Adenovirus Vector

The usefulness of adenovirus type 5 as a vector for homologous recombination was examined in CHO cells by using the adenine phosphoribosyltransferase (aprt) gene. Infection of a hemizygous CHO APRTcell line containing a 3-bp deletion in exon 5 of the aprt gene with a recombinant adenovirus containing the wild-type gene resulted in restoration of the APRT+ phenotype at a frequency of 10'to 10-6 per infected cell. A relatively high frequency (approximately 6 to 20%Yo) of the transductants appears to result from a homologous recombination event. The mutation on the chromosomal aprt gene is corrected in the homologous recombinants, and APRT expression is restored to a normal hemizygous level. Neither adenovirus nor exogenous promoter sequences are detected in the homologous recombinants. The remaining transductants result from random integration of the aprt gene with the adenovirus sequence. A number of adenovirus vectors containing different promoter sequences linked to the hamster aprt gene were constructed. A possible role for the promoter region in the homologous recombination event was indicated by the lack of homologous recombination in constructs lacking an active promoter.

The use of viral vectors as gene transfer agents is an area of increasing interest. A number of successful viral vectors for use in human gene therapy have been recently developed. These include retrovirus and adenovirus vectors. Retrovirus vectors have been used both in vitro (9) and recently in vivo (11,40) and are very efficient agents of gene delivery. However, one of the primary objections to the use of retrovirus in human gene therapy is the randomness of the integration event (14). This may result in insertional mutagenesis (20) or in aberrant regulation of the inserted gene. Adenovirus vectors were originally developed (6) and tested for use as recombinant vaccines, and a number of such vectors containing heterologous viral genes such as those encoding hepatitis B virus surface antigen (41), human immunodeficiency virus envelope proteins (30), and pseudorabies virus glycoproteins (12) have been constructed. More recently, it has been shown that replication-defective recombinant adenoviruses can be used as efficient expression vectors for gene therapy. Rosenfeld et al. (31,32) delivered the cDNAs for the a1-antitrypsin and cystic fibrosis transmembrane conductance regulator genes to cotton rats by insertion of these exogenous DNAs into the ElA region of adenovirus. Stratford-Perricaudet et al. (34) used the ornithine transcarbamylase cDNA in a similar fashion. In all of these experiments, excellent expression of the inserted gene was obtained.
We have previously reported the construction and efficient expression in mouse L cells of an adenovirus type 5 (Ad5) vector containing the hamster adenine phosphoribosyltransferase (aprt) gene (21). Stable transductants of the aprt gene occurred in which the aprt gene was integrated into the mouse chromosome in a random fashion at a frequency of 10-6. In this paper, we report successful detection of targeted integration of the introduced gene, using this vector in CHO cells. These recombinant adenoviruses are replication competent and contain a wild-type CHO aprt gene. Infection of an APRT-CHO cell line with this vector results in correction of a 3-bp deletion mutation on the chromosomal aprt gene. The ratio of targeted recombination to random recombination is 1:5 to 1:14.
MATERLALS AND METHODS Cell lines. CHO-ATS-49tg (from G. M. Adair, M. D. Anderson Cancer Center, University of Texas) is a spontaneous 6-thioguanine-resistant (HPRT-) and 8-aza-adenineresistant (APRT-) CHO cell line (la). It is hemizygotic at the aprt locus. Polymerase chain reaction (PCR) amplification and direct sequencing analysis of the remaining CHO-ATS-49tg aprt gene revealed a 3-bp deletion within exon 5 (1). CHO-ATS-49tg cells were cultured in alpha minimum essential medium with 10% calf serum or 10% fetal calf serum in the cloning experiments.
