Fiber Shaft-Chimeric Adenovirus Vectors Lacking the KKTK Motif Efficiently Infect Liver Cells In Vivo

Abstract

The molecular mechanisms governing the infectivity of adenovirus (Ad) toward specific cell and tissue types in vivo remain poorly understood. The direct Ad binding to hepatic heparan sulfate proteoglycans via the KKTK motif within the fiber shaft domain was suggested to be the major mechanism of Ad liver cell infection in vivo. Here, we describe the generation and in vitro and in vivo infectivity studies of Ad5-based vectors possessing long Ad31- or Ad41-derived fiber shaft domains, which lack the KKTK motif. We found that all the critical early steps of Ad infection, including attachment to the cellular receptor, internalization, and virus genome transfer into the nucleus, occurred with similar levels of efficiency for fiber shaft-chimeric vectors and unmodified Ad5. Upon intravenous delivery into mice, fiber shaft-chimeric vectors accumulated in liver tissue, transduced liver cells, and induced the production of proinflammatory cytokines (tumor necrosis factor alpha and interleukin-6) and the chemokine monocyte chemoattractant protein 1 at levels indistinguishable from those observed for Ad5. Thus, our data provide evidence that the Ad5 fiber shaft amino acid sequence does not play any substantial role in determining adenovirus infectivity toward hepatic cells in vivo. The data obtained contribute to improving our understanding of the molecular mechanisms determining Ad infectivity and biodistribution in vivo and may aid in designing novel Ad-based vectors for gene therapy applications.

Adenovirus (Ad) is a nonenveloped virus with a linear, double-stranded DNA genome. In humans, multiple serotypes of Ad cause a spectrum of illnesses in both normal and immunocompromised hosts (17, 35). There are 51 human Ad serotypes identified to date, which are divided into six species, A to F. Many of these serotypes exhibit diverse cell and tissue tropisms, which are, at least in part, responsible for the variety of observed clinical manifestations caused by Ad infection.

From in vitro studies, it was found that Ad infection starts with virus binding to a high-affinity primary attachment receptor on the cell surface. This initial interaction is mediated by the fiber protein when its distal knob domain binds to a specific cellular receptor. For binding to cells, species A, C, D, E, and F Ads may utilize the coxsackievirus and Ad receptor (CAR); however, the majority of species B Ads utilize CD46 as a high-affinity cellular attachment receptor (4, 5, 9, 33, 34, 37). Fiber-mediated binding of Ad to cells is followed by binding of the viral penton base protein to integrins (50). This interaction induces integrin activation, cytoskeleton rearrangement, and the internalization of the virus particle into the cell (12, 15, 20, 21, 49). Although the early steps of Ad infection in vitro are known in great detail (reviewed in references 10, 23, 24, and 26), there is still minimal understanding of the mechanisms responsible for the efficient natural Ad infection of distal sites and tissues (such as bladder epithelial cells) in vivo. Additionally, the well-accepted two-step model of Ad cell infection fails to explain the infectivity and biodistribution of Ad-based vectors following their intravascular delivery. Specifically, in all the animal models analyzed, including mice, rats, rabbits, dogs, and nonhuman primates, the liver is the predominant organ transduced by Ad5-based vectors (6, 16, 22, 29, 30, 44). Moreover, when non-CAR-binding Ad5 vectors are injected intravenously, they transduce liver cells at levels largely indistinguishable from those observed for CAR-interacting Ads (1, 44).

In recent years, significant attention has been given to the analysis of properties of an Ad5-based vector, AdS*, which has the KKTK amino acid motif within the fiber shaft domain (amino acids 91 to 94) replaced by nonhomologous amino acids GAGA (3, 18, 45, 46). When given intravenously, the AdS* vector demonstrates a remarkable reduction in liver cell transduction in mice, rats, and nonhuman primates compared to Ad5. Based on this data, it was postulated that the KKTK motif is a specific site in the Ad fiber shaft domain responsible for direct Ad attachment to hepatic heparan sulfate proteoglycans (HSPGs), thus enabling efficient Ad liver cell transduction in vivo. Although the idea of the involvement of the KKTK motif in liver infection has been widely discussed (19, 28), there is still no clear evidence that the KKTK motif binds to hepatic HSPGs and mediates liver cell transduction. Moreover, in vitro studies revealed that AdS* is very inefficient at infecting cells expressing high levels of CAR (3, 18, 45, 46). Indeed, to generate sufficient numbers of AdS* particles for in vivo analyses, the virus is propagated on cell lines which provide the fiber protein in trans to maintain virus infectivity (3). Recently, two independent groups demonstrated that AdS* cannot be used for Ad retargeting, mostly due to its inability to transduce susceptible cells (3, 18). To analyze whether the direct binding of the KKTK motif to hepatic HSPGs is the primary mechanism of liver transduction by Ad5-based vectors, in vivo studies of vectors with a mutated KKTK motif would appear to be the most informative. However, due to their attenuated cell transduction phenotype, AdS* vectors cannot be used directly for these in vivo studies. Therefore, indirect alternate strategies should be applied to analyze the contribution of the fiber shaft KKTK motif to efficient Ad transduction of hepatic cells in vivo.

Here, we describe an alternate strategy to evaluate the role of the KKTK motif in liver cell transduction: the generation of fiber shaft-chimeric vectors lacking the KKTK motif. We generated viruses and conducted in vitro and in vivo infectivity studies of Ad5-based vectors possessing long Ad31- or Ad41-derived fiber shaft domains. The fiber knob domains of long-shafted Ad31 and Ad41 fibers can recognize CAR as a virus attachment receptor; however, the amino acid compositions of Ad31 and Ad41 fiber shaft domains are highly diverse and lack the KKTK motif. In contrast to AdS*, our fiber shaft-chimeric vectors were able to bind efficiently to and retain high-level infectivity toward susceptible cells in vitro. When fiber shaft-chimeric vectors were injected intravenously into mice, their accumulation in livers and the efficiency of liver cell infection were undistinguishable from those observed with unmodified Ad5. Our study demonstrates for the first time that the amino acid composition of the Ad5 fiber shaft (which is long) is not highly conserved and that this shaft can be replaced entirely with the fiber shafts from other long-shafted Ad serotypes without any reduction in overall virus infectivity. Our data also strongly suggest that the KKTK motif within the Ad5 fiber shaft plays only a minimal, if any, role in determining adenovirus infectivity toward hepatic cells in vivo. The obtained data contribute to improving our understanding of the molecular mechanisms determining Ad infectivity and biodistribution in vivo and may aid in designing efficient Ad-based vectors for application in humans.

