of better separation of HMW DNA (Fig. 3). Using several enzymes that cut more or less frequently in the primate genome and do not cut in the rAAV genome, we found that the Southern blot profile remained unchanged, suggesting that the majority of rAAV genomes do not integrate into the host cellular genome (Fig. 3).

FIG. 2.

Persistence of rAAV as monomers and as HMW concatemers in skeletal primate muscle. (A) Schematic diagram indicating the enzyme restriction sites and the position of the rAAV-specific probe (WPRE-pA probe) used for the Southern blotting. (B) Southern blot of total DNA extracted from noninjected (primate Mac C1) and injected (Mac 2) tibialis anterior muscles. Twenty micrograms of total DNA purified from primate skeletal muscle were incubated either in the absence of any restriction enzyme (lane 1) or with XbaI (no cut within the vector) (lane 2), EcoRI (one cut within the vector) (lane 3), or BamHI-ClaI (two cuts in the vector) (lane 4). H-H and H-T indicate, respectively, head-to-head or head-to-tail junctions. The 2,911-bp fragment can be generated after EcoRI digestion in the presence of linear (lin) molecules (lane 3). Viral DNA forms are indicated by black arrows. conc, concatemers; sc-mon or r-mon, supercoiled or relaxed monomers, respectively. The HMW-vector-specific signal (open arrowhead) visualized after XbaI digestion (lane 2) was identified further as HMW concatemers (Fig. 3).

FIG. 3.

AAV vector genomes exist as low-molecular-weight and HMW episomes. Twenty micrograms of total DNA extracted from injected primate muscle (Mac 1) were submitted to digestion with several noncutting enzymes: XbaI (4 kbp; lane 2), AvrII (5 kbp; lane 3), and SpeI (8 kbp; lane 4) or were nondigested (lane 1). The average genomic fragment size (kbp) generated by each endonuclease is based on human genome data. Samples were loaded on a multipurpose 0.5% agarose gel intended for HMW DNA separation. After transfer, the membrane was hybridized with the ³²P-labeled rAAV-specific probe (WPRE-pA probe). The open arrow indicates two discrete bands identified as HMW concatemers (HMW conc). sc-mon or r-mon, supercoiled or relaxed monomers, respectively.

rAAV vector genomes exist as episomal circles in NHP muscle.

Schnepp et al. (53, 54) performed molecular analysis on rAAV genomes following transduction in murine skeletal muscle and on wild-type AAV genomes in human tissues. Using the PS-DNase and sensitive PCR assays, they convincingly showed that >99.5% of the rAAV genomes assimilate into episomal circular forms in murine skeletal muscle. Here, purified total DNA from four transduced NHPs was treated with PS-DNase, an exonuclease that hydrolyzes all DNA molecules except circular dsDNA. Additionally, genomic DNA was treated with or without XbaI prior to PS-DNase treatment to ensure digestion of cellular DNA. The products were separated on an agarose gel and probed with a vector-specific WPRE-pA probe. As can be seen in Fig. 4, several PS-DNase-resistant vector-specific bands that represent circular supercoiled and relaxed monomers and circular HMW forms are observed. Monomeric and some concatemeric forms remain even after extensive digestion, i.e., XbaI digestion followed by two 16-hour PS-DNase digestions (Fig. 4A, lane 6), conditions that exceed the time needed to hydrolyze the single-stranded viral DNA, the double-stranded linear DNA (53), and all cellular DNA (Fig. 4C, lane 6). Indeed, when the blot was stripped and probed with a cmEPO probe to detect the endogenous EPO cellular gene, PS-DNase efficiently hydrolyzed the cellular chromosomal DNA, since no EPO signal is detected after two rounds of PS-DNase digestion (Fig. 4B, lane 6). As shown in Fig. 4A, the majority of the rAAV genome signal corresponds to circular monomers as well as some HMW species that are completely resistant to PS-DNase treatment and thus are not associated with the cellular chromosomal DNA. However, we observed that some HMW forms are sensitive to PS-DNase, suggesting (i) the occurrence of concatemer integration in the host genome, (ii) the presence of linear concatemers, or (iii) damage to the HMW episomal molecules during DNA extraction. Nevertheless, linear molecules seem to be a minority, as indicated by the Southern blot data (Fig. 2, lane 3).

