Biology of Ovine Adenovirus Infection of Nonpermissive Cells

Abstract

Nonhuman adenoviruses, including those of the genus Atadenovirus, have the potential to serve as vectors for vaccine and gene therapy applications in humans, since they are resistant to preexisting immunity induced by human adenoviruses in the majority of the population. In this study, we elucidate the outcome of infection by ovine adenovirus type 7 isolate 287 (OAdV) of several nonovine cell types. We show here that OAdV infects a wide range of nonovine cells but is unable to complete its replication cycle in any of them. In nonovine, nonfibroblast cells, viral replication is blocked at an early stage before the onset of, or early in, DNA replication. Some fibroblasts, on the other hand, allow viral DNA replication but block virus production at a later stage during or after the translation of late viral proteins. Late viral proteins are expressed in cells where viral DNA replication takes place, albeit at a reduced level. Significantly, late proteins are not properly processed, and their cellular distribution differs from that observed in infected ovine cells. Thus, our results clearly show that OAdV infection of all nonovine cells tested is abortive even if significant viral DNA replication occurs. These findings have significant positive implications with respect to the safety of the vector system and its future use in humans.

Human adenoviruses have been widely studied over the last 4 decades, and the processes involved in infection and replication have been largely characterized. Abortive infection by human adenoviruses has been investigated in human and several nonhuman cell culture systems (reviewed in reference 35). These studies have provided a better understanding of the molecular biology of human adenoviruses and of events such as the integration of foreign DNA into mammalian genomes and the initiation of malignant transformation by DNA tumor viruses. However, the biology of human adenovirus infection in heterologous cells differs widely for various genotypes, even in the same cell line. For example, baby hamster kidney cells (BHK-21) are semipermissive for adenovirus type 2 (Ad2) and Ad5 replication but block infection with Ad3 and Ad12 at an early step of the viral life cycle (13). The early block in Ad12 replication can be abrogated by constitutive expression of E1 genes from Ad2, -5, or -12 or of the Ad12 pTP gene, but cells still do not allow late protein synthesis and virus production (21, 27, 49). Infection of monkey cells with Ad2 is the best-known example that shows a block in the late stage of adenovirus replication (1, 2, 22, 25, 26). This block, which is caused mainly by a severely reduced translation of the late RNAs, can be overcome by either coexpressing simian virus 40 (SV40) large T antigen in the monkey cells or by infection with a virus mutated in the N terminus of DNA binding protein (DBP) (15, 43).

With the advancing interest in using animal adenoviruses as recombinant vectors, the focus on viral transduction and replication has shifted to these xenogeneic viruses. Animal adenoviruses of canine, porcine, and bovine origin have been developed as gene transfer vectors (reviewed in references 5 and 33). However, these viruses all belong to the genus Mastadenovirus, and their genome arrangement is similar to that of the well-characterized human Ad2 and Ad5. CELO, the best characterized member of the genus Aviadenovirus, has also been used for the generation of gene transfer vectors (38). On the other hand, ovine adenovirus type 7 isolate 287 (OAdV) (6) is the prototype of the recently formed new genus Atadenovirus, which also includes egg drop syndrome virus, several subtypes of bovine adenovirus, and other animal isolates (3, 11, 16, 17). It has been shown previously that OAdV-derived vectors can be used to transfer genes to mice, rats, and rabbits very efficiently (18, 31, 32) and that OAdV therefore has the potential to serve as a gene transfer vector for gene therapy and vaccination.

OAdV is distinguished from members of the mast- and aviadenoviruses by a different genome organization and by the fact that many open reading frames show no detectable homology to those of any other adenoviruses (53-56). Because the biology of OAdV is not well understood, the outcome of infecting nonovine, particularly human, cells with OAdV is of interest. Human adenovirus vectors have at least E1 deleted to render them safe with respect to productive infection of otherwise permissive human cells, although the problem of trans-complementation with wild-type virus still exists. Nevertheless, such vectors still replicate their DNA in a variety of human tumor cell lines (52). In contrast, OAdV lacks the typical E1 genes of mastadenoviruses, and functional homologues have not yet been identified. OAdV vectors used in gene transfer studies so far contain the wild-type genome with an inserted transgene expression cassette. These still replicate productively in permissive ovine cells that do not harbor any viral sequences (45, 59).

Therefore, in this study, we investigated in detail the processes involved in infection of susceptible human and rodent cells with OAdV-derived vectors. We found that the block in virus production occurs at different stages of the viral life cycle in different cell types. We further show that there is no productive infection with OAdV even in some fibroblasts that allow viral DNA replication and expression of late viral genes. These studies add to the safety profile relevant to the use of OAdV vectors as potential gene transfer vehicles in humans.