MATERIALS AND METHODS

Cells and viruses.

HeLa (human cervix carcinoma; ATCC CCL-2.2), 293 (human embryonic kidney; Microbix, Toronto Canada), A549 (human lung carcinoma; ATCC CCL-185), and CHO-K1 (ATCC CCL-61) cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum, 2 mM l-glutamine, and 1× penicillin-streptomycin solution (Invitrogen, Carlsbad, CA). The virus Ad5L, expressing green fluorescent protein (GFP) under the control of the human cytomegalovirus immediate-early promoter and possessing an intact wild-type Ad5 capsid, was previously constructed and is described in detail as Ad5GFP in reference 43. The Ad5-based vectors with fiber shaft domains derived from either Ad31 or Ad41 (Ad31/5 and Ad41/5) were constructed using standard cloning techniques and DNA recombination in Escherichia coli strain BJ5183 (42). To generate chimeric fiber genes encoding the Ad5 tail and knob domains and the shaft domain derived from either Ad31 or long Ad41, we utilized a PCR-based site-directed mutagenesis strategy described previously (42). To obtain sequences corresponding to Ad31 and Ad41 fiber shafts, wild-type Ad31 (American Type Culture Collection stock no. VR-1109) and Ad41 (American Type Culture Collection stock no. VR-930) viruses were grown in 293 cells and purified by centrifugation in CsCl gradients (25). Virus DNA was isolated from purified particles, and sequences corresponding to the fiber shaft domains were amplified using high-fidelity Pfu Ultra DNA polymerase (Stratagene). Following PCR-directed linking of DNA sequences corresponding to the fiber shaft domains and the Ad5 fiber tail and knob domains, the resultant chimeric fiber genes were cloned into a shuttle plasmid for introduction into Ad5 genomic DNA (see Fig. 1A).

FIG. 1.

(A) Schematic representation of fiber structures of fiber shaft-chimeric and unmodified vectors. (B) Amino acid sequence alignment of Ad5, Ad31, and Ad41 fibers. The sequence alignment was done using MultAlin software (http://bioinfo.genopole-toulouse.prd.fr/multalin). The high-level-consensus amino acids are shown in red, low-level-consensus amino acids are shown in blue, and missing amino acids are represented by dashes. The position of the KKTK motif is indicated by a rectangle. The position of the flexible hinge domain in the Ad5 fiber shaft domain is indicated by an arrow. (C) Quantitative Southern blotting of viral genomic DNA from the indicated vectors (upper panel) was conducted as described in Materials and Methods. The position of Ad-specific bands is indicated by the arrowhead. The lower panel shows an ethidium bromide-stained agarose gel with Ad DNA prior to transfer onto a Hybond-N+ membrane.

All viruses were amplified in 293 cells under conditions preventing cross contamination. Viruses were banded in CsCl gradients; viral bands were collected, dialyzed, and aliquoted as described elsewhere (25). Ad genome concentrations were determined using both quantitative Southern blotting (see Fig. 1C) and assays of plaque formation on 293 cells. For each Ad vector used in this study, at least two independent virus stocks were produced and characterized by determining PFU titers on 293 cells and genome titers by Southern blotting. Each produced virus stock was tested for endotoxin contamination using Limulus amebocyte lysate Pyrotell (Cape Cod Inc., Falmouth, MA). For in vivo experiments, only virus preparations confirmed to be free of endotoxin contamination were used.

Analysis of Ad attachment to cells.

Studies of Ad attachment to cells were based on a protocol published elsewhere (42). For these studies, 3.5 × 10⁵ A549 cells were incubated for 1 h on ice with equal amounts of Ad particles, corresponding to a multiplicity of infection (MOI) of 8,000 virus particles per cell, in 100 μl of ice-cold adhesion buffer (DMEM supplemented with 2 mM MgCl₂, 1% bovine serum albumin, and 20 mM HEPES). Cells were then pelleted by centrifugation at 1,000 × g for 4 min and washed two times with 0.5 ml of ice-cold phosphate-buffered saline (PBS). After the last washing step, the cells were pelleted at 1,500 × g, the supernatant was removed, cells were lysed, and total cellular DNA was extracted as described previously (42). To ensure comparable viral loads (see Fig. 2A), following virus addition, cells were immediately lysed and cellular DNA was extracted for subsequent Southern blot analysis as described above.

FIG. 2.

(A) Comparative analysis of Ad vector attachment to A549 cells. Comparable amounts of the indicated Ad vectors were mixed with A549 cells, and one portion of cells was lysed immediately (lanes labeled “Load”), while the other cells were processed to determine the levels of Ad attachment by Southern blotting. The upper panel shows the ethidium bromide-stained agarose gel before Southern blotting to demonstrate the presence of comparable total DNA loads. The membrane was hybridized with an Ad genome-specific ³²P-labeled probe (the HindIII-A fragment of the Ad5 genome), and vector DNA was visualized by autoradiography. Corresponding vector DNA bands are indicated by an arrow. (B) Quantitative representation of Ad attachment to A549 cells determined by phosphorimager analysis of the Ad-specific bands shown in panel A after the adjustment of attachment signal intensities for the load signal intensities for corresponding viruses. (C) Effect of the fiber shaft domain substitution on the rates of internalization of Ad vectors. Virus vectors were allowed to attach to A549 cells on ice. Infection was initiated by cell transfer to 37°C. At the indicated time points, an anti-Ad5 neutralizing antibody was added to cells to prevent further internalization of infectious particles. The resultant levels of gene transfer (as measured by the mean GFP fluorescence intensity) were assessed by flow cytometry 24 h postinfection. The level of infection with each virus when no virus-neutralizing antibody was added (control settings) was taken as 100% (maximum [max]). All infections were done in triplicate, and the data presented are representative of results from two independent experiments. (D to F) Efficiencies of transduction of different cell lines with Ad vectors possessing wild-type and chimeric shaft domain fibers. 293, A549, and HeLa cells were infected at the indicated MOIs (virus particles per cell), and the mean GFP fluorescence intensity was analyzed 24 h later by flow cytometry. The data shown are the averages of results from four experiments. MFI, mean fluorescence intensity.

