Influence of Fiber Detargeting on Adenovirus-Media
http://www.100md.com
病菌学杂志 2005年第18期
Weill Medical College of Cornell University, Hearst Research Foundation, Department of Microbiology and Immunology, Molecular Biology Graduate Program, New York, New York 10021
Molecular Immunology Unit, Institute of Child Health, 30 Guilford Street, University College London, WC1N 1EH, United Kingdom
ABSTRACT
The major adenovirus (Ad) capsid proteins hexon, penton, and fiber influence the efficiency and tropism of gene transduction by Ad vectors. Fiber is the high-affinity receptor binding protein that serves to mediate cell attachment in vitro when using coxsackie-adenovirus receptor (CAR)-containing cell lines. This contrasts with transduction efficiency in macrophages or dendritic cells that lack high concentrations of CAR. To determine how fiber influences gene transduction and immune activation in a murine model, we have characterized Ad type 5 (Ad5) vectors with two classes of chimeric fiber, CAR binding and non-CAR binding. In a systemic infection, Ad5 fiber contributes to DNA localization and vector transduction in hepatic tissue. However, the majority of vector localization is due to Ad5 fiber-specific functions distinct from CAR binding. CAR-directed transduction occurs but at a modest level. In contrast to CAR binding vectors, the F7 and F7F41S non-CAR-binding vectors demonstrate a 2-log decrease in hepatic transduction, with a 10-fold decrease in the amount of vector DNA localizing to the hepatic tissue. To characterize the innate response to early infection using fiber chimeric vectors, intrahepatic cytokine and chemokine mRNAs were quantified 5 hours postinfection. Tumor necrosis factor alpha mRNA levels resulting from Ad5 fiber infections were elevated compared to viruses expressing serotype 7 or 41 fiber. Levels of chemokine mRNA (gamma interferon-inducible protein 10, T-cell activation gene 3, and macrophage inflammatory protein 1?) were 10- to 20-fold higher with CAR binding vectors (Ad5 and F41T) than with non-CAR-binding vectors (F7 and F7F41S). In spite of quantitative differences in vector localization and innate activation, fiber pseudotyping did not significantly change the outcome of anti-Ad adaptive immunity. All vectors were cleared with the same kinetics as wild-type Ad5 vectors, and each induced neutralizing antibody. Although non-CAR-binding vectors were impaired in transduction by nearly 2 orders of magnitude, the level of antitransgene immunity was the same for each of the vectors. Using primary bone marrow-derived macrophages and dendritic cells, we demonstrate that transduction, induction of cytokine/chemokine, and phenotypic maturation of these antigen-presenting cells are independent of fiber content. Our data support a model where fiber-mediated hepatic localization enhances innate responses to virus infection but minimally impacts on adaptive immunity.
INTRODUCTION
Studies using adenovirus (Ad) vectors have provided insight into the major problems that must be addressed when considering viral vectors for gene therapy applications. Issues confronting the use of Ad vectors include the innate and adaptive antiviral immune response, the lack of cell- or tissue-specific targeting, and the need to establish stable long-term gene expression. The Ad capsid plays a major role in two of these issues, tissue-specific targeting and stimulation of the antiviral innate and adaptive immune response. Therefore, efforts to manipulate the viral capsid structure offer an opportunity to control vector targeting as well as influence immune activation.
Three major protein components of the Ad capsid include the structural proteins hexon, penton base, and fiber. Hexon is the main structural element forming the body of the capsid. Penton base and fiber form the structural unit penton, located at each of the 12 vertices of the icosahedral capsid. The capsid proteins are responsible for efficient gene transduction by adenovirus vectors. Adenovirus type 5 (Ad5) uptake into human cell lines occurs through a multistep process that first relies on high-affinity binding of fiber to a cell surface receptor. For the majority of Ad serotypes, fiber binds to the cell surface protein coxsackie-adenovirus receptor (CAR) (1, 2, 19, 28), which is expressed in a variety of tissue types (27). Following fiber/CAR complex formation, penton base binds to v integrins and stimulates endocytosis of viral particles (30). In the absence of fiber-mediated high-affinity binding, transduction can still occur but at significantly reduced levels. How Ad5 recognizes and binds cells in the absence of a fiber-directed pathway is not well understood. Penton base binding to v integrins may contribute to the low efficiency uptake of Ad5 vectors in cells that lack CAR (8). Other nonspecific mechanisms such as phagocytosis and macropinocytosis may be used in certain cell types.
In contrast to in vitro studies of Ad infection, in vivo applications of Ad5-based vectors have revealed a more complex biology of virus-cell interactions that influence gene transduction. Following systemic administration in a murine model, greater than 90% of an Ad5-based vector localizes to the liver (32). However, the majority of virus localized to the liver within the first hours postinfection does not contribute to hepatocyte transduction and is eliminated within 24 h of administration. For the small fraction of Ad5 vector that successfully transduces hepatic tissue, the role of high-affinity fiber/CAR binding is surprisingly small. Depending on the study and vector used, between 10 and 50% of intrahepatic transduction occurs in a fiber/CAR-dependent manner (4, 11, 13, 24). It has been proposed that Ad5 fiber binding to heparan sulfate glycosaminoglycans (HSG) (3, 16, 26) plays an important role in intrahepatic localization as well as transduction following systemic administration of vector in a murine model.
The vast majority of Ad5 vector that localizes to the liver is rapidly eliminated. Kupffer cells are the predominant macrophages of liver, and studies have demonstrated that Kupffer cell depletion impacts on viral localization and transduction (12, 32). Macrophages and other cells of the immune system take up Ad5-based vectors, but these cells often lack CAR and transduction is generally very inefficient (8). Recognition of virus by these cells results in vector clearance and contributes to both the inflammatory and the antiviral adaptive immune responses that arise following exposure to virus (34). For Ad5 vectors, the strong inflammatory response and the antiviral adaptive immune response are major impediments for gene therapy applications but potential assets for anticancer or vaccine applications. These studies highlight both the importance of understanding the complex binding interactions that take place in a systemic infection with Ad vectors and the need to develop effective Ad detargeting strategies.
Several Ad5-based vectors that demonstrate a decreased transduction phenotype in vivo have been constructed. Mutations in Ad5 fiber that abolish CAR binding have been combined with mutations in putative HSG binding domains (16, 25), resulting in diminished liver transduction following systemic administration. Another approach to detargeting Ad5-based vectors has been to replace Ad5 fiber with a fiber moiety from a non-CAR-binding virus. There are several classes of non-CAR-binding Ad fibers. One is the subgroup B class of virus. Recent studies have demonstrated that human CD46 serves as a target receptor for many of the subgroup B viruses (5, 21, 23). A variety of human cell lines express the subgroup B receptors and are efficiently infected by subgroup B vectors or Ad5 vectors pseudotyped with a subgroup B fiber (5, 21). Important to this study, the CD46 receptor is not abundantly expressed in murine tissues, making a subgroup B fiber-containing vector potentially detargeted in a mouse model (5). A second class of non-CAR-binding fiber is derived from the subgroup F adenoviruses. The subgroup F viruses express two fiber genes tandemly arranged (10): a long-shafted fiber that binds CAR and a short-shafted fiber that has no known binding function (19). Recent studies using Ad5 vectors pseudotyped with subgroup F short fiber demonstrate reduced transduction of hepatic tissue following a systemic administration of vector (15, 16, 20).
In the current study we have examined the biology of fiber-modified viruses with respect to viral transduction, viral DNA localization, and the influence on innate and adaptive immune responses to Ad5-based fiber detargeted vectors. We have found that, in Ad5-based vectors, fiber-mediated transduction and localization to liver contribute to a heightened early innate response as determined by induction of chemokine and cytokine mRNAs. In comparison, non-CAR-binding vectors exhibit reduced localization to the liver, reduced gene transduction, and diminished induction of mRNA associated with activation of the inflammatory response. In spite of these beneficial detargeting characteristics, the adaptive arm of the immune system is largely unaffected by these capsid modifications.
MATERIALS AND METHODS
Adenovirus vector stocks and cell lines. Four E1/E3-deleted Ad5-based vectors were used in this study: Ad5, F7, F7F41S, and F41T. Construction and characterization of these vectors have been described in detail elsewhere (20). The F7 vector was generated by genetically swapping the fiber 5 coding region of Ad5 with the entire fiber gene from Ad serotype 7. The F7F41S vector genome contains coding regions from both fiber 7 and the short-shafted fiber from Ad serotype 41 as two terminal exons in tandem. The F41T vector genome contains both short- and long-shafted fibers from Ad serotype 41.
HEK-293 monolayer cultures were kept in Dulbecco modified Eagle medium (DMEM) plus 5% to 10% calf serum. FL83B murine hepatocyte (ATCC CLR-2390) cultures were maintained in F12K plus 10% fetal bovine serum (FBS). P6 macrophages were kindly provided by A. Ding and kept in DMEM plus 10% FBS.
In vitro transduction assays. Cell lines used in this study were seeded into 24-well plates and infected with 1,000 or 5,000 particles/cell in medium without serum for 30 min at 37°C. After infection, virus was aspirated and fresh medium was added back. Infected cells were harvested 24 h postinfection in TEN scrape buffer (40 mM Tris Hcl [pH 7.4], 1 mM EDTA, 0.15 M NaCl) and resuspended in 0.25 M Tris, pH 7.8. Lysates were subjected to three freeze-thaw cycles followed by a 10-min incubation at 65°C to inactivate cellular deacetylases. Chloramphenicol acetyltransferase (CAT) activity assays were performed as described previously (6).
In vivo transduction assays. Six-week-old female B6129/J mice were obtained from Taconic or Jackson Laboratories and maintained in compliance with institutional protocols. Mice were injected retro-orbitally with 1010 virus particles or PBS vehicle alone in a total volume of 100 μl. At indicated time points, animals were sacrificed and organs were harvested, weighed, and resuspended in 2 volumes PBS by weight. Tissue lysates were prepared by homogenizing organs and centrifuging them for 30 min at 3,000 x g, followed by a 10-minute incubation at 65°C. Total tissue CAT activity was assayed as described above.
DNA isolation and Southern analysis. Livers from Ad- or mock-infected animals were harvested, and a portion was used for DNA isolation. Tissue slices were pulverized in Eppendorf tubes and incubated in lysis buffer (10 mM Tris [pH 8.0], 400 mM NaCl, 2 mM EDTA, 0.5% sodium dodecyl sulfate [SDS], 0.4 mg/ml proteinase K) overnight at 55°C. Samples were treated with 2 μl RNase A (10 mg/ml) for 30 min at 37°C. DNA was purified by two rounds of phenol-CHCl3 extraction and a CHCl3-only extraction, followed by ethanol precipitation. For Southern analysis, 10 μg DNA was digested with EcoRI and electrophoresed on a 1% agarose gel. Samples were transferred to a nylon membrane, and blots were probed with either EcoRI-digested Ad5 vector DNA or ?-actin. DNA levels were quantified by PhosphorImager analysis.
RNA isolation and RPA. Bone marrow-derived macrophages (BMMOs) and liver sections from Ad- or mock-infected animals were harvested for RNA isolation with Trizol reagent (Invitrogen) according to the manufacturer's instructions. Five (BMMO) or 10 (liver) μg of total RNA was hybridized to 32P-labeled probes generated from either the mCK3b (cytokine) or mCK5c (chemokine) multiprobe template sets (Pharmingen). A commercially available RNase protection assay (RPA) kit (Pharmingen) was used to RNase treat samples and recover protected fragments. Resolution of protected fragments was carried out by electrophoresis on 10% polyacrylamide sequencing gels. Gels were dried and exposed to a PhosphorImager screen, and mRNA species were quantified by PhosphorImager analysis.
Anti-Ad neutralizing antibody blocking assay. Neutralizing antibody titers were determined by a functional in vitro blocking assay in HeLa cells. Serum samples from mice infected with the various vectors were taken at day 28 and serially diluted in DMEM. Serum dilutions were incubated with 5 x 105 particles of a standard Ad5CAT vector for 1 h at 37°C. The serum/virus mixture was then added to 105 HeLa cells for 30 min at 37°C. Cells were washed with DMEM, and fresh medium was added back. CAT activity in cell extracts was scored 24 h later as described above.
