INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Adeno-associated virus (AAV) type 2 is a nonpathogenic, replication-defective parvovirus that is dependent on superinfection with helper adenovirus for efficient replication (4). The genome is a 4.7-kb single-strand DNA containing coding sequences for nonstructural proteins (Rep78, Rep68, Rep52, and Rep40), structural proteins (VP1, VP2, and VP3), and two 145-bp inverted terminal repeats (ITRs) at either end. The ITRs play essential roles in DNA replication, packaging, and chromosomal integration of the genome. An interesting aspect of AAV is its site-specific integration into a defined locus, called AAVS1, on q13.3 of human chromosome 19 (17). This feature makes recombinant AAV attractive for use as a vector in human gene therapy (5, 24, 25, 42) because vectors capable of targeted gene integration decrease the chance of insertional mutagenesis caused by random integration. A key intermediate step in the integration of AAV appears to be the formation of a complex composed of a large Rep protein (Rep68 or Rep78), ITR, and AAVS1 (41). Unfortunately, this means that in their current form, AAV vectors lack the capacity for site-specific integration because the rep genes are deleted when the therapeutic genes are inserted into the vector. One simple approach to supplying the necessary Rep proteins would be cotransfection of target cells with plasmids encoding AAV and Rep. However, large Rep proteins are cytostatic and/or cytotoxic when constitutively expressed in eukaryotic cells (43), which means that Rep expression must be regulated so that it is only transient.

A variety of methods aimed at accomplishing regulated expression of transgenes have been developed (11). Among them, Cre-mediated recombination has recently been used to accomplish gene activation and inactivation in transgenic mice (40) and in various cultured cells (1, 13, 31). Cre, a bacteriophage P1 recombinase, mediates site-specific recombination between pairs of loxP sites. The loxP element consists of two 13-bp inverted repeats separated by an 8-bp spacer region (10). Cre-mediated recombination between loxP sites in a direct repeat results in excision of the intervening DNA as a circularized molecule (15). Here, we describe a novel transient-expression system based on Cre-loxP recombination. With this system, a transferred gene is activated by Cre recombinase but is expressed only from the circularized episomal form. When the circularized form is linearized, the functional expression unit is disrupted so that it is not stably expressed. We show that transient expression of Rep using this system can support targeted gene integration.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Plasmid construction. The plasmid containing the complete AAV genome (psub201) and the AAV packaging plasmid (pAAV/Ad) have been described previously (32). The Rep expression plasmid, pP5rep, was constructed by removing the ApaI fragment containing the Cap coding region (AA2CG; 2943 to 4040 nucleotides [nt]). The pCALNLw (a generous gift from I. Saito of the Institute of Medical Science, University of Tokyo) contains the CAG promoter (the cytomegalovirus [CMV] immediate-early enhancer and the modified chicken -actin promoters) (26), the Neo^r gene flanked by the two loxP sequences, and the SwaI cloning site (33). The pALRPL was constructed by inserting the p5 promoter (AA2CG; 185 to 315 nt from psub201), the Rep78 coding sequence (AA2CG; 316 to 2194 nt from psub201), the AAV polyadenylation signal fragment (AA2CG; 4214 to 4488 nt from psub201), and the loxP elements into pGEM7Zf(+) (Fig. 1). The relative location and orientation of each component are shown in Fig. 1. The simian virus 40 polyadenylation signal and the loxP sequence were excised from pCALNLw and inserted upstream of the rep78 sequence. The p5 promoter was inserted downstream of the rep78 sequence. The synthetic loxP oligonucleotides were finally inserted downstream of the p5 promoter. The AAV vector plasmid pXF/sub contained the herpes simplex virus-thymidine kinase (TK) promoter-driven Neo^r gene, the CMV promoter-driven alkaline phosphatase cDNA, and ITRs at either end of the tandem expression units. The Cre recombinase expression plasmid, pxCANCre, contained the CAG promoter.

