|Home | About | Journals | Submit | Contact Us | Français|
The in vivo efficacy of adenoviral vectors (AdVs) in gene delivery strategies is hampered by the broad tissue tropism of the virus and its efficient binding to human erythrocytes. To circumvent these limitations, we developed a prototype AdV lacking native binding sites. We replaced the adenoviral fiber with a chimeric molecule consisting of the fiber tail domain, the reovirus σ1 oligomerization domain, and a polyhistidine tag as model targeting moiety. We also abolished the integrin-binding motif in the penton base protein. The chimeric attachment molecule was efficiently incorporated onto AdV capsids, allowed efficient propagation of AdV without requirement for complementing fiber and conferred highly specific tropism to the AdV. Importantly, the targeted AdV exhibited markedly reduced tropism for liver cells. In comparison with control AdV with native tropism, the targeted AdV showed 1000-fold reduced transduction of HepG2 cells and 10,000-fold reduced transduction of mouse liver cells in freshly isolated liver slices. After intravenous inoculation of C57BL/6 mice, the targeted AdV exhibited delayed clearance in comparison with the native AdV, leaving approximately 10-fold greater levels in the blood 2hr after inoculation. For all tissues analyzed, the targeted AdV displayed significantly reduced in vivo transduction in comparison with the native vector. Furthermore, in contrast to the native AdV, the targeted AdV did not bind human erythrocytes. Together, our findings suggest that the targeted AdV design described here provides a promising platform for systemic in vivo gene delivery.
Human adenoviruses of serotypes 2 and 5 are acknowledged as promising gene-delivery vehicles. However, the in vivo utility of these adenoviral vectors (AdVs) is hampered by their promiscuous tropism. The broad array of cell types infected by adenovirus leads to AdV transduction of undesired tissues and reduces the fraction of the administered dose available for transduction of specific target cells. The liver is the major target of AdV transduction in mice after systemic administration and accounts for more than 99% of total transduction (Alemany and Curiel, 2001; Einfeld et al., 2001; Mizuguchi et al., 2002; Koizumi et al., 2003). The liver also is targeted in nonhuman primates, although in these animals a significant fraction of the administered AdV dose transduces the spleen (Smith et al., 2003a). An additional complication of AdV administration in humans is the interaction of the vector with erythrocytes (Cichon et al., 2003; Lyons et al., 2006). In contrast to murine erythrocytes, human erythrocytes are efficiently bound by AdVs, which prevents specific transduction of target cells. In blood samples obtained from a patient enrolled in a clinical gene therapy trial, greater than 98% of the virus was associated with erythrocytes (Lyons et al., 2006). Thus, the native tropism of adenovirus diminishes the in vivo efficacy of AdV-mediated gene delivery. To circumvent this problem and enhance the specificity of AdV transduction, a new generation of AdVs is required. Such vectors should eliminate native adenovirus tropism and incorporate unique target-binding moieties.
The tropism of adenovirus types 2 and 5 is determined by at least three distinct receptor-binding sites on the viral capsid. These sites mediate attachment to the coxsackievirus and adenovirus receptor (CAR), αv integrins, and heparan sulfate glycosaminoglycans (HSGs), respectively (Mathias et al., 1994; Roelvink et al., 1999; Dechecchi et al., 2000, 2001; Kirby et al., 2000; Li et al., 2001). Blood coagulation factor VII (FVII), FIX, FX, and protein C and complement component C4-binding protein (C4BP) have also been shown to play a role in adenovirus tropism by bridging adenovirus to cell surface receptors (Shayakhmetov et al., 2005; Parker et al., 2006). Ablation of the direct receptor interactions can be accomplished by modification or elimination of the receptor-binding sequences, which induces a change in the in vivo biodistribution of AdVs in mice, rats, and nonhuman primates (Einfeld et al., 2001; Koizumi et al., 2003; Nakamura et al., 2003; Smith et al., 2003a, b; Nicol et al., 2004; Koizumi et al., 2006). Remarkably, whereas CAR forms the primary receptor for AdVs in vitro, ablation of the CAR-binding site alone does not substantially affect the biodistribution of AdVs in vivo (Alemany and Curiel, 2001; Einfeld et al., 2001). Instead, alteration of at least two binding sites is required to influence AdV biodistribution. AdVs lacking binding sites for both CAR and αv integrins exhibit significantly decreased transduction of the heart, kidney, liver, and lung (Einfeld et al., 2001; Koizumi et al., 2003). Ablation of the HSG-binding site in combination with alteration of the binding sites for either CAR or αv integrins also reduces transduction of many organs (Koizumi et al., 2003; Smith et al., 2003b). Not unexpectedly, the most substantial reduction in tissue transduction is obtained when all three binding sites are altered (Koizumi et al., 2003, 2006).