Anti-CAT antibody ELISA. To detect serum levels of antitransgene antibody, Maxisorp enzyme-linked immunosorbent assay (ELISA) plates (Nalge Nunc International) were coated with CAT (Sigma) in 50 mM sodium carbonate buffer, pH 9.6, for at least 16 h at 4°C. The CAT solution was removed, and 1% casein (Sigma) in PBS was added as a blocking buffer for 1 h at room temperature. Plates were washed three times in blocking buffer plus 0.05% Tween 20. Serum samples from 28-day-infected mice were serially diluted in blocking buffer and added to ELISA plates for 1.5 h at room temperature. Plates were washed three times as described above. A secondary horseradish peroxidase-conjugated anti-mouse antibody (Amersham) was diluted in blocking buffer plus 0.05% Tween 20 and added to plates for 1.5 h at room temperature. Plates were then washed four times as described above and incubated with the tetramethylbenzidine peroxidase substrate (Sigma). Reactions were stopped with 1 N H2SO4, and plates were read at 450 nm on a spectrophotometer.
BMMOs and bone marrow dendritic cell (BMDC) isolation and infection. Bone marrow macrophages were generated by culturing bone marrow cells in DMEM plus 20% FBS and 30% supernatant derived from L929 confluent cells, replacing two-thirds of the culture volume with fresh granulocyte-macrophage colony-stimulating factor (M-CSF)-containing medium every third day. Cells were used after 7 to 9 days, and all stained positive for CD11b but were negative for CD11c and B220.
BMDCs were obtained by culturing bone marrow cells in DMEM plus 10% FBS in the presence of 15% supernatant derived from J558L cells as a source of GM-CSF, replacing two-thirds of the culture volume with fresh GM-CSF-containing medium every second to third day. Cells were used after 5 days and contained 60 to 90% CD11c+ cells.
Cultures of BMMOs and BMDCs were infected with the indicated virus at 5,000 particles/cell. For transduction experiments, cells were harvested 24 h postinfection for CAT assays. For maturation phenotype, cells were harvested 48 h postinfection and stained on ice with phycoerythrin-conjugated anti-CD86 mouse antibody (Becton Dickinson) for 30 min in PBS containing 1% bovine serum albumin. Cells were fixed in 1% formalin and analyzed on an EPICS XL flow cytometer (Beckman Coulter).
RESULTS
Macrophage and hepatocyte transduction in vitro. Four fiber variants of Ad5 were selected to characterize the influence of fiber on transduction and immune activation. These viruses, Ad5, F7, F7F41S, and F41T, differ only in the nature of the fiber homotrimer(s) present in the viral capsid (Fig. 1A). The parental Ad5 virus is an E1/E3-deleted vector expressing wild-type fiber 5 with the cytomegalovirus immediate-early promoter driving expression of CAT as a reporter gene. The fiber present in this vector contains CAR binding domains as well as the putative HSG binding domain (KKTK) identified as being important for CAR-independent transduction. The F7 virus contains the full-length fiber sequence from the Ad7 subgroup B virus in place of fiber 5 (6). Subgroup B fibers do not bind CAR. The tandem fiber vectors express two distinct fiber proteins, either fiber 41S and fiber 41L (F41T) or fiber 7 and fiber 41S (F7F41S). F41L is a CAR binding fiber but does not contain the KKTK domain present in Ad5 fiber. Fiber 41S has no known binding function.
Viruses were grown on a large scale in 293 cells, followed by two rounds of CsCl banding purification and dialysis to remove salts. Viral particle numbers were quantified by optical density at 260 nm (1012 particles/unit of optical density at 260 nm). To confirm that each vector was quantitatively matched, 1010 particles of each vector were separated by SDS-polyacrylamide gel electrophoresis and analyzed by Western blotting with antihexon antibody (Fig. 1B). The blot was then probed with antifiber antibody (4D2 from Neomarkers) to demonstrate the fiber content of each of the vectors (Fig. 1C). The lack of fiber 7 in F7F41S is due to the preferred expression and incorporation of fiber 41S into virions (20). Based on this assay, the fiber content of F7F41S is essentially F41S.
To determine transduction functions, a murine hepatocyte cell line (FL83B) and a murine macrophage cell line (P6) were infected with the indicated virus at 103 Ad particles/cell. Following a 30-minute incubation, cells were washed twice with fresh medium to remove free virus. Cells were incubated for 24 h and harvested for reporter gene assay. The Ad5 and F41T vectors were highly effective at delivering a transcriptionally active reporter gene (Fig. 2A) into FL83B murine hepatocyte cells. In comparison, transduction through CAR-independent fiber (F7) was decreased 100-fold compared to CAR-targeting viral vector. This contrasts with the efficient transduction mediated by F7 vectors in human cell lines (HeLa, HepG2, and A549) (20; data not shown). The low level of transduction by F7-containing viruses in FL83B hepatocytes is consistent with studies demonstrating low levels of subgroup B virus receptor(s) expression in murine cells (5). CAT gene transduction from F7F41S virus infection was over 200-fold lower than that from Ad5 and 500-fold less than that from F41T. The lack of transduction by F7F41S virus supports previous observations (20) that fiber 41S does not contribute to efficient cell binding and virus uptake.
In contrast to transduction efficiency in the FL83B hepatocyte cell line, transduction of the P6 macrophage cells by the Ad5 vector was found to be very low, approximately 2 logs below transduction in FL83B cells when identical experimental conditions are used (compare Fig. 2A and B). The F7 virus was also unable to transduce the macrophage cell line, arguing that these cells also lack a suitable subgroup B receptor. Each of the tandem fiber vectors had similar and slightly higher transduction efficiencies in macrophage P6 cells. Similar results were found in the RAW264.7 murine macrophage cell line (data not shown). These results indicate that murine macrophage transduction is independent of CAR or subgroup B receptor uptake mechanisms. Because established cell lines do not always reflect the complete phenotype of a primary cell type, we chose to compare the transduction of the P6 macrophage cell line with cultured primary BMMOs isolated from B6129/J mice. Cells (P6 or BMMOs) were infected with a slightly higher dose of virus (5,000 particles/cell), and CAT transduction assays were performed (Fig. 2C). CAT expression in P6 cells was increased approximately fivefold compared to infections with 1,000 particles/cell, reflecting a dose response to each of the fiber chimeric constructs. At 5,000 particles/cell, transduction of the BMMOs was significantly higher than transduction of established macrophage cell lines. At this dose of vector, there was no apparent difference in transduction between the fiber chimeras. This result reveals an indifference of macrophages to the fiber type present in the Ad5 capsid.
Macrophages contribute to the early inflammatory response to pathogens and have been implicated in the antiviral response to Ad vectors. To determine if the fiber chimeras differentially activate macrophages, we assessed the maturation of BMMOs by upregulation of costimulatory molecules (such as CD86, CD40, or major histocompatibility complex class II). At 24 h following virus infection, BMMOs were stained with anti-CD86 antibody and subjected to flow cytometry. Essentially 100% of CD11b-positive BMMOs underwent activation as revealed by upregulation of CD86 regardless of the fiber configuration (Fig. 3A). To assess an early activation response of BMMOs to the chimeric viruses, BMMOs were infected as previously described and harvested 5 h postinfection for total RNA. RNA from each infection was used in an RNase protection assay to quantify induction of cytokine and chemokine mRNA. Using a cytokine multiprobe template set under the experimental conditions described, each fiber chimera was found to induce expression of tumor necrosis factor alpha (TNF-) mRNA (Fig. 3B). Similar expression patterns were seen for gamma interferon (IFN-)-inducible protein 10, macrophage inflammatory protein 1?, and T-cell activation gene 3 using a chemokine multiprobe template set (Fig. 3C). At the level of transduction and activation, the data clearly indicate that each of the chimeric constructs is able to transduce and effectively stimulate primary macrophage cells.
CAR-mediated transduction and viral DNA localization. CAR-dependent infection is a primary mechanism responsible for high-level Ad vector gene transduction in murine hepatocyte cell lines, but transduction in macrophage cells occurs in a fiber/CAR-independent manner. We next asked how the fiber modifications in our vectors influence sequestration and transduction at early times after systemic administration. B6129/J mice were infected with 1010 virus particles by intravenous injection. At 5 and 24 h, animals were sacrificed. Liver and spleen were harvested for reporter gene expression, total DNA, and total RNA. CAT expression at 5 h postinfection represents an efficient uptake and gene expression process. Under our experimental conditions, the CAR-targeted vectors were able to mediate strong levels of CAT expression by 5 h postinfection (Fig. 4A). The results obtained at 5 h are similar to the pattern seen in hepatocytes (Fig. 2A) where CAR-dependent transduction is prevalent, but a blending of hepatocyte and macrophage infections may better represent the overall hepatic expression pattern. At 24 h postinfection the magnitude of expression has increased with all vectors, without an overwhelming fiber-directed bias. Consistent with a detargeting phenotype, the F7 and F7F41S vectors yield 20- to 50-fold-less hepatic transgene expression than Ad5 fiber vectors. Assays characterizing CAT activity in kidney, lung, and intestine showed that activity was well below the low levels found in spleen and therefore the levels were viewed as background (data not shown).
In contrast to transgene expression, a different pattern was found when characterizing the yield of total viral DNA localized to the liver. DNA was isolated from liver harvested 5 and 24 h postinfection with each of the viruses. The CAR binding viruses result in high levels of gene expression but diverge greatly with respect to total viral DNA localized to the liver (Fig. 4B and C). Ad5 vector DNA is present at the greatest level 5 h postinfection whereas F41T is present at the lowest level. Since both of these vectors transduce through CAR at similar efficiencies (Fig. 2A), the observed variation in localization is attributed to non-CAR-binding attributes of fiber 5 compared to fiber 41L or 41S of F41T. As previously mentioned, Ad5 fiber contains a putative HSG binding domain, which is absent in the 41L/S fibers. Although the F7 vector yields the lowest level of transduction, it was second to Ad5 with respect to viral DNA localized to the liver at 5 h postinfection, with F7F41S present at a slightly lower concentration. When total DNA isolated from splenic tissue was probed for viral DNA by Southern blotting, levels were minimal compared to liver for all samples (data not shown); therefore, detargeted vectors were not retargeted to spleen. These data indicate disassociation between viral DNA localization and transgene expression at the 5-hour time point.
At 24 h postinfection, the majority of viral DNA present at 5 h is eliminated. DNA from the Ad5 infection is clearly above background levels (Fig. 4B and C). For Ad5-based vectors, the elimination of vector DNA present at 5 h and absent at 24 h has been attributed to vector clearance by Kupffer cells (31). For both CAR binding and non-CAR-binding vectors, the majority of viral DNA initially localized to the liver is not associated with gene transduction and has been eliminated by 24 h postinfection.
Altered chemokine and cytokine profiles as a function of fiber modification. The fiber-modified viruses are distinguished from one another by the efficiency of gene expression as well as the magnitude of virus localized to the liver. We next asked if these differences in virus targeting would have an impact on immune activation by the host. To characterize the overall early inflammatory response to the fiber-modified vectors, we first characterized serum levels of TNF- and interleukin-6 (IL-6) at 5 and 24 h postinfection by ELISA. Under the experimental conditions described (1010 viral particles/animal) levels of TNF- and IL-6 in serum did not increase over those found in mock-infected animals (data not shown). We then isolated RNA from portions of liver harvested at 5 and 24 h postinfection as described in the previous section. Samples (10 μg) of RNA were used in an RNase protection assay using the mCK3b cytokine template set (BD Biosciences) to determine if the fiber-modified vectors induce novel patterns of cytokine activation. Under our experimental conditions (infection of each animal with 1010 viral particles), Ad-mediated induction of cytokine mRNA from total liver was very modest at 5 h postinfection (Fig. 5A) (and lower at 24 h [data not shown]). Using PhosphorImager analysis of protected liver RNA species (Fig. 5B), we found that TNF- transcripts were induced in animals infected with the Ad5 vector compared to liver isolated from animals infected with the fiber-modified viruses. We also found that mRNAs for two other cytokines, IFN- and macrophage inducing factor, were induced compared to mock-infected animals. Based on the magnitude of induction, the upregulation of cytokine mRNA is roughly consistent with viral DNA localization to the liver (compare Fig. 3C and 5B). Analysis of RNA harvested from animals after 24-hour virus exposure revealed baseline concentrations for all cytokine mRNAs (data not shown).
