Replication Properties of Human Adenovirus In Vivo and in Cultures of Primary Cells from Different Animal Species

Abstract

Oncolytic adenoviruses have emerged as a promising approach for the treatment of tumors resistant to other treatment modalities. However, preclinical safety studies are hampered by the lack of a permissive nonhuman host. Screening of a panel of primary cell cultures from seven different animal species revealed that porcine cells support productive replication of human adenovirus type 5 (Ad5) nearly as efficiently as human A549 cells, while release of infectious virus by cells from other animal species tested was diminished by several orders of magnitude. Restriction of productive Ad5 replication in rodent and rabbit cells seems to act primarily at a postentry step. Replication efficiency of adenoviral vectors harboring different E1 deletions or mutations in porcine cells was similar to that in A549 cells. Side-by-side comparison of the viral load kinetics in blood of swine and mice injected with Ad5 or a replication-deficient adenoviral vector failed to provide clear evidence for virus replication in mice. In contrast, evidence suggests that adenovirus replication occurs in swine, since adenoviral late gene expression produced a 13.5-fold increase in viral load in an individual swine from day 3 to day 7 and 100-fold increase in viral DNA levels in the Ad5-infected swine compared to the animal receiving a replication-deficient adenovirus. Lung histology of Ad5-infected swine revealed a severe interstitial pneumonia. Although the results in swine are based on a small number of animals and need to be confirmed, our data strongly suggest that infection of swine with human adenovirus or oncolytic adenoviral vectors is a more appropriate animal model to study adenoviral pathogenicity or pharmacodynamic and toxicity profiles of adenoviral vectors than infection of mice.

Since the initial description in the 1950s, adenoviruses have been known as a cause of common childhood respiratory illnesses (13, 28, 48). In immunocompetent patients, most of these infections are asymptomatic, mild, or self-limited. In the immunocompromised host, however, adenovirus infections can cause severe localized disease including pneumonitis, colitis, hemorrhagic cystitis, hepatitis, nephritis, encephalitis, or disseminated disease with multiorgan failure (6, 10, 27, 44). Disseminated adenovirus disease has been reported with increasing frequency in patients with AIDS or malignancies (3, 27, 29, 32) and in bone marrow or solid-organ transplant recipients (6, 9, 14, 38, 42).

Recombinant adenoviral vectors have emerged as promising tools for gene therapy as they are able to infect a wide variety of cell types in vitro and in vivo, including postmitotic cells (61). Currently, numerous phase I and II human trials use human adenovirus type 2 (Ad2)- and Ad5-based vectors; most studies attempt to treat various types of cancer by direct intralesional injections (61). Since adenoviral particles are too large to travel within a tumor by diffusion and convection after intralesional injection, intratumoral dispersion of replication-deficient vectors is predominantly confined to close proximity of the injection site (46). Most clinical cancer gene therapy trials using such vectors have shown minimal gene transfer at best and few tumor responses (46, 47, 51). Replication-competent vectors have the potential to overcome this gene delivery barrier, as they travel from the initial site of infection mainly by within-cell and between-cell transport via ongoing cycles of infection, replication, and cell lysis.

The first controlled clinical trials using oncolytic adenoviruses as a single-agent treatment have proved disappointing; however, in combination with chemotherapy encouraging antineoplastic activity was seen (15, 31). Current clinical results with oncolytic adenoviruses suggest strong safety (11, 34). However, their ability to replicate and spread increases the risk for adverse effects compared to their replication-deficient counterparts.

Although there is clear evidence of efficacy, the available clinical data using oncolytic adenoviruses suggest that even with multiple intratumoral injections, there is room for improving the efficacy of treatments to achieve a therapeutic benefit for a larger proportion of patients. To improve the potency of oncolytic viruses, the dynamics between virus replication in the tumor, spread, and extratumoral replication, on one side, and immune-mediated viral clearance, on the other side, have to be better understood (57).

Replication of human adenoviruses has long been thought to be restricted to human cells. However, it is becoming evident that at least certain cell lines from other species support the full replication cycle of adenovirus although at seemingly reduced efficacy. Screening of a panel of murine epithelial cell lines in a viral burst assay revealed that some of the cell lines produced substantial amounts of infectious virus, although at levels approximately 25- to 100-fold lower than the human cells included as positive controls in these studies (17, 25). Intravenous injection of large doses (10⁹ PFU) of adenovirus into mice led to abortive but lytic infection in most cells of the liver (12). However, an increase in PFU in liver from day 1 to 2 by several orders of magnitude also indicated virus replication in at least some cells of this organ. Large doses of about 10¹⁰ PFU were also needed to induce signs of pneumonia after intranasal injection in mice (20), while 10⁸ PFU were sufficient to induce signs of pneumonia after intranasal injection into cotton rats (19). Replication of adenovirus was observed in pulmonary tissue of cotton rats but not of mice. However, even in cotton rats replication of adenovirus was not required for induction of pneumonia (18), indicating that innate host immune responses to the injected viral particle (VP) and/or to expressed early gene products mediate this pneumonia.