Ablation of native tropism (detargeting) must be combined with the introduction of new tropism to create the ultimate AdV with beneficial properties for in vivo application. In general, two strategies have been used to provide AdVs with new tropism: a one-component strategy, which is based on genetic modification of fiber or another capsid component; and a two-component strategy, which relies on the use of bispecific adapter molecules (Barnett et al., 2002). There are advantages and disadvantages to each. The one-component strategy is better defined and more feasible for the generation of clinically applicable targeted vectors. It also is better suited for the design of targeted oncolytic adenoviruses, which rely on efficient in vivo production of targeted viral progeny to produce antitumor effects. Because the fiber molecule principally defines adenovirus tropism, most targeting approaches exploit this capsid protein as a platform for the display of novel binding entities. However, insertion of large and complex folded ligands into the fiber can alter its structure and preclude encapsidation onto the viral particle. Several groups including our own have developed chimeric attachment molecules in which the fiber knob is replaced with an exogenous trimerization domain (van Beusechem et al., 2000; Krasnykh et al., 2001; Magnusson et al., 2001). This strategy allows the introduction of complex receptor-binding moieties and simultaneously causes partial ablation of native AdV tropism by removal of the CAR-binding site. Until more recently, application of such targeted AdVs was severely limited by their impaired propagation efficiency, which necessitated wild-type fiber complementation for their production.
We reported the generation of a new genetically targeted AdV, AdG.L.Tail-T(ii)-MH, which encodes the chimeric attachment molecule Tail-T(ii)-MH in place of the fiber (Schagen et al., 2006). Tail-T(ii)-MH consists of the tail domain of adenoviral fiber, the T(ii) oligomerization domain of reovirus attachment protein σ1, and C-terminal Myc and polyhistidine tags. Therefore, AdG.L.Tail-T(ii)-MH lacks binding sites for CAR and HSG, for the “bridging” factors FIX and C4BP, and most likely for FVII, FX, and protein C as well (Shayakhmetov et al., 2005; Parker et al., 2006). Instead, this AdV is targeted to an artificial histidine tag-binding receptor that allows efficient propagation in packaging cells expressing this receptor without fiber complementation (Schagen et al., 2006). To further reduce native adenovirus tropism for systemic in vivo gene delivery, we generated a derivative of AdG.L.Tail-T(ii)-MH, AdG.L.p*Tail-T(ii)-MH, which lacks the integrin-binding site as well. In this study, we evaluated both Tail-T(ii)-MH-encoding AdVs for targeting specificity, liver cell transduction, and biodistribution after intravenous inoculation of mice. In comparison with control AdV, AdG.L.p*Tail-T(ii)-MH displays substantially enhanced bio-availability and diminished tissue transduction. This new AdV thus offers a promising platform for the design of highly targeted AdVs for numerous gene delivery applications.
The Ad5 E1-transformed human embryonic kidney cell line 293 and the human hepatoma cell line HepG2 were purchased from the American Type Culture Collection (Manassas, VA). The cell line 293.HissFv.rec is a derivative of cell line 293 that stably expresses an artificial histidine tag-binding receptor (Douglas et al., 1999). Cell lines were maintained in F12-supplemented Dulbecco's modified Eagle's medium (DMEM-F12) supplemented to contain 10% fetal calf serum and antibiotics (GIBCO-BRL Life Technologies, Breda, The Netherlands). Medium used for 293.HissFv.rec was supplemented to contain G418 (300μg/ml).
The genetically targeted AdV, AdG.L.Tail-T(ii)-MH, and the control vector, AdG.L, were described previously (Schagen et al., 2006). Both AdVs contain green fluorescent protein (GFP) and luciferase reporter genes in the E1 locus. AdG.L expresses wild-type fiber and penton base genes conferring native adenovirus tropism, whereas AdG.L.Tail-T(ii)-MH contains Tail-T(ii)-MH-encoding sequences in place of the fiber gene and consequently lacks CAR- and HSG-binding sites. To abolish the integrin-binding site of AdG.L.Tail-T(ii)-MH, the integrin-binding motif RGD in the penton base protein was replaced with RGE to generate the new AdV, AdG.L.p*.Tail-T(ii)-MH. Site-directed mutagenesis of the penton base gene was performed with primers 5′-GCCATCCGCGGCGAGACCTTTGCCACAC-3′, 5′-TCACTGACGGTGGTGATGG-3′, 5′-GGCAGAAGATCCCCTCGTTG-3′, and 5′-GTGTGGCAAAGGTCTCGCCGCGGATGGC-3′ and pBHG11 (Bett et al., 1994) as template. The resulting polymerase chain reaction (PCR) product containing a penton base gene with a mutated integrin-binding site, designated p*, was digested with PmeI and AscI and inserted into pBHG11ΔAsc. This derivative of pBHG11 was generated by digestion of pBHG11 with AscI and religation. After insertion of p* into pBHG11ΔAsc, the AscI fragment was reintroduced into that plasmid, generating pBHG11p*. This construct was digested with RsrII, and the penton base gene-containing fragment of 7707 bp was isolated and inserted into the 27,246-bp, RsrII-digested fragment of pAdEasy.Adtail-σ1T(ii)-MH (Schagen et al., 2006). The resultant pAdEasy.p*.Adtail-σ1T(ii)-MH construct was recombined with pAdTrack.CMV.Luc (Schagen et al., 2006) to generate pAdG.L.p*.Tail-T(ii)-MH, which contains the full-length adenoviral genome with GFP and luciferase reporter genes in place of E1, the Tail-T(ii)-MH-encoding sequences in place of the fiber gene, and an RGE-encoding sequence in the penton base gene.