RNA generated from the infections harvested at 5 h was also used to determine Ad induction of chemokines using the mCK5c chemokine template set. Results from this assay revealed a chemokine mRNA induction profile that was fiber dependent and distinct from the cytokine mRNA induction pattern (Fig. 6A and B). The chemokine mRNA expression pattern resembled the pattern of fiber-dependent gene transduction (compare Fig. 3A and 6B). IFN--inducible protein 10 and T-cell activation gene 3 mRNAs were highly induced by both CAR binding viruses, and macrophage inflammatory protein 1? was expressed at more moderate levels. After normalization, the non-CAR-binding viruses F7 and F7F41S were slightly above baseline levels of mock-infected animals but 10- to 20-fold less than the CAR-targeted vectors. These studies demonstrate that fiber-mediated detargeting of otherwise identical vectors has a demonstrable impact on the induction of both cytokines and chemokines associated with early stages of the anti-Ad immune response.
Clearance efficiency and humoral immunity unchanged by fiber modifications. Based on the observations that the fiber-modified viruses were compromised in induction of intrahepatic inflammatory cytokine/chemokine mRNA, we next asked if we could detect a significant difference in the adaptive immune response to virus infection. It is well established that cytotoxic T cells are largely responsible for clearance of Ad-infected cells following an in vivo infection in an immunocompetent host (33). Therefore, if fiber-modified viruses do not activate the cellular immune response as effectively as Ad5, we would predict increased persistence over time of CAT-expressing cells. Similarly if the modified vectors influence the humoral immune response to Ad, we would predict a diminished level of anti-Ad antibody to be present in infected animals. To test these elements of the immune response to fiber-modified vectors, we established a 4-week time course experiment infecting B6129/J mice with 1010 particles of the indicated virus. Animals were sacrificed at the indicated time point, and liver, spleen, and serum were harvested.
The profiles for transgene expression over the 4-week time course for the modified viruses were remarkable in their similarity to Ad5 vector (Fig. 7). In liver, peak expression for each construct occurred at 3 days postinfection (Fig. 7A). CAT expression began to decline at 6 days postinfection, and by 28 days CAT expression had returned to baseline levels in all animals. Similarly, the kinetics of CAT transgene expression and clearance from spleen were essentially identical (Fig. 7B). The only distinguishing feature was the magnitude of transgene expression. At the level of vector clearance, the fiber-modified vectors were eliminated as efficiently as the Ad5 vector.
We next determined the level of anti-Ad neutralizing antibody in serum isolated from animals 28 days postinfection by an antibody-blocking assay as described in Materials and Methods. Each of the fiber-modified viruses induced blocking antibody to Ad5 (Fig. 8A). However, the efficiency of blocking activity was lower in serum generated from infections with the fiber-modified vectors. Because these studies were blocking Ad5 infection and not the dosing vector, this slightly diminished response may reflect differences in antibody directed to serotype-specific fibers.
To determine if the neutralizing antibody generated in response to the fiber-modified vectors was biologically significant, we used an in vivo readministration assay (7, 9). Animals were injected with 1010 particles of each virus, and on day 25 postinfection, when CAT expression is at baseline, each animal was injected with 1010 particles of Ad5CAT. Livers were harvested 3 days postinfection and assayed for CAT gene expression. Regardless of the vector used in the original administration, all animals were able to effectively block Ad5CAT gene transduction in our readministration protocol (Fig. 8B).
The CAR detargeted vectors are significantly less effective in transducing cells in vivo, resulting in a diminished level of overall CAT gene product. To determine if the efficiency of transgene expression correlated with the serum levels of anti-CAT antibody, we quantified anti-CAT antibody by ELISA (Fig. 8C). Serum levels of anti-CAT antibody were lowest for Ad5, which expressed the highest level of CAT. The modified vectors generated similar or slightly higher levels of anti-CAT antibody. In contrast to the different levels of liver and spleen CAT gene expression mediated by the fiber-modified viruses, anti-CAT antibody was found to be essentially the same for each of the viruses examined.
Transduction and maturation of bone marrow-derived dendritic cells by fiber-modified viruses. We have used two basic endpoint assays (clearance of infected cells and anti-Ad humoral immunity) to demonstrate that the adaptive immune response to fiber-modified vectors is largely unaffected by the efficiency of gene transduction or the magnitude of the inflammatory response as characterized by the cytokine/chemokine RNA induction profile. These results would argue that recognition of these vectors by specialized antigen-presenting cells of the immune system (primarily macrophages and dendritic cells) is unaffected by fiber detargeting functions. We have already demonstrated that primary macrophages are responsive to Ad in a fiber-independent manner. Dendritic cells are considered the primary effector cells of the adaptive arm of the immune system and have been shown to respond to adenovirus infection (14, 17, 18, 29). The fiber-modified viruses were used to infect BMDCs (5 x 103 particles/cell) as previously described. Levels of CAT expression in BMDCs were significantly below those previously identified in BMMOs, approximately 10- to 20-fold less under comparable conditions (compare Fig. 3C to 9A). All vectors were above background, and the F7 construct was again found to be the least effective at CAT transduction (Fig. 9A). We next determined if the fiber-modified viruses influenced BMDC maturation/activation. Immature BMDCs express low levels of cell surface markers including CD86. When exposed to a stimulating antigen (Ad5 virus), immature BMDCs undergo maturation characterized by upregulation of cell surface markers including CD86 (17). We found that exposure of BMDCs to Ad5, F41T, and F7F41S resulted in upregulation of CD86 compared to mock-infected BMDCs (Fig. 9B). Upregulation of CD86 following BMDC exposure to the F7 virus was no different from that with mock-infected DCs. The observation that F7 demonstrated diminished transduction and activation in DCs led us to test if increasing doses of F7 would lead to increasing transduction and activation of BMDCs. The BMDCs were incubated with increasing doses of F7 virus, and 24 h postinfection, a portion of each infection was harvested for measurement of CAT gene expression. A control infection of Ad5 at 5,000 particles indicated slightly higher levels of transduction than F7. With increasing doses of F7 virus, we found a corresponding increase in CAT gene expression (Fig. 9C). When the DCs were harvested at 48 h postinfection and characterized with respect to upregulation of the costimulatory molecules CD86 (Fig. 9D) and CD40 (Fig. 9E), we found that increasing concentrations of virus resulted in increased expression of the DC activation markers. Based on the data from this study, we found that fiber-modified viruses were able to induce upregulation of costimulatory molecules in bone marrow-derived dendritic cells.
DISCUSSION
The combination of uncontrolled vector delivery and virus uptake by cells of the immune system has presented major obstacles to the use of Ad vectors in gene therapy applications. In the current study we have used fiber-pseudotyped first-generation Ad vectors to characterize how fiber influences gene transduction, vector localization, and immune stimulation following systemic administration in a murine model. Based on the observations made in this study and consistent with observations from other laboratories (13, 22, 26), Ad5 fiber contributes significantly to liver localization following a systemic administration of vector. When the Ad5 fiber is replaced with either fiber 7 or a combination of fiber 7 and fiber 41S, both vector localization to liver and vector-mediated gene transduction are greatly reduced. These constructs represent "fiber-detargeted" vectors. A major advantage of the fiber-detargeted vectors is a decrease in the magnitude of hepatic cytokine and chemokine mRNA induction. Although both localization to the liver and vector-mediated gene transduction are greatly reduced with these constructs, they are not eliminated. Based on the results presented in this study, we consider residual localization to liver and low-level vector transduction with these constructs to result from fiber-independent transduction. This background may be attributed to low-affinity binding through nonfiber capsid proteins, such as penton base or hexon. Background localization/transduction may also result from nonspecific phagocytic/macropinocytotic uptake of viral particles.
CAR-mediated transduction following a systemic administration of virus is revealed during very early time points (5 h), where an assay of transgene expression depends on efficient virus binding and internalization. Efficient CAT transduction is found with the CAR binding vectors. This result is consistent with in vitro transduction assays in the FL83B hepatocyte cell line (compare Fig. 3A and 2A). Although similar levels of CAT expression are found with both CAR binding vectors at 5 h postinfection, there is five times more Ad5 viral DNA localized to the liver than F41T viral DNA. Two conclusions can be made from this result. Following vector administration, as virus is passing through the hepatic tissue, CAR-mediated binding facilitates hepatic transduction equally with both CAR binding vectors. This conclusion would indicate that both the Ad5 and F41T vectors are initially presented to the hepatic cells in a manner that allows each to target through CAR with equal efficiency. Consistent with this hypothesis, other studies have shown that, regardless of the type of fiber present on Ad, similar amounts of virus are present in the liver within the first 30 min following a systemic infection (22). The second conclusion is that Ad5 fiber, compared to fiber 41L, increases vector residence time within the liver. This conclusion is derived from the observation that Ad5 vector localization to liver is five times that of any of the other vectors at 5 h postinfection. Differences in CAR binding efficiency do not account for the differences in localization between Ad5 and F41T. It is worth noting that the two CAR binding fibers (F5 and F41L) are essentially identical in fiber shaft length. Therefore, we do not consider fiber length to directly account for Ad5 localization to liver. Increased residence time can be attributed to other features of Ad5 fiber, for example, the putative HSG binding domain KKTK present in Ad5 fiber and absent in each of our pseudotyped fiber constructs. This domain has been shown to facilitate Ad5 hepatic localization following a systemic administration in both murine and rat models (16, 26).
At time points of peak gene expression (days 1, 3, and 6) (Fig. 7) levels of Ad5-mediated CAT expression are above those seen with fiber-modified vectors. Transduction associated with the fiber-modified vectors F7 and F7F41S is roughly 1 to 5% of Ad5, whereas F41T-mediated CAT expression is approximately 10% of that found with Ad5. The F41T transduction differential compared to the non-CAR-binding vectors was approximately 10:1. This represents the enhancement of CAR-mediated transduction over fiber-detargeted vectors (F7 and F7F41S). The difference between F41T and Ad5 would be consistent with CAR-independent binding functions of Ad5 fiber. Based on the data presented in this study, the transduction differential between Ad5 and the fiber-modified vectors is consistent with a model in which Ad5 fiber mediated increased hepatic residence time. Therefore, overall levels of transduction are influenced by both CAR binding and non-CAR-binding mechanisms.
In addition to helping define the role of fiber in Ad vector localization and gene transduction, these studies also revealed how fiber can influence the innate response to virus infection. Coincident with increased liver localization, the Ad5 fiber vector facilitates induction of higher levels of intrahepatic TNF- mRNA compared to the pseudotyped vectors. In our studies we have used a comparatively low dose of virus, 1010 particles/animal. At this dose of input virus, the induction of inflammatory response is at the limit of sensitivity for many assays. Serum ELISAs of infected animals for TNF- and IL-6 at 5 and 24 h postinfection showed levels indistinguishable from background levels (data not shown). However, using the RPA we were able to detect TNF- mRNA induction at 5 h postinfection. Induction of cytokines by Ad5 vectors is well documented (at higher doses) and contributes to the strong anti-Ad inflammatory response (33). In a recent study, Shayakhmetov et al. (22) characterized chimeric constructs containing fibers with different shaft and/or knob compositions using 1011 particles/animal. At this dose, when characterizing induction of cytokine mRNA (TNF-), all viruses induced expression at levels that were slightly over background at 6 h postinfection. They also found that constructs that contained the Ad5 fiber shaft induced higher levels of TNF- than viruses expressing Ad35 shaft (Ad35 is a subgroup B virus) by an ELISA. Our studies also support a model where the presence of Ad5 fiber shaft stimulates higher levels of TNF- mRNA. In addition, we demonstrate that a long-shafted fiber such as F41L does not stimulate TNF- in the same manner as the Ad5 fiber. Taken together these results argue that the Ad5 fiber shaft that contains the KKTK domain contributes to TNF- mRNA induction.