Preclinical toxicology studies and host-vector interactions relevant to the oncolytic efficacy and safety of human adenovirus-based vectors are hampered by the lack of an animal model allowing significant replication of human adenovirus serotypes. Therefore, as a model for their human counterparts, studies with animal adenoviruses, such as the canine adenovirus (26), have been conducted in their natural host, which should provide a means to address the principal safety and efficacy issues of oncolytic adenoviruses. However, the conclusion that can be drawn for a particular oncolytic adenovirus for use in humans might be limited.

In order to identify potential alternative animal models for adenoviral pathogenicity studies and evaluation of the safety of oncolytic adenoviral vectors, we performed a comparative study on the replication properties of human Ad5 in a panel of primary cell cultures from different animal species. Efficient virus replication was only supported by porcine cells. In addition, a pilot study suggested replication of human Ad5 in swine and provided initial evidence for adenovirus-induced interstitial pneumonia.

MATERIALS AND METHODS

Cell lines and cell culture.

Permanent cell lines were propagated in D-10, consisting of Dulbecco's modified Eagle medium with high glucose, supplemented with 10% heat-inactivated fetal bovine serum and 50 μg/ml gentamicin. Tissue culture medium and supplements were purchased from Invitrogen/Gibco (Karlsruhe, Germany). Cells were maintained in the logarithmic phase of growth at 37°C in a humidified atmosphere of 95% air and 5% CO₂.

The human embryonic kidney cell line 293 (23) was obtained from Microbix Biosystems, Inc. (Toronto, Ontario, Canada). The cell lines A549 (CCL-185) (36), HeLa (CCL-2) (49), and 3T6 (CLL-96) (53) were purchased from the American Type Culture Collection, Manassas, Va. The porcine kidney cell line PK15 (CCL-33) was kindly provided by Martin Heller, Bundesforschungsanstalt für Viruskrankheiten der Tiere, Jena, Germany. We established primary cultures from kidney, lung, and liver cells from mouse, cotton rat, rabbit, swine, hamster, guinea pig, and woodchuck by enzymatic digestion of fresh tissue samples with collagenase, hyaluronidase, and DNase as described previously (50). Cells were cultured in renal epithelial basal medium (REBM), bronchial epithelial medium (BEGM), or hepatocyte culture medium (HCM) (Cambrex Bio Science Verviers, Verviers, Belgium), respectively, and mixed with Eagle medium-F-12 (1:1) supplemented with 10% fetal bovine serum, gentamicin, and human epidermal growth factor (2 ng/ml) (Seromed, Berlin, Germany). Contaminating fibroblasts were separated from epithelial cells by negative panning and by several rounds of differential trypsinization to enrich cultures for more adherent epithelial cells.

Adenoviruses.

Ad.TK is a first generation replication-defective adenoviral vector encoding herpes simplex virus-1 thymidine kinase (HSV-tk) (58, 60). Both the adenoviral vector Ad.TK^RC(II), which has a deletion of E1B-55K, and the Ad.OW34 vector, which encodes the entire viral E1 region, are replication-competent and carry the HSV-tk gene (60). The vector Ad.OW34Δ24 is identical to the Ad.OW34 vector, but it encompasses an eight-amino-acid deletion in the CR2 domain L₁₂₂TCHEAGF₁₂₉ that is responsible for viral E1A binding to the retinoblastoma protein, as described previously by Fueyo and coworkers (16).

Both the replication-competent Ad.OW94 vector and the replication-defective Ad.GFP vector encode the cDNA for the enhanced green fluorescent protein (GFP) and have been described in detail elsewhere (41, 59). The reference strain VR-5 (ATCC, Manassas, VA) was used as wild-type Ad5. Viruses were propagated in 293 or HeLa cells, purified by two rounds of CsCl density centrifugation (21), dialyzed (Slide-A-Lyzer; Pierce, Rockford, IL) against 1,500 ml of phosphate-buffered saline with 1 mM MgCl₂ and 10% glycerol four times (1 h each) at 4°C, and stored at −80°C until use. Concentration and bioactivity of the adenovirus vectors were determined by measuring absorbency at 260 nm and 50% tissue culture infective doses using 293 cells (39).

Quantitative PCR.