AdG.L.p*.Tail-T(ii)-MH was generated by transfecting PacI-linearized pAdG.L.p*.Tail-T(ii)-MH into 293.HissFv.rec cells, using Lipofectamine Plus (Invitrogen, Carlsbad, CA). Resultant virus was propagated in 293.HissFv.rec cells to reach a scale of 20 T-182 flasks. Harvested AdVs were purified by two successive rounds of CsCl centrifugation, followed by dialysis. Titers of AdVs were quantified on the basis of viral particles (VP) and infective units (IU) (Schagen et al., 2006).
Immunoelectron microscopy was performed as described (Vellinga et al., 2004). CsCl-purified virions were spotted onto carbon-coated copper grids and incubated with antiknob monoclonal antibody (mAb) 1D6.14 (Douglas et al., 1996), anti-Myc mAb 9E10, or anti-(His)5 Penta-His antibody (Qiagen, Hilden, Germany). Grids were subsequently incubated with rabbit anti-mouse serum (Dako Cytomation, Copenhagen, Denmark), followed by gold-labeled protein A (Amersham Biosciences Europe, Freiburg, Germany). After fixation (1.5% glutaraldehyde) and negative staining (1% uranyl acetate), virions were visualized with a Philips EM 410-LS transmission electron microscope.
CsCl-purified virions of each AdV (5×109 VP) were incubated at 95°C for 5min in denaturing sample buffer (62.5mM Tris-HCl [pH 6.8], 10% glycerol, 2% sodium dodecyl sulfate [SDS], and 2.5% 2-mercaptoethanol) and resolved by SDS–10% polyacrylamide gel electrophoresis (PAGE). Viral proteins were transferred to polyvinylidene difluoride (PVDF) membranes and incubated with anti-fiber tail antibody Ab4 (Neomarkers, Fremont, CA) or anti-Myc mAb 9E10 as primary antibody. Bound viral proteins were visualized by chemiluminescence, using rabbit anti-mouse immunoglobulin G conjugated to horseradish peroxidase (RαM HRP; Dako, Glostrup, Denmark) as secondary antibody and Lumi-LightPLUS reagent (Roche, Almere, The Netherlands).
293, 293.HissFv.rec, and HepG2 cells were seeded at a density of 5×104 cells per well in 96-well plates 1 day before infection. AdG.L, AdG.L.Tail-T(ii)-MH, and AdG.L.p*Tail-T(ii)-MH were either untreated or preincubated with 330ng of 1D6.14 mAb or 300ng of Penta-His antibody at room temperature for 1.5hr. After 2hr of adsorption to cells, the inoculum was removed, fresh medium was added, and the cells were incubated for 48hr. GFP expression was assessed by fluorescence microscopy, or cells were lysed in 50μl of reporter lysis buffer (Promega, Madison, WI), and luciferase activity was measured in a chemiluminescence luciferase assay (Promega) with a Berthold luminometer (Berthold, Bad Wildbad, Germany).
Infection of liver tissue was assessed with liver slices prepared from freshly isolated mouse livers. Livers were resected from C57BL/6 mice, placed in ice-cold Krebs buffer (25mM d-glucose, 25mM NaHCO3, and 10mM HEPES), and immediately processed by drilling cores 8mm in diameter. Cores were sliced at a thickness of 200–250μm, using a Krumdieck tissue slicer (Alabama Research and Development, Muntfort, AL) in ice-cold Krebs buffer. Liver slices were transferred to 12-well plates and preincubated for 1hr in Williams' medium E (WME) supplemented to contain 25mM d-glucose and antibiotics (GIBCO-BRL) at 37°C in a 95% O2 and 5% CO2 atmosphere. Slices were then incubated with 108 VP/slice (~100 VP/cell) in oxygenated WME medium at 37°C under continuous rocking in a 95% O2, 5 % CO2 climate chamber. After 72hr of incubation, slices were lysed in 50μl of CCLR lysis buffer (25mM Tris-phosphate, 20mM CDTA, 200mM dithiothreitol [DTT], 10% glycerol, 1% Triton X-100) at room temperature for 15min, followed by three freeze–thaw cycles. Lysates were cleared by centrifugation, and luciferase activity was assessed with 10μl of the supernatant. Luciferase activity was normalized on the basis of protein concentration as determined by the Bio-Rad protein assay (Bio-Rad, Hercules, CA).