FIG. 4.

AAV vector genomes exist as episomal circles. Twenty micrograms of total DNA extracted from injected skeletal muscle (Mac 2) were submitted to digestion with PS-DNase with or without previous digestion with XbaI, which does not cut in the vector. (A) The Southern blot hybridized with an rAAV vector-specific probe is shown. The hydrolysis of rhesus chromosomal DNA by the PS-DNase enzyme was verified by ethidium bromide (EthBr) staining (C) and confirmed after hybridization of the same blot with the cmEPO probe, which is specific for the endogenous cmEPO gene sequence (B).

To confirm that the majority of the vector genomes are episomal and not integrated into cellular chromosomes, we further analyzed DNA derived from the skeletal muscle of five treated NHPs for the presence of vector-cell genome junctions by LAM-PCR (52). We previously demonstrated the potency of our approach in efficiently detecting and sequencing integrated AAV vector forms on transduced human 293 cells (69). Here, the sequencing of a total of 765 LAM-PCR amplicons from five NHPs failed to detect any vector-cell genome junctions (Table 1). Instead, rAAV LAM amplicons predominantly resembled concatemeric variants in the head-to-tail orientation, showing complex deletions and rearrangements of the ITR sequences and the characteristic multiplicity of vector sequence rearrangements previously described by others (40, 64). In addition, massive parallel pyrosequencing of rAAV LAM amplicons reaching a total of approximately 150,000 sequence reads from five macaques again revealed the presence of dominant concatemeric vector forms present in the skeletal muscle of the examined NHPs (data not shown). Our results obtained with NHPs extend those of Schnepp et al. following the transduction of rodent skeletal muscle (53). In summary, these results demonstrate that rAAV genomes are maintained predominantly as episomal circular forms in NHP skeletal muscle.

TABLE 1.

Sequenced LAM-PCR amplicons from NHP skeletal muscle^a

Macaque	rAAV serotype	Time postinjection (mo)	No. of rAAV LAM amplicons	No. of rAAV-cellular junctions	Head-to-tail concatemers	Head-to-head concatemers
Mac 1	2/1	22	283	0	+++	+
Mac 2	2/1	21	130	0	+++	+
Mac 9	2/1	1.5	103	0	+++	+
Mac 5	2/8	12	122	0	+++	+
Mac 6	2/8	10	127	0	+++	+

^aDNA sequencing of rAAV amplicons containing AAV ITR sequences showing predominantly (+++) head-to-tail configurations or, rarely (+), head-to-head amplicons.

Episomal rAAV genomes are in a nucleosomal structure.