Construction of plasmids and recombinant viruses. Standard cloning techniques (3) were used to construct the plasmids. Plasmids pFG-PAR, pFG-PAL, pFG-NPAR, pFG-NPAL, pFG-NPARdE1, and pFG-NPALdE1 were constructed as shown in Fig. 1. Plasmid pHAPRT-2, derived from pHAPRT-1 (24), contains the BamHI fragment of the CHO aprt gene. This plasmid was used for construction of Ad5 viruses containing the aprt gene with its native promoter (Ad5/PAR and AdS/PAL; Fig. 2). Plasmid PUC-aprt has a BamHI-XbaI fragment of the aprt gene. This plasmid was used for the construction of recombinant AdS viruses with a promoterless aprt gene (AdS/NPAR and Ad5/NPAL; Fig. 2) and viruses carrying a deletion mutant of the aprt gene (AdS/NPARdE1 and AdS/NPALdE1). The specific fragments of the aprt gene were excised by digestion with the restriction enzymes indicated in Fig. 1. The ThaI site removes 388 bp from the 5' end of the BamHI fragment, including the promoter sequence, and is only S bp before the ATG start site. The EcoRV site deletes a 606-bp segment from the gene which includes the promoter, exon 1, and intron 1 and part of exon 2. These fragments were converted to an XbaI fragment by XbaI linker addition and cloned into the unique XbaI site of pFG-dxl (19). Plasmid  contains 40% of the right part of the adenoviral genome with a deletion from 78.5 to 84.7 map units. This 1.9-kb deletion removes most of the E3 region but leaves the E3 promoter and termination site. 293 cells (16) were cotransfected with each plasmid and EcoRI-digested AdS DNA to rescue the recombinant viruses. The plaque-purified viruses were analyzed by restriction enzyme digestion. Viruses Ad5/A2 and AdS/Al, which contain a CHO aprt gene with its endogenous promoter replaced by the Moloney murine sarcoma virus (MoMSV) promoter, were constructed as previously described (21).
Selection of transductants. CHO-ATS-49tg (ATS-49tg) cells were seeded at S x 106 cells per 100-mm2 dish on the day prior to the experiment. The cells were infected with Ad5 or recombinant adenoviruses at a multiplicity of infection of 10 PFU. At 12 h postinfection, the cells were trypsinized, split into 5 x 105 cells per dish, and incubated in alpha minimum essential medium for an additional 12 h, at which time the medium was changed to AAA selective medium (25 p.M alanosine, 50 p.M azaserine, 100 p.M adenine) (33). Cells were refed the next day and subsequently every 3 days for about 2 to 3 weeks with AAA medium. One APRT+ clone per plate was picked by using cloning cylinders. The remainder of the APRT+ colonies were counted after crystal violet staining. APRT and HPRT activity assays. The APRT and HPRT assays were performed as described previously (4). Monolayers of cells were washed with phosphate-buffered saline and collected. The cell pellet was lyophilized and resuspended in 200 pul of dilution buffer (50 mM Tris-HCl [pH 7.5], 30 mM KCl, 10 mM MgCl2). A 25-p.l volume of diluted cell extract was mixed with an equal volume of reaction mixture (100 mM Tris-HCl [pH 7.5], 16 M MgCl2, 1.5 mM phosphoribosylpyrophosphate (PRPP), 25 ,uM cold adenine, 50 ,uCi of [3H]adenine per ml, 2 mg of bovine serum albumin per ml) and incubated at 37°C for 10 min. The reaction mixture for the HPRT assay was the same as the APRT reaction mixture except that unlabeled hypoxanthine and 3H-labeled hypoxanthine were used instead of adenine (4). The reaction was terminated by addition of 1 ml of stop buffer (50 mM sodium acetate, 2 mM Na2HPO4). Synthesized [3H]AMP and [3H]IMP were precipitated by LaCl3 and collected on Whatman GF/C filters. Radioactivity was detected by scintillation counting, and the protein concentrations of cell extracts were measured by using the Bio-Rad protein assay reagent (Bio-Rad Laboratories, Richmond, Calif.).