Ad internalization rate analysis.

Briefly, 2 × 10⁵ A549 cells were resuspended in 150 μl of adhesion buffer (DMEM supplemented with 2 mM MgCl₂, 1% bovine serum albumin, and 20 mM HEPES) and chilled for 25 min on ice. Then cells were further incubated on ice for 1 h with equal amounts of Ad particles at an MOI of 1,000 virus particles per cell. Next, unattached virus particles were removed by washing twice with 200 μl of ice-cold PBS. Following the last wash, cells were resuspended in PBS and transferred to a 37°C water bath to initiate virus internalization. At the indicated time points (see Fig. 2C, x axis), 20 μl of polyclonal anti-Ad5 antibody was added to the cells to neutralize noninternalized virus particles (38). In the control settings, no anti-Ad antibody was added to cells in order to determine the maximum level of cell transduction for each virus (taken as 100%). To complete the neutralization of noninternalized vector particles, after the addition of anti-Ad5 antibody, cells were further incubated on ice for 30 min. They were then washed twice with PBS and plated into 1 ml of growth medium in 12-well plates, and the plates were incubated overnight at 37°C. Both the percentage of cells expressing virus-encoded GFP and the mean GFP fluorescence intensity were determined by flow cytometry.

Adenovirus infection in vivo.

All experimental procedures involving animals were conducted in accordance with the institutional guidelines set forth by the University of Washington. C57BL/6 mice (Charles River, Wilmington, MA) were housed in specific-pathogen-free facilities. For the analysis of Ad-mediated gene transfer into liver cells, 10¹¹ Ad particles (corresponding to 2 × 10⁹ PFU of Ad5L vector as determined on 293 cells) in 200 μl of PBS were injected by tail vein infusion. For in vivo transduction studies, mice were sacrificed 1 or 24 h post-virus infusion and livers were processed for histological analyses. For the analysis of Ad genome accumulation in the liver tissue 1 and 24 h after Ad vector administration into the tail vein, the blood was flushed from the liver by cardiac saline perfusion, livers were harvested, and total DNA was purified as described previously (40).

Analysis of adenovirus-Kupffer cell interaction in vivo.

To analyze Ad interactions with Kupffer cells, 10¹¹ virus particles were injected into the tail vein, and 60 min later, livers were flushed with saline via cardiac perfusion, harvested, and immediately frozen in optimal cutting temperature compound. Frozen liver sections were fixed and stained with rat anti-mouse F4/80 primary antibody (BD Biosciences) to detect Kupffer cells. Specific binding of primary antibodies was visualized with secondary anti-rat Alexa Flour 488 antibody (green; Molecular Probes Inc., Eugene, OR). To detect Ad particles, liver sections were stained with antihexon polyclonal antibody (Chemicon). The staining with antihexon primary antibody was developed with secondary phycoerythrin-labeled antibody, which appears red.

Southern blot analyses.

For the analysis of virus genome titers by quantitative Southern blotting, DNA was extracted from purified viral particles and run on agarose gels in serial (twofold) dilutions together with standard DNA of known concentrations (preparatively purified Ad5L DNA). After transfer onto Hybond-N+ nylon membranes (Amersham, Piscataway, NJ), the samples were hybridized with a ³²P-labeled Ad-specific DNA probe (an 8-kb HindIII-A fragment corresponding to the E2 region of the Ad genome) and DNA concentrations were measured by a phosphorimager. These values were used to calculate the genome titer for each virus stock used.

For the analysis of Ad genomic DNA deposition and persistence in mouse livers, the isolation of total liver DNA and Southern analysis were performed as described elsewhere (40). A ³²P-labeled 8-kb HindIII-A fragment corresponding to the E2 region of the adenovirus genome was used for hybridization to specifically detect Ad genomic DNA in livers.

Analysis of cytokine and chemokine levels in mouse plasma.

To analyze levels of proinflammatory cytokines and chemokines in plasma 5 h after the intravenous administration of Ad vectors, blood samples were collected into heparin-treated Eppendorf tubes, cells were pelleted for 5 min at 1,000 × g, and plasma was obtained and stored at −80°C in small aliquots. To analyze the levels of cytokines and chemokines in plasma, a mouse inflammatory cytometric bead array (BD Biosciences, Palo Alto, CA) was used. Briefly, 10 μl of mouse plasma was diluted five times and mixed with cytometric beads capable of binding mouse tumor necrosis factor alpha (TNF-α), interleukin-6 (IL-6), monocyte chemoattractant protein 1 (MCP-1), gamma interferon, IL-12p70, and IL-10. The binding of these proteins was detected with corresponding secondary phycoerythrin-conjugated antibodies and analyzed by flow cytometry along with provided standard proteins. The collected data were processed using the bead array manufacturer's software. Plasma samples obtained from at least three mice (for each Ad vector) were analyzed in duplicate.

Nucleotide sequence accession number.

The sequence of the Ad31 fiber gene determined in this study has been submitted to GenBank under accession number EU029805.

RESULTS

Generation of fiber shaft domain-chimeric adenovirus vectors.

