Pseudotyping of viral vectors has been widely used to enhance viral transduction efficiency. One of the most popular pseudotyping proteins has been the G-protein of the vesicular stomatitis virus, VSV-G. In the present study, we show that the 21-amino-acid ectodomain with transmembrane and cytoplasmic tail domains of VSV-G (VSV-GED) augments baculovirus-mediated gene delivery in vertebrate cells by aiding viral entry. The VSV-GED pseudotyped virus replicated efficiently in insect cells yielding high titers. Five out of six studied cell lines showed improved transduction, as measured by a number of transduced cells or transgene expression level. Nearly 15-fold increase in the transduction efficiency was detected in rat malignant glioma cells as compared to the control virus. In the rat brain, transgene expression could be detected in the walls of lateral ventricles and in subarachnoid membranes. Increased transduction efficiency was also observed in the rabbit muscle. Our results suggest that VSV-GED enhances baculoviral gene transfer by augmenting gp64-mediated endosomal release. Moreover, no cytotoxicity was associated with improved gene transfer efficiency. Thus, VSV-GED pseudotyping provides a simple means to enhance baculovirus-mediated gene transfer in vitro and in vivo.
Baculovirus is a useful vector for gene therapy.1 Baculoviruses do not replicate in mammalian cells; they are easy to manipulate and can house large foreign DNA inserts. They have been shown to transduce various cell types in vitro and in vivo with significant efficiency.2, 3, 4, 5 However, efficient gene delivery still remains a challenge for baculovirus as for other gene therapy vectors in many target cells. Efficient transduction would allow high transgene expression with a lower viral load. Lower multiplicity of infection, MOI, could also decrease possible vector-related immune responses by lowering the total dose administered.
A straightforward method to increase transduction efficiency is viral pseudotyping, which alters viral tropism by modifying, replacing or adding components to virus surface. These components are often derived from other viruses and have specific or wide tropism.6 One of the most widely used pseudotyping tools is G glycoprotein of the vesicular stomatitis virus (VSV-G).7, 8, 9, 10 VSV-G is routinely used to enhance the target range and transduction efficiency of retroviruses11 by providing wider tropism as well as improved viral stability and augmented resistance to the complement inactivation.12 Several reports of VSV-G pseudotyped baculoviruses provide evidence that VSV-G is able to enhance transduction efficiency of baculovirus in vertebrate cells in vitro and in vivo.10, 13 It has been suggested that the improved transduction efficiency is a result of the increased escape of baculovirus nucleocapsids from the endosomes.10, 14
It has been shown that VSV vector expressing a G protein ectodomain (G stem, GS) of 42 amino acids together with the TM and CTD domains confers efficient virus budding probably by inducing membrane curvature at sites of virus assembly.15 Recombinant viruses, having 12 or more membrane proximal residues, produced near wild-type levels of virus particles. Furthermore, Jeetendra et al.16 have demonstrated that the same VSV-G fragment was able to induce hemifusion and potentiate the membrane fusion activity of some heterologous viral envelope proteins when two proteins were coexpressed in BHK-21 cells. Only 14 membrane-proximal residues in addition to TM/CTD were needed for the enhanced fusion activity of the studied viral glycoproteins. Baculovirus major envelope protein gp64 was not, however, included in that study. Previously, it has been demonstrated that the GS-region of VSV-G can be used as a membrane anchor in displaying fusion proteins on baculovirus surface.17, 18 Yet, a vector displaying a fusion protein of IgG-domain (ZZ) of protein A fused to the VSV-GED did not result in increased transduction efficiency.18
In the current study, we constructed a baculovirus vector displaying, in addition to gp64, a 21-amino-acid ectodomain in conjunction with the TM and CTD domains of VSV-G (VSV-GED) and studied its effect on the baculovirus infection and transduction rates in insect and vertebrate cells, respectively. The resulting virus was efficiently produced and resulted in higher transduction efficiency several folds in HeLa, SKOV-3, HepG2, 293T and BT4C cell lines as compared to the control virus. Not only was the number of transduced cells increased but the level of reporter gene expression was also higher in the transduced cells. Increased transduction efficiency was also detected in the rat brain and rabbit muscle in vivo. To the best of our knowledge, this is the first report of augmented gene delivery with VSV-GED pseudotyping. Thus, VSV-GED provides a novel tool to improve baculovirus-mediated gene delivery without compromising high viral titers.