We determined the adenoviral DNA copy number in the samples using a quantitative PCR (QPCR) assay. DNA was isolated using a Genomed JETQUICK Tissue Spin Kit (Löhne, Germany) according to the manufacturer's instructions. DNA concentration was determined spectrophotometrically and adjusted to 200 ng/μl; 1 μl of DNA was used for PCR. The TaqMan PCR amplifying and quantitating the Ad5 E4orf6 region has been described previously (59). Briefly, primer and probe sequences were as follows: E4 forward primer, 5′-GTAATTCACCACCTCCCGGTA-3′, and E4 reverse primer, 5′-GGCTCTCCACTGTCATTGTTC-3′; E4 probe, 5′-(FAM)ACCTCTGATTAAACATGGCGCCATCC(TAMRA)-3′ (where FAM 6-carboxyfluorescein is and TAMRA is 6-carboxytetramethylrhodamine).

For quantitation of the adenoviral load in swine, data were collected and analyzed using a LightCyler II sequence detection system (Roche Diagnostics, Mannheim, Germany). The viral load in mice was quantified with the same PCR assay on a Rotor-Gene RG-3000 system (Corbett Research, Mortlake, NSW, Australia). The lower limit of the linear range of the quantitative assay was determined to be 100 adenovirus DNA copies in the background of cellular genomic DNA, which corresponds to 500 Ad5 DNA copies/μg of DNA and 5,000 Ad5 DNA copies/ml of serum.

Quantitation of productive Ad5 replication in primary cell cultures.

To improve replication and cell release of the GFP encoding replication-competent adenoviral vector Ad.OW94 by complementing for viral E3 functionality (52, 54), this vector was mixed with wild-type Ad5 in a VP ratio of 80:1. Cell monolayers were incubated for 1 h with 10-fold serial dilutions of this virus mixture ranging from 2 × 10¹¹ to 2 × 10⁶ VPs. At the highest concentration, cell monolayers were exposed to 2 × 10¹¹ VPs of Ad.OW94 and 2.5 × 10⁹ VPs of Ad5, which translates to a multiplicity of infection (MOI) of 200. After extensive washing, cell monolayers were incubated with 5 ml of fresh medium. Ten minutes later (70 min postinfection) and 48 h postinfection, an aliquot of the supernatants was collected, centrifuged (1,000 × g for 10 min), and stored without cell pellet at −20°C until it was analyzed for infectious Ad.OW94 vector. One milliliter of the supernatants was transferred onto 1 × 10⁵ subconfluent 293 cells. Twenty-four hours later, GFP expression of 10⁴ cells was analyzed by flow cytometry (FACSCalibur flow cytometer; Becton Dickinson Immunocytometry Systems, Mansfield, MA).

Determination of Ad5 infection efficiency.

Subconfluent cell monolayers were incubated for 1 h with 10-fold serial dilutions of the same viral mixture used for the viral burst assay ranging from 2 × 10¹¹ to 2 × 10⁶ VPs. Twenty-four hours later, GFP expression was analyzed by flow cytometry.

Analysis of adenoviral L3 expression.

Snap-frozen tissue samples were homogenized in 1 ml of Trizol reagent (Invitrogen/Gibco) per 50 mg of tissue with a Polytron PT2100 power homogenizer (Kinematica, Littau-Lucerne, Switzerland). Total RNA extraction proceeded as directed by the Trizol manufacturer. To decrease DNA contamination of the RNA, a second Trizol extraction was performed. Two hundred micrograms of total RNA was used to isolate mRNA with an Oligotex mRNA mini kit from QIAGEN (Hilden, Germany).

The hexon, which is the major structural protein of the adenovirus capsid, is generated by splicing from the 28-kb major late transcript and contains the 150- to 200-nucleotide tripartite leader sequence at its 5′ end (1). Five hundred nanograms of mRNA of the frozen-tissue RNA samples was subjected to reverse transcription-PCR (RT-PCR) analysis in a single-step procedure using a QuantiTect Probe RT-PCR kit (QIAGEN). Reverse transcription was carried out at 50°C with specific primers, followed by hot-start PCR in the same tube. The cycle conditions were the following: initial denaturation incubation at 94°C for 15 min followed by 50 cycles of successive incubation at 94°C for 30 s, 56°C for 1 min, and 72°C for 30 s. Samples omitted from the RT step, starting with the initial denaturation, served as controls. Specific primers were designed to produce a 299-bp amplicon, containing the third leader of the tripartite leader and, partially, the hexon-coding region so as to amplify only cDNA and not contaminating viral DNA. Primer sequences were as follows: tripartite forward primer, 5′-CGAGAAAGGCGTCTAACCAGTC-3′, and hexon reverse primer, 5′-CCCCAGCTAGGGTGAACCG-3′. The 299-bp RT-PCR products were analyzed by agarose gel electrophoresis (2% agarose in 1× Tris-borate-EDTA buffer) and stained with ethidium bromide.

Animal studies.

BALB/c mice, cotton rats (Sigmodon hispidus), New Zealand White rabbits (Oryctolagus cuniculus), hamster (Mesocricetus auratus), and guinea pigs (Cavia aperea) were purchased from Charles River Laboratories (Sulzfeld, Germany). Tissue from woodchuck (Marmota monax) was kindly provided by M. Lu (Institute of Virology, University Duisburg-Essen, Essen, Germany).