MATERIALS AND METHODS

Cells and viruses.

CSL503 (fetal ovine lung cells [45]), HVO156 (ovine embryonal skin fibroblasts [34]), IMR-90 (human lung fibroblasts [42]), and SaOS-2 (ATCC HTB-85) cells were maintained in Dulbecco modified Eagle medium plus 15% fetal calf serum. MCF-7 (ATCC HTB-22), 293 (ATCC CRL-1573), 208F (rat fibroblasts [46]), HepG2 (ATCC HB-8065), HuH7 (human hepatocellular carcinoma [41]), AII (mouse hepatocytes deficient in p53), C2C12 (ATCC CRL-1772), and HepSV40 (mouse hepatocytes expressing SV40 large T antigen [44]) cells were maintained in Dulbecco modified Eagle medium supplemented with 10% fetal calf serum. Primary human hepatocytes were provided by Andreas Nüssler, Humboldt-Universität Berlin, and were kept in William's medium containing 10% fetal calf serum and 10 μg of insulin/ml. Media were generally supplemented with 2 mM glutamine, 100 μg of streptomycin/ml, and 100 U of penicillin/ml.

Wild-type OAdV and OAVhaat (18), OAdV carrying a human α₁-antitrypsin (hAAT) cDNA driven by the Rous sarcoma virus 3′ long terminal repeat, were propagated as described previously (34). The number of viral particles was determined according to the method of Maizel et al. (37) by measuring the absorption at 260 nm. The number of infectious particles was determined as described elsewhere (9) on CSL503 cells. The particle/infectious-particle ratios were generally <10.

Determination of host cell range.

Cells (8 × 10⁴/well) were plated onto 12-well plates and infected the following day with OAVhaat diluted in phosphate-buffered saline (PBS). Infections were carried out in PBS with gentle shaking for 1 h at room temperature. Samples were collected every 24 h and stored at −80°C followed by a change of the tissue culture medium. Quantification of the hAAT content in the tissue culture supernatant was done by an enzyme-linked immunosorbent assay (ELISA) as described previously (10).

Southern blotting.

Cells (5 × 10⁵) were plated onto 6-cm-diameter plates and infected with OAdV as indicated in the figure legends. DNA was isolated at various time points. Detection of the 2,399-bp OAdV DNA fragment at the left-hand end of the genome by Southern blotting has been described previously (18). For Southern blot detection of OAdV DNA from the right-hand end, genomic DNA was digested with EcoRI and StuI, which releases two specific bands of 2,308 and 4,196 bp. After separation of DNA fragments on a 1% agarose gel and transfer to a nylon membrane, hybridization was performed with a probe spanning nucleotides 23063 to 29574 of the OAdV genome.

RT-PCR.

RNA was isolated from infected cells at the time points indicated in the figure legends. Reverse transcription (RT)-PCR was performed using the Titan RT-PCR system (Roche Diagnostics, Mannheim, Germany) in accordance with the manufacturer's protocol. For DBP and E43 transcript amplification, the primers AK28, AK32, AK33, and AK37 (23) were used. For amplification of the late transcripts, a primer residing in the OAdV tripartite leader sequence (5′-CCTCTGGAATTTCCAGCTGTG-3′) was used together with primers annealing to the 5′ region of the respective genes (p52/55, 5′-GGACTAGTTCCAGCAAAATCC-3′; penton, 5′-TGTACCCATTCGTATTTAGGATGT-3′; hexon assembly protein (HAP), 5′-TCCCAAGCTATCATCCCATTT-3′; and fiber, 5′-CATCATTTATTTTTAGGGGAAGTTG-3′). The amplified products come from spliced mRNAs and were in each case clearly distinguishable from products amplified from contaminating virus DNA.

Determination of OAdV MLP activity.

pMLPtpl-luc contains a luciferase gene driven by the OAdV major late promoter (MLP) directly linked to the three tripartite leader exons (55). pCMVluc contains a cytomegalovirus (CMV) promoter-driven luciferase cDNA. Two micrograms of pMLPtpl-luc or pCMVluc was transfected together with 0.2 μg of a Rous sarcoma virus promoter-driven lacZ reporter gene plasmid onto cells seeded in six-well plates using Lipofectamine (Gibco BRL) according to the manufacturer's protocol. The cells were either left untreated or infected with wild-type OAdV 4 h after transfection, and luciferase and lacZ gene expression levels were determined in the cell lysates 48 h later. The relative luciferase activity was calculated as the ratio between luciferase and lacZ gene activities.

