INTRODUCTION

Currently available viral gene transduction agents have both strengths and weaknesses. Some are more useful in one setting, some in another. Retroviral vectors are most frequently used. They are potentially useful for gene transfer to dividing cells, but are limited by their loss of activity on concentration, inability to infect resting cells and other undesired side effects (1, 2, 3, 4, 5). The former problem may be addressed in part by altering retroviral packaging (6). Adenoviral transfer agents infect a wide range of cells and may infect and express their genes well in resting cells from diverse tissues. They can also be concentrated to high titer. However, adenovirus elicits a strong immune response that eliminates infected cells and limits the longevity of expression, often to 10 days or less in immunocompetent hosts (7, 8).

Several newer agents have been proposed for gene transduction, but are not understood well enough to allow definitive conclusions as to their utility. For example, adeno-associated virus is a small DNA virus that may infect both resting and dividing cells. In the latter, it may integrate into host genome at a single site, at a point of considerable genomic instability on chromosome 19 in humans (9), but it may also transduce gene expression without integration (10). Considerable additional work will probably be needed before adeno-associated virus is useful as an efficient transduction agent, particularly for normal cells (11). Other recently described viral agents include a bovine papilloma virus (12), and Herpes simplex virus (13) and, very recently, HIV (14). Analysis of these viruses as potential gene transfer vectors is just beginning. Currently available agents do not, therefore, satisfy the range of likely clinical needs. Additional vectors and strategies are necessary.

We thus sought a viral agent that: (i) could be concentrated to high titers to allow treatment of large organs or cell pools, particularly in vivo; (ii) could infect and express its genes in a wide variety of cells for maximum flexibility in application; (iii) was relatively nonimmunogenic; and (iv) could be made replication-deficient, and handled and packaged with relative ease. Consequently, we devised a gene transfer system based on simian virus-40 (SV40)¹ as a vector.

The rationale for this approach was based on several important aspects of SV40 virus and its activity. SV40 infects a wide range of cell types from humans and other mammals, and expresses its genes in them. Plasmids incorporating SV40 genes and/or promoter may express either transiently or stably in cell lines and in primary cultures. SV40 may integrate into the host genome (15), permitting transmission of expression to daughter cells. The virus is stable to manipulation, and can be concentrated to high titer. Furthermore, the lack of adverse effects was documented in people who received the Salk polio vaccine, early preparations of which contained wild type (wt) SV40 (16, 17). Replication-deficient SV40 can be produced with relative ease. SV40 large T antigen (Tag) is required for virus replication (18, 19). However, Tag expressed by packaging cell lines can support virus replication in trans (20).

MATERIALS AND METHODS

TC7 monkey cells were the kind gift of Dr. Janet S. Butel (Baylor College of Medicine). COS-7 cells were obtained from ATCC. These cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (Hyclone), glutamine, and antibiotics (Life Technologies, Inc.).

pBSV(T)

The system reported here is based on pBSV-1 (kind gift, Dr. J. Butel), in which the SV40 genome had been cloned as a BamHI fragment into pBR322 (see Fig. 1). To make our plasmid, pBSV(T), pBSV-1 was digested with AvrII and BclI, to excise the 2.4-kilobase Tag gene, just upstream from its transcriptional start site, almost to the end of its coding region. A multiple cloning site modified from pGEM7 to delete the BamHI site, was inserted. The pBSV(T) multiple cloning site included unique BstXI, XhoI, and XbaI sites flanked 5 (toward SV40 early promoter) by Sp6 promoter and 3 by T7 promoter. The inclusion of these promoters facilitates sequencing of cloned DNAs by application of standard Sp6 and T7 primers.

Production of pBSV(T).

To evaluate this system, we produced SV40 containing firefly luciferase (Luc) as a reporter. pBSV(T) was opened with XhoI and XbaI. Luc cDNA, + SV40 early promoter, was excised from pGL2-Control Vector® (Promega) as an XhoI-DraI fragment. The XhoI sites of the Luc cDNA and the opened pBSV(T) were compatible. After ligation of the XhoI-digested end of Luc cDNA into the pBSV(T) XhoI site, the linearized plasmid was treated with Klenow and religated to yield pBSVluc. This plasmid contains two copies of the SV40 early promoter. Preliminary data in this system suggested that this double promoter configuration elicited higher levels of transgene expression.²

To produce virus, the viral genome was excised from pBSVluc with BamHI, gel purified, religated, and transfected using calcium phosphate into COS-7 cells (21). Two weeks later, when >50% of the cells showed evidence of infection, virus was harvested by freezing, thawing, and sonicating the cells.