Generation and characterisation of VSV-GED-pseudotyped baculovirus
To improve baculovirus properties as a gene delivery tool, VSV-GED was displayed on the surface of the baculovirus (Figure 1). Successful baculovirus production was studied by immunoblotting cell lysates and concentrated virus samples using gp64 antibody. Correct-sized protein (∼64 kDa) was detected in all samples. The membrane incorporation of the VSV-GED was studied by immunoblotting using a VSV-G antibody against the 15 carboxy-terminal amino acids (497–511) of the VSV-G. The predicted sized VSV-GED (8.6 kDa) and its trimeric form15 were observed with VSV-G antibody in cell lysates and gradient-purified concentrated virus samples. The VSV-GED was found to be efficiently incorporated on the baculovirus membrane (Figure 2).
In order to determine the ratio of total particles to infective particles (tp/ip), immunoblotting with vp39 and gp64 antibodies was performed and the results revealed a similar tp/ip ratio with the VSV-GED virus as with the control virus (Figure 2). Titers of the VSV-GED virus stocks were repeatedly high (after 300 × concentration a typical titer was 2.5 × 10−10 PFU/ml), suggesting no adverse effects of the VSV-GED on the viral replication in insect cells.
Cytotoxicity of the VSV-GED baculovirus was determined by the MTT assay. No cytotoxicity caused by the VSV-GED or the control virus was detected (Table 1). The cytotoxicity of sodium butyrate, however, was evident.
Improved transduction efficiency in vitro
Since the VSV-GED-like VSV-G domain has been reported to potentiate the fusion activity of some heterologous viral envelope proteins, we speculated that the VSV-GED might also aid gp64-mediated baculoviral gene delivery. To study this hypothesis, the transduction efficiency of HeLa, SKOV-3, BT4C, HepG2, EAHY and 293T cells was determined with the VSV-GED virus and control virus using MOIs ranging from 10 to 1000. β-Galactosidase staining was used to calculate the transduction efficiency.
BT4C cells are attractive targets since they reflect the metabolism and subcellular environment of severe brain tumours and provide an established animal model for malignant glioma.19 HepG2 were chosen because of their wide use in studies of baculovirus-mediated gene delivery. They are known to be very susceptible to baculovirus-mediated gene transfer.20, 21 On the contrary, EAHY cell line has been reported to be poorly susceptible to baculovirus transduction.22 Transduction efficiency of HeLa, SKOV-3, BT4C and 293T settles between HepG2 and EAHY.
When compared with the control virus, the VSV-GED-pseudotyped virus resulted in higher transduction efficiency in all cell lines except EAHY (data not shown), where only a neglible gene expression was detected. BT4C cells showed 75% transduction efficiency with the VSV-GED virus while the control virus transduced only 30% of the cells at MOI 50. An almost 15-fold increase in the transduction efficiency was observed at MOI 10 (Figure 3). In general, the increase in the transduction efficiency was most prominent with low viral loads, that is, MOIs under 200 (Figure 4). With higher MOIs, the difference in the transduction efficiency diminished. However, in HepG2 cells, the difference was still notable at higher virus loads, MOI 200 resulting in an increase in the transduction efficiency from 20 to 70% and MOI 1000 from 60% to almost 100%.
The improved transduction was also evident when β-galactosidase enzyme activity was measured in the transduced cell lysates (Figure 3). The results show that the enzyme levels were increased almost 40-fold in BT4C cells at MOI 10. This difference was diminished at higher MOIs. However, in HepG2 cells, the difference was still clearly detectable with MOI 1000 in agreement with the above results showing percentage of transduced cells. In 293T cells, the VSV-GED virus resulted in a six-fold increase in gene expression as compared to the control virus at MOI 10. HeLa showed a marked increase in β-galactosidase level compared to the control virus with MOIs 200 and 1000.