Generation of polyclonal rabbit anti-OAdV serum.

A 12-week-old New Zealand White rabbit was injected with 2 × 10¹⁰ infectious particles of CsCl-purified OAdV, and two booster injections were performed every 3 weeks. Serum was collected 2 weeks after the second boost and tested for the presence of OAdV-specific antibodies on 96-well plates coated with OAdV at 1:200 dilution. After incubation of the wells with horseradish peroxidase (POD)-linked goat anti-rabbit antibody at 1:2,000 dilution, the optical density was plotted against the serum dilution. The antibody titer, determined by interpolating the inflection point of the postimmune graph, was estimated to be 1:40,000.

Detection of late viral proteins.

Cells were infected with OAdV at multiplicities of infection (MOIs) of 2 and 20 as indicated in the figure legends. In experiments where 1-β-d-arabinofuranosylcytosine (araC) (Sigma, Taufkirchen, Germany) was used, it was added to the infected cells at 20 μg/ml at the time of infection and at 12-h intervals until the cells were harvested (7). Samples were harvested by scraping the cells off the tissue culture dish and washing the cell pellet in PBS. Subsequently, the cells were lysed with lysis buffer containing 50 mM HEPES, pH 7.5, 200 mM NaCl, 1 mM EDTA, 2.5 mM EGTA, 10% glycerol, 0.1% Tween 20, 10 mM beta-glycerol-phosphate, 1 mM NaF, 0.1 mM Na₃VO₄, 1 mM dithiothreitol, 0.1 mM phenylmethylsulfonyl fluoride, 10 μg of leupeptin/ml, and 2 μg of aprotinin/ml for 2 h. Samples were then centrifuged for 20 min at 21,000 × g and 4°C, and the supernatants were stored at −80°C.

For detection of late viral proteins by ELISA, MaxiSorp plates (F96; Nunc, Wiesbaden, Germany) were coated with 1 μg of total protein in 100 μl of carbonate-bicarbonate coating buffer (15 mM Na₂CO₃, 35 mM NaHCO₃, pH 9.6)/well overnight at room temperature. The plates were washed five times with PBS containing 0.1% Tween 20 (PBST) and blocked for 1 h at 37°C with PBST-5% skim milk. Following washing of the plates as described above, 150 μl of rabbit serum/well diluted 1:10,000 in PBST was added. The plates were incubated for 2 h at 37°C. After the plates were washed as described above, 100 μl of POD-labeled goat anti-rabbit antibody diluted 1:2,000 in PBST-5% skim milk was added to each well and the plates were incubated for 2 h at 37°C. The plates were washed as described above, and 100 μl of o-phenylenediamine (Sigma)/well was added. Following incubation of the plates for 10 min at room temperature, 100 μl of stop solution (2 N H₂SO₄) was added and the optical density of each well at 490 nm was measured.

For Western blot assays, 20 μg of total protein was separated on a NuPage 4-12% Bis-Tris Gel (Invitrogen, Karlsruhe, Germany) and transferred to a nitrocellulose membrane (Amersham, Freiburg, Germany). Viral proteins were detected using polyclonal rabbit anti-OAdV serum diluted 1:100 in blocking buffer (PBST-5% skim milk-2% bovine serum albumin [BSA]) and subsequent visualization with goat anti-rabbit-immunoglobulin G-POD antibodies by the ECL system (Amersham, Freiburg, Germany).

Immunofluorescence.

Cells were grown on coverslips and infected with OAdV at various MOIs. At the appropriate time points, the cells were fixed for 20 min in −20°C acetone and stored until required. The coverslips were washed once in PBS followed by blocking with PBST-1% BSA-0.5% goat preimmune serum for 20 min at 37°C. Without washing the coverslips, rabbit anti-OAdV serum at a dilution of 1:100 in PBST-1% BSA was added. After incubation for 1 h at 37°C, the coverslips were washed three times in PBST and incubated for 1 h with a Cy3-conjugated donkey anti-rabbit antibody (Jackson Immuno Research, West Grove, Pa.) at 1:500 dilution in PBST. Finally, after being washed as described above, the coverslips were incubated with DAPI (4′,6′-diamidino-2-phenylindole) at 1:1,000 dilution in PBS for 5 min at room temperature and mounted on microscope slides using Fluoromount-G (Southern Biotechnology Associates). Immunofluorescent staining was observed using a fluorescence microscope (Olympus IX70) and the analySIS 3.0 software package (Soft Imaging System, Münster, Germany).