The time course of SVluc production following transfection of COS-7 cells was measured. Culture dishes (60 mm, Falcon Plastics) containing approximately 10⁶ COS-7 cells were transfected with 1 µg (0.5 pmol) of recircularized viral DNA following excision from plasmid and gel purification as described. Cultures were harvested and titered for the generation of infectious SV40 derivative virus every 3-4 days thereafter until 14 days. Results, shown in Fig. 2, indicate that maximum virus yield was obtained at 7 days post-transfection: 3.5 × 10⁵ infectious units MV/0.5 pmol of transfected virus.

Subsequent stocks of replication-deficient SVluc were prepared by infecting COS-7 cells. Virus preparation was carried out in Dulbecco's modified Eagle's medium, 2% fetal bovine serum. When virus from such primary cultures is used to generate a working stock of virus, the 2 ml of lysate derived from the transfected 60-mm dish cultures were used to infect two 75-cm² tissue culture flasks of subconfluent COS-7 cells. Cultures were allowed to proceed for 2 weeks thereafter, and harvested. Cultures were harvested by removing all but 2 ml of culture medium from the flask, freezing and thawing three times, and sonication. In general, these procedures reproducibly yield virus stocks of 10⁹ plaque-forming units/ml. Resulting virus (SVluc) may be band-purified by discontinuous sucrose density gradient ultracentrifugation, and titered using neutral red-agar overlay staining of COS-7 cells, according to standard protocols (22). This increases virus titer 10-100-fold in our hands.

To ascertain that SVluc was replication-deficient, virus was plated, as for titering, on COS-7 and TC7 cells. The latter lack Tag. Titers were determined by plaque-forming units, and compared to titers of wt SV40 (kind gift, Dr. Butel). Each stock of presumed replication-deficient SVluc used in these studies has been tested at least twice for the presence of replication competent virus.

Luciferase was detected in two different ways: by measuring its enzymatic activity and by visualizing Luc-containing cells by immunochemical analysis using affinity-purified anti-Luc antibody (Promega). Luc activity was assayed in frozen tissue samples or cell culture homogenates using a Luciferase Assay Kit (Promega), according to package insert instructions. Frozen tissues (or cultured cell lysates) were weighed and homogenized in luciferase assay buffer (Promega). Light emission was measured in arbitrary units using a Chem-Glo luminometer (Aminco, Inc.). Statistical comparisons of light emission data were performed using Wilcoxson nonparametric comparison.

Luciferase protein was visualized in individual cells using affinity-purified rabbit anti-Luc antibody (anti-Luc antibody, Promega) to treat acetone-fixed cultured cells or frozen tissue sections mounted on glass slides. Slides were treated with anti-Luc (or control normal IgG) antibody, followed by avidin-conjugated goat anti-rabbit Ig, then avidin-peroxidase with a biotin bridge (23). Reactivity was visualized using diaminobenzidine + H₂O₂.

Cultured TC7 or COS-7 cells were infected with SVluc (m.o.i. = 10), or mock infected. RNA was extracted 18 h later (RNAzol, Cinna-BioTex, Friendswood, TX), electrophoresed in formaldehyde gels, blotted to nitrocellulose, and hybridized with ³²P-labeled Luc cDNA (19) (hybridization: 50% formamide, 5 × SSC, 0.1% SDS, 42 °C; final washes: 0.1 × SSC, 0.5% SDS, 42 °C). Hybridization was visualized by autoradiography. As a control probe, rDNA was used (kind gift, Dr. E. Mercer, Jefferson Medical College).

BALB/cJ mice (retired breeders, Jackson Laboratories, Bar Harbor, ME) were given SVluc, 10⁹ plaque-forming units/mouse, in 0.1 ml of saline via intravenous or transoral intratracheal instillation. Mice were sacrificed by cervical dislocation on days 3, 7, 10, 14, 17, and 21. Day 21 was the end of the experiment. Control mice received saline only. Experimental groups consisted of 3 to 5 animals. Control groups consisted of 2 mice per time point. Lungs, liver, kidney, spleen, heart, esophagus, and skin overlying the intravenous inoculation site in the tail were sampled. Half of each organ was used for immunohistochemical analysis to visualize Luc protein using anti-Luc antibody, and half was snap-frozen in liquid nitrogen, to be homogenized for Luc enzyme assay (see above). Negative controls for these assays included tissues from control mice, treated with the anti-Luc antibody, and, as well, experimental and control mouse tissues treated with normal IgG instead of anti-Luc antibody.