View larger version (23K): [in this window] [in a new window]   FIG. 1.   (A) Structures of Rep expression vectors. (B) Schematic representation of the regulated expression system based on Cre-loxP recombination. In pALRPL, Rep expression is silent because the p5 promoter is located downstream of the rep coding sequence. However, after incubation with Cre recombinase, the p5 promoter is linked to the 5' end of the rep78 coding sequence, resulting in induction of Rep78 expression. Open arrows, PCR primers. (C) Inactivation of Rep78 expression after integration. When the circular expression unit is integrated into the genomes of the host cells, expression of the complete form of Rep78 is terminated. SV40, simian virus 40.

Cell culture and transfection. 293 cells, a human embryonic kidney cell line, were grown in Dulbecco's modified Eagle's high-glucose medium supplemented with nonessential amino acids, 10% heat-inactivated (30 min at 56°C) fetal calf serum, and 100 U of penicillin and 100 µg of streptomycin per ml at 37°C under an atmosphere of 5% CO₂-95% air. Cells grown in a monolayer were transfected with plasmid DNA using either the calcium phosphate procedure (10 to 20 µg) (37) or lipofection with cationic liposomes (LipofectAMINE reagent, 10 to 12.5 µg; GIBCO BRL, Gaithersburg, Md.) (39). Forty-eight hours after transfection, the cells were split and replated for selection in medium containing G418 (1 mg/ml, active; GIBCO, Grand Island, N.Y.). After 14 to 28 days of selection, well-isolated colonies were harvested and expanded for further analysis.

Analyses of DNA, RNA, and proteins. Approximately 10⁶ cells were resuspended in 200 µl of TE (10 mM Tris-Cl [pH 7.5], 1 mM EDTA) and then incubated overnight at 37°C with 200 µl of 0.1% proteinase K in buffer containing 10 mM Tris-Cl (pH 7.5), 1% sodium dodecyl sulfate, and 10 mM EDTA. After phenol-chloroform extraction, the DNA was ethanol precipitated and dissolved in 100 µl of TE. Total RNA was isolated using an RNeasy total RNA kit (Qiagen, Inc., Santa Clarita, Calif.), following the procedure recommended by the manufacturer. Southern and Northern analyses were performed (6) using a specific rep78-coding probe corresponding to nt 188 to 814 of the wild-type sequence of AA2CG (GenBank). The AAV rep78 probe was a 630-bp XbaI-SacI fragment from psub201 (32). An anti-Rep polyclonal antibody was prepared in rabbits using a Rep-maltose binding protein fusion protein containing the second proline to the terminal glutamine of Rep78, which was synthesized using a protein fusion and purification system (New England Biolabs, Beverly, Mass.). Proteins were extracted from transfected cells, separated by 5 to 20% (wt/wt) gradient pore sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and analyzed by Western blotting using the rabbit polyclonal antibody.

PCR primers and conditions. The circularized Rep expression unit from pALRPL was detected by PCR using the sense primer CIR1 (5'-TGTGGTCACGCTGGGTATTT-3') and antisense primer CIR2 (5'-TTCTCTTTGTTCTGCTCCTG-3'). The amplification protocol consisted of 30 cycles of 30 s at 94°C, 30 s at 50°C, and 1 min at 72°C. The 3' AAV plasmid/chromosome 19 junction was amplified by nested PCR. The sense primer for the first PCR step (PT-1; 5'-AGTAGCATGGCGGGT) was located upstream of the 3'-AAV ITR within the plasmid, while the antisense primer (CR-1; 5'-CGCGCATAAGCCAGTAGAGAGCC) flanked AAVS1 on chromosome 19. The second PCR step was carried out with a sense primer for the AAV plasmid (PT-2; 5'-GGAATTCAGGAACCCCTAGTGATGG) and an antisense primer for AAVS1 (CR-2; 5'-ACAATGGCCAGGGCCAGGCAG). Using the aforementioned reaction conditions, the first PCR protocol entailed 25 cycles of 1 min at 94°C, 1 min at 55°C, and 2 min at 73°C (8). Two percent of the amplified product was then diluted into a new reaction mixture containing the second set of primers and amplified using the same protocol.