Previous studies have demonstrated that the mutation of the KKTK motif within the human Ad5 fiber shaft domain results in a dramatic reduction of virus infectivity toward cells efficiently expressing the high-affinity primary attachment receptor, CAR (3, 18). In an attempt to modify the fiber shaft domain while preserving Ad infectivity toward susceptible cells, we generated Ad5-based vectors possessing long fiber shaft domains derived from either the Ad31 or Ad41 serotype (Fig. 1A). Ad31 and Ad41 belong to human adenovirus species A and F, respectively, and while the lengths of their fibers are similar to that of the Ad5 fiber, the amino acid compositions of their fiber shaft domains are highly diverse (Fig. 1B). Specifically, the KKTK motif is absent in the fiber shaft domains of both Ad31 and Ad41; therefore, we hypothesized that by analyzing the properties of fiber shaft-chimeric vectors, the role of the KKTK motif in observed virus infectivity in vitro and in vivo could be evaluated. It is also noteworthy that the amino acids comprising the putative flexible fiber hinge domain, which is present in Ad5 fiber but missing in more rigid fibers of short-shafted Ad serotypes (50), are present in both long Ad31 and Ad41 fiber shafts (Fig. 1B). Using PCR-directed mutagenesis, we constructed fiber genes encoding chimeric fiber proteins which possessed the N-terminal Ad5 fiber tail domain, the Ad31 or Ad41 fiber shaft domain, and the C-terminal fiber knob domain derived from Ad5 (Fig. 1A). Sequence analysis of chimeric fiber genes as well as genomic DNA obtained from purified wild-type viruses revealed significant deviation in the nucleotide sequence of the Ad31 fiber compared to that published earlier (32). The alignment of our Ad31 fiber sequence with other available fiber sequences of species A serotypes revealed higher levels of similarity to our newly determined sequence than to the fiber sequence published earlier (32). Therefore, for our fiber-chimeric virus constructs, we selected the newly obtained Ad31 fiber sequence as a reference for further use. The sequence analysis was done on both strands of eight individual plasmid clones containing the Ad31 fiber sequence, and all the sequences obtained from plasmid DNA clones matched the fiber sequence obtained directly from Ad31 genomic DNA. All vectors generated possessed identical GFP gene expression cassettes, which allows for the direct comparison of levels of virus infectivity based on the analysis of GFP gene expression in infected cells. Following virus recovery from a single plaque obtained on 293 cells, virus vectors were grown to high titers and purified using a standard cesium chloride gradient ultracentrifugation technique (25). To obtain genomic titers of fiber shaft-chimeric vectors, we utilized previously optimized quantitative Southern blotting (Fig. 1C). The quantitative analysis of genomic titers demonstrated that all viruses could be amplified to very high titers, ranging from 6.7 × 10¹² virus particles/ml for Ad31/5 to 1.1 × 10¹³ virus particles/ml for Ad5L. The analysis of virus titers demonstrated that the virus yield per cell and genome-to-PFU ratio were comparable for all vectors (Table 1).

TABLE 1.

Virus titers and genome-to-PFU ratios for Ad5L, Ad31/5, and Ad41/5 fiber shaft chimeric-vectors

Virus	Yield of PFU/cell	Virus titer		Genome/PFU ratio
		PFU/mla	Genomes/mlb
Ad5L	98	1.95 × 1011	1.1 × 1013	56.4
Ad31/5	35	7 × 1010	6.72 × 1012	96
Ad41/5	70	1.4 × 1011	7.68 × 1012	54.8

^aThe plaque titer was determined on 293 cells as described previously (42).

^bThe genome titer in preparations of purified Ads was measured by quantitative Southern blotting.

Fiber shaft-chimeric Ads demonstrate efficient attachment to cells, internalization kinetics, and infectivity in vitro.

To evaluate if modification of the fiber shaft domain negatively affected early steps of virus infection, we analyzed the levels of virus attachment and the rate of internalization using A549 cells. Comparable amounts of adenovirus particles for each vector were added in duplicate to A549 cells on ice (42). One set of samples for each vector was immediately lysed (loading control), while the viruses in the second set of samples were allowed to attach to cells for 1 h on ice. Following virus attachment, cells were washed to remove unbound virus particles. Then cells were lysed, total DNA was isolated, and the amounts of virus DNA associated with cells were analyzed by quantitative Southern blotting using an Ad-specific probe. After hybridization, the membrane was exposed to either X-ray film (Fig. 2A) or a phosphorimager screen to quantify the Ad-specific signal on the membrane (Fig. 2B). This analysis showed that all generated Ad vectors attached to cells with similar efficiencies (Fig. 2B), suggesting that the substitution of the entire Ad5 fiber shaft domain with the long shaft from Ad31 or Ad41 did not result in a reduction of the viruses' ability to interact with the primary high-affinity virus attachment receptor.

Since virus binding to the cell represents only the first step of infection, we next analyzed whether fiber shaft substitution could affect the efficiency of virus internalization into cells. Virus particles for all generated vectors were allowed to attach to the cell surface on ice, and after the removal of unbound particles by washing, samples were transferred into a 37°C water bath to initiate internalization. At the time points indicated in Fig. 2C (x axis), polyclonal virus-neutralizing serum was added to the samples to inactivate noninternalized particles. Next, cells were washed free of Ad-specific antibody, resuspended in growth medium, and incubated for 24 h, when virus-encoded GFP expression was quantified by flow cytometry. In control settings, virus-neutralizing antibody was added to cells without incubation at 37°C (0-min time point). The addition of serum to cell-bound virus completely prevented virus infection, observed as a lack of GFP gene expression following cell incubation in growth medium for 24 h (Fig. 2C). To assess the maximal efficiency of virus infection, cells were incubated at 37°C for 30 min without the addition of the virus-neutralizing antibody. Using this technique, we previously demonstrated that the deletion of the penton base RGD motif reduces the rate of internalization of adenovirus particles into susceptible cells (38). However, by analyzing fiber shaft-chimeric vectors, we found that all of them escaped inactivation by virus-neutralizing antibody with similar kinetics (within 5 min following cell transfer to a physiological temperature) (Fig. 2C). These data demonstrate that the modification of the fiber shaft did not affect the viruses' ability to interact with cellular integrins and that, following attachment to cells, all generated vectors underwent internalization with similar kinetics.

Because recently obtained data suggest that Ad vector AdS*, possessing a substitution for the KKTK motif in the fiber shaft domain, exhibits deficiency in gene transfer in postattachment steps of infection (18), we next analyzed the efficiency of gene transfer for fiber shaft-chimeric viruses on 293 (Fig. 2D), A549 (Fig. 2E), and HeLa (Fig. 2F) cells, which were previously used in studies of the potential role of the KKTK motif in Ad infectivity (7, 8, 46). The indicated cell lines were infected with fiber shaft-chimeric vectors at the MOIs indicated in Fig. 2D to F, and the efficiency of gene transfer was assessed by analyzing GFP expression levels using flow cytometry 24 h post-virus infection. This examination revealed that all the generated vectors infected all analyzed cell lines with very high efficiency, independently of the nature of the fiber shaft domain.