In addition, in this study, a total of four sows (Sus scrofa; weight, 23 ± 1 kg) were used in accordance with institutional guidelines. Studies were performed at BioTest Ltd., Konárovice, Czech Republic. To assess liver toxicity, the serum levels of bilirubin, aspartate aminotransferase, alkaline phosphatase, γ-glutamyltransferase (GGT), and glutamate dehydrogenase (GLDH) were determined by VetMedLabor GmbH Ludwigsburg, Germany.

Tissue samples were harvested immediately after euthanization of the animals with separate sets of sterile, DNA-free instruments. All samples were snap-frozen at the time of collection and stored in cryotubes at −80°C until analysis.

Statistical methods.

The software package STATISTICA (version 5.5 for Windows; StatSoft, Inc., Tulsa, OK) was used for statistical data analysis.

RESULTS

Replication efficiency of Ad5 in primary cell cultures.

To screen for laboratory animals that might be permissive to human Ad5 infection, a panel of 14 different primary cell cultures from kidney, lung, and/or liver of seven different species was established and infected with serial dilutions of a replication-competent, E3-deleted Ad5-based vector encoding GFP (Ad.OW94). The amounts of infectious Ad.OW94 virus released from the tested cell cultures 70 min and 48 h postinfection were measured by transfer of the supernatants onto 293 cells and subsequent determination of GFP-positive cells after an additional 24-h incubation period. The difference in the percentage of GFP-positive 293 cells incubated with the supernatant collected 70 min and 48 h postinfection should represent productive viral replication and release (Fig. 1). As a positive control, adenovirus replication was also analyzed in human A549 cells, which revealed substantial production of infectious virus even at an inoculation dose of 10⁸ VPs/flask, corresponding approximately to 10⁶ PFU/flask and an MOI of 0.2 per cell. In contrast, inoculation doses at least 1,000-fold higher were needed in primary cell cultures from mice, cotton rat, rabbit, guinea pig, and woodchuck for production of comparable amounts of infectious Ad.OW94. Infection of primary cell cultures from hamster led to production of infectious virus at an inoculation dose of approximately 10¹⁰ VPs/flask. Surprisingly, the release of infectious adenovirus from primary porcine kidney cells and a permanent kidney cell line (PK15) after 48 h was similar to that of highly permissive human A549 cells.

FIG. 1.

Biological assay to quantitate production of infectious adenovirus. Under the same conditions, subconfluent monolayers of primary cultures from the indicated species and tissues were incubated for 1 h with serial dilutions of a replication-competent adenovirus encoding GFP and Ad5 mixed in a VP ratio of 80:1. An inoculation dose of 2 × 10¹¹ VPs/flask translates to an MOI of 200 per cell. After cell monolayers were extensively washed, an aliquot of the supernatants was taken 70 min and 48 h postinfection (pi). As a measurement of infectious virus release, 293 cells were incubated for 24 h with these supernatants, and GFP expression was analyzed by flow cytometry. The percentages of GFP-positive 293 cells 70 min and 48 h postinfection are presented as the means ± standard deviations of at least three independent experiments. Human A549 cells served as controls.

Infection efficiency of primary cell cultures with Ad5.

The huge differences in release of infectious adenovirus from primary cell cultures from the different species could be due to inefficient transduction (early steps) and/or inefficient late steps of replication. To test whether a block at an early step in the virus replication cycle is responsible for inefficient replication in most rodent cells, primary cell cultures from the different species were incubated with the same viral inoculum used in the adenovirus release assay. One day later the percentage of GFP-positive cells in the primary cell cultures was determined. With the exception of primary cell cultures from cotton rats, inefficient transduction efficiency of rodent or rabbit cells does not seem to be the sole reason for the severe block in virus replication (Fig. 2).

FIG. 2.

Comparative Ad5 transduction efficiency. Indicated cell monolayers were incubated with the same viral inoculum and under the same conditions as described in the legend of Fig. 1. GFP expression was analyzed 24 h later by flow cytometry, and data are presented as the means ± standard deviations of three independent experiments.

Ad5-specific CPE.

Subconfluent monolayers of PK15 and primary cultures of porcine kidney cells were infected at an MOI of 5 with wild-type adenovirus. As shown in Fig. 3, Ad5 caused a nearly complete cytopathic effect (CPE).

FIG. 3.

Adenoviral CPE. Subconfluent monolayers of porcine PK15 cells and primary cultures of porcine kidney cells were infected at an MOI of 5 with Ad5 or mock-infected (control). Seventy-two hours later, cells were photographed through an inverted phase-contrast microscope (original magnification, ×20).

E1-dependent replication of adenoviral vectors in porcine cells.