Integration site analysis. The nested PCR products were resolved on a 1.6% (wt/vol) agarose gel, stained with ethidium bromide, transferred to Hybond N+ paper (Amersham Life Science, Inc., Little Chalfont, Buckinghamshire, United Kingdom), and probed with SIC-415L, which is a previously cloned junction fragment (5'-TCAGGTTCAGGAGAGGGCAGGG-3'; antisense sequence of nt 1149 to 1170 in accession no. S51329, GenBank) labeled using a Megaprime DNA labeling system (Takara Shuzo Co., Ltd., Otsu, Japan). The resultant PCR bands were subcloned into pGEM-T (Promega, Madison, Wis.) and sequenced using the chain termination method with a Prism dye terminator cycle sequencing FS Ready Reaction kit (model 373A; PE Applied Biosystems, Norwalk, Conn.).

Fluorescence in situ hybridization (FISH). CAGSEGFP/TkneoR, an 8.6-kb plasmid containing the neomycin resistance gene cassette, was labeled with digoxigenin using a nick translation kit (Boehringer Mannheim), according to the manufacturer's instructions, and a biotin-labeled human chromosome 19-specific probe was used for chromosome analysis (biotin labeled paint no. 13 [1066-19B]; Cambio, Cambridge, United Kingdom). Chromosome spreads from selected neomycin-resistant 293 cell clones were prepared using standard cytogenetic techniques (22). Visualization of the biotin-labeled probe for chromosome 19 was carried out by repeated incubations with avidin-fluorescein isothiocyanate (FITC), biotinylated FITC, and again with avidin-FITC. The digoxigenin-labeled probe for the AAV vector was detected using mouse antidigoxigenin antibody, digoxigenin-labeled anti-mouse immunoglobulin antibody, and FITC-labeled antidigoxigenin antibody (Boehringer Mannheim). After immunodetection, slides were counterstained with propidium iodide. Photographic images were taken with a color charge-coupled device camera using Adobe Photoshop on a Power Macintosh computer (Apple).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Design of a novel, regulated expression system. To achieve regulated expression of rep, we designed a novel, transient-expression system based on Cre-loxP recombination. We constructed a plasmid, pALRPL, in which the Rep78 coding sequence (rep78) was linked to the 5' end of the p5 promoter (p5) and inserted between two loxP elements (Fig. 1A, panel 3); in this configuration (5'-loxP-rep78-p5-loxP-3'), rep78 was not expressed. However, in the presence of Cre recombinase, the sequence flanked by the loxP elements was precisely excised and self-ligated, yielding a circularized molecule in which the p5 promoter was linked to the 5' end of the rep78 coding sequence (Fig. 1B), thereby enabling its expression.

Regulated expression of the Rep protein. To confirm that Rep expression from pALRPL was regulated by Cre recombinase, 293 cells were cotransfected with pALRPL (5 µg) and pxCANCre, a Cre expression plasmid (5 µg). The DNA was then extracted from the cells, digested with or without KpnI, and subjected to Southern blot analysis using a probe specific for p5-rep. The 5.5-kb band in Fig. 2A corresponds to the full-length pALRPL, whereas the 2.3-kb band corresponds to the excised fragment containing the Rep expression unit. The structure of the isolated DNA was also analyzed by PCR (Fig. 1B). The PCR product from intact pALRPL was a 4.0-kb fragment, while that from the circularized Rep expression unit was a fragment of 0.7 kb, which was consistent with the predicted size (Fig. 2B). Thus, Cre recombinase appears to excise and circularize the Rep expression unit.