Southern hybridization. High-molecular-weight genomic DNA of the transductants was extracted according to Perbal (27). Genomic DNA (20 ,ug) was digested with the desired restriction enzyme. The samples were extracted with phenol-chloroform-isoamyl alcohol (CIA). Digested DNA was run on an 0.8 or 1% agarose gel and transferred to a Nylon membrane (Zeta Probe membrane; Bio-Rad). Prehybridization and hybridization to a 32P-labeled probe were done as recommended by Bio-Rad.
PCR and sequencing. PCR was done by a modification of a method described previously (28). The upstream primer (5'-TAC CTG TGT ClTl TCC TCG TCC CTT CA-3') is the CHO aprt sequence from positions 2132 to 2158 (26), and the downstream primer (5'-CAG AGT ATG TTA GAT GCC CGT CCC AC-3') extends from positions 3417 to 3391. The expected PCR product is around 1.3 kb in length. The reaction was carried out in a total volume of 100 ,ul containing 500 ng of genomic DNA, 1 U of AmpliTaq polymerase (Perkin-Elmer Cetus), 200 ng of each upstream and downstream primer, 50 ,uM each deoxynucleoside triphosphate, 5% acetamide, 0.05% Nonidet P-40, 1 mM MgCl2, 10 mM Tris-HCl (pH 8.3), and 50 mM KCl. Amplification was done in a DNA Thermal Cycler (Perkin-Elmer Cetus). The initial denaturation at 94°C for 4 min was followed by 40 cycles of denaturation at 94°C for 1 min, annealing at 58°C for 1 min, and elongation at 72°C for 1 min. The final elongation was at 72°C for 7 min. The PCR products were further cloned into pBluescript for sequencing. Double-stranded plasmid was directly used for sequencing according to the protocol provided (United States Biochemical, Cleveland, Ohio). Figure 2 presents the structures of a series of recombinant viruses. The viruses initially used in these studies contain the MoMSV promoter juxtapositioned next to the structural aprt gene (38,42).

Structures of the recombinant viruses.
Ad5/A1 is a virus in which the MoMSV-aprt gene is in the opposite orientation to the adenovirus E3 promoter. Ad5/A2 has the MoMSV-aprt gene in the same orientation as the E3 promoter. Viruses AdS/PAR and AdS/PAL contain the CHO aprt promoter and structural gene in either the same (PAR) or opposite (PAL) orientation relative to the adenovirus E3 promoter. Viruses Ad5/NPAR and Ad5/NPAL contain a promoterless aprt gene, with the structural gene about 320 bp from the viral E3 promoter in NPAR. In the case of Ad5/NPAL, the aprt gene is in the opposite orientation to the E3 promoter and thus does not have any functioning promoter sequence. Viral E4 gene transcription terminates before reaching the cloned aprt gene (5). In viruses Ad5/ NPARdE1 and Ad5/NPALdE1, the inserted gene fragment  has a 5'-end deletion which includes the promoter sequence up to several base pairs of exon 2. All virus constructs were assayed for APRT activity in mouse LAT cells and APRT-CHO cells. There was no detectable APRT activity when either cell line was infected with viruses Ad5/NPAL, AdS/ NPARdE1, and Ad5/NPALdE1. Isolation of APRT+ transductants. ATS-49tg cells were infected with the viruses at a multiplicity of infection of 10. Cells were split to low density 12 h postinfection and maintained in normal medium for an additional 12 h before the addition of selective AAA medium. APRT-cells die in AAA medium within 3 days as a result of the blockage of de novo purine biosynthesis by azaserine and an inability to utilize free adenine. Uninfected cells and cells infected with wild-type virus were used as controls. Cells infected with Ad5/NPAR, Ad5/PAR, Ad5/PAL, AdS/Al, and Ad5/A2 survived in AAA medium for about 1 week as a result of transient APRT production from the viral genome. APRT+ colonies were seen after background cells died. APRT+ cells were not found in uninfected cells or cells infected with wild-type virus. They were not detected following infection of 5 x 107 cells with Ad5/NPARdEl, Ad5/NPALdEl, or Ad5/NPAL. Thus, the frequency of APRT+ colonies with these vectors is <5 x 10per cell. The frequencies of APRT+ colonies from AdS/Al, AdS/A2, Ad5/PAR, Ad5/ PAL, and AdS/NPAR infections are shown in Table 1.