Taken together, our data demonstrate that the substitution of the entire Ad5 fiber shaft domain with a corresponding shaft domain from either Ad31 or Ad41 in a virus particle did not result in any detectable deficiency in virus attachment, internalization, or overall gene transfer into susceptible cells in vitro. These data also suggest that while being highly diverse at the amino acid level (Fig. 1B), Ad31 and Ad41 fiber shaft domains can be joined with Ad5 fiber tail and heterologous Ad5 knob domains to form a functional fiber protein.

Coagulation FIX and FX improve infectivity of fiber shaft-chimeric Ads in vitro.

Recently, a novel blood factor-dependent pathway of Ad infection of liver cells in vivo was described (2, 31, 39). This pathway relies, at least in part, on Ad interaction with circulating vitamin K-dependent coagulation factor zymogens which form a bridge between virus particles and HSPGs or low-density lipoprotein receptor-related protein, thus allowing for efficient virus entry into hepatocytes and Kupffer cells in a CAR-independent manner. To evaluate whether the modification of the Ad fiber shaft domain affected the ability of the virus to utilize this pathway of cell entry, we infected CHO-K1 cells with fiber shaft-chimeric vectors in the presence of human coagulation factor IX (FIX) or factor X (FX) (39). CHO-K1 cells are resistant to Ad infection due to their low-level expression of CAR. However, these cells express high levels of HSPGs and can be efficiently infected with Ad in a CAR-independent manner in the presence of vitamin K-dependent coagulation factor zymogens (39). In agreement with the results of earlier studies, when CHO-K1 cells were infected with our fiber shaft-chimeric Ads or corresponding control vector possessing the Ad5 fiber shaft domain, we found that only a small fraction of cells (less than 5%) expressed GFP (Fig. 3A). However, when virus-containing medium was supplemented with 5 (data not shown) or 8 μg of FIX or FX/ml (the physiological concentration of these factors in blood), the levels of infection increased by more than 10-fold, reaching the highest levels with Ad5L and Ad31/5 vectors. Analysis of CHO-K1 cell transduction with fiber shaft-chimeric vectors revealed that the Ad41/5 virus, possessing the Ad41 fiber shaft domain, had a slightly reduced capacity to infect these cells both with and without coagulation factors. While the reason for this reduced infectivity toward CHO-K1 cells is unclear, when growth medium was supplemented with FIX or FX, the proportion of GFP-expressing CHO-K1 cells also increased by more than 10-fold. These data demonstrate that any generated vectors can utilize the blood factor-dependent pathway of cell infection. Importantly, the ability to use FIX or FX for cell entry was shown by analyzed Ads independently of the amino acid sequences of their fiber shaft domains.

FIG. 3.

(A) Transduction of HSPG-expressing CHO-K1 cells with Ad5L, Ad31/5, and Ad41/5 vectors with or without coagulation FIX or FX. Cells were infected at an MOI of 400 virus particles per cell with each indicated virus in the presence of 8 μg of FIX or FX/ml. Two hours after the addition of virus to cells, cells were washed with new medium, and GFP gene expression was analyzed 24 h later by flow cytometry. In control settings, viruses were added to cells in growth medium only. The data shown are the averages of results from four experiments. (B) Ad vectors were incubated with HEPES-buffered heparin (measured as units per milliliter) in PBS for 30 min at room temperature prior to addition to cells at an MOI of 400 virus particles per cell. Following 2 h of incubation, cells were washed twice with PBS and incubated in complete growth medium for 24 h prior to GFP gene expression analysis by flow cytometry. The data shown are the averages of results from four experiments.

Cell infection with fiber shaft-chimeric Ad cannot be blocked by heparin.

The proposed mechanism of Ad entry into cells through direct interaction of the fiber shaft KKTK motif with HSPGs was based, in part, on the observation that cell infection by Ad2 and Ad5, but not Ad3, can be blocked after the preincubation of purified virus particles with heparin (7, 8). In an attempt to evaluate whether the replacement of the Ad5 fiber shaft domain, which possesses the KKTK motif, with a fiber shaft domain derived from Ad31 or Ad41, which lacks it, may have reduced virus sensitivity to the inhibitory activity of heparin, we infected HeLa cells with the chimeric Ads after their preincubation with heparin. Heparin is a negatively charged, highly acidic compound, and Ad particles are known to be unstable at low pHs. A low pH may be one of the stimuli that mimic the intraendosomal environment where the early steps of virus capsid disassembly occur following internalization into the cell (13). To compensate for the potential negative effect of low pH on Ad particles at high concentrations of heparin, we supplemented virus-containing medium with 20 mM HEPES buffer (pH 7.2). Following the preincubation of viruses with HEPES-buffered heparin at increasing concentrations and subsequent cell infection, we found no inhibitory effect of heparin on virus infectivity for all analyzed vectors (Fig. 3B). We also were unable to block the infectivity of any of the analyzed vectors with heparin in PBS without the addition of HEPES (data not shown). These data suggest that the heparin is an inefficient inhibitor of infection for both Ad5L and fiber shaft-chimeric vectors.

Fiber shaft-chimeric Ads are trapped by Kupffer cells and transduce hepatocytes with similar efficiencies.

Although clear evidence for the direct interaction of the fiber shaft KKTK motif with cellular HSPGs in vitro is still not available, several independent studies demonstrated that Ads with mutated KKTK motifs in the fiber shaft domain exhibit a dramatic reduction in liver cell transduction following intravenous delivery, both in rodents and in nonhuman primates (45, 46). To analyze whether fiber shaft-chimeric Ads, which lacked the KKTK motif in their fiber shaft domains, demonstrated reduced capacity to infect liver cells in vivo, we administered 10¹¹ particles of all vectors intravenously to C57BL/6 mice.

Qualitative assessment of the distribution of fiber shaft-chimeric Ads in liver tissue using immunohistochemical analysis revealed that at 1 h post-intravenous virus delivery, all Ads formed large, bright aggregates that were evenly distributed throughout the liver parenchyma (Fig. 4A, Ad hexon panels). The staining of consecutive liver sections with Kupffer cell-specific anti-F4/80 antibody demonstrated that Ad-specific staining overlapped with F4/80-positive staining (Fig. 4A), suggesting that Kupffer cells efficiently trapped all analyzed vectors independently of the amino acid sequences of their fiber shaft domains. Histological evaluation of liver sections 24 h after intravenous virus delivery also revealed that virus-encoded GFP expression was readily detectable in the livers of mice following the administration of all vectors (Fig. 4B).