Improved endosomal release
To determine the pH requirement for viral membrane fusion, a syncytium formation assay was performed (Figure 5). Wild-type baculoviruses and full-length VSV-G-pseudotyped baculoviruses were used as controls. No fusion activity was detected at pH>5.5 with the control virus or the VSV-GED virus. Infection with the virus containing the entire VSV-G protein resulted in extensive syncytia formation during virus preparation in normal medium. The pH of the insect cell medium was 6.2.
To study further the relative pH required for the endosomal release of the viral capsids in mammalian cells, monensin and various concentrations of ammonium chloride were used to prevent endosomal acidification. Treatment of BT4C and HepG2 cells with monensin prevented transduction with the control virus as measured by counting the blue cells, whereas with the VSV-GED virus transduction was partly retained in all cell lines tested (data not shown). A progressive decrease in the transduction efficiency of HepG2 cells was observed with increasing concentrations of ammonium chloride for both viruses. The gene transduction by both viruses was completely inhibited at 8 mM ammonium chloride (Figure 6). These results suggest that the VSV-GED augments baculovirus transduction by enhancing endosomal escape although the pH requirement for fusion remains unaltered.
Improved transduction in vivo
As the VSV-GED baculovirus was able to improve the transduction efficiency in vitro, we wanted also to test its efficiency in different animal models in vivo. We have previously shown that baculovirus-mediated gene delivery into the rat brain results in high production of nuclear-targeted β-galactosidase in cuboidal epithelial cells of the choroid plexus. In addition, some gene expression could be detected in endothelial cells of the microvessels and in the subarachnoidal space.4 In the present study, the LacZ control virus showed expected transduction patterns after intraventricular injections (Figure 7). However, with the VSV-GED virus, a strong LacZ marker gene expression was also found in the epithelial lining of the lateral ventricles, epithelial lining of the cerebral aqueduct and subarachnoidal membrane (Figure 7).
Viruses were injected also into New Zealand white rabbit muscle (Musculus semimembranos). Interestingly, while the control virus expression was observed mainly in nonmuscle cells, for example, pericytes, the VSV-GED virus showed an enhanced transduction in muscle cells. However, even with the VSV-GED virus, only modest transduction efficiency was detected in the rabbit skeletal muscle (data not shown). Thus, the results suggest that VSV-GED pseudotyping provides a simple means to increase baculovirus transduction efficiency also in vivo.
We have produced VSV-GED-pseudotyped baculoviruses, which are able to transduce several cell lines with remarkably higher efficiency than the control viruses. Enhanced gene delivery was also observed in vivo by a wider distribution of β-galactosidase positive cells. The sequence coding the 21 amino-acid ectodomain together with TM/CTD of the VSV-GED was introduced into the baculovirus genome under the strong polyhedrin promoter. VSV-GED display was confirmed from the concentrated viruses by immunoblotting using a VSV-G antibody, which recognises the 15 carboxy-terminal amino acids of the VSV-GED. In agreement with a previous study,15 VSV-GED trimer was also detected on the immunoblot. The quantity of infective particles versus total virus particles (ip/tp) was maintained in the VSV-GED pseudotyped viruses indicating that the quality of the VSV-GED-pseudotyped virus was comparable to that of the control virus. In line with that, the titer of the concentrated virus was constantly high. The small size of the VSV-GED (8.6 kDa) probably contributes favourably to its high-level noninterrupting incorporation into the viral particles.
The increase in the transduction efficiency was remarkable in all studied cell lines except EAHY, which showed negligible β-galactosidase expression. An almost five-fold increase in the transduction rate was achieved in HepG2 cells and nearly 15-fold increase was detected in BT4C cells using the VSV-GED-pseudotyped virus at MOI 10. The VSV-G protein is able to enhance the efficiency of transduction in several cell lines including HeLa10, 14 in line with the results obtained with the VSV-GED. In general, the increase in the transduction efficiency was greatest at low MOIs (<200) and saturated at higher MOIs. The control virus showed a linear dose–response curve while the VSV-GED virus showed an exponential dose–response, that is, fast saturation with increasing MOIs.