CAR binding viruses were also shown to induce higher levels of intrahepatic chemokine mRNA compared to the fiber-detargeted vectors. Upregulation of chemokine mRNA at 5 h postinfection was not a function of total vector DNA localized to the liver, since the F41T vector induced chemokines to a level equal to that induced by Ad5. Rather chemokine induction more closely parallels CAR-mediated transduction. Shayakhmetov et al. (22) found heightened induction of IL-6, MCP-1, and IFN-? with a CAR binding Ad5 fiber shaft construct (Ad5/9L) at 1011 particles/animal compared to their other constructs. They did not characterize a CAR binding virus that lacked the Ad5 fiber shaft and concluded that the shaft length was a major factor in cytokine activation. Our observations are consistent with a role of Ad5 fiber shaft in cytokine activation, but we would also conclude that there exists a distinct role for the CAR binding domain in stimulating elements of the innate immune system. Further investigation will be required to determine if the basis of chemokine induction by the CAR binding vectors is due to gene expression or to efficient early transduction mediated by CAR-directed uptake. In the murine system, the two non-CAR-binding vectors, F7 and F7F41S, represent simple platforms for pursuing a genetic strategy to retarget Ad vectors and offer the benefit of diminished induction of inflammatory cytokines.
The F7 and F7F41S vectors also represent base platforms that can be used to explore the interaction of Ad vectors with cells of the immune system. Based on the decrease in liver (and spleen) localization, the decrease in liver (and spleen) transduction, and the decrease in intrahepatic induction of inflammatory cytokines, we asked if these vectors would also demonstrate a diminished ability to stimulate the adaptive immune response. Characterization of the adaptive immune response was based on straightforward endpoint assays relevant to gene therapy applications: clearance efficiency of transduced cells, antiviral humoral immunity, and the development of a humoral immune response to the delivered transgene CAT. In all cases, regardless of the assay or virus being tested, the adaptive immune response was largely indifferent to fiber. The detargeted vectors that resulted in diminished intrahepatic induction of chemokine mRNAs were as effective as the Ad5 fiber vector in stimulating the adaptive arm of the immune response.
Fiber-detargeted viruses mediate low-level gene transduction both in vitro and in vivo. We have shown that vector activation of antigen-presenting cells (both bone marrow-derived macrophages and dendritic cells) occurs in a fiber-independent manner in vitro. Consistent with in vitro observations, the threshold for activation of the adaptive arm of the immune response has been achieved in these studies with each of the fiber-pseudotyped vectors. In a previous study (17), we found that the RGD motif of the penton base was necessary to optimally activate DCs following exposure to Ad. In order to further compromise the immune activation of the fiber-detargeted vectors, we anticipate that an additional benefit can be obtained by combining fiber detargeting with the penton RGD mutations. Combining these capsid modifications with strategies such as use of tissue-specific promoters for transgene expression, expression of less immunogenic transgene, and use of second- or third-generation Ad vector backbones can in theory provide an Ad vector that has significant and greatly improved stealth characteristics. Alternatively as we learn more about how Ad vectors are recognized by cells of the immune system, we will be in a position to design vaccine or oncolytic vectors that have a significant enhancement in potency with diminished unwanted bystander activity.
ACKNOWLEDGMENTS
This work was supported by NIH grants AI-63142 to E.F.-P. and KO1 HL70438-01 to M.N. GenVec has licensing agreements with Cornell Research Foundation for technology developed in collaboration with E. Falck-Pedersen.
REFERENCES
Bergelson, J. M., J. A. Cunningham, G. Droguett, E. A. Kurt-Jones, A. Krithivas, J. S. Hong, M. S. Horwitz, R. L. Crowell, and R. W. Finberg. 1997. Isolation of a common receptor for Coxsackie B viruses and adenoviruses 2 and 5. Science 275:1320-1323.
Bewley, M. C., K. Springer, Y. B. Zhang, P. Freimuth, and J. M. Flanagan. 1999. Structural analysis of the mechanism of adenovirus binding to its human cellular receptor, CAR. Science 286:1579-1583.
Dechecchi, M. C., P. Melotti, A. Bonizzato, M. Santacatterina, M. Chilosi, and G. Cabrini. 2001. Heparan sulfate glycosaminoglycans are receptors sufficient to mediate the initial binding of adenovirus types 2 and 5. J. Virol. 75:8772-8780.
Einfeld, D. A., R. Schroeder, P. W. Roelvink, A. Lizonova, C. R. King, I. Kovesdi, and T. J. Wickham. 2001. Reducing the native tropism of adenovirus vectors requires removal of both CAR and integrin interactions. J. Virol. 75:11284-11291.
Gaggar, A., D. M. Shayakhmetov, and A. Lieber. 2003. CD46 is a cellular receptor for group B adenoviruses. Nat. Med. 9:1408-1412.
Gall, J., A. Kass-Eisler, L. Leinwand, and E. Falck-Pedersen. 1996. Adenovirus type 5 and 7 capsid chimera: fiber replacement alters receptor tropism without affecting primary immune neutralization epitopes. J. Virol. 70:2116-2123.
Gall, J. G., R. G. Crystal, and E. Falck-Pedersen. 1998. Construction and characterization of hexon-chimeric adenoviruses: specification of adenovirus serotype. J. Virol. 72:10260-10264.
Huang, S., T. Kamata, Y. Takada, Z. M. Ruggeri, and G. R. Nemerow. 1996. Adenovirus interaction with distinct integrins mediates separate events in cell entry and gene delivery to hematopoietic cells. J. Virol. 70:4502-4508.
Kass-Eisler, A., L. Leinwand, J. Gall, B. Bloom, and E. Falck-Pedersen. 1996. Circumventing the immune response to adenovirus-mediated gene therapy. Gene Ther. 3:154-162.
Kidd, A. H., J. Chroboczek, S. Cusack, and R. W. Ruigrok. 1993. Adenovirus type 40 virions contain two distinct fibers. Virology 192:73-84.
Leissner, P., V. Legrand, Y. Schlesinger, D. A. Hadji, M. van Raaij, S. Cusack, A. Pavirani, and M. Mehtali. 2001. Influence of adenoviral fiber mutations on viral encapsidation, infectivity and in vivo tropism. Gene Ther. 8:49-57.
Lieber, A., C. Y. He, L. Meuse, D. Schowalter, I. Kirillova, B. Winther, and M. A. Kay. 1997. The role of Kupffer cell activation and viral gene expression in early liver toxicity after infusion of recombinant adenovirus vectors. J. Virol. 71:8798-8807.
Mizuguchi, H., N. Koizumi, T. Hosono, A. Ishii-Watabe, E. Uchida, N. Utoguchi, Y. Watanabe, and T. Hayakawa. 2002. CAR- or v integrin-binding ablated adenovirus vectors, but not fiber-modified vectors containing RGD peptide, do not change the systemic gene transfer properties in mice. Gene Ther. 9:769-776.
Morelli, A. E., A. T. Larregina, R. W. Ganster, A. F. Zahorchak, J. M. Plowey, T. Takayama, A. J. Logar, P. D. Robbins, L. D. Falo, and A. W. Thomson. 2000. Recombinant adenovirus induces maturation of dendritic cells via an NF-B-dependent pathway. J. Virol. 74:9617-9628.
Nakamura, T., K. Sato, and H. Hamada. 2003. Reduction of natural adenovirus tropism to the liver by both ablation of fiber-coxsackievirus and adenovirus receptor interaction and use of replaceable short fiber. J. Virol. 77:2512-2521.
Nicol, C. G., D. Graham, W. H. Miller, S. J. White, T. A. Smith, S. A. Nicklin, S. C. Stevenson, and A. H. Baker. 2004. Effect of adenovirus serotype 5 fiber and penton modifications on in vivo tropism in rats. Mol. Ther. 10:344-354.
Philpott, N. J., M. Nociari, K. B. Elkon, and E. Falck-Pedersen. 2004. Adenovirus-induced maturation of dendritic cells through a PI3 kinase-mediated TNF-alpha induction pathway. Proc. Natl. Acad. Sci. USA 101:6200-6205.
Rea, D., F. H. Schagen, R. C. Hoeben, M. Mehtali, M. J. Havenga, R. E. Toes, C. J. Melief, and R. Offringa. 1999. Adenoviruses activate human dendritic cells without polarization toward a T-helper type 1-inducing subset. J. Virol. 73:10245-10253.
Roelvink, P. W., A. Lizonova, J. G. Lee, Y. Li, J. M. Bergelson, R. W. Finberg, D. E. Brough, I. Kovesdi, and T. J. Wickham. 1998. The coxsackievirus-adenovirus receptor protein can function as a cellular attachment protein for adenovirus serotypes from subgroups A, C, D, E, and F. J. Virol. 72:7909-7915.
Schoggins, J. W., J. G. Gall, and E. Falck-Pedersen. 2003. Subgroup B and F fiber chimeras eliminate normal adenovirus type 5 vector transduction in vitro and in vivo. J. Virol. 77:1039-1048.
Segerman, A., J. P. Atkinson, M. Marttila, V. Dennerquist, G. Wadell, and N. Arnberg. 2003. Adenovirus type 11 uses CD46 as a cellular receptor. J. Virol. 77:9183-9191.
Shayakhmetov, D. M., Z. Y. Li, S. Ni, and A. Lieber. 2004. Analysis of adenovirus sequestration in the liver, transduction of hepatic cells, and innate toxicity after injection of fiber-modified vectors. J. Virol. 78:5368-5381.
Sirena, D., B. Lilienfeld, M. Eisenhut, S. Kalin, K. Boucke, R. R. Beerli, L. Vogt, C. Ruedl, M. F. Bachmann, U. F. Greber, and S. Hemmi. 2004. The human membrane cofactor CD46 is a receptor for species B adenovirus serotype 3. J. Virol. 78:4454-4462.
Smith, T., N. Idamakanti, H. Kylefjord, M. Rollence, L. King, M. Kaloss, M. Kaleko, and S. C. Stevenson. 2002. In vivo hepatic adenoviral gene delivery occurs independently of the coxsackievirus-adenovirus receptor. Mol. Ther. 5:770-779.
Smith, T. A., N. Idamakanti, J. Marshall-Neff, M. L. Rollence, P. Wright, M. Kaloss, L. King, C. Mech, L. Dinges, W. O. Iverson, A. D. Sherer, J. E. Markovits, R. M. Lyons, M. Kaleko, and S. C. Stevenson. 2003. Receptor interactions involved in adenoviral-mediated gene delivery after systemic administration in non-human primates. Hum. Gene Ther. 14:1595-1604.
Smith, T. A., N. Idamakanti, M. L. Rollence, J. Marshall-Neff, J. Kim, K. Mulgrew, G. R. Nemerow, M. Kaleko, and S. C. Stevenson. 2003. Adenovirus serotype 5 fiber shaft influences in vivo gene transfer in mice. Hum. Gene Ther. 14:777-787.
Tomko, R. P., C. B. Johansson, M. Totrov, R. Abagyan, J. Frisen, and L. Philipson. 2000. Expression of the adenovirus receptor and its interaction with the fiber knob. Exp. Cell Res. 255:47-55.
Tomko, R. P., R. Xu, and L. Philipson. 1997. HCAR and MCAR: the human and mouse cellular receptors for subgroup C adenoviruses and group B coxsackieviruses. Proc. Natl. Acad. Sci. USA 94:3352-3356.
Trevejo, J. M., M. W. Marino, N. Philpott, R. Josien, E. C. Richards, K. B. Elkon, and E. Falck-Pedersen. 2001. TNF-alpha-dependent maturation of local dendritic cells is critical for activating the adaptive immune response to virus infection. Proc. Natl. Acad. Sci. USA 98:12162-12167.
Wickham, T. J., P. Mathias, D. A. Cheresh, and G. R. Nemerow. 1993. Integrins alpha v beta 3 and alpha v beta 5 promote adenovirus internalization but not virus attachment. Cell 73:309-319.
Wolff, G., S. Worgall, N. van Rooijen, W. R. Song, B. G. Harvey, and R. G. Crystal. 1997. Enhancement of in vivo adenovirus-mediated gene transfer and expression by prior depletion of tissue macrophages in the target organ. J. Virol. 71:624-629.
Worgall, S., G. Wolff, E. Falck-Pedersen, and R. G. Crystal. 1997. Innate immune mechanisms dominate elimination of adenoviral vectors following in vivo administration. Hum. Gene Ther. 8:37-44.
Yang, Y., and J. M. Wilson. 1995. Clearance of adenovirus-infected hepatocytes by MHC class I-restricted CD4+ CTLs in vivo. J. Immunol. 155:2564-2570.