FIG. 4.

Interactions of fiber shaft-chimeric Ads with liver cells in vivo. (A) Association of Ad vectors with Kupffer cells. Virus (10¹¹ Ad virus particles) or saline (control) was injected into the tail veins of C57BL/6 mice. One hour later, livers were collected, perfused with PBS, and immediately frozen in optimal cutting temperature compound. To visualize Kupffer cells, fixed liver sections were stained with anti-F4/80 antibody (green). To visualize Ad, liver sections were stained with anti-Ad5 hexon antibody (red). Images of representative fields were taken with red and green filters and were then superimposed to reveal Ad association with Kupffer cells (yellow; indicated by arrows). (B) In vivo transduction of hepatocytes with fiber shaft-chimeric Ads. Twenty-four hours after intravenous Ad injection, livers were collected and serial sections of formalin-fixed tissues were prepared. To visualize GFP fluorescence, images of sections were taken under UV light. Representative fields are shown. Magnification, ×200. Neg., negative.

To analyze the interaction of fiber shaft-chimeric Ads with liver cells in a quantitative manner, we next utilized Southern blot and Western blot analyses to evaluate the levels of Ad genomes and GFP expression in the liver following Ad administration. To ensure that comparable amounts of Ads were delivered to mice for quantitative studies, mice were sacrificed 1 h post-virus delivery, liver tissue was flushed with saline to remove traces of blood, and total DNA was purified and analyzed by Southern blotting as described in Materials and Methods. This experiment revealed that following intravenous delivery, all Ads were trapped in the liver at similar levels (Fig. 5A, lanes labeled 1 h p.i.). Earlier studies have shown that the levels of Ad DNA associated with the liver at this early time point do not correlate with cell transduction or hepatotoxicity, while the levels of Ad DNA retained by the liver 24 to 72 h following virus delivery correlate well with hepatocyte transduction and virus toxicity (40). An analysis of Ad genome levels in the liver 24 h after Ad administration demonstrated no significant differences in the total amounts of vector DNA associated with the liver between the fiber shaft-chimeric vectors and the control Ad5L vector (Fig. 5A, lanes labeled 24 h p.i., and B). Quantitative evaluation of GFP expression by Western blotting further demonstrated that 24 h after intravascular delivery, the amounts of GFP detectable in the liver extracts were similar for all of the constructed fiber shaft-chimeric vectors, independently of the amino acid sequences of their fiber shaft domains (Fig. 5C and D).

FIG. 5.

Quantitative assessment of liver transduction with fiber shaft-chimeric vectors. (A) Southern blot analysis of Ad DNA in the liver at different time points after systemic vector application. At the indicated time points after intravenous Ad injection, livers were removed from mice and total DNA was purified. Ten micrograms of total DNA, digested with HindIII enzyme, was loaded onto the agarose gel. Following transfer to a Hybond-N+ membrane, DNA was hybridized with a mouse β-glucuronidase (GUS)-specific probe to confirm the presence of comparable loads. Subsequently, the membranes were stripped and rehybridized with an Ad-specific probe to visualize the HindIII-A fragment of the Ad5 genome (Ad DNA). Note that the absolute amount of Ad DNA significantly decreased over time (compare the intensities of Ad bands at 1 and 24 h post-virus injection). Control, DNA purified from livers of mice injected with PBS only. Lanes corresponding to 24 h post-virus injection (p.i.) show the results of Southern blot hybridization of DNA samples recovered from two individual mice injected with the indicated vectors. (B) Quantitative representation of Ad accumulation and persistence in livers as determined by phosphorimager analysis of the Ad-specific bands shown in panel A (lanes labeled 24 h p.i.) after the adjustment of Ad DNA signal intensities for the β-glucuronidase gene signal intensities for corresponding viruses. (C) Western blot analysis of GFP expression in the livers of mice 24 h after intravenous Ad delivery. Ten micrograms of total liver protein (recovered from two individual mice injected with the indicated vectors) was applied per lane, and following transfer onto nitrocellulose membrane, the specific GFP bands were developed with an anti-GFP monoclonal antibody (Sigma Co., St. Louis, MO). Control, sample of liver proteins from mice injected with PBS only. The membrane was also developed with an antiactin antibody as a protein loading control. (D) Quantitative representation of variations in GFP expression as determined by Western blotting after densitometric analysis of the GFP-specific bands shown in panel C. The error bars in panels B and D indicate standard deviations of the means.

Taken together, these data demonstrate that Ad5L, Ad31/5, and Ad41/5 were trapped in liver tissue and transduced hepatocytes with similar efficiencies.

Fiber shaft-chimeric Ads induce similar levels of acute systemic toxicity following intravenous delivery to mice.

The induction of an acute antiviral inflammatory response is one of the major consequences of Ad interactions with the host in vivo, and thus, by measuring the levels of inflammatory cytokines in plasma, a quantitative assessment of the interaction of different chimeric Ad vectors with the host can be done (27, 40). To evaluate whether the modification of the Ad fiber shaft domain could affect Ad-host interactions and modulate the systemic toxicity observed following intravenous Ad delivery, we compared the levels of proinflammatory cytokines TNF-α and IL-6 and chemokine MCP-1 in plasma. The levels of these cytokines and this chemokine in plasma were earlier found to be significantly up-regulated in animals following intravenous Ad administration (40, 41). Five hours post-intravenous Ad delivery into mice, plasma samples were collected and analyzed for levels of TNF-α, IL-6, and MCP-1 by using a cytometric bead array (41). This analysis revealed that following intravascular delivery, all analyzed vectors induced similarly elevated levels of TNF-α, IL-6, and MCP-1 in plasma (Fig. 6), suggesting that amino acids comprising the long Ad5 fiber shaft domain are unlikely to be involved in critical virus-host interactions which are responsible for the induction or modulation of the acute anti-Ad systemic inflammatory response.

FIG. 6.