To elucidate the exact mechanism of the transduction enhancement, a syncytium formation assay was performed. According to the literature, a pH5.5 is required to induce gp64-mediated membrane fusion23 and this was confirmed using the LacZ control virus. VSV-G induced membrane fusion occurs, however, at considerably higher pH, 5.8–6.2.24 Indeed, large syncytia formation was observed in infected insect cells with VSV-G baculovirus. The pH of the insect cell medium remained unaltered and was 6.2 after three days of infection. VSV-GED infection resulted in significant syncytia formation only under pH 5.5, indicating that VSV-GED does not share the fusion properties of VSV-G.
In order to test the idea that VSV-GED augments gp64-mediated endosomal escape in vertebrate cells, a study using monensin and ammonium chloride was performed. These compounds prevent the acidification needed for the baculovirus envelope fusion and endosomal escape mediated by gp64. After treatment with monensin, no nuclear-targeted β-galactosidase activity was detected with the control virus, whereas VSV-GED transduced cells showed only a partial reduction in the transduction efficiency. Together with results from the ammonium chloride experiments this suggests that VSV-GED is able to aid gp64-mediated endosomal fusion in vertebrate cells. The results are in agreement with the previous results by Jeetendra et al.,16 who showed that a VSV-GED-like protein domain, GS, is able to induce hemifusion and potentiate membrane fusion activity of the F protein of simian virus 5 and HIV envelope proteins when coexpressed in BHK-21 cells.16 Thus, the enhancement in the transduction efficiency by VSV-GED-pseudotyped baculovirus may be due to faster kinetics of the endosomal release. However, in order to elucidate the exact mechanism, further experiments are needed.
Cytotoxicity of the VSV-G limits retrovirus preparation in vertebrate cells.11 In insect cells, expression of the VSV-G led to cell fusion during virus production. Preparation of the VSV-GED-pseudotyped virus, however, did not show this adverse effect. Indeed, we have found occasional difficulties to produce high-titer VSV-G-pseudotyped baculoviruses while no problems have been associated with the VSV-GED virus production. There have also been reports that the VSV-G included in the viral envelope increases toxicity of the vector25, 26, 27 in addition to the well-known cytotoxicity of VSV-G in packaging cell lines.11, 12, 28 Yet, no cytotoxicity was detected for the VSV-GED-pseudotyped baculoviruses in MTT assay. Furthermore, as mentioned earlier, no adverse effects were seen in the virus production or in the transduction experiments. This strongly supports the beneficial properties of the VSV-GED as compared to the VSV-G in baculovirus pseudotyping.
VSV vector expressing 42 amino-acid ectodomain together with TM and CTD can bind to BHK cell membranes.16 The 21 amino-acid ectodomain of VSV-GED may retain these membrane-binding characteristics and thus mediate a stronger interaction with the target cell membrane.18 This could also explain the difference observed in the transduction pattern in the rat brain of the VSV-GED-pseudotyped virus when compared with the control virus. However, when it is taken into account that baculoviruses are able to enter numerous cell lines,22, 28, 29 the reason for the observed enhanced gene expression in vivo is less likely due to the increased virus attachment than augmented endosomal release.
The rat brain and rabbit muscle were chosen for targets to study properties of the VSV-GED pseudotyped baculovirus in vivo. Interestingly, as the control LacZ baculovirus transduced efficiently cuboid epithelium of the choroids plexus and to some extent epithelial cells in brain microvessels, β-galactosidase expression after the VSV-GED virus injection was also detected in the walls of the lateral ventricles, subarachnoidal space and epithelial lining of the brain. Interestingly, the observed change in the transduced cells is in line with the results obtained by Watson et al.11, 30 using VSV-G-pseudotyped lentiviruses and may indicate that VSV-GED fragment results in a similar transduction pattern as the full-length VSV-G in vivo. Enhancement in the gene transfer was also observed in rabbit M. semimembranosus after intramuscular injection of the VSV-GED virus.