Zhang, Y., N. Chirmule, G. P. Gao, R. Qian, M. Croyle, B. Joshi, J. Tazelaar, and J. M. Wilson. 2001. Acute cytokine response to systemic adenoviral vectors in mice is mediated by dendritic cells and macrophages. Mol. Ther. 3:697-707.(John W. Schoggins, Marcel)
Molecular Immunology Unit, Institute of Child Health, 30 Guilford Street, University College London, WC1N 1EH, United Kingdom
ABSTRACT
The major adenovirus (Ad) capsid proteins hexon, penton, and fiber influence the efficiency and tropism of gene transduction by Ad vectors. Fiber is the high-affinity receptor binding protein that serves to mediate cell attachment in vitro when using coxsackie-adenovirus receptor (CAR)-containing cell lines. This contrasts with transduction efficiency in macrophages or dendritic cells that lack high concentrations of CAR. To determine how fiber influences gene transduction and immune activation in a murine model, we have characterized Ad type 5 (Ad5) vectors with two classes of chimeric fiber, CAR binding and non-CAR binding. In a systemic infection, Ad5 fiber contributes to DNA localization and vector transduction in hepatic tissue. However, the majority of vector localization is due to Ad5 fiber-specific functions distinct from CAR binding. CAR-directed transduction occurs but at a modest level. In contrast to CAR binding vectors, the F7 and F7F41S non-CAR-binding vectors demonstrate a 2-log decrease in hepatic transduction, with a 10-fold decrease in the amount of vector DNA localizing to the hepatic tissue. To characterize the innate response to early infection using fiber chimeric vectors, intrahepatic cytokine and chemokine mRNAs were quantified 5 hours postinfection. Tumor necrosis factor alpha mRNA levels resulting from Ad5 fiber infections were elevated compared to viruses expressing serotype 7 or 41 fiber. Levels of chemokine mRNA (gamma interferon-inducible protein 10, T-cell activation gene 3, and macrophage inflammatory protein 1?) were 10- to 20-fold higher with CAR binding vectors (Ad5 and F41T) than with non-CAR-binding vectors (F7 and F7F41S). In spite of quantitative differences in vector localization and innate activation, fiber pseudotyping did not significantly change the outcome of anti-Ad adaptive immunity. All vectors were cleared with the same kinetics as wild-type Ad5 vectors, and each induced neutralizing antibody. Although non-CAR-binding vectors were impaired in transduction by nearly 2 orders of magnitude, the level of antitransgene immunity was the same for each of the vectors. Using primary bone marrow-derived macrophages and dendritic cells, we demonstrate that transduction, induction of cytokine/chemokine, and phenotypic maturation of these antigen-presenting cells are independent of fiber content. Our data support a model where fiber-mediated hepatic localization enhances innate responses to virus infection but minimally impacts on adaptive immunity.
INTRODUCTION
Studies using adenovirus (Ad) vectors have provided insight into the major problems that must be addressed when considering viral vectors for gene therapy applications. Issues confronting the use of Ad vectors include the innate and adaptive antiviral immune response, the lack of cell- or tissue-specific targeting, and the need to establish stable long-term gene expression. The Ad capsid plays a major role in two of these issues, tissue-specific targeting and stimulation of the antiviral innate and adaptive immune response. Therefore, efforts to manipulate the viral capsid structure offer an opportunity to control vector targeting as well as influence immune activation.
Three major protein components of the Ad capsid include the structural proteins hexon, penton base, and fiber. Hexon is the main structural element forming the body of the capsid. Penton base and fiber form the structural unit penton, located at each of the 12 vertices of the icosahedral capsid. The capsid proteins are responsible for efficient gene transduction by adenovirus vectors. Adenovirus type 5 (Ad5) uptake into human cell lines occurs through a multistep process that first relies on high-affinity binding of fiber to a cell surface receptor. For the majority of Ad serotypes, fiber binds to the cell surface protein coxsackie-adenovirus receptor (CAR) (1, 2, 19, 28), which is expressed in a variety of tissue types (27). Following fiber/CAR complex formation, penton base binds to v integrins and stimulates endocytosis of viral particles (30). In the absence of fiber-mediated high-affinity binding, transduction can still occur but at significantly reduced levels. How Ad5 recognizes and binds cells in the absence of a fiber-directed pathway is not well understood. Penton base binding to v integrins may contribute to the low efficiency uptake of Ad5 vectors in cells that lack CAR (8). Other nonspecific mechanisms such as phagocytosis and macropinocytosis may be used in certain cell types.
In contrast to in vitro studies of Ad infection, in vivo applications of Ad5-based vectors have revealed a more complex biology of virus-cell interactions that influence gene transduction. Following systemic administration in a murine model, greater than 90% of an Ad5-based vector localizes to the liver (32). However, the majority of virus localized to the liver within the first hours postinfection does not contribute to hepatocyte transduction and is eliminated within 24 h of administration. For the small fraction of Ad5 vector that successfully transduces hepatic tissue, the role of high-affinity fiber/CAR binding is surprisingly small. Depending on the study and vector used, between 10 and 50% of intrahepatic transduction occurs in a fiber/CAR-dependent manner (4, 11, 13, 24). It has been proposed that Ad5 fiber binding to heparan sulfate glycosaminoglycans (HSG) (3, 16, 26) plays an important role in intrahepatic localization as well as transduction following systemic administration of vector in a murine model.
The vast majority of Ad5 vector that localizes to the liver is rapidly eliminated. Kupffer cells are the predominant macrophages of liver, and studies have demonstrated that Kupffer cell depletion impacts on viral localization and transduction (12, 32). Macrophages and other cells of the immune system take up Ad5-based vectors, but these cells often lack CAR and transduction is generally very inefficient (8). Recognition of virus by these cells results in vector clearance and contributes to both the inflammatory and the antiviral adaptive immune responses that arise following exposure to virus (34). For Ad5 vectors, the strong inflammatory response and the antiviral adaptive immune response are major impediments for gene therapy applications but potential assets for anticancer or vaccine applications. These studies highlight both the importance of understanding the complex binding interactions that take place in a systemic infection with Ad vectors and the need to develop effective Ad detargeting strategies.
Several Ad5-based vectors that demonstrate a decreased transduction phenotype in vivo have been constructed. Mutations in Ad5 fiber that abolish CAR binding have been combined with mutations in putative HSG binding domains (16, 25), resulting in diminished liver transduction following systemic administration. Another approach to detargeting Ad5-based vectors has been to replace Ad5 fiber with a fiber moiety from a non-CAR-binding virus. There are several classes of non-CAR-binding Ad fibers. One is the subgroup B class of virus. Recent studies have demonstrated that human CD46 serves as a target receptor for many of the subgroup B viruses (5, 21, 23). A variety of human cell lines express the subgroup B receptors and are efficiently infected by subgroup B vectors or Ad5 vectors pseudotyped with a subgroup B fiber (5, 21). Important to this study, the CD46 receptor is not abundantly expressed in murine tissues, making a subgroup B fiber-containing vector potentially detargeted in a mouse model (5). A second class of non-CAR-binding fiber is derived from the subgroup F adenoviruses. The subgroup F viruses express two fiber genes tandemly arranged (10): a long-shafted fiber that binds CAR and a short-shafted fiber that has no known binding function (19). Recent studies using Ad5 vectors pseudotyped with subgroup F short fiber demonstrate reduced transduction of hepatic tissue following a systemic administration of vector (15, 16, 20).
In the current study we have examined the biology of fiber-modified viruses with respect to viral transduction, viral DNA localization, and the influence on innate and adaptive immune responses to Ad5-based fiber detargeted vectors. We have found that, in Ad5-based vectors, fiber-mediated transduction and localization to liver contribute to a heightened early innate response as determined by induction of chemokine and cytokine mRNAs. In comparison, non-CAR-binding vectors exhibit reduced localization to the liver, reduced gene transduction, and diminished induction of mRNA associated with activation of the inflammatory response. In spite of these beneficial detargeting characteristics, the adaptive arm of the immune system is largely unaffected by these capsid modifications.
MATERIALS AND METHODS
Adenovirus vector stocks and cell lines. Four E1/E3-deleted Ad5-based vectors were used in this study: Ad5, F7, F7F41S, and F41T. Construction and characterization of these vectors have been described in detail elsewhere (20). The F7 vector was generated by genetically swapping the fiber 5 coding region of Ad5 with the entire fiber gene from Ad serotype 7. The F7F41S vector genome contains coding regions from both fiber 7 and the short-shafted fiber from Ad serotype 41 as two terminal exons in tandem. The F41T vector genome contains both short- and long-shafted fibers from Ad serotype 41.
HEK-293 monolayer cultures were kept in Dulbecco modified Eagle medium (DMEM) plus 5% to 10% calf serum. FL83B murine hepatocyte (ATCC CLR-2390) cultures were maintained in F12K plus 10% fetal bovine serum (FBS). P6 macrophages were kindly provided by A. Ding and kept in DMEM plus 10% FBS.
In vitro transduction assays. Cell lines used in this study were seeded into 24-well plates and infected with 1,000 or 5,000 particles/cell in medium without serum for 30 min at 37°C. After infection, virus was aspirated and fresh medium was added back. Infected cells were harvested 24 h postinfection in TEN scrape buffer (40 mM Tris Hcl [pH 7.4], 1 mM EDTA, 0.15 M NaCl) and resuspended in 0.25 M Tris, pH 7.8. Lysates were subjected to three freeze-thaw cycles followed by a 10-min incubation at 65°C to inactivate cellular deacetylases. Chloramphenicol acetyltransferase (CAT) activity assays were performed as described previously (6).
In vivo transduction assays. Six-week-old female B6129/J mice were obtained from Taconic or Jackson Laboratories and maintained in compliance with institutional protocols. Mice were injected retro-orbitally with 1010 virus particles or PBS vehicle alone in a total volume of 100 μl. At indicated time points, animals were sacrificed and organs were harvested, weighed, and resuspended in 2 volumes PBS by weight. Tissue lysates were prepared by homogenizing organs and centrifuging them for 30 min at 3,000 x g, followed by a 10-minute incubation at 65°C. Total tissue CAT activity was assayed as described above.
DNA isolation and Southern analysis. Livers from Ad- or mock-infected animals were harvested, and a portion was used for DNA isolation. Tissue slices were pulverized in Eppendorf tubes and incubated in lysis buffer (10 mM Tris [pH 8.0], 400 mM NaCl, 2 mM EDTA, 0.5% sodium dodecyl sulfate [SDS], 0.4 mg/ml proteinase K) overnight at 55°C. Samples were treated with 2 μl RNase A (10 mg/ml) for 30 min at 37°C. DNA was purified by two rounds of phenol-CHCl3 extraction and a CHCl3-only extraction, followed by ethanol precipitation. For Southern analysis, 10 μg DNA was digested with EcoRI and electrophoresed on a 1% agarose gel. Samples were transferred to a nylon membrane, and blots were probed with either EcoRI-digested Ad5 vector DNA or ?-actin. DNA levels were quantified by PhosphorImager analysis.
RNA isolation and RPA. Bone marrow-derived macrophages (BMMOs) and liver sections from Ad- or mock-infected animals were harvested for RNA isolation with Trizol reagent (Invitrogen) according to the manufacturer's instructions. Five (BMMO) or 10 (liver) μg of total RNA was hybridized to 32P-labeled probes generated from either the mCK3b (cytokine) or mCK5c (chemokine) multiprobe template sets (Pharmingen). A commercially available RNase protection assay (RPA) kit (Pharmingen) was used to RNase treat samples and recover protected fragments. Resolution of protected fragments was carried out by electrophoresis on 10% polyacrylamide sequencing gels. Gels were dried and exposed to a PhosphorImager screen, and mRNA species were quantified by PhosphorImager analysis.
Anti-Ad neutralizing antibody blocking assay. Neutralizing antibody titers were determined by a functional in vitro blocking assay in HeLa cells. Serum samples from mice infected with the various vectors were taken at day 28 and serially diluted in DMEM. Serum dilutions were incubated with 5 x 105 particles of a standard Ad5CAT vector for 1 h at 37°C. The serum/virus mixture was then added to 105 HeLa cells for 30 min at 37°C. Cells were washed with DMEM, and fresh medium was added back. CAT activity in cell extracts was scored 24 h later as described above.