Levels of proinflammatory cytokines and chemokines in plasma following intravenous administration of fiber shaft-chimeric Ad vectors. Six hours post-intravenous delivery of 10¹¹ virus particles of the indicated Ad vectors, plasma samples from three individual mice per virus group were collected and analyzed in duplicate for levels of TNF-α, IL-6, and MCP-1 by using cytometric bead array analysis as described previously (41). n.s., not statistically significant.

DISCUSSION

Unraveling of the molecular mechanisms governing the infectivity of a virus toward a particular cell or tissue type in vivo is crucial for understanding virus pathogenesis and the development of antiviral drugs, as well as novel virus-based vectors for application in humans to correct or ameliorate numerous genetic or acquired diseases. The inability to reconcile the molecular mechanisms of Ad infection known from in vitro studies and the Ad infectivity and biodistribution observed in vivo stimulated extensive research for novel pathways which may mediate cell type-specific Ad infectivity in vivo. The liver is the predominant organ transduced by Ad vectors following intravascular delivery into mice, rats, rabbits, dogs, and nonhuman primates (6, 16, 22, 29, 30, 44). Shortly after the discovery of Ad, Hartwell et al. reported that human Ad serotype 5 could be isolated at a high frequency (in 27 out of 30 evaluated cases) from the blood clots of patients with infectious hepatitis (14). In a later study of the epidemiology of simian immunodeficiency virus, Gravell et al. reported the isolation of Ad type 11 from plasma, urine, feces, and kidneys of monkeys with and without signs of simian acquired immune deficiency syndrome (11). There are accumulating reports of clinical cases of fatal disseminated Ad infection in pediatric patients that was manifested as Ad hepatitis in which levels of Ad genomes in plasma exceeded 10⁹ per ml (47). Collectively, these data suggest that in humans, even upon natural infection, at least species C and B Ads are capable of reaching circulating blood at certain stages of viral infection, in addition to the fact that human hepatic cells per se have all the necessary receptors and coreceptors which allow for the establishment of productive Ad infection in liver tissue. Despite advances in the understanding of Ad-host interactions on a cellular level, the specific pathways and molecular mechanisms enabling different Ad serotypes to reach particular cell and tissue types in vivo remain unclear.

After the identification of CAR as a high-affinity virus attachment receptor for the majority of human Ad serotypes, it was surprising to find that following intravenous delivery, non-CAR-binding Ads efficiently transduce liver cells and cause acute systemic toxicity at levels largely indistinguishable from those observed for CAR-interacting Ads (1, 44). This observation has led to the identification of the blood factor-dependent pathway of Ad liver cell infection (2, 31, 39). This pathway allows for CAR-independent Ad cell entry and relies on virus binding to blood factors, including vitamin K-dependent coagulation FIX and FX, which form a bridge between virus particles and hepatic HSPGs or low-density lipoprotein receptor-related protein (2). It is worth noting that when mouse liver was cleared of blood and then perfused with Ad5 vector-containing saline without any plasma proteins, subsequently purified hepatocytes expressed high levels of virus-carried transgene protein (39). However, when liver perfusion was done with saline containing the non-CAR-binding vector AdF* instead of Ad5, the hepatocyte transduction was markedly reduced, suggesting that Ad5 hepatocyte transduction in intact liver can also occur via the CAR-dependent pathway. This observation allows for speculation that under native conditions in the presence of blood, CAR-interacting Ads are likely to retain the capacity to infect hepatic cells via interaction with CAR.

In recent years, significant attention has been given to the analysis of properties of an Ad5-based vector, AdS*, possessing a replacement of the KKTK amino acid motif within the fiber shaft domain (amino acids 91 to 94) by nonhomologous amino acid motif GAGA (45, 46). When injected intravenously, AdS* vector demonstrates a remarkable reduction in liver cell transduction in mice, rats, and nonhuman primates compared to Ad5 (3, 18, 45, 46). Based on these data, together with the earlier finding that the preincubation of Ad particles with heparin can reduce the infectivity of Ad2 and Ad5, which possess the KKTK motif in their fiber shaft domains, but not that of Ad3, which lacks it (7, 8), it was postulated that the KKTK motif is a specific site in the Ad fiber shaft domain responsible for direct Ad attachment to hepatic HSPGs, thus enabling efficient Ad liver cell transduction in vivo. In addition to the dramatic reduction in liver transduction observed following the intravenous administration of AdS* compared to that observed after the administration of unmodified Ad5 vectors, in vitro studies revealed that AdS* is very inefficient at infecting cells expressing high levels of CAR (46). Moreover, when Arg-Gly-Asp (RGD) or QPEHSST peptides targeting cellular integrins or endothelial cell surface receptors, respectively, were introduced into AdS* fibers, they failed to improve virus infectivity by redirecting the virus through alternative cell entry pathways (3, 18). While these studies clearly demonstrate that Ads with a mutated KKTK motif cannot be used for cell-specific virus targeting, they provide no evidence regarding the involvement of the fiber shaft KKTK motif in mediating efficient Ad liver cell entry and further confirm earlier data on the attenuated nature of AdS* cell transduction both in vitro and in vivo.