In conclusion, VSV-GED pseudotyping is able to significantly enhance baculovirus-mediated gene transfer into vertebrate cells in vitro and in vivo. VSV-GED display has several advantages compared to the VSV-G pseudotyping and may also provide a useful tool to augment gene transfer of other vectors.
Materials and methods
Generation of the recombinant baculovirus
In order to replace gp64 and avidin sequences with VSV-GED, two linkers were introduced to the Baavi31 transfer plasmid. The first linker (AAATAGATCTC-CTAGGAGATCTATTT) containing the BglII site was ligated to SwaI/AvrII cut Baavi vector in order to remove one of the three SmaI-sites. The two remaining SmaI-sites flanking the gp64 gene were used to remove this sequence from the vector. The removal of one of the three PstI sites resulted in two intact PstI sites flanking avidin sequence, enabling its elimination. The PstI site was removed by a combination of PstI partial digestion and SwaI digestion followed by ligation of a second linker (ATGCATTT – AAATGCATTGCA) containing a unique NsiI restriction site. The gene encoding VSV-G ectodomain17 was amplified with 5′ primer GGGGTGATACTGGGCTATCCAA and 3′ primer AGATCTTTACTTTCCAAGTCGGTTCA (BglII site underlined) and transferred into the SmaI site of the vector. All steps were confirmed with restriction enzyme digestion. The LacZ control virus was prepared as described earlier.2 Several independent batches of VSV-GED and control virus were generated by BVBoost system32 and purified as described earlier.2 To verify the titer of the virus stocks, end point dilution was used. Virus preparations were tested for sterility and analysed for mycoplasma contamination.
Ten million plaque-forming units (PFU) of gradient purified viruses were diluted 1:4 to sample buffer (0.125 M Tris HCl/pH 6.8/4% SDS/20% glycerol/0.004% bromophenol blue/10% 2-mercaptoethanol) and samples were denaturated at 100°C for 10 min prior to SDS-PAGE and immunoblotting. Samples were run on reducing 10% SDS-PAGE, transferred onto a nitrocellulose membrane (Trans-Blot, Bio-Rad, USA) and the blots were probed with mouse anti-gp64 mAb (1:1000; Insight Biotechnology, Webley, UK), mouse anti-VSV-G (1:1000) or vp39 antibody (1:2000) as described.22 Finally, primary antibodies were detected with alkaline phosphatase conjugated secondary antibodies (1:2000; Bio-Rad), followed by a colour reaction (NBT/BCIP, Roche, Basel, Switzerland).
Cells were seeded at 7500 cells per well on 48-well plates (transduction efficiency) or 15 000 cells per well on 24-well plates (β-galactosidase enzyme assay) in their recommended medium. After 24 h, the medium was removed and fresh complete medium containing virus dilutions was added. Following 2 h incubation at 37°C, 5% CO2, 5 mM of sodium butyrate was added to all but HepG2 cells, which had 2.5 mM of sodium butyrate. After 48 h incubation, cells were fixed with 1.25% glutaraldehyde, stained with X-gal to visualise β-galactosidase-expressing cells33 and blue cells were counted.
β-galactosidase enzyme assay
Luminescent β-galactosidase enzyme assay (Clontech, BD Biosciences) was used to analyse the amount of enzyme expressed in the transduced cells according to the manufacturer's instructions. The luminescence was measured with black luminometer 96-well plates (Black Isoplate™ TC Wallac, Turku) and Victor2 luminometer (Wallac, Turku). Coomassie Plus protein assay (Bio-Rad) was used to equalise the protein amounts from lysed cell samples according to the manufacturer's instructions.