Female C57BL/6 mice weighing 20–25g were obtained from Harlan-CPB (Harlan, Horst, The Netherlands), fed a standard laboratory diet, provided water ad libitum, and allowed to acclimatize for at least 1 week before the initiation of experiments. Mice were randomly divided into three groups of at least five animals each and inoculated with 1010 VP of AdG.L, AdG.L.Tail-T(ii)-MH, or AdG.L.p*Tail-T(ii)-MH in 200μl of phosphate-buffered saline (PBS) via a lateral tail vein. At 2, 5, 10, 20, 30, 60, and 120min after inoculation, blood was obtained by tail vein venipuncture and placed into heparinized micropipettes (Marienfeld, Lauda-Königshofen, Germany). AdV concentration in blood samples was quantified by luciferase expression after transduction of 293.HissFv.rec cells. One day before assay, 293.HissFv.rec cells were seeded at a density of 5×104 cells per well in 96-well plates. Cells were inoculated with a mixture of 2μl of whole blood and 50μl of DMEM–F12. After 48hr of incubation, 293HissFv.rec cells were lysed in reporter lysis buffer (Promega), and luciferase activity was determined. AdV concentration was calculated by comparison with a standard curve, which was generated with known concentrations of the corresponding AdV.
Transduction of mouse tissues with AdVs was determined 48hr after intravenous administration. Mice were killed and liver, spleen, kidneys, lungs, and heart were resected. Tissues were snap-frozen in liquid nitrogen and homogenized with a mortar and pestle. A 25-mg aliquot of tissue was lysed in 100μl of CCLR lysis buffer by incubation at room temperature for 10min, followed by two freeze–thaw cycles. Lysates were cleared by centrifugation, and transduction of each tissue was determined by luciferase activity in 20μl of lysate. Luciferase activity was normalized for protein concentration as determined by the Bio-Rad protein assay. Differences in tissue transduction by the three AdVs were evaluated by one-tailed Mann–Whitney test, using GraphPad InStat 3.0 (GraphPad Software, San Diego, CA). The experimental protocols adhered to the rules outlined in the Dutch Animal Experimentation Act and the Guidelines on the Protection of Experimental Animals (Council of the European Community, 1986) and were approved by the Committee on Animal Research of the VU University Medical Center (Amsterdam, The Netherlands).
Blood of mice, rats, or humans was collected in EDTA tubes and mixed with a 1-vol equivalent of Alsever solution (23mM trisodium citrate, 114mM glucose, 55mM NaCl, and 3mM citric acid [pH 6.1]). Cells were centrifuged at 1200×g for 10min and washed three times by repeated resuspension in 2-vol equivalents of Alsever solution and centrifugation at 1200×g for 10min. The final pellet was resuspended in Alsever solution to generate a 30% packed-cell suspension. Hemagglutination was assayed with a 1% erythrocyte suspension, which was generated by dilution of the 30% packed-cell suspension in HA buffer (PBS, 0.005% bovine serum albumin [BSA]). A 50-μl volume of 1% erythrocyte suspension was prelaid in wells of a concave bottom-shaped 96-well plate and gently mixed with 50μl of a dilution series of each AdV (stock concentration, 1012 VP/ml). After 2hr of gravitational sedimentation, plates were photographed and analyzed for hemagglutination.
Binding of AdV virions to erythrocytes was quantified by determining the number of adherent AdV genomes, using quantitative PCR (qPCR). A 30% packed-cell suspension of human erythrocytes was diluted in PBS to a physiological concentration of 8.4×108 erythrocytes per 250μl. Erythrocytes were incubated with 8.4×107 VP at 37°C for 60min. Virions bound to erythrocytes were separated from unbound virions by centrifugation at 1200×g for 14min. The erythrocyte pellet was washed twice with 10-vol equivalents of PBS. Adenoviral DNA in bound and unbound viral fractions was isolated with a QIAamp DNA blood mini kit (Qiagen) according to the manufacturer's protocol. The content of viral genomes was quantified with a LightCycler 480 (Roche Diagnostics, Mannheim, Germany) with the LightCycler 480 SYBR Green I master kit, 20 pmol of forward hexon primer 5′-ATGATGCCGCAGTGGTCTTA-3′, and 20 pmol of reverse hexon primer 5′-GTCAAAGTACGTGGAAGCCAT-3′. A standard curve was generated with 10-fold serial dilutions of adenoviral DNA.