Deletion of different reading frames of E1 was used to modulate the replication efficiency of oncolytic adenoviral vectors. Therefore, the DNA replication efficiency of adenoviral vectors with differing E1 deletions in porcine PK15 cells and human A549 cells was compared. Both cell lines were infected (MOI of 5) with an E1-deleted Ad.TK vector, an E1A⁺, E1B-55K-deleted adenovirus Ad.TK^RC(II), and an E1⁺ but partially E1A-CR2-deleted vector Ad.OW34Δ24. The E1⁺ adenoviral vector Ad.OW34 served as a control. The quantity of virus recovered from medium was assayed 4 and 48 h after infection by QPCR. As shown in Fig. 4, in the supernatants of A549 cell monolayers infected with Ad.TK^RC(II), Ad.OW34Δ24, or Ad.OW34, the median Ad5 copy number increased 1,266-, 2,643-, and 3,344-fold, respectively. In the supernatant of PK15 cells infected with these adenoviral vectors, the median Ad5 copy number increased 915-, 2,841-, and 3,314-fold, respectively. Comparing the replication of the recombinant adenoviral vectors in A549 and PK15 cells, there was no statistically significant difference. Furthermore, the replication of the Ad.OW34Δ24 and Ad.OW34 vectors was similar in the tested human and porcine cell lines. In both cell lines, the E1B-55K-deleted mutant Ad.TK^RC(II) grew less efficiently than the E1B-55K-positive vectors.

FIG. 4.

Quantitation of adenovirus progeny production in porcine PK15 and human A549 cells by QPCR. Cell monolayers were infected at an MOI of 5 with an E1-deleted Ad.TK vector, an E1B-55K-deleted adenovirus Ad.TK^RC(II) vector, and an E1⁺ but partially E1A-CR2-deleted vector Ad.OW34Δ24. An E1⁺ adenoviral vector Ad.OW34 served as a control. After cell monolayers were extensively washed, the quantity of virus recovered from the medium was assayed 4 and 48 h after infection by QPCR. The increase in adenovirus copy numbers is shown.

Viral load after intravenous injection of Ad5 in swine and mice.

The relevance of the in vitro permissivity of porcine cells to productive Ad5 infection for the in vivo situation was analyzed in a pilot study by intravenous administration of 2 × 10¹² Ad5 VPs to three swine. Animals were sacrificed on day 1, 4, or 7 after infection, and DNA was extracted from three independently excised samples of each organ.

As shown in Fig. 5A, the mean Ad5 DNA copies/μg of DNA in the liver were 45,547, 5,554, and 9,809 on days 1, 4, and 7, respectively. At the same time points, 2.77 × 10⁶, 5.15 × 10⁵, 5.81 × 10⁵ Ad5 DNA copies/μg of DNA were detected in the lung, and 761, 1,161, and 504 DNA copies/μg of DNA in the kidney, respectively. In the heart, skeletal muscle, brain, and gonads at all time points, less than 1,000 Ad5 copies were found. Furthermore, in the liver, the detectable Ad5 copy number dropped from day 1 to days 4 and 7, but there was no obvious difference between days 4 and 7. In the lung, there was no obvious difference in the Ad5 copy number comparing all three time points. However, since only a single animal was analyzed at each time point, we cannot exclude the possibility that the differences observed are due to variations between animals rather than reflecting changes over time or that differences over time are missed due to variations between animals.

FIG. 5.

Biodistribution of adenovirus in swine and mice. Animals received intravenously wild-type Ad5 or a first-generation replication-deficient adenovirus (Ad.TK) at day 0. At 1, 4, and 7 days after virus administration, viral DNA copy numbers in the indicated organs were determined independently in triplicate (swine) or in groups of four mice using QPCR (A). Data are presented as means ± standard deviations. In panel B, the median adenoviral DNA levels determined in triplicate in the blood of individual swine or groups of four mice are shown.

In contrast to viral load measurement in organs, drawing serum samples allowed longitudinal analyses (Fig. 5B). On days 1, 3, 5, and 7 of the porcine study, the median Ad5 copy number per ml of serum was 2.89 × 10⁶, 4.31 × 10⁴, 1.18 × 10⁵, and 5.38 × 10⁵, respectively. Blotting serum viral load levels for each single animal reveals a rapid drop of viral load from day 1 to day 3 and a subsequent increase of viral load by more than a factor of 13.5 between day 3 and day 7 for the animal with the longest observation period. To exclude the possibility that this increase in serum viral load reflects a redistribution phenomenon, one swine was injected with 2 × 10¹² VPs of Ad.TK, a replication-deficient adenoviral vector. A progressive decline of Ad.TK viral load in the serum was observed. On day 7 of the study, the Ad5 concentration in serum was about 100-fold higher than that of Ad.TK.