Gene targeting following AdS/Al and Ad5/A2 infection. To determine whether the APRT+ clones arose as the result of homologous recombination or random integration, DNA from a total of 63 APRT+ clones from the Ad5/Al and AdS/A2 transductants was examined by Southern analysis. Figure 3A illustrates the restriction map of the mutant aprt gene in ATS-49tg cells and the wild-type CHO aprt gene introduced by the adenovirus vectors. The APRT-phenotype results from a 3-bp deletion in the hemizygous aprt gene, resulting in a loss of an MboII restriction site in exon 5 (la). Figure 3B shows typical MboII restriction patterns of DNA from ATS-49tg and wild-type CHO cells probed with the full-length CHO aprt sequence. A 2.0-kb fragment is found in ATS-49tg DNA instead of the 0.5-and 1.5-kb fragments found in wild-type CHO DNA. As shown in Fig.  4, following MboII digestion of the transductant DNA from the first 17 clones isolated (10 from Ad5/A2-infected clones and 7 from Ad5/Al-infected clones), two clones, Al-1 and A2-5, show a restriction pattern identical to that of wild-type CHO cells. That the pattern in clones Al-1 and A2-5 was the same as the wild-type pattern was confirmned byXbaI, KjpnI, HindIll, and EcoRI digestions ( Fig. 4B and C; HindIII and EcoRI digestions are not shown). The other 15 clones appear to result from random integration of the aprt gene. Following XbaI digestion, an extra 2.5-kb fragment can be seen in 12 clones (Fig. 4B). It represents an integrated copy of the gene (a 2.5-kb XbaI fragment of MoMSV-aprt). Three other clones show a fragment larger than 2.5 kb, indicating that part of the long terminal repeat (LTR)-aprt sequence, including at least one of the XbaI sites, was deleted during the integration process. Clone A2-8 has two extra XbaI fragments, which suggests multiple copies of gene integration. A total of 8 clones were isolated from 63 screened clones that possess the same restriction pattern as do Al-1 and A2-5 (Fig. 5). Six of 37 clones isolated following Ad5/A2 infection and 2 of 26 isolated following AdS/Al infection appeared to result from homologous replacement of the endogenous deletion. The other 55 clones showed the presence of both 2.0-and 1.5-kb bands and additional variable flanking fragments. This result indicates that in these cells, transduction occurred via random integration of the introduced gene, with retention of the endogenous mutant aprt gene.
HPRT deficiency in the homologous recombinants. Since ATS-49tg cells are also defective in the hprt gene, the clones presumed to be due to homologous recombination were examined for HPRT deficiency in order to verify that they did not arise from contamination of the culture with wildtype CHO cells. All eight presumptive homologous recombinants died in hypoxanthine-aminopterin-thymidine medium, and HPRT enzyme activity could not be detected in extracts of these cells (Table 2). Spontaneous revertants have never been detected with ATS-49tg (la). In agreement, we never detected revertants in uninfected cells or cells infected with wild-type virus, nor were APRT+ colonies found following infection with recombinant viruses carrying a deletion mutant of the aprt gene (AdS/NPARdE1 and AdS/NPALdE1) or lacking a promoter in the opposite orientation to E3 (Ad5/NPAL) ( Table 1).