To achieve complete ablation of native adenovirus tropism, we generated a new AdV, AdG.L.p*Tail-T(ii)-MH, which lacks all known adenovirus receptor interaction sites. AdG.L.p*Tail-T(ii)-MH was derived from AdG.L.Tail-T(ii)-MH (Schagen et al., 2006). Both vectors encode the chimeric attachment molecule Tail-T(ii)-MH in place of fiber. However, AdG.L.p*Tail-T(ii)-MH incorporates the nonbinding sequence R340GE342 in its penton base (Obara et al., 1988), in place of the integrin-binding motif R340GD342. AdG.L.Tail-T(ii)-MH and AdG.L.p*Tail-T(ii)-MH were generated with 293.HissFv.rec cells (Douglas et al., 1999), which express an artificial receptor comprising an anti-histidine tag single-chain antibody. Both vectors replicated efficiently after infection of these cells without requirement for fiber complementation. In fact, their particle titers were similar to that of the native control vector AdG.L after infection of 293.HissFv.rec cells (Table 1).
Tropism mediated by the histidine tag allowed us to assess targeted transduction independent of native infection pathways. AdG.L.Tail-T(ii)-MH and AdG.L.p*Tail-T(ii)-MH preparations exhibited similar VP-to-IU ratios, which were approximately 10-fold higher than the VP-to-IU ratio of the native control vector (Table 1). The lower functional titers of the targeted vectors might be attributable to differences in the expression levels of CAR and the histidine tag-binding receptor on the surface of 293.HissFv.rec cells, or to differences in vector affinity for these receptors. The similar VP-to-IU ratios observed for the targeted AdVs with and without an intact RGD sequence suggest that histidine tag-mediated infection does not require penton base interactions with integrins.
To assess physical incorporation of Tail-T(ii)-MH in the capsids of targeted AdVs and to study accessibility of the Myc and histidine tags for binding, CsCl-purified virions were analyzed by immunoelectron microscopy (Fig. 1A). Immunogold labeling demonstrated anti-Myc mAb association with AdG.L.Tail-T(ii)-MH and AdG.L.p*Tail-T(ii)-MH virions but not with control vector AdG.L. Similar results were obtained with an anti-histidine mAb (data not shown). In contrast, incubation of the AdVs with an anti-fiber knob mAb did not reveal any association of gold particles with AdG.L.Tail-T(ii)-MH or AdG.L.p*Tail-T(ii)-MH, whereas AdG.L was clearly stained (Fig. 1A). Together, these findings confirm that wild-type fiber in the native vector has been replaced with the chimeric attachment protein in AdG.L.Tail-T(ii)-MH and AdG.L.p*Tail-T(ii)-MH.
The attachment protein incorporated in the AdVs was further characterized by immunoblotting (Fig. 1B). In a lysate of purified AdG.L particles, a fiber tail-specific antibody detected a 64-kDa protein, which corresponds to wild-type fiber. In AdG.L.Tail-T(ii)-MH and AdG.L.p*Tail-T(ii)-MH lysates, a 22-kDa protein was detected that corresponds to the expected size of Tail-T(ii)-MH. To corroborate the identity of this protein, we reprobed the blot with a Myc-specific mAb. This antibody bound to the 22-kDa protein of AdG.L.Tail-T(ii)-MH and AdG.L.p*Tail-T(ii)-MH and did not recognize any capsid component of the control vector. These findings confirm the exclusive incorporation of Tail-T(ii)-MH in the capsids of AdG.L.Tail-T(ii)-MH and AdG.L.p*Tail-T(ii)-MH.
To assess the effect of ablation of native adenovirus tropism and introduction of a binding sequence that engages an artificial histidine tag-binding receptor, we compared AdV transduction of parental 293 cells and 293.HissFv.rec cells (Fig. 2A). Cells were inoculated with AdVs at various multiplicities of infection (MOIs), incubated for 48hr, and assayed for luciferase activity. We observed dose-dependent transduction of both cell lines with all three viruses. As expected, native control vector AdG.L transduced both cell lines with almost equivalent efficiency. In contrast, the chimeric AdVs exhibited reduced transduction of 293 cells in comparison with 293.HissFv.rec cells. Over the range of MOIs tested, AdG.L.Tail-T(ii)-MH exhibited an average reduction of ~400-fold. Additional alteration of the RGD sequence in AdG.L.p*Tail-T(ii)-MH yielded a decrease in transduction of ~1200-fold. However, in comparison with AdG.L, AdG.L.Tail-T(ii)-MH and AdG.L.p*Tail-T(ii)-MH exhibited only 16- and 25-fold reductions, respectively, in the transduction of 293.HissFv.rec cells. Thus, ablation of native adenovirus binding sites substantially diminishes AdV infection of 293 cells and incorporation of a histidine tag-binding moiety largely restores infection of 293.HissFv.rec cells.