View larger version (33K): [in this window] [in a new window]   FIG. 2.   Molecular analysis of cells transfected with pALRPL and pxCANCre. (A) Southern blot analysis of DNA from 293 cells transfected with pALRPL alone (lane 1) or with pALRPL plus pxCANCre (lane 2). Total DNA was digested with KpnI and blot hybridized with the p5/rep-specific probe. The 5.5-kb band represents nonrearranged pALRPL, while the 2.3-kb band is the circularized rep expression unit. (B) PCR analysis of the circularized molecules. The structure of the rep expression unit was also confirmed by electrophoresis of PCR products. DNAs were the same as those used in Southern blot analysis. The 4.0-kb band corresponds to intact pALRPL, and the 0.7-kb band corresponds to the circularized molecules. (C) Northern blot analysis of rep expression in transfected 293 cells. (Top panel) RNA was extracted from nontreated 293 cells (lane 1) and 293 cells transfected with pxCANCre (lane 2), pALRPL (lane 3), or pxCANCre plus pALRPL (lane 4) and blot hybridized with the rep78-specific probe. The 2.0-kb band represents full-length rep78 transcripts. (Bottom panel) The ethidium bromide-stained gel from which the blot in the top panel was made. MW, molecular weight markers. (D) Western blot analysis of rep expression in transfected 293 cells. Protein was extracted from nontreated 293 cells (lane 1) or 293 cells transfected with pALRPL (lane 2), pP5Rep (lane 4), pAAV/Ad (lane 5), or pxCANCre plus pALRPL (lane 3) and subjected to Western analysis using a polyclonal anti-Rep antibody.

Rep-mediated integration of AAV vector DNA. We next examined whether regulated expression of Rep78 would support site-specific integration of AAV vector DNA. 293 cells (5 × 10⁶ cells) were transfected with pXF/sub (2.5 µg), an AAV plasmid containing the TK promoter-driven Neo^r gene, the CMV promoter-driven alkaline phosphatase gene, ITRs at either end of the tandem expression units, various Rep expression plasmids (5 µg), and pxCANCre or pUC8X (5 µg). The transfected cells were selected by culture with G418, and both individual and pooled clones were prepared. DNA was extracted from these cells, and the integration sites of the AAV vector sequence were analyzed using a PCR-based assay system for detection of integration at the AAVS1 region. Genomic DNA from pooled clones was subjected to two rounds of PCR amplification. In both reactions, one primer from the AAVS1 sequence and a second primer from the AAV vector sequence were used to amplify the junction sequence. PCR products were separated on a 1.5% agarose gel (Fig. 3A) and blot hybridized with an AAVS1 probe (Fig. 3B). DNA from 293 cells infected with wild-type AAV served as a positive control and gave a strong signal (lanes 8). On the other hand, no signal was detected from cells transfected with AAV vector plasmid pXF/sub plus pUC8X (lanes 3 and 10). We also analyzed pooled clones transfected with AAV vectors lacking rep sequences and, as expected, no AAVS1 signal was detected from these cells (data not shown).

View larger version (51K): [in this window] [in a new window]   FIG. 3.   PCR-based assay for integration at AAVS1. 293 cells were transfected with pXF/sub and various rep expression plasmids. Shown are pP5Rep (lane 4), pAAV/Ad (lane 5), pxCANCre (lane 11), pALRPL (lane 12), pxCANCre plus pALRPL (lanes 6 and 13), and pxCANCre plus pUC8X (lanes 3 and 10), nontreated 293 cells (lane 1), and HeLa cells infected with wild-type AAV (lane 8). Lanes 2, 7, 9, and 14 show molecular weight markers. Genomic DNA was extracted from pooled G418-resistant clones and subjected to two rounds of PCR (A, the ethidium bromide-stained agarose gel) followed by Southern blot analysis (B), using SIC-415L as a probe (see Materials and Methods).

View larger version (13K): [in this window] [in a new window]   FIG. 4.   FISH of Cre-loxP clone 5. Metaphase spreads were first hybridized with a chromosome 19 painting probe (A, yellow signal) and then, after stripping, rehybridized with a Neor probe (B, yellow double dots indicated by the arrow). Each panel was counterstained with propidium iodide (red).