Since mouse models are frequently used to assess the efficacy and safety of oncolytic adenoviral vectors, we sought to determine whether there are fundamental differences in the organ distribution and viral load kinetics of Ad5 in mice to those of swine. Since we expected lower levels of replication in mice, the Ad5 inoculation dose per kg of body weight chosen for mice was approximately 10-fold higher than the dose used for infection of swine. Groups of four BALB/c mice each were therefore injected intravenously with 2 × 10¹⁰ VPs of the same virus stocks of wild-type Ad5 or replication-deficient Ad.TK used in swine. Viral loads in liver, lung, and kidney were determined 1, 4, and 7 days after administration of Ad5 or Ad.TK (Fig. 5A).

In mice that received Ad5, the mean viral loads on day 1 in liver, lung, and kidney were 1.20 × 10⁷, 8.92 × 10⁶, and 1.33 × 10⁵ Ad5 DNA copies/μg of DNA, respectively, with little change over time. Similar viral load levels were obtained after injection of the Ad.TK vector. Logarithmic transformation of the data and one-way analysis of variance did not reveal significant differences of viral loads in the organs of mice that received Ad5 or Ad.TK. In contrast to the viral load in the organs, in serum a significant decline of viral load levels from day 1 to day 7 was observed for both Ad5 and Ad.TK (Fig. 5B).

Analysis of adenoviral L3 expression.

Since adenoviral late gene expression has been reported to be indicative of productive adenoviral replication (7), RT-PCR was used to measure hexon mRNA levels in tissues of swine given Ad5. In samples treated with reverse transcriptase, viral hexon transcripts were detected on day 7 in lung and liver specimens of the animal infected with Ad5 (Fig. 6). Without reverse transcriptase treatment, no product was detectable, indicating that the detectable transcripts were from hexon mRNA and not from hexon DNA. In the positive control (mRNA from A549 spiked with 1 ng of mRNA from A549 cells infected with Ad5), hexon transcripts were detectable only in the sample treated with reverse transcriptase. As expected, no hexon transcripts could be amplified in noninfected PK15 cells.

FIG. 6.

Ad5 L2 expression in swine. Seven days after intravenous Ad5 injection, viral hexon expression was analyzed by RT-PCR in lung and liver tissue. mRNA from noninfected PK15 cells and A549 cells infected with Ad5 served as controls. No hexon transcripts were detected in the absence of reverse transcription.

Histopathology of swine and mice given Ad5.

Livers and lungs of swine and mice that received Ad5 or Ad.TK were histologically scored for inflammatory changes according to the following scale: 0, normal or insignificant; 1, slight; 2, moderate; and 3, severe inflammatory changes. Representative histological sections are shown in Fig. 7. None of the swine given adenovirus showed significant histological changes in the liver or kidney parenchyma (score of ≤1). However, scores above 2 were obtained for the lungs of all swine given Ad5. On day 7, marked multifocal inflammatory changes in the lung, affecting bronchi, bronchioli, and lung parenchyma, were observed (score of 3). Furthermore, we observed focal hyaline membrane formation, hemorrhage, and edema in this animal. On day 7, histological changes of the animal that received Ad.TK were comparable to those seen in the Ad5-injected animal on day 1 (score of 2). Clinically, no apparent distress was observed in any of the animals.

FIG. 7.

Histology. Liver and lungs of swine and mice were histologically evaluated seven days after intravenous Ad5 or Ad.TK injection. Representative micrographs are shown (hematoxylin and eosin staining; original magnification, ×200 (liver) or ×100 (lung).

In contrast, histological examination of liver samples of mice at days 4 and 7 revealed numerous portal and intralobular infiltrations with lymphocytes and histiocytes (score of 2). Furthermore, some very prominent nuclei with signs of nuclear damage were seen. However, on day 1, we did not observe significant histological changes in liver sections of mice given Ad5. In contrast to the findings in the swine, mice given Ad5 showed only minimal peribronchiolar lymphocytic interstitial infiltration of the lung at days 1, 4, and 7 (score of ≤1).

Comparative histological analysis of liver and lung sections of swine and mice given adenovirus revealed significantly greater changes in the livers of mice and lungs of swine, respectively.

Liver biochemistry of swine.

Since hepatic injury has been reported in humans and nonhuman primates after high doses of intravenously administered adenoviruses (4, 8), we analyzed serum levels of bilirubin, aspartate aminotransferase, alkaline phosphatase, GGT, and GLDH. As shown in Fig. 8, the GLDH level (reference range of <1.8 U/liter) of all three animals rose to 2.6, 4.0, and 2.3 U/liter 24 h after virus injection. In the following days, GLDH dropped to normal levels and rose to 7.3 U/liter 7 days after virus injection. There was no increase in GLDH levels in the animal injected with a replication-defective adenovirus. None of the other monitored hepatic markers significantly changed during the course of the study.

FIG. 8.