Electron microscopy.

Electron microscopic images were obtained as described previously (19). Briefly, cells were infected at various MOIs and harvested 72 h postinfection by scraping them off the tissue culture plates. Following washing in PBS, the cells were fixed in PBS containing 3% glutaraldehyde and embedded in Epon 812. Thin sections were cut on an Ultrotom (LKB). Analyses were carried out, and micrographs were taken using a JEM 100 CX (JEOL) electron microscope.

RESULTS

Infection of human cell lines with OAdV-derived vector.

Cell lines of ovine origin permissive for OAdV infection (CSL503 and HVO156) have a fibroblast phenotype. These cells, as well as several human cell lines (293, MCF-7, and IMR-90), were infected with OAVhaat, and the concentration of the transgene product, hAAT, was measured in the cell culture supernatant (Fig. 1A). Although weak transgene expression was observed with 293 and MCF-7 cells at 24 h postinfection (p.i.), no altered hAAT levels were observed for the period between 48 and 72 h p.i. In contrast, the fibroblast cells, IMR-90, showed hAAT gene expression comparable to that observed with permissive cell lines after 24 h, which increased about fivefold between 48 and 72 h p.i., suggesting that viral DNA is replicated in IMR-90 cells. To extend this observation to nonhuman cell types, mouse liver (AII) and rat fibroblast (208F) cell lines were infected with OAVhaat as described above, and the concentration of the transgene product in the supernatant was again determined (Fig. 1B). Transgene expression again increased ∼5-fold in the fibroblasts but remained unchanged in the liver epithelial cells.

FIG. 1.

Determination of susceptibility of cells to OAdV infection. Ovine and human (A) as well as rodent (B) cell lines were infected at an MOI of 2 (shaded bars) or 20 (solid bars). The concentrations of hAAT produced between 0 and 24 and between 48 and 72 h p.i. were determined in cell culture supernatants by ELISA. Background levels of hAAT were less than 2 ng/ml in the supernatants of all cell lines tested. The error bars indicate standard deviations.

Replication of viral DNA.

Based on the above results, we wished to determine if the elevated transgene expression correlated with viral DNA replication. Cells were infected at an MOI of 2 or 20 infectious particles per cell for ovine and nonovine cells, respectively, and DNA was isolated at various time points. Viral DNA was detected by Southern blotting using a specific probe spanning nucleotides 1968 to 3408 of the OAdV genome (Fig. 2). Strong replication was observed in permissive CSL503 cells, with significant replication in the fibroblasts IMR-90 (84-fold increase in DNA copy numbers between 1 and 72 h p.i.) and 208F (47-fold increase) (Fig. 2A), although there is at least a 100-fold difference in viral DNA replication between ovine and nonovine fibroblasts (10-fold-lower MOI at infection and 10-fold-smaller amount of DNA applied to Southern blotting with CSL503 cells than with IMR-90 and 208F cells). Comparable results were also obtained with a probe from the right-hand end of the viral genome, indicating that the whole OAdV genome was replicated (data not shown). In contrast, by the same analyses, replication of viral DNA was much lower in MCF-7 cells (8.5-fold increase). Weak or no viral DNA replication was also observed in other cell lines of human and rodent origin (HepG2, HuH7, SaOS2, 293, AII, C2C12, and HepSV40), as well as in primary human hepatocytes (Fig. 2B). We conclude that most cell lines of nonovine origin are not capable of efficiently replicating OAdV DNA after being infected but that certain cells of fibroblast origin support viral DNA replication to a certain extent.

FIG. 2.

OAdV DNA replication. (A) Southern blot analyses of fibroblast cells infected with OAdV. Samples were harvested at the indicated time points. The position of the expected 2,399-bp EcoRI fragment from OAdV is indicated. EcoRI-cut OAdV DNA equivalent to one and five OAdV genome copies per cell served as standards. (B) Southern blot analyses of nonfibroblast cells infected with OAdV.

Detection of early and late viral transcripts.

We next asked whether the limited OAdV DNA replication in nonovine fibroblasts was due to opportunistic DNA replication or whether, as expected, the viral DNA replication machinery was active in these cells. Since early gene products of the E2 region are responsible for adenovirus DNA replication, we determined whether transcripts specific for DBP, the major E2 product, could be detected in permissive CSL503 cells as well as in IMR-90 and 208F cells. DBP was detectable after 9 h in all three cell lines, and transcript levels appeared to be at a maximum by 24 h in both ovine and nonovine fibroblasts (Fig. 3B). In contrast, in MCF-7 cells, which had shown very weak DNA replication, the onset of DBP RNA expression was clearly delayed, reaching maximum levels only after 72 h. Similar results were observed with E43 transcripts (Fig. 3C). This finding is in agreement with an earlier study (24) in which the E2 promoter was shown to be active in several nonovine cell lines, although activation of the E2 promoter occurred much more slowly in these cells than in IMR-90 and 208F cells, which show a strong E2 gene expression at only 9 h p.i.