APRT activity in the transductants. The specific activity of APRT of the transduced clones was measured ( Table 2). In the clones resulting from homologous recombination, the level of APRT activity was expressed at a hemizygous level about 50% of the wild-type CHO level. In contrast, the random integrants exhibited APRT specific activity varying from 9 to 160% of the wild-type level. Detection of vector sequences in homologous recombinants. To further confirm that the homologous recombinants arose from a double-crossover event or gene conversion within the  aprt gene, the Southern blots ofXbaI digests were rehybridized with MoMSV and adenovirus sequences which could detect almost all of the XbaI fragments of the adenovirus genome (Fig. 6). Neither adenoviral or MoMSV sequences can be detected in the presumptive homologous recombinants (Fig. 7). In contrast, the other clonal transductants have both adenovirus and MoMSV sequences (Fig. 7). Sequencing of PCR products of homologous and nonhomologous recombinant clones. As shown in Fig. 8, the 3-bp deletion in the exon 5 of the aprt gene is at the site of a direct TTC repeat (1) and within the MboII recognition site. From the sequence analysis, the restoration of the MboII site can also be obtained by a single T insertion. However, the frameshift mutation of the aprt gene resulting from this single-base insertion should produce a nonfunctional APRT (10). Since the homologous recombinants are APRT+ and have APRT activity at hemizygous levels, it is unlikely that the restoration of the MboII site in those recombinants results from the single T insertion. To further exclude this possibility, PCR amplification of the aprt gene of the homologous recombinants and random integrants was carried out by using two primers. The downstream primer hybridizes only to the sequence which is present in the chromosomal aprt gene but not in the introduced aprt gene, so that with use of this pair of primers, the randomly integrated aprt gene cannot be amplified. In Fig. 9, Al-1 (an example of homologous recombinants) displays a complete wild-type sequence and A2-8 (an example of random integrants) has the same sequence as do ATS-49tg cells.
Role of the promoter region in targeted recombination.
Since the constructs originally used for the targeting experiment contain the MoMSV promoter, we investigated whether this promoter played a role in the integration process. A number of viral constructs containing the aprt gene either with its native promoter or lacking a promoter (Fig. 2) were used to infect ATS-49tg cells. The results of these experiments are presented in Table 1. Approximately the same overall frequency of transduction to APRT+ occurred for all vectors tested except for Ad5/NPARdE1, Ad5/NPALdE1, and Ad5/NPAL, the vectors which had been shown to lack transient APRT activity (Table 1). These three viruses did not give rise to any APRT+ colonies. Twenty-nine APRT+ colonies were isolated and screened after AdS/PAR and AdS/PAL infection. One of fifteen colonies from AdS/PAL was confirmed as a homologous recombinant. Fourteen individual APRT+ colonies were also picked following AdS/PAR infection. Homologous recombinants were not detected among this population. However screening of more recombinants would probably have led to the isolation of homologous recombinants, since orientation of the insert (   the danger of inactivation of an essential gene or activation of an oncogene. The adenovirus/aprt vector reduces the frequency of random integration but appears to retain the same or a higher frequency of homologous recombination by and by the increasing the ratio of homologous recombination to random .ts (homolointegration by about 2 to 3 orders of magnitude.
higher by a The adenovirus vector and the MoMSV sequences are not introduced detectable in the homologous recombinants, suggesting that ilthough the the homologous recombination event is a double crossover on does not or gene conversion. In the random integrants, the various ion system. levels of APRT activity might reflect the impact of the site of integration or the integration of multiple copies of the aprt gene.
The factors that influence homologous recombination or ZO random integration are basically unknown in mammalian systems. It has been shown that homologous recombination between the adenoviruses does not require ElA gene expression (13). It is also known that the ElA gene is required   examined in the current experiments, there was no obvious loss of any one specific sequence (Fig. 7), but rather there appeared to be random loss of the viral DNA. Moreover, since the cells used in these experiments were not permissive for virus production, only early adenovirus gene functions are expressed. Thus, proteins normally associated with cytopathic effect (coat proteins or adenovirus fiber proteins) are not synthesized under these conditions.