Levels of GLDH in serum in swine. Blood was collected at the indicated time points from swine that received Ad5 or Ad.TK intravenously at day 0. Of the markers indicative of hepatic injury, only GLDH levels changed significantly over the course of the study. The shaded area indicates the reference range of GLDH in swine.

DISCUSSION

Oncolytic adenoviral vectors represent a rapidly expanding novel therapeutic option for cancer. However, clinical application has been thwarted by high viral immunogenicity, lack of cell-specific infectivity, poor viral distribution within the tumor, and the lack of a permissive animal model in which parameters can be experimentally modified and optimized.

To screen for laboratory animals that might be permissive to human Ad5 infection, a panel of primary cell cultures from seven different species was established. Using a single-cycle virus burst assay, the replication properties of human Ad5 in these primary cell cultures were compared side by side. Serial dilutions of an E3 deletion mutant of Ad5 encoding GFP (Ad.OW94) were used to infect the various primary cell cultures. To avoid reduced replication due to the lack of adenovirus death protein (E3-11.6K) expression (52, 54), Ad.OW94 was mixed in a VP ratio of 80:1 with wild-type Ad5 complementing the E3 deletion after coinfection of cells. A comparison of results based on low input multiplicities with those obtained at high input multiplicities needs to take into account that at low input multiplicities not all cells were infected with wild-type Ad5. The amount of infectious Ad.OW94 virus released from the primary cell cultures after a 48-h incubation period was measured by infection of 293 cells with supernatants of cells from the primary culture and subsequent determination of the percentage of GFP-positive indicator cells after an additional 24-h incubation period. Since a pilot experiment did not reveal a significant difference in virus production 48 or 72 h postinfection by burst assay and QPCR, in all further experiments virus production was evaluated 48 h postinfection. However, since we did not evaluate virus production at much later time points and since we did not analyze cell lysates, we cannot formally exclude delayed or low-level replication in the rodent cells analyzed. After the cell monolayers were washed extensively 1 h postinfection, an aliquot of the supernatants was taken to discriminate between input virus and virus produced by cells from the primary cell culture. The input virus could only be detected consistently at the highest inoculation dose. The amount of input virus detected at 4 h after infection mirrored the levels detected at 1 h after infection (data not shown). The amount of virus released at these early time points from the various cells tested did not correlate with the amount of virus released at 2 days after infection. At this time point, the highest levels of infectious virus were produced and released from porcine renal cell cultures. Inoculation of these cells at an MOI of approximately 0.02 led to release of substantial amounts of infectious virus, while renal cell cultures from the other animal species tested released lower levels of virus even at inoculation doses that were 10,000-fold higher. This extends a study by Torres et al. (55), who reported highly efficient and productive Ad5 infection of an epithelial porcine testicle and an intestinal cell line. The lack of a side-by-side comparison to murine or cotton rat cells made an evaluation of the relative replication efficiency in different species difficult, particularly since selection of cell lines can lead to biased results. Further analyses showed that Ad5-infected porcine cells showed a CPE analogous to human cells. As in human cells, there was no detectable progeny production of porcine cells infected with an E1-deleted Ad.TK vector. Furthermore, the E1B-55K-deleted adenoviral vector Ad.TK^RC(II) did replicate less efficiently than a similar vector, Ad.OW34, encoding the E1B-55K gene product (60). Interestingly, the vector Ad.OW34Δ24 replicated efficiently in human A549 cells and porcine PK15 cells, as well as primary porcine cells, despite the fact that it should replicate only in cells with down-regulated retinoblastoma pRb function or abnormal cell cycle control (16) as it carries a deletion in the pRb protein-binding region of the E1A gene product. Thus, replication properties of different E1 deletion mutants in a porcine cell line reflect those seen in human A459 cells.

After demonstrating in vitro that primary porcine cells also are permissive to productive Ad5 infection, we examined whether intravenous injection of Ad5 will result in a productive infection of swine. In this pilot study we used Ad5 since it contains the entire E3 region and is not replication attenuated. Since the E3 region is dispensable for the growth and propagation of adenovirus in cell culture (30), this region has been often deleted in adenovirus-based gene transfer vectors to increase the packaging capacity for transgene and regulatory elements (22). However, the E3 region encodes, among others, the E3-19K gene product, which is involved in combatting cellular immune responses by retaining major histocompatibility complex antigens in the endoplasmic reticulum (33), and the E3-11.6K gene product, which is required for efficient cells lysis and release of adenoviral progeny from infected cells (54).

Systemic administration of Ad5 to swine resulted in a widespread tissue distribution because of its ability to transduce many cell types. However, despite the lack of significant differences in the adenoviral transduction efficiency of kidney, lung, and liver cells in vitro (data not shown), the Ad5 concentration after intravenous injection in the lung was 67-fold higher than in the liver and 1,780-fold higher than in the kidney. The ratio did not change significantly over the course of the study. Since the ratio of DNA concentrations in liver and lung in the Ad.TK-injected swine was similar (data not shown), this probably reflects differences in the uptake rather than in the replication efficiency.