RESULTS

To ascertain that the virus produced is replication-deficient in cell lines that do not supply Tag, TC7 cells were used. In contrast to wt SV40, a Tag-deficient SV40 mutant should not yield a lytic, productive infection in cells lacking Tag. That is, SVluc should not produce plaques in TC7 cells. SV40 and SVluc titers using TC7 and COS-7 cultures were compared at multiplicities of infection (m.o.i.) from 10³ to 10. Inocula of up to 10⁷ plaque-forming units were used. SVluc did not produce plaques in TC7 cells at any m.o.i., while wt SV40 did. Both viruses produced plaques on COS-7 cells. Our current stocks of SVluc have been passed 11 times in culture, without evidence of replication competence.

Infection of cultured COS-7 and TC7 cells was done to determine whether Luc could be detected. SVluc infected (m.o.i. = 1), cultured cells 18 h post-infection were assayed for Luc gene expression by Northern analysis and for Luc enzyme activity using a standard Luc assay kit (Promega) and a luminometer (Aminco, Inc.). Luciferase mRNA was found in both COS-7 and TC7 cells, but not in mock-infected control cultures (Fig. 3). Luciferase enzyme activity was also detected in SVluc-infected cells: SVluc-infected TC7 and COS7 cells averaged 116 (±49) light emission units/50-µl aliquot of culture protein (75 µg), while mock-infected cells averaged 10 ± 14 light emission units/culture (p = 0.05). Therefore, SVluc successfully transferred Luc production to cultured cells.

The transduction efficiency of the SV40-derivative virus was examined. At m.o.i. = 1, 52% of cultured cells (either TC7 or COS-7) cells stained positively for transgene expression. At m.o.i. = 0.1, the percentage of positive cells was 4.1%. Interestingly, at m.o.i. = 10, only 57% of cells were positive for transgene expression, suggesting that for these cells, that percentage represents the maximum proportion of these cell types that will express the transgene.

To test the ability of SVluc to act in vivo as a gene transduction agent, BALB/c mice received SVluc (or saline only) as described under ``Materials and Methods.'' The animals were sacrificed on days 3, 7, 10, 14, 17, and 21. Selected tissues were tested for Luc expression using anti-Luc antibody and luminometry.

Cells producing immunohistochemically detectable Luc were seen (Fig. 4) in the liver, kidney, spleen, in the large conducting airways of the lung, and in the skin at the inoculation site. The periphery of the lung and the heart were negative. Cells bound by anti-Luc antibody were first noted by day 3, but Luc protein production was most pronounced on day 21. Intravenous and intratracheal instillation yielded similar patterns of Luc expression in the tissues examined, save that intratracheal treatment produced many Luc-expressing cells in conducting airways and esophagus, while intravenous inoculation did not do so in the course of this 21-day study. In the liver, hepatocytes and Kupffer cells contained Luc protein. In the spleen, both lymphocytes and mononuclear phagocytes produced it. Splenic megakaryocytes also produced Luc (insets, Fig. 4). In the skin at the inoculation site, Luc was detected principally in keratinocytes, while in the kidney it was mainly found in glomeruli.

No inflammatory reaction was found in any organ that was positive for Luc, at any time point studied (days 3-21). Thus, SVluc-infected cells, as recognized by anti-Luc antibody, were normal in appearance and were not associated with any inflammation (Fig. 4).

We used several negative controls for immunochemical studies: control mouse tissues treated with anti-Luc antibody, and both experimental and control mouse tissues treated with normal IgG instead of anti-Luc antibody. All these controls yielded negligible staining (Fig. 4).

These same selected tissues were homogenized and assayed for Luc enzymatic activity using a luminometer (Fig. 5). SVluc recipients showed strong Luc activity in the liver, kidney, and spleen beginning on day 14, and lasting through the end of the study on day 21. Heart and the lungs (excluding large airways) were negative at all times. Although slight increases in luciferase activity appeared in samples of heart and lungs toward the end of these studies, these increases were not statistically significant (p > 0.2). Skin at inoculation sites was strongly positive (data not shown). The level of activity observed was relatively constant from day 14 onwards, and did not deteriorate during the time frame studied (Fig. 5). Tissues from all organs of control mice (receiving vehicle only) were negative at all times for Luc activity.

DISCUSSION

The utility of SV40 or any other virus as a gene transfer agent requires that certain conditions be met. The virus must: (i) be replication-deficient under normal circumstances; (ii) be safe for its intended recipients; and (iii) transfer gene expression effectively.

The requirement for replication incompetence reflects a need for limited exposure to any virus and a concern over potential transmissibility of an infectious therapeutic agent. We studied this question using a titer assay in Tag+ and Tag cells. Plaques require that virus complete a lytic replicative cycle in infected cells, lyse those cells, spread to adjacent cells, and repeat the process. At a range of m.o.i. from 10³ to 10, SVluc did not produce plaques in Tag TC7 cells. Wild type SV40 did. Thus, SVluc could not replicate detectably in cells lacking Tag.