To confirm that transduction by the targeted AdVs is dependent on the binding activity of Tail-T(ii)-MH, we incubated the vectors with an anti-histidine tag mAb before inoculation of 293.HissFv.rec cells. As controls, we used an anti-fiber knob mAb and native vector AdG.L. After 48hr of incubation, AdV-mediated GFP transduction was assessed by fluorescence microscopy (Fig. 2B). The anti-fiber knob mAb neutralized transduction by AdG.L but not by the targeted AdVs. Conversely, the anti-histidine mAb neutralized transduction by the targeted AdVs but not by the native control vector. Therefore, replacement of native adenovirus-binding sites with a new binding moiety ablates native tropism and results in highly stringent targeting specificity.
Because the majority of intravenously administered adenovirus is captured in the liver either via transduction of hepatocytes or uptake in Kupffer cells, the primary goal of adenovirus detargeting for systemic administration is to limit liver sequestration. To determine whether the targeted AdVs display altered hepatic tropism, we first analyzed transduction efficiency using the human hepatoma cell line HepG2. HepG2 cells were inoculated with 1000 VP/cell, and transduction efficiency was determined 48hr after incubation by quantifying luciferase activity (Fig. 3A). AdG.L was capable of efficient transduction of HepG2 cells, whereas transduction of these cells by both targeted AdVs was substantially reduced. In comparison with AdG.L, AdG.L.Tail-T(ii)-MH and AdG.L.p*Tail-T(ii)-MH displayed~100- and 1000-fold reductions in transduction of HepG2 cells, respectively.
We next determined whether the targeted AdVs were altered in transduction of fresh liver slices prepared from C57BL/6 mice. Each slice was inoculated for 72hr with 108 VP of AdG.L, AdG.L.Tail-T(ii)-MH, or AdG.L.p*Tail-T(ii)-MH. As an internal control for liver cell viability, each slice was also inoculated with 108 VP of an AdV with native tropism and expressing DsRed (AdDsRed). After 72hr of incubation, AdDsRed transduction was equivalent in all slices (data not shown). The transduction efficiency of the test vectors was quantified by luciferase activity in liver slice lysates (Fig. 3B). As expected, AdG.L showed efficient liver cell transduction. In contrast, ablation of native binding sites in AdG.L.Tail-T(ii)-MH and AdG.L.p*Tail-T(ii)-MH was associated with a 10,000-fold reduction in liver transduction. In fact, the level of luciferase expression achieved by the targeted AdVs did not differ significantly from that of control slices transduced with AdDsRed, which does not express luciferase. These findings suggest that the targeted AdVs display markedly diminished tropism for mouse liver.
To directly evaluate the effect of ablation of native adenovirus tropism in vivo, we inoculated C57BL/6 mice with 1010 AdV particles intravenously. During the first 2hr after inoculation, blood samples were drawn from the tail vein, and AdV titers were determined with 293.HissFv.rec cells (Fig. 4A). All three AdVs exhibited a biphasic clearance pattern with a rapid decrease during the first 30min, followed by a slower decrease over the remaining 90min. AdG.L exhibited the most rapid clearance, with less than 1% of the administered dose detectable at 10min and approximately 0.1% at 20min. Although the clearance profile of AdG.L.Tail-T(ii)-MH suggests somewhat improved bioavailability in the first 30min after administration, clearance of this AdV did not differ significantly from that of AdG.L in the interval thereafter. In contrast, the half-life of AdG.L.p*Tail-T(ii)-MH immediately after inoculation was prolonged, with approximately 10% of the administered dose detected in the circulation at 10min and 1% at 30min. The virus then exhibited a rate of clearance identical to that of the other two vectors but at an approximately 10-fold higher concentration. Thus, AdG.L.p*Tail-T(ii)-MH displays enhanced bioavailability in the blood in the first 2hr after inoculation in comparison with the other AdVs.
To determine transduction efficiency of mouse tissues after intravenous administration of AdVs, mice were killed 2 days postinoculation, and organs were resected and processed for luciferase assay (Fig. 4B). As expected, AdG.L preferentially transduced the liver, consistent with several previous reports (Einfeld et al., 2001; Koizumi et al., 2003, 2006; Nakamura et al., 2003; Smith et al., 2003a, b; Nicol et al., 2004). Removal of the native adenovirus-binding sites in the targeted AdVs significantly reduced liver transduction. In comparison with the control vector, AdG.L.Tail-T(ii)-MH exhibited a 70% (p<0.05) reduction in liver transduction and AdG.L.p*Tail-T(ii)-MH exhibited an 83% (p<0.01) reduction. Transduction of other tissues by AdG.L.Tail-T(ii)-MH also was reduced, reaching statistical significance for the heart (75%; p<0.05) and lung (50%; p<0.05). Moreover, transduction with AdG.L.p*Tail-T(ii)-MH was significantly diminished for all tissues tested. Transduction of the heart and spleen with AdG.L.p*Tail-T(ii)-MH was reduced 75% (p<0.05) and 66% (p<0.01), respectively, and transduction of the kidney and lung was reduced to background levels (p<0.01).