FIG. 3.

OAdV early transcripts. (A) Transcription map of early OAdV genes. The asterisks indicate forward and reverse primer positions. (B and C) Cells were infected at an MOI of 20, and RNA was isolated at the time points indicated. RT-PCR was performed with 1 μg of total RNA using primers specific for DBP (B) and E43 (C) transcripts as described in Materials and Methods. The specific PCR products come from spliced mRNA species and are indicated by arrows on the right. n.i., not infected.

We next determined whether transcripts of the late transcription unit were present in nonovine fibroblasts. The late transcription unit (Fig. 4A) codes for structural proteins, and all late transcripts are expressed from the MLP. In several cell types tested, the OAdV MLP showed only little activity when cloned together with the tripartite leader sequence in front of a reporter gene, and MLP activity could not be enhanced by OAdV infection (24) (Fig. 4B). However, fiber gene transcripts were readily detected in the fibroblasts IMR-90 and 208F (Fig. 4C). Addition of araC, a general DNA synthesis inhibitor that also blocks replication of OAdV DNA, markedly reduced transcription of the fiber gene in all cell lines, indicating that the MLP of OAdV requires DNA replication in order to be active, as is the case for mastadenoviruses. Cloning and sequencing of the fiber-specific RT-PCR products from nonovine fibroblasts revealed that the fiber gene was correctly spliced to the tripartite leader sequence (54) in these cell lines. When the temporal production of the fiber mRNA was examined, we found that fiber gene transcription in IMR-90 lagged behind that of CSL503 (Fig. 4D). Transcripts for p52/55, HAP, and penton were also readily detected in nonovine fibroblasts (Fig. 4E). Although the RT-PCR data do not represent strictly quantitative results, a reduced rate of transcription of late genes in the nonpermissive fibroblasts seemed to be apparent. In contrast, none of the late transcripts investigated here were detectable in MCF-7 cells up to 72 h p.i. (data not shown).

FIG. 4.

OAdV MLP activity and late transcript synthesis. (A) Transcription map of the late OAdV genes. The positions of the MLP, the tripartite leader (TPL) exons, and the late gene open reading frames are given by arrows. The positions of primers used for RT-PCRs are indicatedby asterisks. (B) Activity of the isolated OAdV MLP. pMLPtpl-luc and pCMVluc were transfected onto cell lines of ovine and human origin, respectively, and either left uninfected (solid bars) or infected (shaded bars) with wild-type OAdV at an MOI of 2 (CSL503) or 20 (other cell lines). Reporter gene expression obtained with the CMV promoter-driven luciferase gene (hatched bars) was set to 100%. MLP activity was less than 0.2% of CMV promoter activity in all cell lines tested. Shown are mean values of two independent experiments. (C through E) Synthesis of OAdV late transcripts. Cells were infected as for Fig. 3 with or without the addition of 20 μg of araC/ml, and RNA was isolated at the time points indicated. RT-PCR was performed with 1 μg of total RNA using the primers described in Materials and Methods. The arrows on the right indicate the positions of the transcripts specific for late gene products. n.i., not infected. (C) araC inhibits fiber transcript synthesis in CSL503, IMR-90, and 208F cells. (D) Time course of fiber transcript synthesis in CSL503 and IMR-90 cells. (E) Synthesis of transcripts for p52/55, HAP, and penton genes at 48 h p.i. in CSL503, IMR-90, and 208F cells.

Detection of late viral proteins.

Since late transcripts were produced in nonovine fibroblasts, we questioned whether late protein synthesis occurred. For this purpose, a rabbit serum was raised against purified OAdV and used in ELISA and Western blotting experiments. It was assumed that the serum reacted mainly with structural proteins of the virus, which are expressed from the MLP late in infection. Late viral proteins produced by IMR-90 and 208F cells were clearly detected in an ELISA (Fig. 5A), albeit at a much reduced level compared to permissive CSL503 or HVO156. The reduction in the amount of late viral proteins was estimated to be about 30-fold, and addition of araC reduced late protein to background levels (data not shown). Western blot experiments (Fig. 5B) revealed that the pattern of late viral proteins differed between permissive and nonpermissive cells. IMR-90 cells produced no hexon or fiber/IIIa protein and hardly any penton, proteins that comprise the major components of the viral capsid. However, three new virus-specific bands appeared at approximately 70, 40, and 35 kDa, indicating that late viral proteins were not properly synthesized in these cells. These bands were sensitive to araC, indicating that they corresponded to newly synthesized late-viral-gene products. The same pattern of virus-specific bands was also observed in other cells, such as SW1116, HuH7, HepSV40, and FE8 (data not shown).