Structure of junctions between AAV vector and AAVS1 sequences. The detailed structure of the junctions was analyzed by sequencing the amplified PCR products (Fig. 5A). In all cases examined, the breakpoints of the AAV vector DNA mapped within the ITR sequence, whereas those of the AAVS1 were within the 5' 1.5-kb fragment. These breakpoints closely resembled those identified at the junctions between the wild-type AAV provirus genome and chromosomal DNA. Except for Cre-loxP clones 5 and 18, spacer sequences of various sizes were inserted between the ITR and AAVS1 sequences. These spacer sequences appeared not to be related to either the AAV vector or plasmid sequences. These structural features of the junctions suggest that AAV vector DNA derived from double-strand plasmid DNA is integrated into the AAVS1 locus through a common pathway also used for integration of wild-type AAV (8, 16, 23).

View larger version (37K): [in this window] [in a new window]   FIG. 5.   Structure of junctions between AAV vector and AAVS1 sequences. (A) Sequences of the junctions. Uppercase letters represent the AAVS1 sequences (nucleotide numbers are from accession no. S51329, GenBank), while italic letters are the ITR sequences (numbers from PvuII site of 3'-ITR in psub201). Lowercase letters are the spacer sequences unrelated to either ITR or AAVS1. (B) Maps of junction breakpoints.

Effects of transient Rep expression on Neo^r colony formation. Finally, we compared the effects of different Rep expression systems on colony formation (Table 2). When 293 cells were cotransfected with pXF/sub and various Rep expression plasmids and then selected for G418 resistance, the number of G418-resistant colonies from cells transfected with pAAV/Ad or pP5Rep was about 25% of that generated by transfection with pXF/sub alone. Reverse transcription-PCR analysis revealed that there was no expression of Rep proteins in G418-resistant clones. Cotransfection of pXF/sub with pALRPL plus pxCANCre also decreased the number of colonies. Control experiments showed that the number of G418-resistant colonies with pALRPL alone was 87%, while that with pxCANCre alone was 64%, suggesting that while Rep52/40 proteins expressed from intact pALRPL may be slightly cytotoxic, the decline in colony formation is mainly caused by expression of Cre recombinase.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

A number of approaches to site-specific gene integration have utilized elements of AAV. One simple method is to cotransfect cells with plasmids containing the ITR and encoding Rep (2, 36), though more efficient transduction is achieved with hybrid vector systems in which ITR-flanked DNA and a Rep expression unit are inserted into either a baculoviral vector (27) or a helper-dependent adenoviral vector (30). It was also recently demonstrated that cointroduction of an ITR expression vector with purified Rep68 or Rep78 protein could support targeted gene integration (21; Y. Hirai, W. Satoh, and T. Shimada, unpublished results). In each of these examples, however, problems arise due to the cytotoxic effects of Rep proteins. Expression of the AAV rep gene inhibits cellular transformation mediated by various oncogenes (3, 14) as well as cellular proliferation assessed as a function of colony formation efficiency (reference 43 and this work). It has also been reported that Rep78 moderately inhibits DNA synthesis (43) and suppresses various Sp1-dependent promoters through direct interaction with Sp1 (9), but the precise mechanism by which Rep inhibits growth is still unclear.