Tag is the SV40 transforming protein (24). It was excised in our shuttle vector, greatly reducing potential risks of applying this vector to therapeutic systems. These findings do not rule out the possibility that SVluc may reacquire Tag during passaging. We have not yet detected SVluc replication in TC7 cells even 11 passages beyond the initial generation of the virus. The observed replication incompetence of SVluc suggests that reacquisition of Tag happens very rarely or not at all in our system. The possible occurrence of such an event indicates a need for caution in using this virus as a gene transfer agent. There have been reports of SV40 genome detection in DNA from some rare forms of human tumors (25, 26). However, the most convincing report used tumor specimens from patients who had no known contact with the virus and were too young to have received live virus during immunization for polio (25). Furthermore, the careful epidemiological studies of Salk vaccine recipients showed that inoculation of wt SV40 does not have significant harmful sequellae for humans (16, 17). As I have been unable to detect replication competent revertants, however, such concerns at this point are mostly theoretical.

Effective gene transfer using SV40 as a vehicle was first described by Gething and Sambrook (27) and by Asano and co-workers (28). These studies used replication-deficient SV40 variants to transfer gene expression to cultured cells. This article describes the first application of SV40 to gene transfer in live animals. The ability of a mutant SV40 to transfer Luc reporter activity to mice was demonstrated using two detection techniques: immunochemistry with affinity-purified anti-Luc antibody, and assay for reporter protein enzymatic activity. Initial studies indicated that Luc transduced by SVluc was detectable by immunochemistry, and that it was functionally active.

When we tested SVluc in animals, Luc-producing cells were detected immunohistochemically 3 days after administration, and enzyme activity was demonstrated by luminometry 14 days post-inoculation. This time lag may be needed for enzyme levels to accumulate sufficiently to be detectable in tissue homogenates. Subsequently, Luc activity persisted undiminished. Increases in Luc activity with time cannot be explained by the presence of replication-competent revertants in our stock of SVluc. Even if some such recombinants were to have occurred and escaped our detection, mice are not permissive hosts for wt SV40.

Some gene transfer agents, adenovirus, for example, elicit strong immune responses that eliminate infected cells, usually within 7 to 10 days (7, 8). In that context, we chose 21 days as the duration of these studies to test whether the uncompromised host immune system would eliminate the SV40-derived viral vector. We found no evidence that host reactivity to SVluc or SVluc-infected cells altered the course of reporter gene expression or eliminated cells infected with SVluc. That is, Luc activity did not decrease during this study, and there was no histologic suggestion of an inflammatory reaction versus Luc-producing cells. Time course studies of sera from SVluc infected mice to detect neutralizing anti-SVluc antibody are in progress. However, immunologic responsiveness to SV40 has been shown to be relatively limited: both humoral and cellular immune responses have been found to be focused primarily on Tag, with scant or no reactivity detected versus viral structural proteins (29, 30). The lack of inflammatory response in our tissue sections is in keeping with these reports, since Tag is not expressed in this system. This finding also suggests that luciferase protein does not elicit a cytotoxic immune response as produced by transduced cells. This may reflect its intracellular localization so that it does not present a cell membrane target for the immune system, or perhaps a strong susceptibility to degradation.

Clearly, additional work is needed to understand this system better. The size limitations for inserts in SV40 are not yet known. The Tag gene is 2.4 kilobases, comparable in size to the Luc cDNA. Smaller inserts (300 base pairs to 1 kilobase pair) have been packaged in recombinant SV40 (preliminary data), but the upper size limits of this system are not yet clear. We have obtained expression of -galactosidase in this system (3.6 kilobases), but have observed that packaging does not proceed as efficiently for this as for smaller inserts.³ Current studies in progress include excising SV40 late genes to see whether additional cloning capacity can be added in this fashion. Studies for longer time periods, involving multiple inoculations of virus, and including analysis of more organs will assess more definitively the longevity and distribution of reporter gene expression, and whether immune responsiveness may eliminate infected cells. However, stable transgene expression in vivo over 21 days without evidence of inflammation is encouraging. Preliminary studies using organ homogenates suggest that SVluc DNA integrates into the host genome.⁴ The ability of wt SV40 DNA to integrate into the host genome is well documented (15).

In conclusion, recombinant, Tag-deficient SV40 can transfer active gene function that is stable, at least for 21 days. This system may have potential for application to therapeutic gene transfer in vivo.