FIG. 5.

Synthesis of late viral proteins. Ovine or nonovine cells were infected with OAdV at an MOI of 2 or 20, respectively, with or without the addition of araC. Production of late viral proteins was monitored in the cell lysates at 72 h p.i. (A) ELISA analysis of late protein synthesis in ovine (CSL503 and HVO156) and nonovine (IMR-90 and 208F) fibroblasts. OD₄₉₀, optical density at 490 nm. The error bars indicate standard deviations. (B) Western blotting of late OAdV proteins from CSL503 (lanes 1 to 3), HVO156 (lanes 4 to 6), and IMR-90 (lanes 7 to 9) with (+; lanes 2, 5, and 8) or without (−; lanes 3, 6, and 9) 20-μg/ml araC treatment. Lane 10, protein from approximately 10⁷ infectious particles of a double CsCl gradient-purified OAdV preparation. n.i., not infected. The arrows indicate the positions of newly synthesized OAdV proteins of an unexpected size in IMR-90 cells. The positions of the protein size markers are given on the left.

Finally, synthesis of late viral proteins was also detected by immunofluorescence (Fig. 6). Once again, compared with uninfected cells (Fig. 6A and E), no significant level of late viral proteins was detected in cells treated with araC, indicating that input capsid proteins do not produce a signal and confirming that the serum detected mainly late proteins (data not shown). Cells not treated with araC showed signals for late viral proteins at a reduced rate in IMR-90 cells compared to CSL503 cells (Fig. 6B and F). In addition, as the intensity of the DAPI staining was much higher in CSL503 cells than in any of the other cell lines tested (Fig. 6C and G), this confirmed that the extent of viral DNA replication was reduced in IMR-90 cells. Finally, as shown by the superimposition of images, in CSL503 cells, late viral proteins localized to the nucleus, where virus assembly takes place, but they remained mainly cytoplasmic in IMR-90 cells (Fig. 6D and H).

FIG. 6.

Detection of OAdV proteins by immunofluorescence at 72 h p.i. CSL503 cells (A, B, C, and D) were grown on coverslips and infected at an MOI of 2 PFU/cell. IMR-90 cells (E, F, G, and H) were infected at an MOI of 20 PFU/cell. Negative controls (not infected) are depicted in panels A and E. Nuclei (C and G) were stained with DAPI. Superimpositions of immunofluorescence and DAPI staining are shown in panels D and H.

Abortive infection of human and animal cells.

We next asked whether any infectious virus was produced in nonovine cells, especially where late promoter activity was detected. OAdV was passaged twice on either CSL503, IMR-90, or 208F cells and a third time on CSL503 cells. In each case, 10% of the freeze-thaw lysate of the preceding passage was used for reinfection. No viral DNA was detected in Hirt extracts of passage 3 cells when either IMR-90 or 208F cells were used initially (Fig. 7A). Importantly, whereas the virus titer (determined on permissive CSL503 cells) remained constant in lysates from CSL503 cells up to passage 3, it decreased dramatically (by >99%) in IMR-90 and 208F cells after only one passage. No virus was detected from passage 2 onwards by the method applied for titer determination, which corresponds to <20 infectious viral particles per ml (Fig. 7B). Finally, to check for the presence of premature or defective viral particles not detectable as infectious viruses in the previous experiment, electron microscopy was applied to IMR-90 cells infected with OAdV. OAdV icosahedral viral capsids were readily detected in the nuclei of CSL503 and HVO156 cells, but no viral particles were visible in IMR-90 cells (Fig. 8). Taken together, our findings indicate that, despite DNA replication and expression of early and late genes, no infectious virus is produced in IMR-90 and 208F cells.

FIG. 7.

Passaging of OAdV through ovine and nonovine cells. Cells were infected with OAVhaat as indicated and harvested 72 h p.i. by three freeze-thaw cycles; 10% of the lysates were used to reinfect cells of the same origin. At 72 h p.i., the cells were once again harvested, and the lysates were subsequently used to infect permissive CSL503 cells. (A) Hirt extracts and subsequent BamHI restriction digest after the final passage through CSL503. The sizes (in base pairs) of the expected OAdV BamHI fragments are given on the left. M, 1-kb DNA ladder. (B) Percent virus yield after every passage determined by an endpoint dilution assay on CSL503 cells. The cells were initially infected at an MOI of 20, corresponding to 100% (1e + 2).