In the present study, we attempted to develop a novel, regulated expression system that minimized the cytotoxic effects of Rep. Among the various strategies for regulated gene expression, we utilized the Cre-loxP system, which has been used previously to regulate gene expression in a variety of protocols. In this protocol, the activation switch consisted of a stuffer DNA flanked by two loxP elements inserted between the promoter and the coding sequence to inhibit translation. Expressed Cre recombinase removes the stuffer, activating gene expression. As an inactivation switch, a pair of loxP elements were inserted within the gene. In this case, Cre recombinase disrupted the gene, thus blocking its expression. Furthermore, if a circularized DNA fragment containing a single loxP site and a Cre expression plasmid are cotransfected, targeted insertion of the DNA fragment into a loxP site in the genome is possible (34, 38). We therefore designed a system in which gene expression was activated by Cre recombinase-mediated circularization of DNA composed of the promoter, the coding sequence, and a polyadenylation signal. This circular DNA contained the minimal essential elements required for gene expression and did not contain the replication origin. Consequently, it is highly unlikely that the gene could be integrated into the chromosome in a functional linear form, making its activity self-limiting.

In this study, we used the p5 promoter for expression of Rep. We have previously shown that the p5 promoter is weak but active in 293 cells in the absence of adenovirus infection (37). It is known that a high concentration of Rep is deleterious to cells. Since p5 is negatively regulated by Rep (20, 29), the promoter in the recombined expression cassette may also effectively be shut down, precluding further cytotoxic effects. The weak and potentially self-limiting promoter activity of p5 appears to be favorable for targeted integration.

Transient expression of Rep is particularly important for site-specific integration of genes, but as discussed above, Rep proteins inhibit cell proliferation (19, 43). In addition, stable expression of Rep, even at a low level, may induce rearrangement of the AAVS1 region (36) and excise integrated AAV vector sequences upon adenoviral infection. The half-life of the circular DNA in mammalian cells is not known, but since this expression unit does not contain a replication origin, the effects of Rep proteins should be temporally limited in dividing cells, making this expression system potentially useful for transient expression of other cytotoxic molecules.

The efficiency of site-specific integration was determined to range from 10 to 40%, which is somewhat lower than efficiencies reported in earlier studies. For example, the efficiency achieved with wild-type AAV was 68 to 82% (18, 35), while 40 to 75% of clones transfected with AAV plasmids along with Rep expression units contained provirus at AAVS1 (27, 36). It is important to recognize, however, that site-specific integration was assayed in those previous studies using genomic Southern analysis or FISH, whereas we used a PCR-based assay that may have underestimated the true integration efficiency. Recent studies have shown that the junction sites in both the AAVS1 and AAV genome ITRs are quite heterogeneous (7, 36). Furthermore, the viral junction is often not within the ITR sequence but is instead at an internal site. Consequently, depending on their actual structure, the junctions may not be detected by the primer sets we used, or the products may be too long for PCR amplification. In addition, the sequence near the junction is often highly variable (36). In separate experiments, we found that following integration, mutations introduced during the replication-mediated recombination process occurred at 5 to 7 of the 27 nucleotides in the sequence corresponding to the AAVS1 hybridization probe, resulting in no signal being detected by our PCR-based assay.

Although Rep78 expressed in our transient-expression system supported site-directed integration of the gene, the overall transduction efficiency was low. Several scenarios might account for this impaired colony formation. For instance, cytotoxic effects of Rep78 early on may be sufficient to seriously inhibit cell growth, and/or Rep52 and Rep40 may negatively affect cell proliferation. It has been reported that Rep52 modestly inhibits various promoter activities (12) and adenovirus replication (30). Since the internal promoter p19 is active in pALRPL, Rep52 and Rep40 are constitutively expressed, regardless of the structure of the Rep expression unit. Furthermore, we unexpectedly found that Cre recombinase also decreased the efficiency of colony formation. While the mechanism of this antiproliferative effect is unknown, cell lines stably expressing Cre have been established, making it likely that its cytotoxic effects are not very strong (28). On the other hand, the activity of the CAG promoter used for Cre expression is very high in 293 cells (26), making it probable that high concentrations of Cre recombinase induce nonspecific recombination within cellular genomes.

We conclude that site-specific integration based on AAV components is a potentially useful approach to targeted gene therapy. Before such a system can be used in a clinical setting, however, it is essential that complete understanding of the functions of Rep proteins be achieved and then applied to the development of a precisely regulated gene expression system.