Here, we describe the generation of fiber shaft-chimeric Ad5-based vectors possessing long shaft domains lacking the KKTK motif in their fibers, which were derived from either Ad31 or Ad41. In vitro studies demonstrated that despite the replacement of the entire Ad5 fiber shaft domain with heterologous sequences derived from Ad31 or Ad41, the fiber shaft-chimeric vectors were able to attach to cells, get internalized into cells, and transduce all tested cell lines with high levels of efficiency similar to those of unmodified Ad5 (Fig. 2). Our data are in contrast to the previous finding that the replacement of the KKTK motif within the Ad5 fiber shaft with GAGA results in a dramatic reduction in virus infectivity toward susceptible cell lines in vitro. While the specific reasons for the reduced in vitro infectivity of AdS* remain to be determined, it is clear that the mutation of the KKTK motif to GAGA (18), or just the change from KKTK to GATK (3), is sufficient to dramatically affect either the overall fiber structure or its stability, or both. In a recent study by Kritz et al., it was shown that despite efficient attachment to CAR, AdS* with a mutated KKTK motif demonstrates abnormal intracellular trafficking in vitro. Specifically, the virus does not accumulate at the nuclear membrane following efficient internalization into the cell (18). These data allow us to speculate that the KKTK fiber shaft motif is likely to be necessary for the efficient processing of one of the postinternalization steps of virus infection, which include endosome trafficking, endosome escape, and intracytoplasmic transport of internalized virus particles toward the nucleus. Clearly, while further studies are necessary to determine the exact role of the KKTK motif in Ad cell infection, it appears that the KKTK motif is more important for postinternalization steps of virus infection than for initial Ad cell binding. Although we cannot formally exclude that other areas of the long Ad31 and Ad41 shaft domains may compensate for the lack of the KKTK motif, the amino acid compositions of the fiber shafts from these serotypes are highly diverse and possess no obvious stretches of positively charged amino acids similar to the KKTK motif of Ad5 (Fig. 1B). Moreover, recent data obtained by Stone et al. demonstrated that intravenous injection of species B (Ad3, Ad11p, and Ad35), C (Ad5), E (Ad4), or F (Ad41) wild-type viruses into mice results in equally efficient accumulations of virus genomes in the livers shortly after virus administration (48). Because none of these viruses, except Ad5, possess the KKTK motif within their fiber shafts, these data may suggest that the KKTK fiber shaft motif unlikely mediates early Ad-liver cell interaction.

It is worth mentioning that the only other known example of deliberate Ad5 fiber shaft point mutation (the mutation of a putative flexible fiber hinge region) also resulted in a reduction of virus infectivity in vitro (51). These data suggest that within the Ad fiber, there is a functional interdependence between the shaft and receptor-recognizing knob domain that may play a crucial role in determining Ad infectivity. Indeed, in vivo analyses of all previously reported Ad fiber-chimeric vectors demonstrated that the infectivity of such vectors toward hepatic cells in vivo had been reduced compared to that of Ad5. For instance, liver transduction by the Ad5/41s fiber-chimeric vector was reduced by 58-fold compared to that by Ad5 (29); liver transduction by Ad5/7 was reduced by 100-fold (36) and that by Ad5/35 and Ad5/11 was reduced by 20- to 30-fold compared to liver transduction by Ad5 (27). Here we report for the first time the generation of fiber shaft-chimeric vectors that fully retained their capacity to efficiently transduce liver cells following intravascular delivery, thus demonstrating that in the context of Ad5 capsid, the exact amino acid composition of the fiber shaft is not essential and the length of the fiber shaft domain is likely to be responsible for the reduced infectivity of previously studied fiber shaft-chimeric vectors toward hepatic cells in vivo (40, 42).

In addition to Ad interaction with a high-affinity attachment receptor, virus entry into hepatocytes in vivo can be mediated by interaction with blood factors, including coagulation FIX and FX (2). To analyze whether fiber shaft-chimeric vectors are still able to utilize the blood factor-dependent pathway for cell entry, we infected CHO-K1 cells with Ads in the presence of FIX or FX. Our studies demonstrated that the infectivity of all fiber shaft-chimeric viruses increased more than 10-fold under these conditions compared to that in the absence of coagulation factors (Fig. 4A). These data showed that the lack of the KKTK motif within the fiber shaft domain derived from Ad31 or Ad41 did not affect the virus ability to enter cells in a FIX- or FX-dependent manner. In our heparin inhibition studies, we preincubated chimeric and control Ad vectors with heparin at increasing concentrations. In contrast to findings in earlier reports (7, 8), we found no inhibition of virus infectivity following Ad incubation with heparin, independently of the nature of the virus fiber shaft domain (Fig. 4B). Importantly, Bayo-Puxan et al. also recently reported finding no inhibiting effects of heparin on Ad5 vectors in their studies of AdS* vector in vitro (3). Although the reasons for an apparent discrepancy between our observations and the earlier data on the effect of heparin on Ad infectivity are presently unclear, by conducting extensive in vitro analyses we demonstrated that our fiber shaft-chimeric Ads, which lack the KKTK motif within the fiber shaft domain, were as efficient as unmodified vector with an Ad5 fiber shaft domain at all steps of Ad infection in vitro.

To further address the issue of potential involvement of the fiber shaft KKTK motif in mediating Ad liver cell infection following intravascular virus delivery, we administered our fiber shaft-chimeric vectors to C57BL/6 mice. Southern blot analysis of the accumulation of virus particles in liver tissue 1 h post-virus injection, which is a fiber-independent event, at least in mice (40), revealed that all analyzed viruses were trapped in liver tissue at similar levels (Fig. 5A). The analysis of Ad DNA in the liver 24 h post-virus injection, which at this time point correlates with liver cell transduction (39), also demonstrated that similar levels of genomic DNA from all analyzed vectors were present in the liver, independently of the amino acid sequences of the vector fiber shaft domains. These data imply that the lack of the KKTK motif within fiber shaft-chimeric vectors did not affect their ability to enter liver cells following intravascular delivery. To further evaluate whether the replacement of the Ad5 fiber shaft domain with heterologous sequences derived from Ad31 or Ad41 adversely affected Ad-liver cell interactions, we analyzed Ad trapping in Kuppfer cells and virus-encoded GFP expression in hepatocytes (Fig. 4 and 5). In agreement with data obtained by Southern blotting, these analyses showed that all fiber shaft-chimeric vectors were efficiently trapped by Kuppfer cells and transduced hepatocytes following their intravascular delivery. Also, in agreement with these findings, the analysis of markers of acute toxicity demonstrated that all analyzed vectors induced similar levels of IL-6, TNF-α, MCP-1, and gamma interferon secretion in mouse plasma following intravenous Ad administration.

Taken together, our data demonstrate that fiber shaft-chimeric Ad vectors lacking the KKTK motif efficiently infect liver cells in vivo. This finding suggests that the proposed direct interaction between the KKTK fiber shaft motif and hepatic HSPGs is unlikely to be the mechanism of efficient liver cell infection by Ad when virus is given intravenously. Our study contributes to better defining the molecular mechanisms responsible for Ad infectivity and biodistribution in vivo. The obtained data may also prove useful for further studies of the role of the Ad fiber shaft domain in determining Ad infectivity as well as for the design of Ad-based vectors for application in humans.