We also propose that adenoviral replication might play a role in the gene targeting observed in the adenovirus/aprt system. The recipient CHO cells are semipermissive for adenovirus replication, leading to a blockage in a late adenovirus function (23). However, the viral DNA replicates as efficiently as in a permissive HeLa cell, resulting in the production of thousands of copies of the introduced gene. The large numbers of DNA molecules may facilitate the finding of homologous sequences in the cellular genome. However, this hypothesis is not in agreement with the work of Zheng and Wilson (43). Their experimental evidence, based on nonreplicative plasmid transfection, has shown that homologous recombination is independent of the copy number of the introduced DNA or targeted gene. However, the instability of transfected plasmid should be taken into account in this suggestion. The replication of adenovirus in CHO cells is delayed and prolonged after infection (23). The viral genome is protected by the viral core proteins so that it is not readily accessible to DNase. These features might influence the time course of homologous recombination events and create more opportunities for the introduced DNA molecules to find their counterpart chromosomal genes.
One possibility that should be considered in considering the mechanism of homologous recombination in the adenovirus/aprt system is the involvement of the single-stranded DNA substrate. During adenovirus replication, singlestranded DNA is generated by the displacement of parental strands during DNA replication (22,36,37). Studies with Escherichia coli and mammalian cells using plasmid vectors indicate that single-stranded DNA is an intermediate in homologous recombination (8,29).
Our initial strategy for studying the absolute frequency of homologous recombination in the adenovirus vector system was to use Ad5/NPARdE1 and Ad5/NPALdE1 for the targeting experiments. These viruses carry a mutant aprt gene which is nonfunctional if integrated in a random fashion. APRT+ colonies formed following infection by either of these virus constructs would arise only as a result of homologous recombination. Infection with the promoterless Ad5/NPAR and Ad5/NPAL viruses should permit direct analysis of the ratio of homologous recombination to nonhomologous recombination. Both homologous and nonhomologous recombinants could arise following infection with Ad5/NPAR, and only homologous recombinants could arise following Ad5/NPAL infection. Thus, the number of APRT+ colonies derived from Ad5/NPAL infection versus the number of APRT+ colonies derived from Ad5/NPAR infection in selective medium is the ratio of nonhomologous to homologous recombination.
The failure to detect homologous recombination after infection with viruses having either the 5'-end deletion of the aprt gene (Ad5/NPARdE1 and Ad5/NPALdE1) or the promoterless aprt genes (Ad5/NPAR and Ad5/NPAL) implies a role for promoter sequences in initiating or enhancing the homologous recombination. The MoMSV LTR sequence might contain specific sequences which are recognized by cellular recombinase-like activities (25) or might serve as a transcriptionally active region which enhances homologous recombination. The latter has been suggested for both mammalian and yeast systems (7,35,39) except that the transcriptionally active region is on the recipient homologous DNA sequences. The ratios of homologous to nonhomologous recombination obtained in AdS/A1 and Ad5/PAL infections are at similar levels, indicating that promoter sequence specificity is unlikely since the MoMSV promoter and the aprt promoter do not share sequence similarity.
The proximity of a promoter sequence to the targeting gene on the vector may also play a role in homologous recombination. Following infection of Ad5/NPAR, in which the adenoviral E3 promoter is located about 320 bp upstream of the aprt gene, no homologous recombinants are found, although the level of aprt gene transcription initiated from the adenovirus promoter is high (data not shown). The 320 bp of heterologous sequence (adenovirus sequence) residing between the promoter and the structural gene might interfere with the homologous recombination process initiated from an active region.
We show in vitro that an adenovirus vector may have potential in human gene therapy and that it offers one feature that is not available in the retrovirus vector system, namely, targeted integration by homologous recombination. It may also be an excellent model for studying the mechanism of homologous recombination in mammalian cells. VOL. 13,1993 3"l on April 30, 2019 by guest http://mcb.asm.org/