Anti-CAT antibody ELISA. To detect serum levels of antitransgene antibody, Maxisorp enzyme-linked immunosorbent assay (ELISA) plates (Nalge Nunc International) were coated with CAT (Sigma) in 50 mM sodium carbonate buffer, pH 9.6, for at least 16 h at 4°C. The CAT solution was removed, and 1% casein (Sigma) in PBS was added as a blocking buffer for 1 h at room temperature. Plates were washed three times in blocking buffer plus 0.05% Tween 20. Serum samples from 28-day-infected mice were serially diluted in blocking buffer and added to ELISA plates for 1.5 h at room temperature. Plates were washed three times as described above. A secondary horseradish peroxidase-conjugated anti-mouse antibody (Amersham) was diluted in blocking buffer plus 0.05% Tween 20 and added to plates for 1.5 h at room temperature. Plates were then washed four times as described above and incubated with the tetramethylbenzidine peroxidase substrate (Sigma). Reactions were stopped with 1 N H2SO4, and plates were read at 450 nm on a spectrophotometer.
BMMOs and bone marrow dendritic cell (BMDC) isolation and infection. Bone marrow macrophages were generated by culturing bone marrow cells in DMEM plus 20% FBS and 30% supernatant derived from L929 confluent cells, replacing two-thirds of the culture volume with fresh granulocyte-macrophage colony-stimulating factor (M-CSF)-containing medium every third day. Cells were used after 7 to 9 days, and all stained positive for CD11b but were negative for CD11c and B220.
BMDCs were obtained by culturing bone marrow cells in DMEM plus 10% FBS in the presence of 15% supernatant derived from J558L cells as a source of GM-CSF, replacing two-thirds of the culture volume with fresh GM-CSF-containing medium every second to third day. Cells were used after 5 days and contained 60 to 90% CD11c+ cells.
Cultures of BMMOs and BMDCs were infected with the indicated virus at 5,000 particles/cell. For transduction experiments, cells were harvested 24 h postinfection for CAT assays. For maturation phenotype, cells were harvested 48 h postinfection and stained on ice with phycoerythrin-conjugated anti-CD86 mouse antibody (Becton Dickinson) for 30 min in PBS containing 1% bovine serum albumin. Cells were fixed in 1% formalin and analyzed on an EPICS XL flow cytometer (Beckman Coulter).
RESULTS
Macrophage and hepatocyte transduction in vitro. Four fiber variants of Ad5 were selected to characterize the influence of fiber on transduction and immune activation. These viruses, Ad5, F7, F7F41S, and F41T, differ only in the nature of the fiber homotrimer(s) present in the viral capsid (Fig. 1A). The parental Ad5 virus is an E1/E3-deleted vector expressing wild-type fiber 5 with the cytomegalovirus immediate-early promoter driving expression of CAT as a reporter gene. The fiber present in this vector contains CAR binding domains as well as the putative HSG binding domain (KKTK) identified as being important for CAR-independent transduction. The F7 virus contains the full-length fiber sequence from the Ad7 subgroup B virus in place of fiber 5 (6). Subgroup B fibers do not bind CAR. The tandem fiber vectors express two distinct fiber proteins, either fiber 41S and fiber 41L (F41T) or fiber 7 and fiber 41S (F7F41S). F41L is a CAR binding fiber but does not contain the KKTK domain present in Ad5 fiber. Fiber 41S has no known binding function.
Viruses were grown on a large scale in 293 cells, followed by two rounds of CsCl banding purification and dialysis to remove salts. Viral particle numbers were quantified by optical density at 260 nm (1012 particles/unit of optical density at 260 nm). To confirm that each vector was quantitatively matched, 1010 particles of each vector were separated by SDS-polyacrylamide gel electrophoresis and analyzed by Western blotting with antihexon antibody (Fig. 1B). The blot was then probed with antifiber antibody (4D2 from Neomarkers) to demonstrate the fiber content of each of the vectors (Fig. 1C). The lack of fiber 7 in F7F41S is due to the preferred expression and incorporation of fiber 41S into virions (20). Based on this assay, the fiber content of F7F41S is essentially F41S.
To determine transduction functions, a murine hepatocyte cell line (FL83B) and a murine macrophage cell line (P6) were infected with the indicated virus at 103 Ad particles/cell. Following a 30-minute incubation, cells were washed twice with fresh medium to remove free virus. Cells were incubated for 24 h and harvested for reporter gene assay. The Ad5 and F41T vectors were highly effective at delivering a transcriptionally active reporter gene (Fig. 2A) into FL83B murine hepatocyte cells. In comparison, transduction through CAR-independent fiber (F7) was decreased 100-fold compared to CAR-targeting viral vector. This contrasts with the efficient transduction mediated by F7 vectors in human cell lines (HeLa, HepG2, and A549) (20; data not shown). The low level of transduction by F7-containing viruses in FL83B hepatocytes is consistent with studies demonstrating low levels of subgroup B virus receptor(s) expression in murine cells (5). CAT gene transduction from F7F41S virus infection was over 200-fold lower than that from Ad5 and 500-fold less than that from F41T. The lack of transduction by F7F41S virus supports previous observations (20) that fiber 41S does not contribute to efficient cell binding and virus uptake.
In contrast to transduction efficiency in the FL83B hepatocyte cell line, transduction of the P6 macrophage cells by the Ad5 vector was found to be very low, approximately 2 logs below transduction in FL83B cells when identical experimental conditions are used (compare Fig. 2A and B). The F7 virus was also unable to transduce the macrophage cell line, arguing that these cells also lack a suitable subgroup B receptor. Each of the tandem fiber vectors had similar and slightly higher transduction efficiencies in macrophage P6 cells. Similar results were found in the RAW264.7 murine macrophage cell line (data not shown). These results indicate that murine macrophage transduction is independent of CAR or subgroup B receptor uptake mechanisms. Because established cell lines do not always reflect the complete phenotype of a primary cell type, we chose to compare the transduction of the P6 macrophage cell line with cultured primary BMMOs isolated from B6129/J mice. Cells (P6 or BMMOs) were infected with a slightly higher dose of virus (5,000 particles/cell), and CAT transduction assays were performed (Fig. 2C). CAT expression in P6 cells was increased approximately fivefold compared to infections with 1,000 particles/cell, reflecting a dose response to each of the fiber chimeric constructs. At 5,000 particles/cell, transduction of the BMMOs was significantly higher than transduction of established macrophage cell lines. At this dose of vector, there was no apparent difference in transduction between the fiber chimeras. This result reveals an indifference of macrophages to the fiber type present in the Ad5 capsid.
Macrophages contribute to the early inflammatory response to pathogens and have been implicated in the antiviral response to Ad vectors. To determine if the fiber chimeras differentially activate macrophages, we assessed the maturation of BMMOs by upregulation of costimulatory molecules (such as CD86, CD40, or major histocompatibility complex class II). At 24 h following virus infection, BMMOs were stained with anti-CD86 antibody and subjected to flow cytometry. Essentially 100% of CD11b-positive BMMOs underwent activation as revealed by upregulation of CD86 regardless of the fiber configuration (Fig. 3A). To assess an early activation response of BMMOs to the chimeric viruses, BMMOs were infected as previously described and harvested 5 h postinfection for total RNA. RNA from each infection was used in an RNase protection assay to quantify induction of cytokine and chemokine mRNA. Using a cytokine multiprobe template set under the experimental conditions described, each fiber chimera was found to induce expression of tumor necrosis factor alpha (TNF-) mRNA (Fig. 3B). Similar expression patterns were seen for gamma interferon (IFN-)-inducible protein 10, macrophage inflammatory protein 1?, and T-cell activation gene 3 using a chemokine multiprobe template set (Fig. 3C). At the level of transduction and activation, the data clearly indicate that each of the chimeric constructs is able to transduce and effectively stimulate primary macrophage cells.
CAR-mediated transduction and viral DNA localization. CAR-dependent infection is a primary mechanism responsible for high-level Ad vector gene transduction in murine hepatocyte cell lines, but transduction in macrophage cells occurs in a fiber/CAR-independent manner. We next asked how the fiber modifications in our vectors influence sequestration and transduction at early times after systemic administration. B6129/J mice were infected with 1010 virus particles by intravenous injection. At 5 and 24 h, animals were sacrificed. Liver and spleen were harvested for reporter gene expression, total DNA, and total RNA. CAT expression at 5 h postinfection represents an efficient uptake and gene expression process. Under our experimental conditions, the CAR-targeted vectors were able to mediate strong levels of CAT expression by 5 h postinfection (Fig. 4A). The results obtained at 5 h are similar to the pattern seen in hepatocytes (Fig. 2A) where CAR-dependent transduction is prevalent, but a blending of hepatocyte and macrophage infections may better represent the overall hepatic expression pattern. At 24 h postinfection the magnitude of expression has increased with all vectors, without an overwhelming fiber-directed bias. Consistent with a detargeting phenotype, the F7 and F7F41S vectors yield 20- to 50-fold-less hepatic transgene expression than Ad5 fiber vectors. Assays characterizing CAT activity in kidney, lung, and intestine showed that activity was well below the low levels found in spleen and therefore the levels were viewed as background (data not shown).
In contrast to transgene expression, a different pattern was found when characterizing the yield of total viral DNA localized to the liver. DNA was isolated from liver harvested 5 and 24 h postinfection with each of the viruses. The CAR binding viruses result in high levels of gene expression but diverge greatly with respect to total viral DNA localized to the liver (Fig. 4B and C). Ad5 vector DNA is present at the greatest level 5 h postinfection whereas F41T is present at the lowest level. Since both of these vectors transduce through CAR at similar efficiencies (Fig. 2A), the observed variation in localization is attributed to non-CAR-binding attributes of fiber 5 compared to fiber 41L or 41S of F41T. As previously mentioned, Ad5 fiber contains a putative HSG binding domain, which is absent in the 41L/S fibers. Although the F7 vector yields the lowest level of transduction, it was second to Ad5 with respect to viral DNA localized to the liver at 5 h postinfection, with F7F41S present at a slightly lower concentration. When total DNA isolated from splenic tissue was probed for viral DNA by Southern blotting, levels were minimal compared to liver for all samples (data not shown); therefore, detargeted vectors were not retargeted to spleen. These data indicate disassociation between viral DNA localization and transgene expression at the 5-hour time point.
At 24 h postinfection, the majority of viral DNA present at 5 h is eliminated. DNA from the Ad5 infection is clearly above background levels (Fig. 4B and C). For Ad5-based vectors, the elimination of vector DNA present at 5 h and absent at 24 h has been attributed to vector clearance by Kupffer cells (31). For both CAR binding and non-CAR-binding vectors, the majority of viral DNA initially localized to the liver is not associated with gene transduction and has been eliminated by 24 h postinfection.
Altered chemokine and cytokine profiles as a function of fiber modification. The fiber-modified viruses are distinguished from one another by the efficiency of gene expression as well as the magnitude of virus localized to the liver. We next asked if these differences in virus targeting would have an impact on immune activation by the host. To characterize the overall early inflammatory response to the fiber-modified vectors, we first characterized serum levels of TNF- and interleukin-6 (IL-6) at 5 and 24 h postinfection by ELISA. Under the experimental conditions described (1010 viral particles/animal) levels of TNF- and IL-6 in serum did not increase over those found in mock-infected animals (data not shown). We then isolated RNA from portions of liver harvested at 5 and 24 h postinfection as described in the previous section. Samples (10 μg) of RNA were used in an RNase protection assay using the mCK3b cytokine template set (BD Biosciences) to determine if the fiber-modified vectors induce novel patterns of cytokine activation. Under our experimental conditions (infection of each animal with 1010 viral particles), Ad-mediated induction of cytokine mRNA from total liver was very modest at 5 h postinfection (Fig. 5A) (and lower at 24 h [data not shown]). Using PhosphorImager analysis of protected liver RNA species (Fig. 5B), we found that TNF- transcripts were induced in animals infected with the Ad5 vector compared to liver isolated from animals infected with the fiber-modified viruses. We also found that mRNAs for two other cytokines, IFN- and macrophage inducing factor, were induced compared to mock-infected animals. Based on the magnitude of induction, the upregulation of cytokine mRNA is roughly consistent with viral DNA localization to the liver (compare Fig. 3C and 5B). Analysis of RNA harvested from animals after 24-hour virus exposure revealed baseline concentrations for all cytokine mRNAs (data not shown).