The experiment was performed according to Zhang et al.34 except for the infection time, which was 48 h. Briefly, Sf9 cells were infected with LacZ control virus, a virus displaying full VSV-G5 or the virus displaying VSV-GED, all with MOI 10. At 48 h postinfection, the growth medium (Insect-Xpress, Bio Whittaker) was removed, and the cells were washed once with PBS at pH 7.4. The cells were then exposed to PBS with pH varying from 5.0 to 7.4 for 20 min. The PBS was removed and the cells were washed twice with PBS at pH 7.4 and returned to the growth medium. After 4 h incubation at 28°C, cells were fixed with 1.25% glutaraldehyde in PBS for 20 min and examined for syncytia formation.
Blocking of endocytosis
Experiments were performed using monensin22 and ammonium chloride.35 Cells were incubated in a medium supplemented with 0.5 μM monensin or containing 0, 1, 2, 4, 6, 8 or 12 mM ammonium chloride for 30 min at 37°C. The medium was removed and the viral dilutions in the same medium (MOI 200 for monensin and MOI 250 for ammonium chloride) were added on the cells and incubated for 48 h at 37°C. Finally, the monensin-treated cells were fixed and stained as described above. Luminescent β-galactosidase enzyme assay was used to determine the relative percentage of LacZ expression with various concentrations of ammonium chloride.
Cytotoxicity of VSV-GED was determined by an MTT-assay, CellTiter 96® Aqueous One Solution Cell Proliferation Assay (Promega) according to the manufacturer's instructions. The measurements were performed with a minimum of five replicates and absorbance was measured at 492 nm. Survival percentage was calculated by comparison to the absorbance in the no virus/no butyrate or no virus/butyrate wells (100% survival). Results were compared to the control LacZ virus.
Gene delivery into rat brain
Female inbred BDIX rats (200–250 g, n=13) were anesthetised intraperitoneally with a solution (0.150 ml/100 g) containing fentanyl-fluanisone (Janssen-Cilag, Hypnorm®, Beerce, Belgium) and midazolame (Roche, Dormicum®, Basel, Switcherland) and placed in a stereotaxic apparatus (Kopf Instruments). Plaque-forming units (2 × 108) of the virus in PBS were injected with a Hamilton syringe and a 27-gauge needle either into the right ventricle (coordinates: 1.0 mm caudal to bregma, 1.5 mm right to sutura sagittalis, and into a depth of 3.5 mm; n=10), or into a malignant glioma tumor inoculated into the rat brain (coordinates: 1.0 mm caudal to bregma, 1.0 mm right to sutura sagittalis, and into a depth of 2.5 mm; n=3). Rats were killed on day 4 and perfused with PBS intracardially and fixed with X-gal fixative (4% PFA in phosphate buffer, pH 7.2) for 30 min.19 After the fixative, the brains were rinsed for 2 h in PBS and imbedded in OCT.
Transduction of rabbit muscle
Rabbits (n=3) were anesthetised with medetomidine-ketamine (Domitor 0.7 ml s.c.; Orion Pharma, Espoo, Finland and Ketalar 0.9 ml s.c.; Pfizer, NY, USA). A total dose of 109 PFU of either the control baculovirus with LacZ-marker gene or the VSV-GED baculovirus was injected in a volume of 50 μl into M. semimembranosus by 10 injections. The rabbits were killed 6 days after the gene transfer and muscles were fixed as described above, washed in PBS for 2 h, frozen in liquid nitrogen-cooled isopentane and cryosectioned. X-gal staining was performed as described,4 followed by Mayer's Carmalum counterstaining. For statistical analysis, 25 slides with two tissue sections on each slide were prepared from each animal. All positive cells were counted, areas were equalised and means were calculated.
Prism™ 4 from GraphPad was used to analyse the results with an unpaired t-test to determine whether the differences between the subgroups were statistically significant.
We are grateful to Loy Volkman (University of California, Berkeley, CA, USA) for antibodies against vp39. We thank Tarja Taskinen, Erik Peltomaa, Mervi Nieminen, Riina Kylätie, Riikka Eisto, Tiina Koponen and Seija Sahrio for excellent technical assistance. This work was supported by the Finnish Academy, Sigrid Juselius Foundation and Ark Therapeutics Ltd.
About this article
Virologica Sinica (2012)