AdVs with native tropism bind and agglutinate erythrocytes of human and rat but not murine origin (Cichon et al., 2003; Nicol et al., 2004). Interactions with human erythrocytes constitute a major roadblock for therapeutic application of AdVs. Therefore, we tested the targeted AdVs for erythrocyte binding and agglutination (Fig. 5A and Table 2). Native AdV AdG.L produced hemagglutination of human and rat erythrocytes but not mouse erythrocytes, as anticipated. However, neither AdG.L.Tail-T(ii)-MH nor AdG.L.p*Tail-T(ii)-MH produced hemagglutination of any of the erythrocytes tested.
To corroborate these findings, we determined the fraction of each AdV bound to human erythrocytes. AdVs were incubated with human erythrocytes, and AdV genomes in cell (bound) and supernatant (unbound) fractions were quantified by qPCR (Fig. 5B). AdV with native tropism bound human erythrocytes efficiently, leaving less than 5% of the inoculum in the supernatant. In sharp contrast, the affinity of AdG.L.Tail-T(ii)-MH and AdG.L.p*Tail-T(ii)-MH for human erythrocytes was markedly reduced, with greater than 90% of the inoculum remaining in the supernatant. Thus, ablation of native adenovirus tropism substantially compromises the capacity of targeted AdVs to interact with human erythrocytes.
Targeting AdVs to specific cells by complete abrogation of native adenovirus tropism and introduction of a unique binding moiety should improve the in vivo applicability and efficacy of these vectors. We previously generated a genetically targeted AdV, AdG.L.Tail-T(ii)-MH, by replacing the adenoviral fiber molecule with a prototype chimeric attachment protein, which lacks CAR- and HSG-binding sites and confers histidine tag-dependent tropism (Schagen et al., 2006). In addition, deletion of the fiber knob removed putative binding sites for coagulation factors FVII, FIX, FX, protein C, and complement component C4BP, which have been shown to link adenovirus to alternative cell surface receptors (Shayakhmetov et al., 2005; Parker et al., 2006). This vector design has the potential to incorporate large and complex proteins as ligands and allows vector propagation without the need for complementing fiber.
In this study, we substantially improved the AdG.L.Tail-T(ii)-MH platform to facilitate systemic in vivo administration by ablating the final known adenovirus attachment site, the integrin-binding motif, from the viral capsid. Importantly, this modification did not affect efficient propagation of the vector in cells that express an artificial histidine tag-binding receptor. In fact, the new genetically targeted vector, AdG.L.p*Tail-T(ii)-MH, replicates in histidine tag receptor-expressing cells efficiently and achieves viral particle yields that approximate those of an AdV with native tropism. We conclude that binding of the chimeric attachment protein to the artificial histidine tag-binding receptor is sufficient for AdV internalization and delivery of the viral genome to the nucleus, thus allowing efficient virus replication and transgene expression. This suggests that retention of the presumably physiologically inert histidine tag in derivative targeted AdVs incorporating selective binding ligands of choice will allow their efficient production using histidine tag receptor-expressing cells. Therefore, production of a variety of genetically targeted AdVs with different binding specificities should be possible with a single packaging cell line.
Many studies have shown that intravenous administration of native AdV expressing luciferase usually results in high luciferase activity per milligram of protein in the liver, with approximately 100-fold lower levels in other tissues (e.g., Reynolds et al., 1999; Alemany and Curiel, 2001; Einfeld et al., 2001; Koizumi et al., 2003, 2006). However, we found approximately 70% of AdG.L activity in the liver and 10–15% in lungs and spleen. This was more in line with a previous observation by Seki and coworkers (2002), who found a similar transduction efficiency of the liver and even more efficient transduction of the spleen. We can only speculate as to why the biodistribution profile of AdV with native tropism differed considerably between individual studies using similar methods. Perhaps this could be partially explained by differences in the injected AdV dose. Seki and coworkers and we injected 1010 particles per animal. In most studies the injected dose was higher, reaching up to 1011 particles per animal. It has been reported that liver cell transduction efficiency after systemic AdV administration in mice is nonlinear, exhibiting a viral dose threshold effect at approximately 3×1010 particles (Tao et al, 2001). Thus, less effective gene expression by the mouse liver as seen by us might not be unexpected after low-dose AdV infusion.