To correlate the persistence of rAAV in vivo and its molecular structure, we wanted to determine if the episomal viral forms were assimilated into chromatin. Nuclei were isolated from freshly excised NHP skeletal muscle that was transduced for 10 to 22 months and treated with MNase. MNase is an endonuclease that hydrolyzes native chromatin in the linker DNA region between the nucleosomal cores to reveal the nucleosomal pattern (2). A titration of MNase was capable of digesting cellular chromatin into a typical nucleosomal ladder, as revealed by ethidium bromide staining (Fig. 5A) and when probed with the endogenous cmEPO probe (Fig. 5C). When the gel was probed with the AAV vector-specific WPRE-pA probe (prior to the cmEPO probe), a nucleosomal ladder was observed in Mac 1, the injected primate, but not in Mac C2, the nontransduced control animal (Fig. 5B). The major persistent rAAV forms in NHP skeletal muscle, i.e., the supercoiled monomeric and HMW concatemeric circles, were found to be susceptible to MNase digestion. However, two particular forms identified as open-circular and linear monomers were less sensitive to MNase before being digested into single nucleosomes (Fig. 5B). At the highest MNase concentration, a distinct nucleosomal pattern was observed, demonstrating that all forms of rAAV are associated with histones on the target sequence. Analysis of additional NHPs revealed that this initial observation was extended to all individuals analyzed, independently of the rAAV vector serotype (rAAV1 or rAAV8), but not in the nontransduced control animals (Fig. 6 and data not shown). Using a vector-specific RSV promoter probe, we demonstrated that the vector promoter region in the rAAV episomes is also associated with histones (Fig. 6C). Analysis of the nucleosomal unit size was done after an intermediate MNase condition treatment where mononucleosomes, dinucleosomes, and trinucleosomes are visualized. It revealed a unit length of 170 ± 10.4 bp for the cellular chromosomal DNA (cmEPO probe) (Fig. 6D) and of 162 ± 8.8 bp after hybridization with the 867-bp, vector-specific WPRE-pA probe (Fig. 6B). Similarly, the RSV probe that corresponds to a 646-bp sequence in the RSV promoter showed an MNase pattern where nucleosomes were regularly spaced with a unit length of 159 ± 11.3 bp (Fig. 6C). All together, these data reveal that cellular histones are regularly positioned along the rAAV episomes, creating a chromatin-like structure that is similar to the cellular chromosomal DNA nucleosome pattern (EPO probe) (Fig. 5C and 6D).

FIG. 5.

Episomal rAAV genomes are in a chromatin-like structure. An MNase digestion assay was performed on nuclei isolated from an injected primate muscle (Mac 1) more than a year after rAAV administration of serotype 2/1 AAV RSV-LEA29Y-WPRE-pA particles. Muscle biopsies were performed on a noninjected macaque (Mac C2), which was used as a negative control. Isolated nuclei were treated with increasing concentrations of MNase (0 to 50 U/ml, with “no” indicating no MNase) at 37°C and deproteinized, and DNA fragments were purified. After electrophoresis on a 0.8% TAE-agarose gel, the MNase pattern of the digested native chromatin was checked by ethidium bromide staining (A). Following transfer to the membrane, the vector-specific signal was detected by hybridization with the ³²P-labeled WPRE-pA probe (B). The blot was stripped and hybridized with the ³²P-labeled cmEPO probe specific for the endogenous cmEPO gene sequence (C). rAAV forms are indicated as HMW concatemers (HMW conc), open-circular monomers (oc-mon), linear monomers (lin-mon), and supercoiled monomers (sc-mon). Black arrows indicate the mono-, di-, tri-, and tetra-nucleosomes visualized after MNase digestion and characteristic of a classical chromatin pattern.

FIG. 6.

Different regions of the rAAV vector genomes are in a chromatin-like structure. An MNase digestion assay was performed as described in the legend to Fig. 5. Equivalent doses of serotype 2/1 (Mac 1 and 2) or serotype 2/8 (Mac 6) AAV RSV-LEA29Y-WPRE-pA particles were administrated by intramuscular injections. Muscular biopsies were performed on a noninjected macaque (Mac C2), which was used as a negative control. Isolated myocyte nuclei were treated with increasing concentrations of MNase (0 to 50 U/ml, with “no” indicating no MNase) to digest native chromatin, and the chromatin fragments were purified. After electrophoresis on a 1.4% TBE agarose gel, the MNase profile of the digested chromatin was checked by ethidium bromide staining (A). Following transfer to the membrane, the vector-specific signal was detected by hybridization with the ³²P-labeled WPRE-pA probe (B). The blot was stripped and hybridized with a vector-specific RSV promoter probe (C). Lastly, the blot was hybridized with the ³²P-labeled cmEPO probe specific for the endogenous cmEPO gene sequence (D). Black arrows indicate the mono-, di-, and trinucleosomes visualized after MNase digestion and characteristic of a classical chromatin pattern.