RNA generated from the infections harvested at 5 h was also used to determine Ad induction of chemokines using the mCK5c chemokine template set. Results from this assay revealed a chemokine mRNA induction profile that was fiber dependent and distinct from the cytokine mRNA induction pattern (Fig. 6A and B). The chemokine mRNA expression pattern resembled the pattern of fiber-dependent gene transduction (compare Fig. 3A and 6B). IFN--inducible protein 10 and T-cell activation gene 3 mRNAs were highly induced by both CAR binding viruses, and macrophage inflammatory protein 1? was expressed at more moderate levels. After normalization, the non-CAR-binding viruses F7 and F7F41S were slightly above baseline levels of mock-infected animals but 10- to 20-fold less than the CAR-targeted vectors. These studies demonstrate that fiber-mediated detargeting of otherwise identical vectors has a demonstrable impact on the induction of both cytokines and chemokines associated with early stages of the anti-Ad immune response.
Clearance efficiency and humoral immunity unchanged by fiber modifications. Based on the observations that the fiber-modified viruses were compromised in induction of intrahepatic inflammatory cytokine/chemokine mRNA, we next asked if we could detect a significant difference in the adaptive immune response to virus infection. It is well established that cytotoxic T cells are largely responsible for clearance of Ad-infected cells following an in vivo infection in an immunocompetent host (33). Therefore, if fiber-modified viruses do not activate the cellular immune response as effectively as Ad5, we would predict increased persistence over time of CAT-expressing cells. Similarly if the modified vectors influence the humoral immune response to Ad, we would predict a diminished level of anti-Ad antibody to be present in infected animals. To test these elements of the immune response to fiber-modified vectors, we established a 4-week time course experiment infecting B6129/J mice with 1010 particles of the indicated virus. Animals were sacrificed at the indicated time point, and liver, spleen, and serum were harvested.
The profiles for transgene expression over the 4-week time course for the modified viruses were remarkable in their similarity to Ad5 vector (Fig. 7). In liver, peak expression for each construct occurred at 3 days postinfection (Fig. 7A). CAT expression began to decline at 6 days postinfection, and by 28 days CAT expression had returned to baseline levels in all animals. Similarly, the kinetics of CAT transgene expression and clearance from spleen were essentially identical (Fig. 7B). The only distinguishing feature was the magnitude of transgene expression. At the level of vector clearance, the fiber-modified vectors were eliminated as efficiently as the Ad5 vector.
We next determined the level of anti-Ad neutralizing antibody in serum isolated from animals 28 days postinfection by an antibody-blocking assay as described in Materials and Methods. Each of the fiber-modified viruses induced blocking antibody to Ad5 (Fig. 8A). However, the efficiency of blocking activity was lower in serum generated from infections with the fiber-modified vectors. Because these studies were blocking Ad5 infection and not the dosing vector, this slightly diminished response may reflect differences in antibody directed to serotype-specific fibers.
To determine if the neutralizing antibody generated in response to the fiber-modified vectors was biologically significant, we used an in vivo readministration assay (7, 9). Animals were injected with 1010 particles of each virus, and on day 25 postinfection, when CAT expression is at baseline, each animal was injected with 1010 particles of Ad5CAT. Livers were harvested 3 days postinfection and assayed for CAT gene expression. Regardless of the vector used in the original administration, all animals were able to effectively block Ad5CAT gene transduction in our readministration protocol (Fig. 8B).
The CAR detargeted vectors are significantly less effective in transducing cells in vivo, resulting in a diminished level of overall CAT gene product. To determine if the efficiency of transgene expression correlated with the serum levels of anti-CAT antibody, we quantified anti-CAT antibody by ELISA (Fig. 8C). Serum levels of anti-CAT antibody were lowest for Ad5, which expressed the highest level of CAT. The modified vectors generated similar or slightly higher levels of anti-CAT antibody. In contrast to the different levels of liver and spleen CAT gene expression mediated by the fiber-modified viruses, anti-CAT antibody was found to be essentially the same for each of the viruses examined.
Transduction and maturation of bone marrow-derived dendritic cells by fiber-modified viruses. We have used two basic endpoint assays (clearance of infected cells and anti-Ad humoral immunity) to demonstrate that the adaptive immune response to fiber-modified vectors is largely unaffected by the efficiency of gene transduction or the magnitude of the inflammatory response as characterized by the cytokine/chemokine RNA induction profile. These results would argue that recognition of these vectors by specialized antigen-presenting cells of the immune system (primarily macrophages and dendritic cells) is unaffected by fiber detargeting functions. We have already demonstrated that primary macrophages are responsive to Ad in a fiber-independent manner. Dendritic cells are considered the primary effector cells of the adaptive arm of the immune system and have been shown to respond to adenovirus infection (14, 17, 18, 29). The fiber-modified viruses were used to infect BMDCs (5 x 103 particles/cell) as previously described. Levels of CAT expression in BMDCs were significantly below those previously identified in BMMOs, approximately 10- to 20-fold less under comparable conditions (compare Fig. 3C to 9A). All vectors were above background, and the F7 construct was again found to be the least effective at CAT transduction (Fig. 9A). We next determined if the fiber-modified viruses influenced BMDC maturation/activation. Immature BMDCs express low levels of cell surface markers including CD86. When exposed to a stimulating antigen (Ad5 virus), immature BMDCs undergo maturation characterized by upregulation of cell surface markers including CD86 (17). We found that exposure of BMDCs to Ad5, F41T, and F7F41S resulted in upregulation of CD86 compared to mock-infected BMDCs (Fig. 9B). Upregulation of CD86 following BMDC exposure to the F7 virus was no different from that with mock-infected DCs. The observation that F7 demonstrated diminished transduction and activation in DCs led us to test if increasing doses of F7 would lead to increasing transduction and activation of BMDCs. The BMDCs were incubated with increasing doses of F7 virus, and 24 h postinfection, a portion of each infection was harvested for measurement of CAT gene expression. A control infection of Ad5 at 5,000 particles indicated slightly higher levels of transduction than F7. With increasing doses of F7 virus, we found a corresponding increase in CAT gene expression (Fig. 9C). When the DCs were harvested at 48 h postinfection and characterized with respect to upregulation of the costimulatory molecules CD86 (Fig. 9D) and CD40 (Fig. 9E), we found that increasing concentrations of virus resulted in increased expression of the DC activation markers. Based on the data from this study, we found that fiber-modified viruses were able to induce upregulation of costimulatory molecules in bone marrow-derived dendritic cells.
DISCUSSION
The combination of uncontrolled vector delivery and virus uptake by cells of the immune system has presented major obstacles to the use of Ad vectors in gene therapy applications. In the current study we have used fiber-pseudotyped first-generation Ad vectors to characterize how fiber influences gene transduction, vector localization, and immune stimulation following systemic administration in a murine model. Based on the observations made in this study and consistent with observations from other laboratories (13, 22, 26), Ad5 fiber contributes significantly to liver localization following a systemic administration of vector. When the Ad5 fiber is replaced with either fiber 7 or a combination of fiber 7 and fiber 41S, both vector localization to liver and vector-mediated gene transduction are greatly reduced. These constructs represent "fiber-detargeted" vectors. A major advantage of the fiber-detargeted vectors is a decrease in the magnitude of hepatic cytokine and chemokine mRNA induction. Although both localization to the liver and vector-mediated gene transduction are greatly reduced with these constructs, they are not eliminated. Based on the results presented in this study, we consider residual localization to liver and low-level vector transduction with these constructs to result from fiber-independent transduction. This background may be attributed to low-affinity binding through nonfiber capsid proteins, such as penton base or hexon. Background localization/transduction may also result from nonspecific phagocytic/macropinocytotic uptake of viral particles.
CAR-mediated transduction following a systemic administration of virus is revealed during very early time points (5 h), where an assay of transgene expression depends on efficient virus binding and internalization. Efficient CAT transduction is found with the CAR binding vectors. This result is consistent with in vitro transduction assays in the FL83B hepatocyte cell line (compare Fig. 3A and 2A). Although similar levels of CAT expression are found with both CAR binding vectors at 5 h postinfection, there is five times more Ad5 viral DNA localized to the liver than F41T viral DNA. Two conclusions can be made from this result. Following vector administration, as virus is passing through the hepatic tissue, CAR-mediated binding facilitates hepatic transduction equally with both CAR binding vectors. This conclusion would indicate that both the Ad5 and F41T vectors are initially presented to the hepatic cells in a manner that allows each to target through CAR with equal efficiency. Consistent with this hypothesis, other studies have shown that, regardless of the type of fiber present on Ad, similar amounts of virus are present in the liver within the first 30 min following a systemic infection (22). The second conclusion is that Ad5 fiber, compared to fiber 41L, increases vector residence time within the liver. This conclusion is derived from the observation that Ad5 vector localization to liver is five times that of any of the other vectors at 5 h postinfection. Differences in CAR binding efficiency do not account for the differences in localization between Ad5 and F41T. It is worth noting that the two CAR binding fibers (F5 and F41L) are essentially identical in fiber shaft length. Therefore, we do not consider fiber length to directly account for Ad5 localization to liver. Increased residence time can be attributed to other features of Ad5 fiber, for example, the putative HSG binding domain KKTK present in Ad5 fiber and absent in each of our pseudotyped fiber constructs. This domain has been shown to facilitate Ad5 hepatic localization following a systemic administration in both murine and rat models (16, 26).
At time points of peak gene expression (days 1, 3, and 6) (Fig. 7) levels of Ad5-mediated CAT expression are above those seen with fiber-modified vectors. Transduction associated with the fiber-modified vectors F7 and F7F41S is roughly 1 to 5% of Ad5, whereas F41T-mediated CAT expression is approximately 10% of that found with Ad5. The F41T transduction differential compared to the non-CAR-binding vectors was approximately 10:1. This represents the enhancement of CAR-mediated transduction over fiber-detargeted vectors (F7 and F7F41S). The difference between F41T and Ad5 would be consistent with CAR-independent binding functions of Ad5 fiber. Based on the data presented in this study, the transduction differential between Ad5 and the fiber-modified vectors is consistent with a model in which Ad5 fiber mediated increased hepatic residence time. Therefore, overall levels of transduction are influenced by both CAR binding and non-CAR-binding mechanisms.
In addition to helping define the role of fiber in Ad vector localization and gene transduction, these studies also revealed how fiber can influence the innate response to virus infection. Coincident with increased liver localization, the Ad5 fiber vector facilitates induction of higher levels of intrahepatic TNF- mRNA compared to the pseudotyped vectors. In our studies we have used a comparatively low dose of virus, 1010 particles/animal. At this dose of input virus, the induction of inflammatory response is at the limit of sensitivity for many assays. Serum ELISAs of infected animals for TNF- and IL-6 at 5 and 24 h postinfection showed levels indistinguishable from background levels (data not shown). However, using the RPA we were able to detect TNF- mRNA induction at 5 h postinfection. Induction of cytokines by Ad5 vectors is well documented (at higher doses) and contributes to the strong anti-Ad inflammatory response (33). In a recent study, Shayakhmetov et al. (22) characterized chimeric constructs containing fibers with different shaft and/or knob compositions using 1011 particles/animal. At this dose, when characterizing induction of cytokine mRNA (TNF-), all viruses induced expression at levels that were slightly over background at 6 h postinfection. They also found that constructs that contained the Ad5 fiber shaft induced higher levels of TNF- than viruses expressing Ad35 shaft (Ad35 is a subgroup B virus) by an ELISA. Our studies also support a model where the presence of Ad5 fiber shaft stimulates higher levels of TNF- mRNA. In addition, we demonstrate that a long-shafted fiber such as F41L does not stimulate TNF- in the same manner as the Ad5 fiber. Taken together these results argue that the Ad5 fiber shaft that contains the KKTK domain contributes to TNF- mRNA induction.
CAR binding viruses were also shown to induce higher levels of intrahepatic chemokine mRNA compared to the fiber-detargeted vectors. Upregulation of chemokine mRNA at 5 h postinfection was not a function of total vector DNA localized to the liver, since the F41T vector induced chemokines to a level equal to that induced by Ad5. Rather chemokine induction more closely parallels CAR-mediated transduction. Shayakhmetov et al. (22) found heightened induction of IL-6, MCP-1, and IFN-? with a CAR binding Ad5 fiber shaft construct (Ad5/9L) at 1011 particles/animal compared to their other constructs. They did not characterize a CAR binding virus that lacked the Ad5 fiber shaft and concluded that the shaft length was a major factor in cytokine activation. Our observations are consistent with a role of Ad5 fiber shaft in cytokine activation, but we would also conclude that there exists a distinct role for the CAR binding domain in stimulating elements of the innate immune system. Further investigation will be required to determine if the basis of chemokine induction by the CAR binding vectors is due to gene expression or to efficient early transduction mediated by CAR-directed uptake. In the murine system, the two non-CAR-binding vectors, F7 and F7F41S, represent simple platforms for pursuing a genetic strategy to retarget Ad vectors and offer the benefit of diminished induction of inflammatory cytokines.