Targeted vector AdG.Lp*Tail-T(ii)-MH exhibited histidine tag receptor-specific transduction and substantially diminished infectivity of liver cells in vitro. Most importantly, intravenous inoculation of mice with the targeted vector resulted in reduced transduction of all tested tissues in comparison with the native vector. Interestingly, transduction of kidney and lung with AdG.Lp*Tail-T(ii)-MH was diminished to undetectable levels. In contrast, AdG.LTail-T(ii)-MH, which contains an intact integrin-binding site, exhibited only a modest reduction in transduction of those tissues. Thus, interaction of the capsid with integrins is essential for AdV transduction of kidney and lung, consistent with previous observations (Einfeld et al., 2001; Koizumi et al., 2003). Because the kidney and lung sequester only a minor fraction of systemically administered AdV, effective detargeting of those tissues might not increase target cell transduction or substantially prolong circulation half-life. In absolute terms, the modest reduction in liver transduction is likely to have a larger impact on bioavailability than the complete detargeting of kidney and lung. However, effective detargeting of kidney and lung establishes a framework for the development of highly selective target cell transduction of those tissues by isolated organ perfusion. This approach might be an attractive option for AdV oncolytic treatment of primary malignancies originating from kidney or lung or metastatic lesions involving those organs. Such approaches could be considered as alternatives for systemic gene delivery if liver and spleen transduction cannot be entirely abolished.
Reduced uptake of AdG.Lp*Tail-T(ii)-MH in tissues was associated with increased vector levels in the circulation. We think it important that the assay used to quantify virus in the circulation measured particles capable of transducing target cells. Therefore, AdVs detected in the circulation represented bioavailable AdVs. For the first 30min after intravenous inoculation, AdG.Lp*Tail-T(ii)-MH exhibited slower clearance from the blood than did native AdG.L and AdG.LTail-T(ii)-MH. This delay in clearance resulted in approximately 10-fold higher levels of circulating AdV during the majority of the observation interval. Although replacing fiber with the Tail-T(ii)-MH chimeric attachment molecule in AdG.LTail-T(ii)-MH removed most known native adenovirus-binding sites, this modification did not increase the circulation half-life. Instead, additional deletion of the integrin-binding motif from AdG.LTail-T(ii)-MH to generate AdG.Lp*Tail-T(ii)-MH was essential for prolonged AdV maintenance in the circulation. Sustained bloodstream persistence of comparable doses of intravenously administered AdVs was previously achieved by coating AdVs with polyethylene glycol or polymer to inhibit cellular uptake or by depleting Kupffer cells via GdCl3 treatment before AdV infusion (Alemany et al., 2000; Akiyama et al., 2004; Green et al., 2004; Ogawara et al., 2004). It is possible that the combined effects of genetic modification of the vector and previously employed coating strategies might result in an even more prolonged circulation half-life.
Interactions of AdV with human erythrocytes pose a significant limitation to the utility of AdVs for systemic administration. After intravenous inoculation of humans, greater than 98% of the AdV dose in the blood is associated with erythrocytes, which significantly hampers infection of target cells (Lyons et al., 2006). Moreover, this phenomenon prevents extrapolation of AdV bioavailability in mouse studies to humans, because human AdVs do not bind murine erythrocytes. Replacement of fiber with Tail-T(ii)-MH in the targeted AdVs used in this study abrogated interactions with human red blood cells. Both AdG.LTail-T(ii)-MH and AdG.Lp*Tail-T(ii)-MH exhibited a 90% reduction in erythrocyte binding and loss of detectable hemagglutination. These findings indicate that the integrin-binding site does not mediate erythrocyte interactions, confirming a previous observation by Lyons and coworkers (2006). Rather, our findings localize the erythrocyte-binding site to the fiber shaft or knob domain, which is in agreement with a previous report that mapped hemagglutination capacity to the CAR-binding site (Nicol et al., 2004). Thus, targeted AdVs encoding Tail-T(ii)-MH attachment molecules evade potential sequestration by human erythrocytes, suggesting that AdVs based on this platform would have prolonged bioavailability in the circulation of humans.
In this study, we developed a prototype targeted AdV, AdG.L.p*Tail-T(ii)-MH, which incorporates many desirable features for systemic gene delivery. Although caution should be taken in extrapolating observations based on mouse models to humans, the highly specific transduction profile, improved bioavailability, and absence of erythrocyte binding of AdG.L.p*Tail-T(ii)-MH make the systemic application of AdV-based gene delivery more viable, and further investigation is warranted.
The authors thank Dr. Joanne Douglas (UAB, Birmingham, AL) for 293HissFv.rec cells and Jan Dekker for expert technical assistance. This research was supported by a grant from the VU University Stimuleringsfonds (USF99/25), the Dutch Digestive Diseases Foundation (WS02-31), the Pasman Foundation, Public Health Service award R37 AI38296 from the National Institute of Allergy and Infectious Diseases (T.S.D.), and the Elizabeth B. Lamb Center for Pediatric Research (T.S.D.). Victor van Beusechem was supported by a research fellowship of the Royal Netherlands Academy of Arts and Sciences (KNAW). Jort Vellinga was supported by the Sixth Framework Program GIANT from the European Union (contract 512087).
No competing financial interests exist.