HDAC inhibitor does not affect transgene expression.

Sodium phenylbutyrate is used clinically to treat sickle cell anemia patients by stimulating the expression of fetal hemoglobin, which undergoes programmed inactivation during development (18, 19, 28). As inhibition of HDAC was shown to increase the transgene expression mediated by rAAV in vitro (46), we wanted to assess the effect of such a drug on rAAV-mediated expression in NHPs. Mac 9, which received rAAV2/1 RSV-LEA29Y-WPRE-pA, expressed LEA29Y for more than 11 months in the classic pattern; transgene expression increased during the first 3 months, followed by a decrease until it reached a stable level 6 months following rAAV delivery (Fig. 7). We administered sodium phenylbutyrate, an HDAC inhibitor, per os three times a day for 2 months at a medium dose (540 mg/kg/day) and then for 1 month at a higher dose (1 g/kg/day). The two doses were well tolerated by the primate; however, transgene expression remained unchanged after sodium phenylbutyrate treatment (Fig. 7). This result suggests that the episomal rAAV forms that persist in skeletal muscle are associated with cellular histones that are positioned or modified in a manner that allows transgene expression.

FIG. 7.

Transgene expression mediated by rAAV cannot be modulated by an HDAC inhibitor. Mac 9 injected with AAV2/1 RSV-LEA29Y-WPRE-pA was treated by sodium phenylbutyrate at 5.5 months postinjection (black arrowhead) at a dose of 540 mg/kg/day and for 2 months. The treatment was prolonged for one supplementary month at a dose of 1 g/kg/day (gray arrowhead) and stopped (open arrowhead). The serum concentration of LEA29Y was measured by an enzyme-linked immunosorbent assay.

DISCUSSION

The present study (Fig. 1) and previous work (51) show the persistence of functional rAAV genomes for several years after one single IM administration in NHPs. Persistence appears to be independent of the immunosuppressive LEA29Y transgene used here since previous long-term studies of NHPs have utilized a constitutively expressed tetracycline regulator, a rapamycin regulator, and EPO (11, 51). Therefore, there must be a mechanism by which episomal and nonreplicating vector genomes associate with nuclear components to achieve stability and functionality in quiescent tissues. In this paper, we provide evidence that in a nondividing cell population such as normal skeletal muscle, rAAV episomal genomes are remarkably stable and persist mainly as supercoiled monomeric and concatemeric circles and, most notably, in a chromatin-like structure.

As determined by pulsed-field gel electrophoresis, fluorescent in situ hybridization, isolation of vector-cellular DNA fragments, and in vivo selection of hepatocytes, rAAV vectors can integrate in vivo in the liver (8, 16, 40-42), although extrachromosomal rAAV genomes represent ∼90% of the vector genomes and are primarily responsible for stable liver transduction (43). Hepatectomy-induced liver regeneration shows that in specific experimental settings, rAAV genomes are diluted and do not persist (26), similar to the dilution seen in neonatal mice (66). It seems that no-end, linear, double-stranded rAAV genomes are more easily degraded and lost by hepatocyte division than circular and concatemeric genomes after rAAV administration in mice (30). In contrast, no integration event was detected in murine muscle tissue (53). Indeed, based on the demonstrated sensitivity of the B1-PCR assay complemented by a PS-DNase digestion assay, Schnepp et al. concluded that episomes correspond to more than 99.5% of vector DNA (53). Additional work showed that the majority of the rAAV vector DNA in muscle persisted over time as episomes (21, 57, 64). Moreover, circular rAAV genomes have also been detected in humans who received an rAAV-FIX vector intramuscularly (35). Several studies of muscle suggested that extrachromosomal, circular rAAV dsDNA genomes are likely responsible for long-term transgene expression (20, 53, 68). Altogether, our molecular analysis confirms that circular rAAV episomes drive transgene expression in primate skeletal muscle. This finding is supported by the fact that LAM-PCR and high-throughput sequencing (data not shown) analysis failed to detect integration events in NHP skeletal muscle. The combinatorial approach of high-throughput LAM-PCR and massive parallel pyrosequencing, which is dispensable for bacterial subcloning steps (29, 40), point to a very low integration frequency of rAAV vectors in NHP skeletal muscle and confirm that the rAAV genomes found to be associated with histones are predominantly present as episomal forms. We cannot exclude the possibility of the presence of rare integrated forms following the transduction of NHP skeletal muscle, as was seen in mice by Inagaki et al. (29). Nevertheless, our analyses demonstrate that the majority of the rAAV genomes are episomal. Further experiments are needed to give exact estimations of the integration frequency in diverse rAAV-transduced tissues from clinically relevant studies, as well as to carefully consider dose-dependent influences.