FIG. 8.

Electron microscopic images of OAdV-infected cells. (A) CSL503 cells infected at an MOI of 20 and harvested 72 h p.i. Magnification, ×37,400. (B) HVO156 cells infected at an MOI of 20 and harvested 72 h p.i. Magnification, ×37,400. (C) IMR-90 cells infected at an MOI of 100 and harvested at 72 h p.i. Magnification, ×47,000.

DISCUSSION

Nonhuman adenoviruses are gaining increasing importance as vectors for vaccine and gene therapy applications. The realization that preexisting immunity to human adenoviruses in the majority of the population poses a major problem for the use of these vectors prompted us and others (18, 28, 29, 32, 34, 38, 47) to explore adenoviruses of nonhuman origin. Many of the results are promising and suggest that nonhuman adenoviruses are a useful alternative and addition to the currently used human adenovirus-derived vectors. However, safety aspects, which are rooted in the biology of the virus itself, have hardly been looked at. In the present study, we investigated the biology of an ovine-adenovirus-derived vector in human and other nonovine cells in some detail. We showed that viral DNA replication and expression of viral genes occurred in some nonovine cells but that productive replication was limited to permissive ovine cells. Moreover, infection with OAdV was blocked at different stages of the viral life cycle in different cell types.

In this as well as in other studies (24, 31), it was shown that OAdV recombinants infect a wide spectrum of nonovine cells. By analysis of reporter gene expression following infection, we found a clear preference of OAdV for fibroblast-derived cell lines. As both known producer cell lines for OAdV, the ovine cell lines CSL503 and HVO156, also have a fibroblast phenotype, such cells may express higher levels of surface proteins that serve as the OAdV receptor. The OAdV receptor is not yet known but is distinct from the CAR receptor, which is utilized by some human adenoviruses (4, 58). Since the OAdV penton lacks the RGD motif typical of human adenoviruses (54), molecules other than α_v-integrins (57) may serve as coreceptors for OAdV. Alternatively, no secondary receptor may be required, as changing only the cell-binding domain of the fiber protein profoundly influenced OAdV tropism (58). Additionally, fibroblasts may also contain cellular factors involved in certain steps of virus replication.

Infection with OAdV was blocked at different stages of the viral life cycle depending on the cell type (summarized in Table 1). Whereas the increased reporter gene expression over time observed for the fibroblasts IMR-90 and 208F corresponded to relatively strong DNA replication in these cells, no significant or only weak replication of viral DNA was observed in most other cell lines, including human tumor cells of different origins and primary human hepatocytes, as well as rodent liver and myoblast cells. These results are in accordance with previously published results (24), where no DNA replication was observed in several human cell lines. Interestingly, OAdV did not replicate its DNA in human 293 cells, which contain the E1 region of Ad5. Therefore, Ad5 E1 gene products are not capable of inducing OAdV DNA replication in cells that do not support OAdV growth; hence, there is no interaction between the two viruses, which is in accordance with recently published results (30).

TABLE 1.

Steps of OAdV replication and their efficacies in cell lines investigated^a

Cell line	Origin	Cell type	DNA replication	Early transcription	Late transcription	Late protein synthesis	Virus production
CSL503	Ovine	Fibroblast	++++	++++	++++	++++	++++
HVO156	Ovine	Fibroblast	++++	++++	++++	++++	++++
IMR-90	Human	Fibroblast	++	++	+	Faulty	−
208F	Rat	Fibroblast	++	++	+	Faulty	−
MCF-7	Human	Epithelial	+/−	Delayed	−	−	−

^a++++, very strong; ++, strong; +, clearly detectable; +/−, weak; −, nondetectable.