Following intravenous administration of Ad5, the average ratio of Ad5 DNA copies in liver and lung was 3.25 in mice and 0.01 in swine, suggesting more efficient uptake of Ad5 by the murine liver compared to the porcine. Preferential uptake of adenovirus in the liver (24, 63) has been previously linked to a dose-dependent hepatotoxicity (8, 37, 40). This high liver uptake is due to the efficient phagocytosis of Ad5 by Kupffer cells (35, 62), resulting in rapid clearance of Ad5 from the bloodstream, with a half-life of less than 2 min (2). However, when Kupffer cells are blocked, 90% of the adenovirus genome is cleared from the liver within 24 h (64). Based on the lower levels of adenovirus uptake by the porcine liver, hepatotoxicity is predicted to be less problematic in swine. Consistently, only marginal histological changes were observed in this organ after adenovirus injection. However, an increase in the GLDH level, which is indicative for hepatic injury, was observed in the wild-type-infected animal on day 7 but not in the Ad.TK-treated animal. The elevated GLDH serum level might be caused directly by the adenoviral CPE or immune mediated. It is known that the expression of viral late proteins elicits major histocompatibility complex class I-restricted CD8⁺ cytotoxic T cell-mediated destruction of infected cells (65-68). Because of the patchy Ad5 organ distribution, it is conceivable that these changes have not been picked up in the histological examination of the liver. Other hepatic tests used in humans for diagnosis and monitoring of hepatic injury, i.e., aminotransferases, alkaline phosphatase, and GGT did not show significant changes in swine. Additional studies in a larger number of animals and a longer follow-up period are needed to address the question of hepatotoxicity.

A striking observation of this pilot study in swine was the degree of inflammation seen in the lungs of animals given Ad5 intravenously, reflecting interstitial pneumonia in immunocompromised humans with disseminated adenovirus disease. It remains to be determined whether replication is necessary, but the lower pathology score in the swine given Ad.TK suggests that it is. In contrast to previously used animal models of pneumonia in cotton rats (45) or mice (20), pneumonia was observed after systemic infection. If this can be confirmed in further experiments, infection of swine with human adenoviruses might be a relevant animal model to study the pathogenesis of adenovirus-induced pneumonia.

Using RT-PCR we were also able to detect the expression of the viral hexon gene in lung and liver tissue 7 days after Ad5 injection, indicating replication of the viral genome (7). This approach has been used recently by Wadler et al. to assess whether there is not only presence of the oncolytic E1B-55K-deleted adenovirus dl1520 (ONYX-015) after intralesional injection in cancer patients but also replication (56).

In swine, Ad5 DNA concentrations in serum decreased during the first 3 days after injection. From day 3 to day 7, a 13.5-fold increase was observed in the Ad5-infected animal, while a progressive decline of serum viral load was seen in the Ad.TK-injected animal. Seven days after injection, viral load in serum was 100-fold higher in the Ad5-infected animal than in the animal receiving Ad.TK. Although this has to be confirmed in a larger number of animals, the longitudinal increase in viral load in an individual animal and higher level of viremia in the animal infected with replication-competent virus strongly suggests replication of Ad5 in vivo.

More direct evidence for replication of human adenovirus in swine comes from a study on replication-competent adenoviral vectors as vaccines inducing mucosal immune responses. Simultaneous intranasal and intraperitoneal injection of adenoviral vectors with a deletion in E3 into miniature swine led to a 100-fold increase in the infectious titer of lung homogenates from 12 h to 48 h postinfection (55). Thereafter, the titers progressively declined. Similar kinetics were observed after injection of wild-type Ad5 by the same routes, but infectious titers were at least 10-fold higher.

Persistent detection of adenoviral DNA in the serum of Ad5-infected swine is very different from the short half-life determined in rodents, but the data obtained in swine are congruent with clinical studies where patients with neoplastic disease received intratumoral injections of a replication-defective (69) or replication-competent adenoviral vector (43). Inefficient replication of human adenovirus in primary cell cultures from rodents and rabbits and similar viral load kinetics in the plasma of mice injected with replication-competent and replication-deficient adenoviruses indicate that Ad5 replication is severely restricted in mice. Since the transduction efficiency of primary murine cells is affected to a much lesser extent than the replication efficiency, this restriction seems to act primarily at a postentry step as reported previously (5).

In summary, efficient productive replication of human Ad5 in porcine primary cell cultures, a longitudinal increase of viral load in an individual animal, late gene expression, higher viral load after infection with replication-competent than replication-deficient adenovirus, and severe pathological alterations in the lung indicate that infection of swine with human adenoviruses or adenoviral vectors might be a more appropriate animal model to study adenoviral pathogenicity and pharmacodynamic and toxicity profiles of adenovirus vectors than rodents.