The concept of a viral minichromosome was formulated to describe the organization of the circular SV40 genome with cellular histones in the nuclei of infected permissive cells (2, 13, 25). The linear adenovirus genome has been shown to be complexed with the virus-derived histone-like protein VII, which condenses the 36,000-bp adenoviral genome in the virion, but this protein is released from the viral DNA for chromatin assembly a few hours after infection (59). The parvoviruses AAV and MVM were also described as organized in chromatin structure within hours after infection in cell culture (3, 36), and in the case of wild-type AAV did not require the presence of a helper virus or DNA replication. In all cases, the presence of viral nucleosomes is believed to be an essential step in the life cycle of the wild-type virus, which prompted us to determine whether rAAV genomes are arranged in situ like minichromosomes. As shown here, our MNase digestion assay data indicate that the rAAV vector genome, devoid of sequences coding for viral proteins, is a substrate for chromatization, with a structure that is similar to the structure of cellular chromatin. This was further confirmed when histones were removed from the episomal rAAV vector genomes during DNA extraction to yield negatively supercoiled circular molecules that were relaxed by Escherichia coli topoisomerase I (data not shown). Indeed, removal of histones from DNA during extraction produces negative supercoiling (45, 24). DNA supercoiling could also be a mark of transcription activity (31).

The fact that histones are complexed with episomal circular plasmid following naked plasmid injection in vivo, which in turn regulates transgene expression via epigenetic modifications (50), suggests that DNA replication and/or viral sequence motifs are not a prerequisite for the generation of rAAV chromatin, at least in quiescent tissues and in vitro (6). Furthermore, the generation of circular rAAV genome intermediates is responsible for long-term episomal persistence in muscle tissue (21). In our study, it appears that circularization and chromatinization of the rAAV genomes may be critical for in vivo persistence in a quiescent tissue, mechanisms that are conserved with other viral genomes maintained during latency. However, we do not exclude the possibility that some gene expression may originate from a minority population of linear or naked circular molecules. The unique aspect of episomal rAAV vector chromatin is that it remains in a configuration that allows high-level, long-term gene expression.

As we demonstrated that double-stranded rAAV genomes are a substrate for chromatinization in vivo, transgene expression could be regulated by histone epigenetic modifications. In the cell lines stably harboring a single integrated vector, transgene expression can be reactivated following the treatment of cells with HDAC inhibitors (9, 10). These in vitro studies do not reflect the status of the rAAV vector DNA in vivo. In this report, we did not analyze the status of the rAAV vector genomes at early time points following transduction, so it is unclear when histones associate with rAAV episomes in vivo. It has been shown that within a few weeks postinjection in the skeletal muscle and liver, the number of rAAV genomes is dramatically reduced (64, 65). This loss of vector genomes may be due to degradation of virions by the proteasome following rAAV entry into the target cell (17). Alternatively, processing of rAAV genomes has been shown to be dependent upon DNA damage/repair pathways (7, 12). While most vector genomes are lost, a minority are converted to stable, double-stranded circular and concatemeric forms. It is possible that most genomes do not participate in these pathways or are assimilated into nonproductive structures that are degraded by cellular nucleases. The clearance of most of the vector DNA is concomitant with the rise of transgene expression levels. Therefore, less than a month after intramuscular administration, only a few genomes are stabilized and established as transcriptionally active. Interestingly, this time frame is similar to the one necessary in vivo for naked plasmid DNA to associate with histones, resulting in hetero/euchromatin (50). Our data indicate that chromatinization occurs on episomal double-stranded circular rAAV when such a substrate is available.