In nonfibroblast cells, the lack of viral DNA replication was not generally due to a block of early gene expression or to early promoter inactivity, since MCF-7 cells still produced transcripts of the early regions 2 and 4. However, the expression of the DBP gene lagged behind that observed in ovine and nonovine fibroblasts. Diminished expression of E2 genes was suggested as a major cause of the early block in Ad12 replication in hamster cells, although reports about the levels of single E2 gene products are contradictory (21, 36). Reduced amounts of E1 mRNAs were also reported for abortive Ad12 infection (35). In addition to the expression of certain viral genes, the availability of cellular factors might be crucial for viral DNA replication. In the case of Ad2 and -5, the involvement of NF1 and Oct-1/NFIII in DNA replication has been well established (reviewed in reference 12), and a low affinity of the Ad12 origin for NFIII present in hamster cells and/or low nuclear concentration of NFIII may contribute to the abortive character of Ad12 infection in BHK cells (20, 49). However, the involvement of both of these transcription factors in stimulating the replication of other than A and C group adenoviruses has not yet been established, and the absence of potential NFI and Oct-1 binding sites from the OAdV origin may therefore be of minor importance. Nevertheless, the participation of other cellular factors in OAdV replication initiation is likely, and higher expression of these factors in ovine and nonovine fibroblasts might contribute to the observed levels of viral DNA replication in these cells.

In contrast to most tumor-derived cell lines, we clearly detected early and late OAdV transcripts in fibroblasts of human and rodent origin. However, compared to permissive ovine cells, late protein synthesis was strongly hampered in these nonpermissive fibroblasts. Similar results with respect to late mRNA synthesis were obtained for Ad2 infection of monkey cells and Ad12 infection of hamster cells stably transfected with the Ad5 E1 region (2, 22, 26, 49). In both systems, a reduction in late mRNA levels was observed. In the case of Ad2, a delayed onset of late transcription of L3 and L5 genes was reported, and reduced mRNA levels were due to decreased RNA synthesis and not to diminished RNA stability in the nonpermissive cell. At least for the Ad12 system, no late protein synthesis was observed, whereas several reports showed low expression of a limited number of late genes after Ad2 infection of rodent cells (14, 51). We detected some OAdV late protein synthesis in nonpermissive fibroblasts in ELISA, Western blotting, and immunofluorescence experiments. However, we found that (i) the level of late OAdV proteins was markedly reduced, (ii) only a few proteins of an unexpected size were synthesized, and (iii) the intracellular localization of late proteins was different from that seen after infection of permissive ovine cells, which resembled the situation in Ad2-infected monkey kidney cells (8). Major late gene products, such as hexon and fiber proteins, were not detected at the expected positions in Western blot experiments. Therefore, translation and/or posttranslational modification of OAdV late proteins seems to be defective in nonovine fibroblasts, and in consequence, no virus was produced, as shown by repeated passaging experiments. For human adenoviruses, the block to late protein synthesis, especially to fiber production, was explained by premature termination of translation at an attenuator site and by splicing defects. Specifically, the absence of the ancillary leaders x and y in fiber transcripts, as well as a high number of incompletely spliced fiber transcripts, was reported for abortive Ad2 infection of monkey cells (1, 26, 50). However, cloning and sequencing of OAdV fiber transcripts from ovine and nonovine cells revealed no difference in the leader sequences (data not shown), and longer fiber-specific transcripts that could be caused by incomplete splicing were not detected by RT-PCR. Therefore, an altered structure, especially of the fiber mRNA, may not be the major cause of the late block to OAdV infection. We rather speculate that the single protein at ∼70 kDa in abortively infected IMR-90 cells corresponds to an incorrectly glycosylated fiber protein. Unfortunately, there is a lack of specific antibodies to individual late OAdV proteins needed to further investigate this hypothesis.

The impact of viral DNA replication and synthesis of OAdV gene products on the in vivo characteristics of OAdV-derived vectors is not yet clear. Earlier experiments showed that, despite the high-level transgene expression, OAdV did not produce detectable transcripts of early and late genes after intramuscular injection in mice, whereas early transcripts were produced after infection with an Ad5 vector with E1 deleted (32). However, transgene expression from OAdV-derived vectors is transient, and vector genomes are rapidly cleared from the infected tissues after systemic administration (18). The mechanism of in vivo vector clearance, therefore, remains to be elucidated. The expression of viral genes and the presentation of their products to the immune system was a major barrier to long-term expression after gene transfer with vectors derived from human adenoviruses, and only the development of vectors with all expressed viral sequences deleted has overcome this problem (39, 40, 48). Similar strategies will be necessary for OAdV-based vectors if long-term expression of the transferred genes is required. Therefore, studies are in progress to attenuate the virus and, as a first step towards this goal, to define viral functions involved in control of DNA replication and early gene expression. On the other hand, we conclude that even in the presence of viral DNA replication and synthesis of early and late gene products, OAdV replication remains blocked in nonovine cells. Together with the lack of transforming ability of OAdV (60) and the lack of trans-complementation of OAdV by Ad5 (30), our data have important positive implications for the biosafety of OAdV vectors.