The F7 and F7F41S vectors also represent base platforms that can be used to explore the interaction of Ad vectors with cells of the immune system. Based on the decrease in liver (and spleen) localization, the decrease in liver (and spleen) transduction, and the decrease in intrahepatic induction of inflammatory cytokines, we asked if these vectors would also demonstrate a diminished ability to stimulate the adaptive immune response. Characterization of the adaptive immune response was based on straightforward endpoint assays relevant to gene therapy applications: clearance efficiency of transduced cells, antiviral humoral immunity, and the development of a humoral immune response to the delivered transgene CAT. In all cases, regardless of the assay or virus being tested, the adaptive immune response was largely indifferent to fiber. The detargeted vectors that resulted in diminished intrahepatic induction of chemokine mRNAs were as effective as the Ad5 fiber vector in stimulating the adaptive arm of the immune response.
Fiber-detargeted viruses mediate low-level gene transduction both in vitro and in vivo. We have shown that vector activation of antigen-presenting cells (both bone marrow-derived macrophages and dendritic cells) occurs in a fiber-independent manner in vitro. Consistent with in vitro observations, the threshold for activation of the adaptive arm of the immune response has been achieved in these studies with each of the fiber-pseudotyped vectors. In a previous study (17), we found that the RGD motif of the penton base was necessary to optimally activate DCs following exposure to Ad. In order to further compromise the immune activation of the fiber-detargeted vectors, we anticipate that an additional benefit can be obtained by combining fiber detargeting with the penton RGD mutations. Combining these capsid modifications with strategies such as use of tissue-specific promoters for transgene expression, expression of less immunogenic transgene, and use of second- or third-generation Ad vector backbones can in theory provide an Ad vector that has significant and greatly improved stealth characteristics. Alternatively as we learn more about how Ad vectors are recognized by cells of the immune system, we will be in a position to design vaccine or oncolytic vectors that have a significant enhancement in potency with diminished unwanted bystander activity.
ACKNOWLEDGMENTS
This work was supported by NIH grants AI-63142 to E.F.-P. and KO1 HL70438-01 to M.N. GenVec has licensing agreements with Cornell Research Foundation for technology developed in collaboration with E. Falck-Pedersen.
REFERENCES
Bergelson, J. M., J. A. Cunningham, G. Droguett, E. A. Kurt-Jones, A. Krithivas, J. S. Hong, M. S. Horwitz, R. L. Crowell, and R. W. Finberg. 1997. Isolation of a common receptor for Coxsackie B viruses and adenoviruses 2 and 5. Science 275:1320-1323.
Bewley, M. C., K. Springer, Y. B. Zhang, P. Freimuth, and J. M. Flanagan. 1999. Structural analysis of the mechanism of adenovirus binding to its human cellular receptor, CAR. Science 286:1579-1583.
Dechecchi, M. C., P. Melotti, A. Bonizzato, M. Santacatterina, M. Chilosi, and G. Cabrini. 2001. Heparan sulfate glycosaminoglycans are receptors sufficient to mediate the initial binding of adenovirus types 2 and 5. J. Virol. 75:8772-8780.
Einfeld, D. A., R. Schroeder, P. W. Roelvink, A. Lizonova, C. R. King, I. Kovesdi, and T. J. Wickham. 2001. Reducing the native tropism of adenovirus vectors requires removal of both CAR and integrin interactions. J. Virol. 75:11284-11291.
Gaggar, A., D. M. Shayakhmetov, and A. Lieber. 2003. CD46 is a cellular receptor for group B adenoviruses. Nat. Med. 9:1408-1412.
Gall, J., A. Kass-Eisler, L. Leinwand, and E. Falck-Pedersen. 1996. Adenovirus type 5 and 7 capsid chimera: fiber replacement alters receptor tropism without affecting primary immune neutralization epitopes. J. Virol. 70:2116-2123.
Gall, J. G., R. G. Crystal, and E. Falck-Pedersen. 1998. Construction and characterization of hexon-chimeric adenoviruses: specification of adenovirus serotype. J. Virol. 72:10260-10264.
Huang, S., T. Kamata, Y. Takada, Z. M. Ruggeri, and G. R. Nemerow. 1996. Adenovirus interaction with distinct integrins mediates separate events in cell entry and gene delivery to hematopoietic cells. J. Virol. 70:4502-4508.
Kass-Eisler, A., L. Leinwand, J. Gall, B. Bloom, and E. Falck-Pedersen. 1996. Circumventing the immune response to adenovirus-mediated gene therapy. Gene Ther. 3:154-162.
Kidd, A. H., J. Chroboczek, S. Cusack, and R. W. Ruigrok. 1993. Adenovirus type 40 virions contain two distinct fibers. Virology 192:73-84.
Leissner, P., V. Legrand, Y. Schlesinger, D. A. Hadji, M. van Raaij, S. Cusack, A. Pavirani, and M. Mehtali. 2001. Influence of adenoviral fiber mutations on viral encapsidation, infectivity and in vivo tropism. Gene Ther. 8:49-57.
Lieber, A., C. Y. He, L. Meuse, D. Schowalter, I. Kirillova, B. Winther, and M. A. Kay. 1997. The role of Kupffer cell activation and viral gene expression in early liver toxicity after infusion of recombinant adenovirus vectors. J. Virol. 71:8798-8807.
Mizuguchi, H., N. Koizumi, T. Hosono, A. Ishii-Watabe, E. Uchida, N. Utoguchi, Y. Watanabe, and T. Hayakawa. 2002. CAR- or v integrin-binding ablated adenovirus vectors, but not fiber-modified vectors containing RGD peptide, do not change the systemic gene transfer properties in mice. Gene Ther. 9:769-776.
Morelli, A. E., A. T. Larregina, R. W. Ganster, A. F. Zahorchak, J. M. Plowey, T. Takayama, A. J. Logar, P. D. Robbins, L. D. Falo, and A. W. Thomson. 2000. Recombinant adenovirus induces maturation of dendritic cells via an NF-B-dependent pathway. J. Virol. 74:9617-9628.
Nakamura, T., K. Sato, and H. Hamada. 2003. Reduction of natural adenovirus tropism to the liver by both ablation of fiber-coxsackievirus and adenovirus receptor interaction and use of replaceable short fiber. J. Virol. 77:2512-2521.
Nicol, C. G., D. Graham, W. H. Miller, S. J. White, T. A. Smith, S. A. Nicklin, S. C. Stevenson, and A. H. Baker. 2004. Effect of adenovirus serotype 5 fiber and penton modifications on in vivo tropism in rats. Mol. Ther. 10:344-354.
Philpott, N. J., M. Nociari, K. B. Elkon, and E. Falck-Pedersen. 2004. Adenovirus-induced maturation of dendritic cells through a PI3 kinase-mediated TNF-alpha induction pathway. Proc. Natl. Acad. Sci. USA 101:6200-6205.
Rea, D., F. H. Schagen, R. C. Hoeben, M. Mehtali, M. J. Havenga, R. E. Toes, C. J. Melief, and R. Offringa. 1999. Adenoviruses activate human dendritic cells without polarization toward a T-helper type 1-inducing subset. J. Virol. 73:10245-10253.
Roelvink, P. W., A. Lizonova, J. G. Lee, Y. Li, J. M. Bergelson, R. W. Finberg, D. E. Brough, I. Kovesdi, and T. J. Wickham. 1998. The coxsackievirus-adenovirus receptor protein can function as a cellular attachment protein for adenovirus serotypes from subgroups A, C, D, E, and F. J. Virol. 72:7909-7915.
Schoggins, J. W., J. G. Gall, and E. Falck-Pedersen. 2003. Subgroup B and F fiber chimeras eliminate normal adenovirus type 5 vector transduction in vitro and in vivo. J. Virol. 77:1039-1048.
Segerman, A., J. P. Atkinson, M. Marttila, V. Dennerquist, G. Wadell, and N. Arnberg. 2003. Adenovirus type 11 uses CD46 as a cellular receptor. J. Virol. 77:9183-9191.
Shayakhmetov, D. M., Z. Y. Li, S. Ni, and A. Lieber. 2004. Analysis of adenovirus sequestration in the liver, transduction of hepatic cells, and innate toxicity after injection of fiber-modified vectors. J. Virol. 78:5368-5381.
Sirena, D., B. Lilienfeld, M. Eisenhut, S. Kalin, K. Boucke, R. R. Beerli, L. Vogt, C. Ruedl, M. F. Bachmann, U. F. Greber, and S. Hemmi. 2004. The human membrane cofactor CD46 is a receptor for species B adenovirus serotype 3. J. Virol. 78:4454-4462.
Smith, T., N. Idamakanti, H. Kylefjord, M. Rollence, L. King, M. Kaloss, M. Kaleko, and S. C. Stevenson. 2002. In vivo hepatic adenoviral gene delivery occurs independently of the coxsackievirus-adenovirus receptor. Mol. Ther. 5:770-779.
Smith, T. A., N. Idamakanti, J. Marshall-Neff, M. L. Rollence, P. Wright, M. Kaloss, L. King, C. Mech, L. Dinges, W. O. Iverson, A. D. Sherer, J. E. Markovits, R. M. Lyons, M. Kaleko, and S. C. Stevenson. 2003. Receptor interactions involved in adenoviral-mediated gene delivery after systemic administration in non-human primates. Hum. Gene Ther. 14:1595-1604.
Smith, T. A., N. Idamakanti, M. L. Rollence, J. Marshall-Neff, J. Kim, K. Mulgrew, G. R. Nemerow, M. Kaleko, and S. C. Stevenson. 2003. Adenovirus serotype 5 fiber shaft influences in vivo gene transfer in mice. Hum. Gene Ther. 14:777-787.
Tomko, R. P., C. B. Johansson, M. Totrov, R. Abagyan, J. Frisen, and L. Philipson. 2000. Expression of the adenovirus receptor and its interaction with the fiber knob. Exp. Cell Res. 255:47-55.
Tomko, R. P., R. Xu, and L. Philipson. 1997. HCAR and MCAR: the human and mouse cellular receptors for subgroup C adenoviruses and group B coxsackieviruses. Proc. Natl. Acad. Sci. USA 94:3352-3356.
Trevejo, J. M., M. W. Marino, N. Philpott, R. Josien, E. C. Richards, K. B. Elkon, and E. Falck-Pedersen. 2001. TNF-alpha-dependent maturation of local dendritic cells is critical for activating the adaptive immune response to virus infection. Proc. Natl. Acad. Sci. USA 98:12162-12167.
Wickham, T. J., P. Mathias, D. A. Cheresh, and G. R. Nemerow. 1993. Integrins alpha v beta 3 and alpha v beta 5 promote adenovirus internalization but not virus attachment. Cell 73:309-319.
Wolff, G., S. Worgall, N. van Rooijen, W. R. Song, B. G. Harvey, and R. G. Crystal. 1997. Enhancement of in vivo adenovirus-mediated gene transfer and expression by prior depletion of tissue macrophages in the target organ. J. Virol. 71:624-629.
Worgall, S., G. Wolff, E. Falck-Pedersen, and R. G. Crystal. 1997. Innate immune mechanisms dominate elimination of adenoviral vectors following in vivo administration. Hum. Gene Ther. 8:37-44.
Yang, Y., and J. M. Wilson. 1995. Clearance of adenovirus-infected hepatocytes by MHC class I-restricted CD4+ CTLs in vivo. J. Immunol. 155:2564-2570.
Zhang, Y., N. Chirmule, G. P. Gao, R. Qian, M. Croyle, B. Joshi, J. Tazelaar, and J. M. Wilson. 2001. Acute cytokine response to systemic adenoviral vectors in mice is mediated by dendritic cells and macrophages. Mol. Ther. 3:697-707.(John W. Schoggins, Marcel)