The conversion of linear AAV genomes to double-stranded supercoiled circles has been seen in vivo for (i) single-stranded AAV vectors, (ii) double-stranded AAV vectors that are also called self-complementary rAAV, (iii) multiple serotypes (AAV1, -2, and -8), (iv) different target tissues (skeletal muscle, liver, heart), (v) animals of different levels of maturity (neonates and adults), (vi) early and late time points posttransduction, (vii) different vector copy numbers, and (viii) different species (mice, NHPs) (references 44, 64, 65, and 66 and data presented here). In skeletal muscle, cellular histone assembly on rAAV genomes could trigger stable transcription of the transgene from monomeric and concatemeric circles and alternatively may catalyze and/or protect the initial structural changes, including the slow conversion of double-stranded monomers to larger concatemers (21). Several histones are known to be implicated in the DNA repair machinery, such as H2AX and H1R variants (22, 27, 47), or in intermolecular recombination (22, 58) that could favor rAAV circularization or concatemerization. Further studies are required to address the existence and the nature of such a relationship between early changes of rAAV molecular structures and the chromatinization of rAAV genomes.

Changes in the dynamic chromatin structure are notably affected by posttranslational modifications of histones via alteration of their interaction with DNA and nuclear protein. Histone modifications act in diverse biological processes, such as gene expression regulation, DNA repair, and DNA condensation. Unlike in the in vitro studies using transformed cell lines where rAAV-mediated gene expression can be silenced due to integration into the cellular genome in a repressive chromatin environment and restored by HDAC inhibitors (9, 10), in our in vivo studies, the episomal chromatin structure of rAAV allows constitutive expression that can be sustained for years. Consequently, we wanted to determine if some persistent forms of rAAV are susceptible to epigenetic regulation through the use of an HDAC inhibitor (sodium phenylbutyrate). The literature does not clearly demonstrate the effect of such drugs on rAAV-mediated transgene expression. Indeed, using hydroxyurea, a cell cycling drug that is shown to act on histone acetylation, Lo et al. were unable to enhance transgene expression in the brains of mice transduced with rAAV (33). With our experimental settings, we did not manage to modify the LEA29Y expression level by a 3-month sodium phenylbutyrate treatment, nor did Rivera et al. in some primates transduced intramuscularly with rAAV (51). The inefficiency of sodium phenylbutyrate treatment on transgene expression could be due to instability and low retention of this drug (39). Nevertheless, sodium phenylbutyrate is used in clinical studies to treat sickle cell anemia patients by stimulating the expression of fetal hemoglobin (18, 19, 28) or to increase SMN gene expression in spinal muscular atrophy patients (4). As mentioned in references 3 and 48, epigenetic drugs may therefore be effective only at early times after rAAV injection, i.e., during or just after (i) the establishment of dsDNA circles, (ii) the organization of episomes into a chromatin-like structure, and (iii) appropriate intranuclear compartmentalization. If this is the case, it would provide a possible explanation for the unresponsiveness of the rAAV genome to the sodium phenylbutyrate treatment at late time points when stable vector structures have already been established.

In summary, we have demonstrated that rAAV vector genomes persist in extrachromosomal circular forms in the skeletal muscles of NHPs and that these episomes assimilate into active chromatin that mediates long-term transgene expression. In future studies, we will examine the epigenetic regulation of rAAV genomes in situ.